quantitative research title about online selling in pandemic

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• Elsevier - PMC COVID-19 Collection

Managing the effectiveness of e-commerce platforms in a pandemic

Given the severe impacts of the Covid-19 pandemic on business activities, this study presents a systematic framework to examine the effect of the perceived effectiveness of e-commerce platforms (PEEP) on consumer's perceived economic benefits in predicting sustainable consumption. This study adopted uses and gratification theory to base the conceptual model while adding a boundary condition of pandemic fear. The primary research method of this study is a quantitative survey and analysis. Using a sample of 617 online consumers with PLS analytical technique, this study finds a positive moderating effect of pandemic fear on the relationships among PEEP, economic benefits, and sustainable consumption. The contribution of this study is its examination of how economic benefit mediates the PEEP and sustainable consumption relationship which is dependable on the levels of pandemic fear. Implications for managers and theory are also discussed.

1. Introduction

Recent decades have witnessed an escalation of global pandemics, e.g., SARS in 2003, H5N1 in 2006, and H1N1 in 2009, each of which hampered business activities and economic growth ( Chung, 2015 ). For example, the Covid-19 virus caused a 13.5% drop in China's industry production and a 20.5% decline in retail sales in the first two months of 2020 while the U. S's stock market lost trillions of dollars, leading to a negative wealth effect and lower GDP ( Pesek, 2020 ). As scientists have long warned, infectious diseases can force business activities into a new reality that severely impacts operations and one where managers have unclear guidance about how to effectively respond ( Hudecheck et al., 2020 ). As a result of the Covid-19 impact, for instance, consumers are increasingly turning to online purchases; thus, managers need to be innovative in seeking alternative forms of supplies which raise the interest in the facilitation between firms and consumers.

The extant literature has situated the importance of e-commerce platforms that facilitate virtual interactions and include informative and insightful product information ( Chandna and Salimath, 2018 ; Li et al., 2020 ) that consumers might consider to be economic benefits ( Zhang et al., 2017 ). Such that, firms can increase sales revenue as e-commerce platforms help leverage existing consumers and attract new ones, build social groups, and ensure compatibility with legacy systems ( Lee et al., 2018 ), especially in pandemic periods ( Nielsen, 2020a , Nielsen, 2020b ). In fact, Covid-19 has disrupted and changed the business landscape as managers have been thrust into the position of operating online supplies due to the reduction of in-person contact. However, the literature has yet to address how e-commerce platforms can benefit both firms and consumers during the pandemic. As a result, managers who seek to implement online selling currently have no specific guidance to improve consumer intention for sustainable consumption over time and circumstance ( Guillen-Royo, 2019 ; Hernant and Rosengren, 2017 ).

Therefore, this study provides an exploratory study in which the author develops a model of perceived effectiveness of e-commerce platform (PEEP) and analyzes its effectiveness on consumer's sustainable consumption behavior. In so doing, the author examines if economic benefits are a potential mediator of the PEEP and sustainable consumption relationship. The author, drawing on uses and gratification theory (UGT), also takes pandemic fear into account to identify contingencies for the framework. UGT explains why consumers use a specific medium to satisfy their needs ( Kaur et al., 2020 ), providing valuable insight into the nascence of e-commerce platform adoption, information-seeking behavior, and social interactions ( Abid and Harrigan, 2020 ). In fact, rather than being a cause to worry, the Covid-19 pandemic appears to be a catalyst to testify a firm's effectiveness in operating its business activities more sustainably.

2. Conceptual background and hypothesis development

2.1. uses and gratification theory.

Uses and gratification theory (UGT) refers to an influential sociological paradigm that explains which social and psychology needs motivate consumers to select particular platforms ( Li et al., 2018 ). Advances in Internet technologies have enabled e-commerce platforms to transform retail and logistics operations that create more economic benefits as reductions of costs and delays, triggering the roles of UGT in explaining usage motivations ( Luo et al., 2011 ). For example, consumers use e-commerce platforms to interact and purchase products from retailers while retailers can provide consumers with various online shopping opportunities ( Wagner et al., 2020 ). In this regard, UGT fundamentally helps to examine what role consumer behavior and motivation plays in consumers' online interactions with firms ( Huang et al., 2014 ). If firms can effectively fulfill consumer needs with products via online platforms, consumers will be more willing to continue to interact with firms ( Li et al., 2018 ). The current study argues that UGT helps to capture how consumers adopt and utilize e-commerce platforms to satisfy their purchase needs. In terms of usability and functionality, UGT addresses the patterns and motivations of online platform's application in seeking information, interacting with contents and communities, and sustaining purchase decisions for specific situations ( Korhan and Ersoy, 2016 ). For example, the Covid-19 pandemic increased consumers purchase intention towards e-commerce platforms due to the medium's perceived health and safety benefits in contrast to those of traditional retailers. UGT helps to explain why, in light of pandemic fear, consumers increasingly engaged in online platforms to make relevant purchase considerations. In addition, UGT enables managers to identify the reasons behind consumer choice and product and service ratings ( Ray et al., 2019 ). UGT also provides managers with importance of differentiated content strategy based on transactional data and information exchange to develop content strategy more effectively ( Lim and Kumar, 2019 ). Therefore, this study employs UGT to explore the interactive linkages of PEEP, consumers' perceived economic benefits, pandemic fear, and sustainable consumption.

2.2. Perceived effectiveness of e-commerce platforms (PEEP)

E-commerce literature has well-documented that advances in Internet technology allow firms to directly sell products to consumers through e-commerce platforms ( Fan et al., 2020 ), resulting in increased sales ( Lee et al., 2018 ). However, consumers cannot physically examine products when buying online, which leads to product uncertainty ( Song et al., 2020 ). The literature also indicates that online customer information is collected and tracked for data-driven marketing efforts but are inadequately safeguarded ( Bandara et al., 2020 ). For example, security risk (e.g., personal information leakage) has been found to have strong negative effects on online transactions ( Hubert et al., 2017 ). This is important because today's e-commerce platforms offer various online products and services from electronic devices to high-tech products, health and beauty to food, and fashion to sport and travel with a large volume of financial transactions. When consumers use e-commerce platforms to book a hotel room or flight ticket with online payments, they must provide their personal and bank account information which needs to be protected from leakage and fault behaviors or consumers will probably face with potential risks and dangers, otherwise ( Hubert et al., 2017 ). Based on UGT, consumer perception of online transactional safety increases continuance intention and motivation to write positive reviews of products and services. Consequently, online safeguards are critical for personal and transactional data ( Liao and Shi, 2017 ).

Furthermore, the literature recommends that online safeguards include credit card guarantees and privacy protection ( Plangger and Watson, 2015 ; Wang et al., 2019 ). For instance, when receiving online payments, e-commerce platforms have privacy policies and agreements in place with financial companies (e.g., banks) to protect customers against fraudulent actions ( Chang and Chang, 2014 ; Fang et al., 2014 ; Kaur et al., 2020a ; Liao and Yang, 2020 ). Guided by UGT, online consumers often express concern for privacy and security when participating in the on-going brand relationship ( Simon, 2017 ). This theory also suggests that consumers only interact with firms when triggering gratification from firm commitment ( Simon, 2017 ). Therefore, the author conceptualizes PEEP as an online customer's perception that e-commerce platforms have safeguard mechanisms in place to protect online transactions from potential risks in both regular and pandemic times.

2.3. Economic benefits

In e-commercial literature, the conventional meaning of economic benefit largely refers to the consumer perception that e-commerce platforms offer price discounts, promotions, or other preferential activities ( Liu et al., 2019 ). Liu et al. (2019) indicate that economic benefit derived from e-commerce platforms can generate positive emotional responses, which in turn leads to online purchase intention. In line with this logic, Wang and Herrando (2019) suggest that e-commerce platforms enhance interactions between seller and consumers and among consumers, as these interactions impact increasing online shopping behavior and generate economic benefits. Based on UGT, economic benefits also trigger consumer intention to engage in sustainable consumption during uncertain situations. Benefits, for instance, may include coupons, cash-backs, and discounts ( Ray et al., 2019 ). UGT also suggests that if commercial offerings do not match consumers' expectations, the commercial quid pro quo relationship may be triggered by consumers ( Simon, 2017 ). Additionally, this study adds that economic benefits pertain to consumers’ perceived economic benefits (e.g., cost savings and discounts) for purchasing from e-commerce platforms during pandemic periods.

2.4. Pandemic fear

A pandemic refers to a new disease that most people do not have an immunity to and one that spreads worldwide ( WHO, 2010 ), and pandemics are becoming one of the biggest threats for the world today ( Harvard Global Health Institute, 2020 ). According to the Harvard Global Health Institute (2020) , an infectious disease could rapidly cause millions of deaths globally, destabilize governments, and restrict trade and travel. The recent Covid-19 virus, for instance, has infected 422,945 people worldwide, resulting in 18,907 deaths at the time of writing (March 25, 2020), and accounting for 3.4% of death in comparison with seasonal flu (1.0%) ( Worldometer, 2020 ). This fear of contagion becomes an emerging issue, such that it is imperative to learn how pandemic fear influences consumer spending and purchase behavior ( Khan and Huremović, 2019 ). The consumer behavior literature indicates that fear refers to the negative consequences of a specific event that can lead to changes in consumer behavior and attitude ( Solomon, 2017 ). In this regard, the Covid-19 pandemic has changed consumer purchasing behavior as consumers fear contagion ( Laato et al., 2020 ; Prentice et al., 2020 ). For example, a prompt survey by Nielsen, 2020a , Nielsen, 2020b indicates that 45% of Vietnamese consumers purchased products intended for storage, and 25% purchased these products online, while Taiwanese consumers purchased additional instant noodles. In addition, Vietnam firms have changed their support programs (e.g., home delivery and masks and sanitizers at stable prices), the result of which is a notable growth in sales ( Vietnam News, 2020 ). Collectively, this study conceptualizes pandemic fear as consumer contagion fear, a belief which influences the way consumers use e-commerce platforms to purchase products.

2.5. Extending the concept of sustainable consumption

Sustainable consumption refers to consumers' adoption of green lifestyle to satisfy their needs without damaging the earth's resources or putting future generation at risks ( Sharma and Jha, 2017 ). In line with this logic, previous studies have synthesized common threads of sustainable consumption as care for nature, self, and community, factors at the heart of contemporary marketing discipline ( Lim, 2017 ). Much of the literature documents a wide range of positive outcomes for customers to include happiness and life satisfaction resulting from product and service interactions ( Guillen-Royo, 2019 ). Conversely, the existing literature also evidences that unsustainable consumption patterns cause adverse social, environmental, and economic side effects ( Sharma and Jha, 2017 ). For example, Covid-19 has sparked worldwide alarm, as it spreads rapidly, through human-to-human contact ( Harvard Medical School, 2020 ). The virus can be spread through small droplets as the infected person coughs or exhales ( WHO, 2020 ). To reduce the spread of the virus, consumption practices should turn to e-commerce platforms as this can increase efficiency of shopping, enable new purchases, and facilitate information access and online communication between consumers and sellers ( Guillen-Royo, 2019 ). Given the existing literature and Covid-19's impact, this study extends the concept of sustainable consumption as “ purchasing products and services from e-commerce platforms to satisfy needs and wants, and to increase health safety for selves and community during a pandemic period .”

2.6. Hypotheses

This section describes the development of the research model to explain how PEEP affects economic benefits which subsequently lead to sustainable consumption under the boundary condition of pandemic fear. In doing so, this study argues that pandemic fear positively moderates (1) the relationship between PEEP and economic benefits, and (2) does so in the relationship between economic benefits and sustainable consumption. Fig. 1 presents the proposed relationships.

Proposed model.

PEEP plays a key part in the strategy to create online seller product and service credibility. This is because the online customer is unable to physically interact with the seller in the virtual environment, and this might trigger privacy risk concerns ( Liao and Shi, 2017 ). With PEEP, the customer can be assured that personal and transactional data collected from e-commerce platforms are protected. According to Pappas (2016) , higher levels of trust in the online platforms safety and security help to construct a consumer belief in e-vender credibility which ultimately increases the likelihood of a sale. In an online situation, e-commerce platforms provide customers various benefits, e.g., privacy and security, information search and provision, product reviews, promotion, and order fulfillment ( Qin et al., 2020 ). Plangger and Watson (2015) added to the literature that while advances in information technology have facilitated business operations to be more effective, e.g., less costly and data rich, firms also need to protect customer privacy and avoid risks that can threaten the long-term relationships. From the UGT perspective, e-commerce platforms generate economic benefits for customers such as privacy and cost savings ( Huang et al., 2014 ). In addition, the Covid-19 pandemic, and related fear of contagion, has created worldwide chaos, e.g., hoarding, market gyration, and travel restrictions ( McNulty, 2020 ). The pandemic has triggered infection fears and social distance in which the normal supplying routines are insufficient while consumers are increasingly turning to online purchase whereas raising the important role of e-commerce platforms. The literature also situates that e-commerce platforms provide fruitful benefits for a wide range of business from restaurants to florist and laundry to medicines while customers are becoming more adaptive with online products ( Zhang et al., 2019 ). As the Covid-19 virus can be spread via person-to-person contact, and effective medications or vaccines have yet to be found ( Smith and Prosser, 2020 ), consumers can be expected to make more online purchases to reduce exposure ( Nielsen, 2020a , Nielsen, 2020b ). The pandemic had become an unprecedented grand challenge that created many social, health, and economic problems to society at large ( Bacq et al., 2020 ). For example, the pandemic has led to the lower accessibility of in-store selling retailers due to higher health concerns of consumers which triggered a sudden increase in demands for alternative distribution channels ( Pantano et al., 2020 ). In such situations, the author believes that pandemic fear will increase the relationship between PEEP and economic benefit. As so:

Online customer pandemic fear positively moderates the relationship between PEEP and economic benefits, such that PEEP influences economic benefits more strongly when pandemic fear is higher.

Furthermore, the issue of economic benefits facilitating consumer sustainable consumption intention has been explicitly addressed in the literature ( Dabbous and Tarhini, 2019 ). The rationale behind this is that consumers perceived economic well-being, a comparison of current situation to a past situation when make a purchasing decision ( Verma and Sinha, 2018 ). Some researchers consider economic benefits to be as important as product attributes when exploring online purchase intention ( Lee et al., 2018 ), while others consider it to be a driver of sustainable consumption ( Dabbous and Tarhini, 2019 ). This relationship is often based on a consumer responsibility, e.g., social, environmental, and ethical concerns ( Lim, 2017 ), or emotional responsibility ( Luchs et al., 2015 ). According to UGT, consumers often reciprocate the benefits they receive in the form of feelings of appreciation ( Simon, 2017 ). However, the current Covid-19 outbreak has added health concerns to this and the need for mediums where firms and consumers can interact without in-person contact. Covid-19 has killed thousands and set millions in quarantine ( Worldometer, 2020 ), as it spreads easily with breathtaking speed. The fear of Covid-19 stirs up the role of e-commerce that supports social interaction and stakeholders’ contributions to provide online buying and selling of products and services ( Addo et al., 2020 ). Pantano et al. (2020) indicated that while the Covid-19 pandemic has produced anxiety, depression and stress in society, consumers are increasingly purchasing products and services through online platforms as they perceived the safety offered by the Internet and online technologies. Accordingly, the Covid-19 pandemic has increased health concerns as well as the need for online transactions to protect sellers and buyers from infection. This study proposes that pandemic fear facilitates the relationship between perceived economic benefits and sustainable consumption in the context of e-commerce platforms. Therefore:

Online customer pandemic fear positively moderates the relationship between economic benefits and sustainable consumption, such that economic benefits influence sustainable consumption more strongly when pandemic fear is higher.

3.1. Sample

The sample consists of 617 online consumers from the Vietnam-based market. Sampling online consumers was adopted because the recent Covid-19 pandemic has changed the shopping behavior in Vietnam. A survey by Nielsen Vietnam (2020) shows that 95% of consumers feel afraid of the Covid-19 virus, resulting in a 25% increase in online shopping purchases. As consumers increasingly use online methods to prepare for a possible emergency, firms need to ensure smooth, frictionless, and fast experiences on their e-commerce platforms that meet customer needs and expectations ( Abramovich, 2020 ). Thus, a study of online consumer behavior in a pandemic context can provide firms with useful information for their business operations.

3.2. Data collection

Data were collected from a sample of 1000 online consumers using Facebook. A valid online consumer is identified as one who had online shopping experience during the Covid-19 outbreak (January 1 to March 15, 2020). The author sent the questionnaire to fifty key social networkers whose Facebook accounts have at least 500 friends and followers, asking them to contact 20 subsequent respondents from their Facebook lists. This technique is useful when respondents are difficult to identify and contact while the survey needs to be conducted expediently ( Cooper and Schindler, 2013 ). The author tracked the survey's progress daily using the key networkers' Facebook walls, tags, and messengers. Out of the initial 1000 invitees, 617 valid responses were obtained (61.7% response rate). The author assessed nonresponse bias by comparing early and late responses through t-tests of means among the research constructs which of results showed no differences (ranging from −1.06 to 0.24, p  > .05). A chi-square test for PEEP ( χ 2  = 14.827, p  = .53), economic benefits ( χ 2  = 1.884, p  = .75), pandemic fear ( χ 2  = 10.884, p  = .28), and sustainable consumption ( χ 2  = 16.656, p  = .16) indicated no differences at the 95% significance level. Accordingly, nonresponse bias and common method bias were not a concern in this study. The respondents (227 males and 390 females) were on average 27.5 years of age (ranging 18 to 64, SD  = 8.85), with an average monthly income of 13, 309, 141VND ( SD  = 11,730,842). Finally, most of the consumers (71.4%) held a bachelor's degree, 19.6% held a master's degree or above, and 9.0% held a college degree or below. The detailed demographic characteristics are provided in Table 1 .

Demographic characteristics (n = 617).

3.3. Instrument

This study designed online questionnaire using Google Forms for the data collection. The scales of the research constructs were adopted, revised, and translated from English to Vietnamese as the survey was administered in Vietnam. The survey was pretested with 20 online consumers to detect possible weaknesses in design and instrumentation. This is important to avoid negative consequences to the survey operation such as ambiguities, confusions, and offensive questions ( Cooper and Schindler, 2013 ). On the basis of the consumer's comments, this study refined some items and included the final questionnaire form. Responses were recorded using a 5-point agreement scale.

3.4. Measures

First, measures of PEEP were adopted from Fang et al. (2014) . Second, economic benefits and sustainable consumption were measured using the scales developed by Dabbous and Tarhini (2019) . Finally, pandemic fear was operationalized on the basis of Chatterjee et al. (2019) . The author provides details of the constructs in Appendix 1 .

3.5. Control variables

Demographic characteristics, such as age, gender, education, and income were included as control variables to capture possible effects on sustainable consumption. This was also done to ensure that the results were not biased by not including control variables ( Cooper and Schindler, 2013 ).

3.6. Analytical technique

This study tested the proposed model using the two-stage partial least squares method which allows for an examination of the causal relationship among latent constructs ( Hair et al., 2011 ). To examine the moderating effect, Vinzi et al. (2010) procedure was followed. This study first centered each indicator of moderator and predictor variables and then multiplied them to create interaction terms. The strength of the moderating effects was determined by comparing R-square changes through effect size ( f 2 ) ( Vinzi et al., 2010 ).

4.1. Measurement quality

First, the author assessed construct reliability and validity using composite reliability (CR) and average variance extracted (AVE). As shown in Table 2 , factor loadings (0.71-0.90), CRs (0.81-0.91), and AVEs (0.59-0.78) appeared to be higher than the thresholds of 0.70, 0.70, and 0.50 ( Hair et al., 2011 ), respectively. These results confirm indicator reliability, internal consistency reliability, and convergent validity. In addition, the variance inflation factors ranging from 1.35 to 2.42 were below the cutoff of 5 ( Hair et al., 2011 ), removing the multicollinearity threat.

Measurement model.

Second, the author assessed the discriminant validity using Fornell-Larcker Criterion and Heterotrait-Monotrait Ratio (HTMT) ( Henseler et al., 2015 ). More specifically, the correlation matrix in Table 3 shows the highest correlation was 0.57, less than the 0.71 cutoff ( MacKenzie et al., 2011 ) while all construct correlations were lower than the square root of AVE of their own constructs ( Hair et al., 2011 ). Appendix 1 also shows that all items loaded well onto their own construct and poorly on others. In addition, the largest ratio of HTMT (0.68), shown in Fig. 2 , was far below the 0.90 benchmark recommended by Henseler et al. (2015) . Collectively, both Fornell-Larcker Criterion and HTMT results confirm discriminant validity.

Descriptive statistics and correlations.

Note: The bold values on diagonal are the square root of average variance extracted, while the others represent correlation matrix.

Heterotrait-Monotrait ratio.

4.2. Hypothesis testing results

As shown in Fig. 3 , the model explains 20% of the variance in economic benefits and 37.85% of the variance in sustainable consumption. H1 , which stated that pandemic fear positively moderates the relationship between PEEP and economic benefits, was supported ( β  = .07, t  = 1.85, p  < .10). Fig. 4 depicts this moderating effect. As predicted, at high levels of pandemic fear ( Mean  + 1 SD ), economic benefits increase as PEEP increases; however, low levels of pandemic fear ( Mean – 1 SD ) slightly enhance the relationship between PEEP and economic benefits. H2 , which stated that pandemic fear positively moderates the relationship between economic benefits and sustainable consumption, was also supported ( β  = .12, t  = 3.00, p  < .01). As plotted in Fig. 4 , sustainable consumption increases rapidly as economic benefits increase when pandemic fear is higher ( Mean  + 1 SD ). At low levels of pandemic fear ( Mean −1 SD ), sustainable consumption does not increase regardless of the effect of economic benefits. In addition, the relationship between PEEP and economic benefits was positive and significant ( β  = .42, t  = 12.2, p  < .01) as was the relationship between economic benefits and sustainable consumption ( β  = .56, t  = 17.87, p  < .01), thus confirming the mediating role of economic benefits in the relationship between PEEP and sustainable consumption.

Results of proposed model.

Moderating effects of pandemic fear.

This study further determines the substantivity of the moderating effects by comparing R-square changes in the model through f 2 (effect size) ( Vinzi et al., 2010 ). In doing so, the author, as illustrated in Appendix 2 , runs two models in which Model 1 includes only direct effects and Model 2 adds interaction effects and control variables. As shown in Appendix 2 , PEEP significantly increased R 2 of economic benefits by 1.6%, indicating a small effect size ( f 2  = 0.02). The interaction effect of pandemic fear with economic benefits also significantly increased R 2 of sustainable consumption by 4.4%, indicating a small-to-medium effect size ( f 2  = 0.07). Thus, the interaction terms increased R 2 significantly with f 2 values surpassing the threshold of 0.02 suggested by Vinzi et al. (2010) , thus confirming the substantive significance of the moderating effects.

5. Discussion

The findings support the research hypotheses that for firms operating in pandemic periods, PEEP is an important determinant of sustainable consumption through economic benefits. The results further suggest that pandemic fear increases the positive effect of PEEP on economic benefits which subsequently exerts a strong effect on sustainable consumption. This study adds to the literature by including the boundary effect of pandemic fear in the conceptual model to provide more insight about how to increase managerial effectiveness.

5.1. Theoretical and managerial implications

This study holds several theoretical and managerial implications. The author adopts UGT ( Huang et al., 2014 ; Simon, 2017 ) to understand and extend the concept of PEEP ( Fang et al., 2014 ) and how sustainable consumption is triggered in uncertain environments (the Covd-19 pandemic). The results of the mediating effect indicate that PEEP is important to sustainable consumption through economic benefits. The results indicate that personal and transactional safeguards of PEEP are conductive to facilitate a consumer's perceived economic benefits in online business environments. This adds to the study of Fang et al. (2014) by examining the focal role of PEEP in predicting changes in consumer consumption behavior. Second, during pandemic periods, consumers are increasingly concerned more about engaging in sustainable consumption through e-commerce platforms as the findings signify economic benefits influence sustainable consumption. These findings are consistent with Dabbous and Tarhini (2019) who suggest that economic benefits influence consumer sustainable consumption intention. Thus, the current study, through the lens of UGT, offers a new theoretical view of the relationship among PEEP, economic benefits, and sustainable consumption. Furthermore, pandemic fear is an important moderator as it increases the impacts of PEEP and economic benefits which ultimately enhance sustainable consumption. This finding extends the fear theory of Chatterjee et al. (2019) , who suggest that fear negatively influences consumer buying intention in risky or uncertain situations.

Given the findings, firms should reallocate online resources to employ e-commerce platforms to serve customers, as e-commerce platforms can facilitate communication among sellers and buyers on the basis of building trust and credibility to establish and maintain the customer-seller relationship. In a pandemic condition like Codvid-19, high levels of pandemic fear motivate consumers to rely more on PEEP for transactional activities which requires trust-building and risk-reducing potential. As such, firms should seek opportunities to provide consumers effective online safeguards. For example, e-wallet (e.g., MoMo: https://momo.vn/ ) secure online transactional information, as it only requires consumers to enter their user name and password at the end of the purchase, and all data is encrypted. Business managers should reinforce the proliferation of technological application with suppliers, retailers, and legal institutions to build reliable platforms to deliver products and services while utilizing online platforms as a strategic marketing orientation to promote online transactions to reduce the infections of the Covid-19 pandemic. Firms also need to require credit card guarantees from financial institutions (e.g., banks) to compensate online consumers in case of fraudulent seller behavior ( Fang et al., 2014 ). It is critical for firms to increase PEEPs and enhance interactive communication with consumers by providing online safeguards effectively. For instance, they can offer specific guarantees and privacy protection by publishing customer reviews, as reviews generate consumer confidence products and services ( Mattison Thompson et al., 2019 ). Note that when buying online, consumers also focus on economic benefits, and this positively affects sustainable consumption during pandemic periods. Firms should emphasize the consumer savings (e.g., lower costs in contrast with the traditional counterparts) and improved transaction performance and efficiency that result from the use of e-commerce platforms ( Dabbous and Tarhini, 2019 ). In addition, consumer's fear of the Covid-19 pandemic boost online purchase intention. This requires that firms not only provide online products and services but also e-commerce platforms that include an online care center to ensure online consumers' health and safety (e.g., Lazada offers a Covid-19's online help center, https://www.lazada.vn/ ).

5.2. Limitations and research direction

In this exploratory study, there appears to be several limitations which provide opportunities for future research. First, the conceptual model was tested with online consumers in an emerging economy (Vietnam), and the respondents viewed e-commerce platforms from their own perspectives. However, their characteristics may differ remarkedly from consumers in more advanced economies (the U.S.). Therefore, subsequent research should cross-validate the empirical model by obtaining data from different economies in order to provide more insightful findings. Future explorations should also assess the e-commerce platforms' operation model from the perceptions of firm managers that would be more valuable. Second, the conclusions of this study should be interpreted as preliminary inasmuch as pandemic fear was observed for a brief period (January 1 to March 15, 2020) whereas consumer perception and behavior may vary over time and be dependable on Covid-19's lifespan. In this regard, a longitudinal framework which includes a time-series database that examines the boundary effect of pandemic fear on the PEEP, economic benefits, and sustainable consumption relationship can provide more details about probable causation which can then increase strategic effectiveness. Another limitation of this study is the target group. As reviewed, experts situated that a subculture associated with social networking sites (e.g., Facebook) should be considered because this may influence economic benefits. Many issues can increase countable problems in supply chains, lack of assortment, and delayed deliveries that led to the change in behavioral and attitude assessment of consumer toward e-commerce. Given this, future research should focus on multiple cultures and social networking sites to compare different perspectives on supply chains, lack of assortment, and delayed deliveries will provide more insights about the effectiveness of e-commerce platforms. In addition, the change in consumer wealth is also as a function of time during a pandemic time while the purchasing power of money is changing. Therefore, future research might focus on consumer-specific financial situations, which will be valuable in explaining consumer purchasing behavior.

6. Conclusion

This study is among the first to provide a systematic framework to examine the PEEP, economic benefits, and sustainable consumption relationship under the boundary condition of pandemic fear. For researchers, this study extended the concept of economic benefit and empirically found support for its mediational role in the relationship between PEEP and sustainable growth while confirming the moderating effect of pandemic fear. Drawing on the UGT, the current study adds to the retailing literature by introducing a mediation and moderating-based model based on the UGT in a global pandemic. Future research should be replicated in diverse environments and over time to increase the power of the conceptual model and theory. For managers, the findings showed that high levels of pandemic fear required PEEP to build an effective mechanism to protect online personal and transactional data such as information leakage and credit card fraud and economical offerings to increase customer consumptions. This study suggests that managers see pandemics (Covid-19 virus), which can happen at any time, as catalysts to prepare and respond more effectively. In addition, firms must strategically build e-commerce platforms and operate in conjunction with offline methods for supplies because consumers turn to online sources to avoid infectious diseases and are increasingly engaging in sustainable consumption behaviors. Overall, this study provides a new perspective of the critical roles of PEEP and economic benefits and sets a supplementing point for future research to further explore sustainable consumption behavior of customers in the boundary condition of pandemic fear.

Acknowledgments

The author would like to extend his appreciation to the Editor-in-Chief, Professor Harry Timmermans for his efforts in handling the review process, the anonymous reviewers for their developmental feedback, and James Kang, Ruth Kang, and John Baker, for their valuable comments on the early draft of the study.

Lobel Trong Thuy Tran is a lecturer in the Faculty of Business Administration at Ton Duc Thang University in Vietnam. He received his PhD in the Department of Business Administration at Asia University in Taiwan. His research interests include marketing strategy, consumer and social media, education marketing, tourism and hospitality innovation, decision making under uncertainties, and organizational behavior and HRM.

Appendix 1. Items and cross-loadings.

Appendix 2. results details..

Note: * p  < .10, ** p  < .01. PEEP: Perceived effectiveness of e-commerce platforms.

• Abid A., Harrigan P. An exploration of social media-enabled voter relationships through uses and gratifications theory, psychological contract and service-dominant orientation. Australasian Marketing Journal (AMJ) 2020 doi: 10.1016/j.ausmj.2020.02.002. accessed 12 March 2020. [ CrossRef ] [ Google Scholar ]
• Abramovich G. 2020. How COVID-19 Is Impacting Online Shopping Behavior, Adobe Blog. https://theblog.adobe.com/how-covid-19-is-impacting-online-shopping-behavior/ Adobe. accessed 26 March 2020. [ Google Scholar ]
• Addo P.C., Jiaming F., Kulbo N.B., Liangqiang L. COVID-19: fear appeal favoring purchase behavior towards personal protective equipment. Serv. Ind. J. 2020; 40 (7–8):471–490. [ Google Scholar ]
• Bacq S., Geoghegan W., Josefy M., Stevenson R., Williams T.A. The COVID-19 virtual idea blitz: marshaling social entrepreneurship to rapidly respond to urgent grand challenges. Bus. Horiz. 2020 doi: 10.1016/j.bushor.2020.05.002. avaiable at. [ PMC free article ] [ PubMed ] [ CrossRef ] [ Google Scholar ]
• Bandara R., Fernando M., Akter S. Explicating the privacy paradox: a qualitative inquiry of online shopping consumers. J. Retailing Consum. Serv. 2020; 52 [ Google Scholar ]
• Chandna V., Salimath M.S. Peer-to-peer selling in online platforms: a salient business model for virtual entrepreneurship. J. Bus. Res. 2018; 84 :162–174. [ Google Scholar ]
• Chang J.-S., Chang W.-H. Analysis of fraudulent behavior strategies in online auctions for detecting latent fraudsters. Electron. Commer. Res. Appl. 2014; 13 (2):79–97. [ Google Scholar ]
• Chatterjee S., Gao X., Sarkar S., Uzmanoglu C. Reacting to the scope of a data breach: the differential role of fear and anger. J. Bus. Res. 2019; 101 :183–193. [ Google Scholar ]
• Chung L.H. Impact of pandemic control over airport economics: reconciling public health with airport business through a streamlined approach in pandemic control. J. Air Transport. Manag. 2015; 44–45 :42–53. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Cooper D.R., Schindler P. 12 ed. McGraw-Hill Education; 2013. Business Research Methods. [ Google Scholar ]
• Dabbous A., Tarhini A. Assessing the impact of knowledge and perceived economic benefits on sustainable consumption through the sharing economy: a sociotechnical approach. Technol. Forecast. Soc. Change. 2019; 149 [ Google Scholar ]
• Fan X., Yin Z., Liu Y. The value of horizontal cooperation in online retail channels. Electron. Commer. Res. Appl. 2020; 39 [ Google Scholar ]
• Fang Y., Qureshi I., Sun H., McCole P., Ramsey E., Lim K.H. Trust, satisfaction, and online repurchase intention: the moderating role of perceived effectiveness of e-commerce institutional mechanisms. MIS Q. 2014; 38 (2):407–A409. [ Google Scholar ]
• Guillen-Royo M. Sustainable consumption and wellbeing: does on-line shopping matter? J. Clean. Prod. 2019; 229 :1112–1124. [ Google Scholar ]
• Hair J.F., Ringle C.M., Sarstedt M. PLS-SEM: indeed a silver bullet. J. Market. Theor. Pract. 2011; 19 (2):139–152. [ Google Scholar ]
• Harvard Global Health Institute . Harvard University; 2020. Pandemic Preparedness & Global Health Security. https://globalhealth.harvard.edu/pandemics-and-health-security accessed 26 March 2020. [ Google Scholar ]
• Harvard Medical School . Harvard Health Publishing; 2020. As Coronavirus Spreads, Many Questions and Some Answers, Coronavirus Resource Center. https://www.health.harvard.edu/diseases-and-conditions/coronavirus-resource-center#Terms accessed 24 March 2020. [ Google Scholar ]
• Henseler J., Ringle C.M., Sarstedt M. A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Market. Sci. 2015; 43 (1):115–135. [ Google Scholar ]
• Hernant M., Rosengren S. Now what? Evaluating the sales effects of introducing an online store. J. Retailing Consum. Serv. 2017; 39 :305–313. [ Google Scholar ]
• Huang L.-Y., Hsieh Y.-J., Wu Y.-C.J. Gratifications and social network service usage: the mediating role of online experience. Inf. Manag. 2014; 51 (6):774–782. [ Google Scholar ]
• Hubert M., Blut M., Brock C., Backhaus C., Eberhardt T. Acceptance of smartphone‐based mobile shopping: mobile benefits, customer characteristics, perceived risks, and the impact of application context. Psychol. Market. 2017; 34 (2):175–194. [ Google Scholar ]
• Hudecheck M., Sirén C., Grichnik D., Wincent J. MIT Sloan Management Review; 2020. How Companies Can Respond to the Coronavirus. March 09, 2020) [ Google Scholar ]
• Kaur P., Dhir A., Bodhi R., Singh T., Almotairi M. Why do people use and recommend m-wallets? J. Retailing Consum. Serv. 2020; 56 doi: 10.1016/j.jretconser.2020.10209. [ CrossRef ] [ Google Scholar ]
• Kaur P., Dhir A., Chen S., Malibari A., Almotairi M. Why do people purchase virtual goods? A uses and gratification (U&G) theory perspective. Telematics Inf. 2020 [ Google Scholar ]
• Khan S., Huremović D. Springer; 2019. Psychology of the Pandemic, Psychiatry of Pandemics; pp. 37–44. [ Google Scholar ]
• Korhan O., Ersoy M. Usability and functionality factors of the social network site application users from the perspective of uses and gratification theory. Qual. Quantity. 2016; 50 (4):1799–1816. [ Google Scholar ]
• Laato S., Islam A.K.M.N., Farooq A., Dhir A. Unusual purchasing behavior during the early stages of the COVID-19 pandemic: the stimulus-organism-response approach. J. Retailing Consum. Serv. 2020; 57 doi: 10.1016/j.jretconser.2020.102224. [ CrossRef ] [ Google Scholar ]
• Lee J.-Y., Fang E., Kim J.J., Li X., Palmatier R.W. The effect of online shopping platform strategies on search, display, and membership revenues. J. Retailing. 2018; 94 (3):247–264. [ Google Scholar ]
• Li X., Zhao X., Xu W., Pu W. Measuring ease of use of mobile applications in e-commerce retailing from the perspective of consumer online shopping behaviour patterns. J. Retailing Consum. Serv. 2020; 55 doi: 10.1016/j.jretconser.2020.102093. [ CrossRef ] [ Google Scholar ]
• Li Y., Yang S., Chen Y., Yao J. Effects of perceived online–offline integration and internet censorship on mobile government microblogging service continuance: a gratification perspective. Govern. Inf. Q. 2018; 35 (4):588–598. [ Google Scholar ]
• Liao Z., Shi X. Web functionality, web content, information security, and online tourism service continuance. J. Retailing Consum. Serv. 2017; 39 :258–263. [ Google Scholar ]
• Liao S.-H., Yang L.-L. Mobile payment and online to offline retail business models. J. Retailing Consum. Serv. 2020; 57 doi: 10.1016/j.jretconser.2020.102230. [ CrossRef ] [ Google Scholar ]
• Lim H., Kumar A. Variations in consumers' use of brand online social networking: a uses and gratifications approach. J. Retailing Consum. Serv. 2019; 51 :450–457. [ Google Scholar ]
• Lim W.M. Inside the sustainable consumption theoretical toolbox: critical concepts for sustainability, consumption, and marketing. J. Bus. Res. 2017; 78 :69–80. [ Google Scholar ]
• Liu Y., Li Q., Edu T., Jozsa L., Negricea I.C. Mobile shopping platform characteristics as consumer behavior determinants. Asia Pac. J. Market. Logist. 2019 doi: 10.1108/apjml-05-2019-0308. In press. [ CrossRef ] [ Google Scholar ]
• Luchs M.G., Phipps M., Hill T. Exploring consumer responsibility for sustainable consumption. J. Market. Manag. 2015; 31 (13–14):1449–1471. [ Google Scholar ]
• Luo M.M., Chea S., Chen J.-S. Web-based information service adoption: a comparison of the motivational model and the uses and gratifications theory. Decis. Support Syst. 2011; 51 (1):21–30. [ Google Scholar ]
• MacKenzie S.B., Podsakoff P.M., Podsakoff N.P. Construct measurement and validation procedures in MIS and behavioral research: integrating new and existing techniques. MIS Q. 2011; 35 (2):293–334. [ Google Scholar ]
• Mattison Thompson F., Tuzovic S., Braun C. Trustmarks: strategies for exploiting their full potential in e-commerce. Bus. Horiz. 2019; 62 (2):237–247. [ Google Scholar ]
• McNulty E.J. 2020. Leading through Covid-19. MIT Sloan Management Review. https://sloanreview.mit.edu/article/leading-through-covid-19/ accessed 23 March 2020. [ Google Scholar ]
• Nielsen . 2020. Nielsen Investigation: “Pandemic Pantries” Pressure Supply Chain amid Covid-19 Fears. https://www.nielsen.com/us/en/insights/article/2020/nielsen-investigation-pandemic-pantries-pressure-supply-chain-amidst-covid-19-fears/ accessed 22 March 2020. [ Google Scholar ]
• Nielsen Vietnam. 2020. Nielsen Survey: Impact of Covide-19 on Vietnamese. Nielsen Vietnam. https://www.facebook.com/NielsenVietnam/photos/nielsen-survey-impact-of-covid-19-on-vietnameseappearing-at-the-end-of-2019-coro/2919284404799369/ accessed 25 March 2020. [ Google Scholar ]
• Pantano E., Pizzi G., Scarpi D., Dennis C. Competing during a pandemic? Retailers' ups and downs during the COVID-19 outbreak. J. Bus. Res. 2020; 116 :209–213. [ PMC free article ] [ PubMed ] [ Google Scholar ]
• Pappas N. Marketing strategies, perceived risks, and consumer trust in online buying behaviour. J. Retailing Consum. Serv. 2016; 29 :92–103. [ Google Scholar ]
• Pesek W. 2020. China's Recession Will Hurt as Coronavirus Crisis Spreads to U.S. And Europe, Forbes. https://www.forbes.com/sites/forbes-personal-shopper/2020/03/16/disney-amazon-prime-video--more-best-tv-shows-to-stream-during-your-time-home/#3336a7ee5882 accessed 18 March 2020. [ Google Scholar ]
• Plangger K., Watson R.T. Balancing customer privacy, secrets, and surveillance: insights and management. Bus. Horiz. 2015; 58 (6):625–633. [ Google Scholar ]
• Prentice C., Chen J., Stantic B. Timed intervention in COVID-19 and panic buying. J. Retailing Consum. Serv. 2020; 57 doi: 10.1016/j.jretconser.2020.102203. [ CrossRef ] [ Google Scholar ]
• Qin X., Liu Z., Tian L. The strategic analysis of logistics service sharing in an e-commerce platform. Omega. 2020; 92 [ Google Scholar ]
• Ray A., Dhir A., Bala P.K., Kaur P. Why do people use food delivery apps (FDA)? A uses and gratification theory perspective. J. Retailing Consum. Serv. 2019; 51 :221–230. [ Google Scholar ]
• Sharma R., Jha M. Values influencing sustainable consumption behaviour: exploring the contextual relationship. J. Bus. Res. 2017; 76 :77–88. [ Google Scholar ]
• Simon F. Relationship norms and media gratification in relational brand communication. J. Bus. Res. 2017; 79 :12–22. [ Google Scholar ]
• Smith T., Prosser T. Elsevier; 2020. COVID-19 Drug Therapy–Potential Options. https://www.elsevier.com/__data/assets/pdf_file/0007/988648/COVID-19-Drug-Therapy_Mar-2020.pdf accessed 23 March 2020. [ Google Scholar ]
• Solomon M.R. 12 ed. Pearson; 2017. Consumer Behavior: Buying, Having, and Being. [ Google Scholar ]
• Song W., Li W., Geng S. Effect of online product reviews on third parties' selling on retail platforms. Electron. Commer. Res. Appl. 2020; 39 [ Google Scholar ]
• Verma P., Sinha N. Integrating perceived economic wellbeing to technology acceptance model: the case of mobile based agricultural extension service. Technol. Forecast. Soc. Change. 2018; 126 :207–216. [ Google Scholar ]
• Vietnam News . 2020. How COVID-19 Affects Vietnamese Consumers' FMCG Spend and Purchase Behaviours. https://vietnamnews.vn/economy/654138/how-covid-19-affects-vietnamese-consumers-fmcg-spend-and-purchase-behaviours.html accessed 26 March 2020. [ Google Scholar ]
• Vinzi V.E., Chin W.W., Henseler J., Wang H. Springer; 2010. Handbook of Partial Least Squares. [ Google Scholar ]
• Wagner G., Schramm-Klein H., Steinmann S. Online retailing across e-channels and e-channel touchpoints: empirical studies of consumer behavior in the multichannel e-commerce environment. J. Bus. Res. 2020; 107 :256–270. [ Google Scholar ]
• Wang D., Chen B., Chen J. Credit card fraud detection strategies with consumer incentives. Omega. 2019; 88 :179–195. [ Google Scholar ]
• Wang Y., Herrando C. Does privacy assurance on social commerce sites matter to millennials? Int. J. Inf. Manag. 2019; 44 :164–177. [ Google Scholar ]
• WHO . WHO; 2010. What Is a Pandemic? World Health Organization. https://www.who.int/csr/disease/swineflu/frequently_asked_questions/pandemic/en/ accessed 23 March 2020. [ Google Scholar ]
• WHO . World Health Organization; 2020. Q&A on Coronaviruses (COVID-19) https://www.who.int/news-room/q-a-detail/q-a-coronaviruses WHO. accessed 24 March 2020. [ Google Scholar ]
• Worldometer . 2020. Coronavirus. https://www.worldometers.info/coronavirus/ accessed 25 March 2020. [ Google Scholar ]
• Zhang S., Pauwels K., Peng C. The impact of adding online-to-offline service platform channels on firms' offline and total sales and profits. J. Interact. Market. 2019; 47 :115–128. [ Google Scholar ]
• Zhang T., Li G., Cheng T.C.E., Lai K.K. Welfare economics of review information: implications for the online selling platform owner. Int. J. Prod. Econ. 2017; 184 :69–79. [ Google Scholar ]

ORIGINAL RESEARCH article

Online consumer satisfaction during covid-19: perspective of a developing country.

• 1 Antai College of Economics & Management, Shanghai Jiao Tong University, Shanghai, China
• 2 School of Management, Zhejiang Shuren University, Hangzhou, China
• 3 Faculty of Management Sciences, Riphah International University, Faisalabad Campus, Punjab, Pakistan
• 4 Department of Professional Psychology, Bahria University, Islamabad, Pakistan

A conceptual model based on the antecedents and consequences of online consumer satisfaction has been proposed and empirically proved in this study. Data were collected during Smart Lockdown of COVID-19 from 800 respondents to observe the difference between perceived and actual, and direct and indirect e-stores. Confirmatory factor analysis was used to observe the validity of the data set. The structural equation modeling technique was used to test the hypotheses. The findings indicated that consumers feel more satisfied when they shop through direct e-store than indirect e-store, whereas their perception and actual experience are different. Implications have also been added to the study.

Introduction

Online shopping is the act of buying a product or service through any e-stores with the help of any website or app. Tarhini et al. (2021) stated that shopping through online channels is actively progressing due to the opportunity to save time and effort. Furthermore, online shopping varies from direct e-store and indirect e-store about their perception against the actual experience. Developing countries still face various conflicts and issues while promoting and utilizing e-commerce to the maximum compared with the developed countries ( Rossolov et al., 2021 ). In the developing countries, the difference between the perception and actual experience of the consumers varies when buying from indirect e-store compared to the direct e-store. On the contrary, as the world has been suffering from the COVID-19 pandemic, it has brought drastic changes globally in many sectors, business being one of them. De Vos (2020) stated that a large-scale lockdown was imposed worldwide to prevent the virus from spreading.

To survive, switching traditional shopping or trade toward digital was one factor that captured the attention across the globe on a larger scale. In April 2020, Walmart reported a 74% increase in online sales even though they faced a low customer walk-in at stores ( Nassauer, 2020 ; Redman, 2020 ). This upsurge of swift adoption of online channels has led this research to ask a few questions. First, what will be the difference between the perceived and the actual product purchased online? A recent study has documented that consumers bear actual risk after shopping through online channels ( Yang et al., 2020 ). Research reported that 30% of the products through online channels get returned and are not according to their perception ( Saleh, 2016 ). The same author also showed that the return and complaint rates are getting higher when consumers shop through an online channel.

Second, is there any difference between the perceived and the actual product purchase online from a direct e-store or an indirect e-store? Direct e-store means the online brand store, for example, Walmart, and indirect e-store means third-party stores such as Amazon, Alibaba, Jingdong (JD), and Daraz. The direct e-store strives hard to maintain a clear, potent perception in the mind of its buyer ( Grewal et al., 2009 ). Kumar and Kim (2014) stated that a brand strengthening its relationship with its consumer satisfies its needs through the actual product or services. In the literature ( Olotewo, 2017 ; Rossolov et al., 2021 ), it is stated that the shopping patterns of buyers from direct and indirect e-stores vary greatly, especially in the developing countries. In this way, when shopping through a direct e-store, consumers may easily recognize the difference in buying from a direct and indirect e-stores ( Mendez et al., 2008 ).

Third, a conceptual framework from a consumer perspective, antecedents and consequences of customer satisfaction, has been proposed and empirically proved. The literature ( Alharthey, 2020 ) discussed different risk types in online shopping. Three main types of risk, perceived uncertainty, perceived risk, and price, are addressed in this model. To the best of the knowledge of the authors, no such investigation directed specific circumstances, particularly in the developing countries. Therefore, it is necessary to look for the antecedents and consequences of customer satisfaction to promote online shopping in the developing countries. The degree of consumer satisfaction defines his/her experience and emotions about the product or service purchased through the online channel. Recent studies ( Guzel et al., 2020 ; Mamuaya and Pandowo, 2020 ) stated that the intention of the consumers to repurchase and their electronic-word-of-mouth (e-WOM) depends on their degree of satisfaction. In light of these heavy investments in online shopping, there is an exciting yet unexplored opportunity to comprehend better how the purchasing experiences of consumers through online channels influence their satisfaction level.

The study contributed to the current marketing literature in several ways. First, this study has highlighted that the perceived risk is very high when shopping through online channels, mainly the indirect e-stores. Therefore, the managers should develop strategies that reduce the perceived risk for the online consumer to shop more. Second, the study also disclosed that the perceived uncertainty in shopping through the online channel is high. While shopping online, the website design, graphics, and color scheme make the product more attractive than the actual one. Therefore, the managers must balance the visual appearance of the product on the website with the actual appearance of the product. This would increase the confidence and satisfaction of the consumer. Third, this study has also revealed that people are more satisfied while shopping from direct e-stores than indirect e-stores. Because the focal brands officially sponsor the direct e-stores, they pay more attention to their quality to retain consumers and maintain their brand reputation. Fourth, an indirect e-store works as a third party or a retailer who does not own the reputation of the product. This study exhibited the difference between the perception of the consumer being very high and the actual experience of using that product being quite different when shopping from the indirect channel. Last but not the least, this study is the first to report pre- and post-purchase consumer behavior and confirmed the perceived and the actual quality of a product bought from (i) direct e-store and (ii) indirect e-store.

Literature Review

Theoretical review.

Literature shows that when consumers get influenced to buy a particular product or service, some underlying roots are based on their behavior ( Wai et al., 2019 ). Appraisal theory significantly explains consumer behavior toward shopping and provides an opportunity to analyze the evaluation process (e.g., Roseman, 2013 ; Kähr et al., 2016 ; Moors et al., 2017 ; Ul Haq and Bonn, 2018 ). This research, aligned with the four dimensions of appraisal theory as the first stage, clearly defines the agency stage that either of the factors is responsible for customer satisfaction. The second stage explains that consumer's degree of satisfaction holds great importance and refers to novelty in the literature. The third stage of the model briefly explains the feelings and emotions of the consumers about the incident, aligning with the certainty phase. The last step explains whether the consumers have achieved their goal or are not aligned with the appetitive purpose.

Cognitive appraisal researchers stated that various emotions could be its root cause ( Scherer, 1997 ); it could be the reaction to any stimulus or unconscious response. On the contrary, four dimensions of appraisal theory are discussed in this research ( Ellsworth and Smith, 1988 ; Ma et al., 2013 ). Agency (considering themselves or objects are answerable for the result of the circumstance) ( Smith and Ellswoth, 1985 ; Durmaz et al., 2020 ); novelty (assessing the difference between the perception of an individual and his actual experience) ( Ma et al., 2013 ); certainty (analysis of the apparent probability of a specific outcome and its effect on the emotions of the buyer) ( Roseman, 1984 ), and appetitive goal (judging the degree to what extent the goal has been achieved) ( Hosany, 2012 ).

Hypotheses Development

Perceived risk and consumer satisfaction.

Perceived risk is the perception of shoppers having unpleasant results for buying any product or service ( Gozukara et al., 2014 ). Consumers who buy a specific product or service strongly impact their degree of risk perception toward buying ( Jain, 2021 ). Buyers who tend to indulge in buying through online channels face perceived risk characterized by their perception compared to the actual uncertainty involved in it ( Kim et al., 2008 ). Literature ( Ashoer and Said, 2016 ; Ishfaq et al., 2020 ) showed that as the risk of buying is getting higher, it influences the degree of consumers about information about their buying, either purchasing from the direct or indirect e-shop. Johnson et al. (2008) stated that consumer judgment that appears due to their experience strongly impacts their satisfaction level. Jin et al. (2016) said that as the ratio of risk perception of their consumer decreases, it enhances customer satisfaction. Thus, from the above arguments, it is hypothesized as follows:

H 1 : Perceived risk has a significant negative impact on consumer satisfaction—direct vs. indirect e-store; perceived vs. actual experience .

Perceived Uncertainty and Consumer Satisfaction

Uncertainty is defined as a time that occurs in the future that comprises the predictable situation due to the asymmetry nature of data ( Salancik and Pfeffer, 1978 ). Consumers may not expect the outcome of any type of exchange conducted as far as the retailer and product-oriented elements are concerned ( Pavlou et al., 2007 ). Therefore, uncertainty initiates that retailers may not be completely predictable; on the contrary, consumers tend to analyze and understand their actions about decision making ( Tzeng et al., 2021 ). Thus, the degree of uncertainty involved in buying through online channels influences that degree of customer satisfaction. In addition, when the performance of any particular product or service matches the degree of expectations, he gets satisfied and, hence, repeats his decision of buying ( Taylor and Baker, 1994 ). But if the product quality fails to meet the requirements, it negatively affects the degree of satisfaction ( Cai and Chi, 2018 ).

H 2 : Perceived uncertainty has a significant negative impact on consumer satisfaction—direct vs. indirect e-store; perceived vs. actual experience .

Price Value and Consumer Satisfaction

Oliver and DeSarbo (1988) suggested that the price value is the proportion of the result of the buyer to the input of the retailer. It is defined as an exchange of products/services based on their quality against a price that is to be paid ( Dodds et al., 1991 ). Consumers look for a higher value in return; consumers are willing to pay a higher price ( Pandey et al., 2020 ). Yet, it leads to higher dissatisfaction when they receive a lower degree of profitable products. Besides, the buyers associate such type of product/service they use as less favorable or not according to their needs and desires. Hence, the buyers regret their decision-making degree for choosing that particular product ( Zeelenberg and Pieters, 2007 ). Aslam et al. (2018) indicated that a product/service price influences the satisfaction of a buyer. Afzal et al. (2013) recommended that the price is among those factors that hold great significance for the degree of satisfaction of the consumer. If the price value of any product/service differs from consumer to consumer, consumers tend to switch brands. Hence, it is hypothesized that:

H3 : Price value has a significant positive impact on consumer satisfaction—direct vs. indirect e-store; perceived vs. actual experience .

Consumer Satisfaction With Consumer Delight, Consumer Regret, and Outrage

Satisfaction is defined as how a consumer is pleased with a particular brand or view about a product/service that matches requirements. It is an essential factor that triggers when the product or service performance exceeds the expectation and perception of the customers ( Woodside et al., 1989 ). The decision of the buyer significantly affects their satisfaction toward the product or service ( Park et al., 2010 ). If buyers are satisfied with the product/service they purchased online, this degree of satisfaction significantly affects their repurchase intention and WOM ( Butt et al., 2017 ). Tandon (2021) stated that a consumer satisfied with the product/service would get delighted. Consumer satisfaction, when exceeding the expectations, leads to consumer delight ( Mikulić et al., 2021 ). Mattila and Ro (2008) recommended that the buyer gets disappointed by anger, regret, and outrage. It also defines that negative emotions have a significant effect on the purchasing intention of the consumers. Oliver (1989) stated that unsatisfied buyers or products that do not fulfill the needs of the customers can create negative emotions. Sometimes, their feelings get stronger and result in sadness and outrage. Bechwati and Xia (2003) recommended that the satisfaction of the consumers influences their behavior to repurchase; outraged consumers due to dissatisfaction sometimes want to hurt the company. Besides deciding to purchase, consumers mostly regret their choices compared to other existing choices ( Rizal et al., 2018 ). Hechler and Kessler (2018) investigated that consumers who are outraged in nature actively want to hurt or harm the company or brand from which they got dissatisfied or hurt. Thus, it is proposed that:

H 4 : Consumer satisfaction has a significant negative impact on (a) consumer delight, (b) consumer regret, (c) consumer outrage—direct vs. indirect e-store; perceived vs. actual experience .

Consumer Delight and E-WOM

Oliver et al. (1997) recommended that a degree of delight in a buyer is termed as a positive emotion. Consumers purchase a product/service that raises their degree of expectation and gets them delighted ( Crotts and Magnini, 2011 ). Delighted buyers are involved in sharing their experiences with their friends and family and spreading positive WOM to others ( Parasuraman et al., 2020 ). Happy buyers generally share their opinions while posting positive feedback through social media platforms globally ( Zhang, 2017 ). A positive WOM of the buyer acts as a fundamental factor in spreading awareness about the product/service and strongly impacts other buyers regarding buying it ( Rahmadini and Halim, 2018 ). Thus, it is proposed that:

H5 : Consumer delight has a significant positive impact on E-WOM—direct vs. indirect e-store; perceived vs. actual experience .

Consumer Delight and Repurchase Intention

Delighted consumers tend toward brand loyalty; thus, they increase their buying intention of the service or product ( Ludwig et al., 2017 ; Ahmad et al., 2021 ). Customers can understand the objective of loyalty in purchasing a similar product or a new one from the same company. Delighted consumers tend to indulge in a higher degree of an emotional state that leads them to higher purchase intentions; it eliminates the switching of brands ( Parasuraman et al., 2020 ). Kim et al. (2015) stated that consumers delighted with a product or service of a brand become loyal to it, and the possibility of switching brands gets very low. Research ( Loureiro and Kastenholz, 2011 ; Tandon et al., 2020 ) shows that delighted consumers are more eager to purchase the same product again. Hence, it is proposed that:

H6 : Consumer delight has a significant positive impact on his repurchase intention—direct Vs. indirect e-store; Perceived Vs. actual experience

Consumer Regret and E-WOM

Regret is considered a negative emotion in reaction to an earlier experience or action ( Tsiros and Mittal, 2000 ; Kumar et al., 2020 ). Regret is when individuals frequently feel pity, disgrace, shame, or humiliation after acting in a particular manner and afterward try to amend their possible actions or decisions ( Westbrook and Oliver, 1991 ; Tsiros and Mittal, 2000 ). Regret is that specific negative emotion the buyers feel while making a bad decision that hurts them; their confidence level is badly affected. They blame themselves for choosing or creating a terrible decision ( Lee and Cotte, 2009 ). Li et al. (2010) suggested that buyers quickly start regretting and find their way to express their negative emotions. When they feel betrayed, they tend to spread negative WOM (NWOM) as a response to their anxiety or anger. Jalonen and Jussila (2016) suggested that buyers who get dissatisfied with their selections get involved in negative e-WOM about that particular brand/company. Earlier research says that buyers suffering from failure to buy any product/services tend to participate actively and play a role in spreading NWOM due to the degree of regret after making bad choices. Whelan and Dawar (2014) suggested that consumers sense that business has treated them unreasonably, and many consumers complain about their experience, resulting in e-WOM that may reduce consumer repurchase intention. Thus, it can be stated that:

H7 : Consumer regret has a significant negative impact on e-WOM—direct vs. indirect e-store; perceived vs. actual experience .

Consumer Regret and Repurchase Intention

Regret has a substantial influence on the intentions of the consumers to not entirely be measured by their degree of happiness ( Thibaut and Kelley, 2017 ). Results may not be evaluated by matching the structured degree of expectation but are also linked to alternatives reachable in the market. Therefore, such sort of evaluation and assessments will probably influence repurchase intention. For example, suppose the skipped reserve overtakes the picked alternative. In that case, the customer might change the replacement for the future purchase, regardless of whether the individual is profoundly happy with the picked option ( Liao et al., 2017 ). According to the researchers, there is a negative relationship between regret and consumer repurchase intention ( Liao et al., 2017 ; Durmaz et al., 2020 ). Furthermore, Unal and Aydin (2016) stated that perceived risk negatively impacts regret, influencing the repurchase intention of the consumers. Thus, it can be stated that:

H8 : Customer's regret has a significantly negative influence on his repurchase intention—direct vs. indirect e-store; perceived vs. actual experience .

Consumer Outrage and E-WOM

The disappointment of the consumers is a negative response to a product or a service ( Anderson and Sullivan, 1993 ). Outrage is the negative emotion a consumer experience when he purchases something totally against his requirements ( Lindenmeier et al., 2012 ). Besides, when the perception of the buyer is infringed, such behaviors occur. According to Torres et al. (2019) , enraged consumers get involved in communicating their outrage through e-WOM. Outraged consumers actively hurt the firm or brand from which they got hurt ( Hechler and Kessler, 2018 ). Consumers give e-WOM online reviews to decrease the negative emotions from the experiences of the consumer and re-establish a calm mental state to equilibrium ( Filieri et al., 2021 ). Thus, such consumers tend to give negative comments about the brand or product, which failed to match their expectations. NWOM has been characterized as negative reviews shared among people or a type of interpersonal communication among buyers concerning their experiences with a particular brand or service provider ( Balaji et al., 2016 ). Hence, it is hypothesized that:

H9 : Consumer outrage has a significant negative impact on e-WOM—direct vs. indirect e-store; perceived vs. actual experience .

Consumer Outrage and Repurchase Intentions

Repurchase intentions are characterized as the expressed trust of a buyer that they will or will not purchase a specific product and service again in the future ( Malhotra et al., 2006 ). Establishing relations with buyers should result in the repurchase. Negative disconfirmation ensues dissatisfaction or a higher level of outrage ( Escobar-Sierra et al., 2021 ). When a service/product fails and is not correctly addressed, the negative appraisal is overstated. Hence, “it may be more difficult to recover from feelings of victimization than to recover from irritation or annoyance” typically associated with dissatisfaction ( Schneider and Bowen, 1999 , p. 36). Therefore, consumers get outraged from buying such a product that fails to match their perception. When the experience of a consumer prompts a negative disconfirmation, the purchaser will also have a higher urging level through outrage. Therefore, consumers will probably have negative intentions to repurchase and do not want to indgule in making the same decision repeatedly ( Wang and Mattila, 2011 ; Tarofder et al., 2016 ). Therefore, it is proposed that:

H10 : Consumer outrage has a significant negative impact on repurchase intention—direct vs. indirect e-store; perceived vs. actual experience .

Methodology

This research explores the difference between the perception of the consumers and the actual online shopping experience through direct and indirect e-stores. It was an experimental design in which online shopping was studied in the developing countries. Data were collected from those individuals who shop from online channels; direct e-store and indirect e-store. Taking care of COVID-19 standard operating procedures, only 50 respondents were gathered two times, every time in a university auditorium after obtaining the permission from the administration. The total capacity of the auditorium was 500, as the lockdown restrictions were lifted after the first wave of the coronavirus.

Data Collection Tool

A questionnaire was used for the survey. The questionnaire was adapted in English to guarantee that the respondents quickly understood the questions used. A cross-sectional study technique was used for this research. A cross-sectional study helps in gathering the data immediately and collects data from a large sample size. The total number of distributed questionnaires was 1,250, out of which 800 were received in the usable form: 197 incomplete, 226 incorrect, and dubious responses, and 27 were eliminated. Thus, a 64% response rate was reported. Research showed that a 1:10 ratio is accepted ( Hair et al., 1998 ) as far as the data collection is concerned; for that instance, this study data fell in the acceptable range.

Indirect E-Store

Consumers who prefer to shop through online channels were gathered in an auditorium of an institute. Only those consumers were eligible for this experiment, who themselves buy through e-stores. A few products were brought from an indirect e-store, and later on, those products were shown to the respondents from the website of that indirect e-store. After showing products, we asked the respondents to fill the survey as per their perception of the product. Then we asked them to fill out another questionnaire to ascertain the difference between the perception and actual experience when purchasing from an indirect e-store. Once all the respondents completed the survey, we have shown them the actual products they have selected by seeing the website of the indirect e-store.

Direct E-Store

The second experiment was carried out on those consumers who shop from direct e-stores. For that purpose, a few popular reviewed clothing articles were purchased from the e-store. As in the case of an indirect e-store, respondents were also shown these articles from the websites of these direct e-stores. We then asked the respondents to fill the survey to confirm their perception of the products. Once all the respondents completed the survey, we showed them the actual product and asked them to fill out another questionnaire according to their actual purchasing experience from the direct e-store. The primary purpose of this experiment was to compare buying from direct e-store and indirect e-store.

Construct Instruments

The total number of items was 34, which were added in the earlier section of the questionnaire. These items were evaluated with the help of using a five-point Likert scale that falls from strongly disagree (1) to strongly agree (5). The items used in the study were empirically validated. Table 2 carries the details of the items of the questionnaire. The price value was evaluated using three items used by Venkatesh et al. (2012) . The perceived uncertainty was one of the independent variables that carry four items derived from Pavlou et al. (2007) . Perceived risk was the third independent variable used, held three items; thus, its scale was derived from Shim et al. (2001) . Wang (2011) validated consumer satisfaction carrying three items; consumer delight was measured by a 3-item scale proposed by Finn (2012) ; consumer regret was measured by the scale proposed by Wu and Wang (2017) . It carries a three-item scale. Consumer outrage was measured by Liu et al. (2015 ); it has six items. Repurchase intention was measured through a scale adapted from Zeithaml et al. (1996) , which carries four items. e-WOM was validated by the scale adapted from Goyette et al. (2010) ; it has five items.

Demographics of the Respondents

A total of 800 questionnaires were filled, and the respondents expressed their perception and actual experience from direct e-store and indirect e-store. Respondents belonged to different age groups from 18 to 50 years and above. There were 49% women and 51% men who took part in filling this survey. The income level of the respondents was grouped in different categories from “above 10,000 to above 50,000. The majority (56%) of the respondents were single, and 44% were married (Details can be viewed in Figure 1 ; Table 1 ). Data for both direct and indirect e-store was collected equally; 50% each to compare each category better.

Figure 1 . Proposed conceptual framework.

Table 1 . Demographics of the respondents.

Reliability and Validity

Reliability evaluates with the help of composite reliability (CR). All CR values fall into the range of 0.7–0.9, which is acceptable ( Hair et al., 2011 ). Convergent and discriminant validity has been observed through confirmatory factor analysis as recommended by some researchers ( Fornell and Larcker, 1981 ; Hair et al., 2010 ).

Convergent Validity

Convergent validity is evaluated with the help of two standards mentioned in the literature earlier, factor loading and average variance extracted (AVE), both the values should be >0.5 ( Yap and Khong, 2006 ). The values are mentioned in Table 2 .

Table 2 . Reliability and convergent validity.

Discriminant Validity

Discriminant validity is evaluated based on two conditions that are required to evaluate it. First, the correlation between the conceptual model variables should be <0.85 ( Kline, 2005 ). Second, the AVE square value must be less than the value of the conceptual model ( Fornell and Larcker, 1981 ). Table 3 depicts the discriminant validity of the construct of the study.

Table 3 . Discriminant validity.

Multi-Group Invariance Tests

Multi-group confirmatory factor analysis was conducted as the pre-requisites for the measurement model. The multi-group analysis was used to investigate a variety of invariance tests. Different invariance tests were performed to guarantee the items working precisely in the same manner in all the groups. In this research, the following are the model fit indexes, that is, CMIN/dF =2.992 CFI = 0.915, TLI = 0.906, and RMSEA = 0.071. Byrne (2010) and Teo et al. (2009) stated that CFI gives more accurate results, especially when comparing variables in different groups.

Hypotheses Testing

Scanning electron microscope technique was used to run and test the proposed hypotheses for the conceptual model. First, all the hypotheses proposed were checked, from which eight were initially accepted. Later, the multi-group test was utilized to test the proposed hypotheses and compare the shopping experience from direct e-store with indirect e-store and consumer perception with actual experience. Table 4 explains this in detail.

Table 4 . Hypotheses results.

Discussion and Implications

This research offers a remarkable number of facts for practitioners. This study can benefit marketing strategists by reducing the perceived risk, decreasing the intensity of perceived uncertainty, stabilizing the price, enhancing consumer satisfaction, promoting delighting consumers, accepting the negative behavior of the consumers, consumer retention, and establishing a positive e-WOM.

Reducing Risks

Certain factors play a role in antecedents of consumer satisfaction; they are particularly those that resist consumers to shop from any online channel, neither direct e-store nor indirect e-store. Perceived risk, perceived uncertainty, and the price are some of those antecedents that play a significant role in affecting the degree of satisfaction of the consumers, resulting in either to retain a consumer or to outrage a consumer. This study aligns with the existing literature. Tandon et al. (2016) ; Bonnin (2020) and Pandey et al. (2020) showed that consumers seek to shop from an e-store without bearing any risk. Consumers feel more confident about an e-store when the perceived risk is less than shopping from traditional ones as consumers want to feel optimistic about their decision. Second, e-vendors should ensure that the quality of a product is up to the mark and according to the consumer needs. Therefore, vendors should offer complete details about the product/service and its risks to the consumers. Moreover, this study suggests that e-stores must align the visuals of a product with its actual appearance. This would help them to increase customer satisfaction and confidence in the e-store.

Focus on Consumer Satisfaction

Consumer satisfaction is the deal-breaker factor in the online sector. Literature ( Shamsudin et al., 2018 ; Hassan et al., 2019 ) showed that organizations prioritize their consumers by fulfilling their requirements and required assistance. As a result, consumers are more confident and become satisfied consumers in the long run. This study adds to the literature that the degree of satisfaction of the consumers plays an essential role in shopping from an e-store. Consumers feel more confident in shopping from a direct e-store than an indirect e-store as the difference in the perception of consumers and the actual experience varies. Therefore, online vendors should focus on satisfying their consumers as it plays a remarkable role in retaining consumers.

Value Consumer Emotions

Online, retaining, and satisfying consumers are the most vital factor that directly affects the organization. This research aligns with the existing literature ( Jalonen and Jussila, 2016 ; Hechler and Kessler, 2018 ; Coetzee and Coetzee, 2019 ); when the retailer successfully fulfills its requirements, the consumer gets delighted repeating his choice to repurchase. On the other hand, if the online retailer fails to serve the consumer, the consumer regrets and, in extreme cases, becomes outraged about his decision. The negative emotions of the consumers threaten the company from many perspectives, as the company loses its consumer and its reputation in the market is affected. Therefore, first, market practitioners should avoid ignoring the requirements of consumers. Second, online vendors should pay special attention to the feedback of the consumers and assure them that they are valued.

Consumer Retention

The ultimate goal is to retain its consumers, but e-vendors should make proper strategies to satisfy their consumers as far as the online sector is concerned. The earlier studies of Zhang et al. (2015) and Ariffin et al. (2016) contributed to the literature that consumer satisfaction is a significant aspect in retaining a consumer. This research has also suggested that the satisfaction of the consumers plays a vital role in retaining them. Moreover, online shoppers provide the fastest spread of the right WOM about the product/ service. Second, consumers should feel valued and committed to vendors.

Pre- and Post-buying Behavior

This study contributed to a conceptual model that deals with consumer pre- and post-purchase behavior from the direct and indirect e-stores. With the help of experimental design, this study has reported its finding, highlighted how a satisfied customer is delightful and shares e-WOM, and showed repurchase intention. However, if the customer is not satisfied with the flip of a coin, he may feel regretted or outraged and cannot share e-WOM or have a repurchase intention.

Conclusions

This research concludes that online shopping has boomed during this COVID-19 pandemic period, as the lockdown prolonged in both the developed and the developing countries. The study further supports the difference between shopping from a direct e-store and an indirect e-store. The perception of the consumers shopping from direct e-store is more confident, and their degree of satisfaction is much higher, as the actual experience of the consumers aligns with their perceptions. Instead, consumers feel dissatisfied or outraged to choose an indirect e-store for shopping. Indirect e-store makes false promises and guarantees to its buyers, and eventually, when the consumers experience the product, it is against their perception.

This research fills the literature gap about the antecedents that lead to online shopping growth in the developing countries. This study aligns with Hechler and Kessler's (2018) earlier research, which stated that dissatisfied consumers threaten the reputation of the organization. Furthermore, Klaus and Maklan (2013) , Lemon and Verhoef (2016) suggested that handling the experience and satisfaction of the buyers plays a significant role in surviving among its competitors. Grange et al. (2019) recommended that e-commerce develops and attracts consumers by fulfilling their needs and requirements quickly. This study aligned with the existing literature by adding factors influencing the shopping preferences of the consumers from an e-store.

Limitations and Future Research

Despite its significant findings, this research has some limitations and scope for future research. First, this research only examined a few risks involved in online shopping. Future research studies should analyze other risks, for example, quality risk and privacy risk. Second, this study focused on shopping through direct e-stores and indirect e-stores. Future research can implement a conceptual model of a specific brand. Third, this study can be implemented in other sectors, for example, tourism, and hospitality. Fourth, it may be fascinating to look at other fundamentals, such as age, gender, education, relation with the retailer, or the degree of involvement with online shopping to differentiate other factors.

The proposed framework can be utilized in other developing countries, as every country faces different problems according to its growth and development. The model can be examined among specific direct e-stores to compare new customers and loyal customers. Future studies can explore indirect relationships along with adding mediators and moderators in the proposed model.

Data Availability Statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.

Ethics Statement

The studies involving human participants were reviewed and approved by This study involving human participants was reviewed and approved by the Ethics Committee of the Department of Management Sciences, Riphah International University, Faisalabad Campus, Faisalabad, Pakistan. The participants provided their written informed consent to participate in this study. The patients/participants provided their written informed consent to participate in this study.

Author Contributions

AS contributed to the conceptualization and writing the first draft of the research. JU contributed to visualizing and supervising the research. All authors who contributed to the manuscript read and approved the submitted version.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Afzal, S., Chandio, A. K., Shaikh, S., Bhand, M., and Ghumro, B. A. (2013). Factors behind brand switching in cellular networks. Int. J. Asian Soc. Science 3, 299–307.

Google Scholar

Ahmad, W., Kim, W. G., Choi, H. M., and Ul Haq, J. (2021). Modeling behavioral intention to use travel reservation apps: a cross-cultural examination between US and China. J. Retail. Consum. Serv. 63:102689. doi: 10.1016/j.jretconser.2021.102689

CrossRef Full Text | Google Scholar

Alharthey, B. (2020). The role of online trust in forming online shopping intentions. Int. J. Online Market. 10, 32–57. doi: 10.4018/IJOM.2020010103

Anderson, E. W., and Sullivan, M. W. (1993). The antecedents and consequences of customer satisfaction for firms. Market. Sci. 12, 125–143. doi: 10.1287/mksc.12.2.125

Ariffin, S., Yusof, J. M., Putit, L., and Shah, M. I. A. (2016). Factors influencing perceived quality and repurchase intention towards green products. Proc. Econ. Finan. 37, 391–396. doi: 10.1016/S2212-5671(16)30142-3

Ashoer, M., and Said, S. (2016). “The impact of perceived risk on consumer purchase intention in Indonesia; a social commerce study,” in Proceeding of the International Conference on Accounting, Management, Economics and Social Sciences . 1–13.

Aslam, W., Arif, I., Farhat, K., and Khursheed, M. (2018). The role of customer trust, service quality and value dimensions in determining satisfaction and loyalty: an Empirical study of mobile telecommunication industry in Pakistan. Market-TrŽište 30, 177–194. doi: 10.22598/mt/2018.30.2.177

Balaji, M. S., Khong, K. W., and Chong, A. Y. L. (2016). Determinants of negative word-of-mouth communication using social networking sites. Inform. Manage. 53, 528–540. doi: 10.1016/j.im.2015.12.002

Bechwati, N. N., and Xia, L. (2003). Do computers sweat? the impact of perceived effort of online decision aids on consumers' satisfaction with the decision process. J. Consum. Psychol. 13, 139–148. doi: 10.1207/S15327663JCP13-1andamp;2_12

Bonnin, G. (2020). The roles of perceived risk, attractiveness of the online store and familiarity with AR in the influence of AR on patronage intention. J. Retail. Consum. Serv. 52:101938. doi: 10.1016/j.jretconser.2019.101938

Butt, M. M., Rose, S., Wilkins, S., and Haq, J. U. (2017). MNCs and religious influences in global markets: drivers of consumer-based halal brand equity. Int. Market. Rev . 12:277. doi: 10.1108/IMR-12-2015-0277

Byrne, B. M. (2010). Structural Equation Modeling With AMOS: Basic Concepts, Applications, and Programming , 2nd Edn. New York, NY: Routledge.

Cai, R., and Chi, C. G. Q. (2018). The impacts of complaint efforts on customer satisfaction and loyalty. Serv. Industr. J. 38, 1095–1115. doi: 10.1080/02642069.2018.1429415

Coetzee, A., and Coetzee, J. (2019). Service quality and attitudinal loyalty: the mediating effect of delight on retail banking relationships. Glob. Bus. Econ. Rev. 21, 120–138. doi: 10.1504/GBER.2019.096856

Crotts, J. C., and Magnini, V. P. (2011). The customer delight construct: is surprise essential? Ann. Tourism Res. 38, 719–722. doi: 10.1016/j.annals.2010.03.004

De Vos, J. (2020). The effect of COVID-19 and subsequent social distancing on travel behavior. Transport. Res. Interdisciplin. Perspect. 5:100121. doi: 10.1016/j.trip.2020.100121

PubMed Abstract | CrossRef Full Text | Google Scholar

Dodds, W. B., Monroe, K. B., and Grewal, D. (1991). Effects of price, brand, and store information on buyers' product evaluations. J. Market. Res. 28, 307–319. doi: 10.1177/002224379102800305

Durmaz, Y., Demira,g, B., and Çavuşoglu, S. (2020). Influence of regret and regret reversing effort on dissatisfaction and repurchase intention after purchasing fashion products. Preprints. doi: 10.20944/preprints202003.0280.v1

Ellsworth, P. C., and Smith, C. A. (1988). Shades of joy: patterns of appraisal differentiating pleasant emotions. Cogn. Emot. 2, 301–331. doi: 10.1080/02699938808412702

Escobar-Sierra, M., García-Cardona, A., and Vera Acevedo, L. D. (2021). How moral outrage affects consumer's perceived values of socially irresponsible companies. Cogent Bus. Manage. 8:1888668. doi: 10.1080/23311975.2021.1888668

Filieri, R., Galati, F., and Raguseo, E. (2021). The impact of service attributes and category on eWOM helpfulness: an investigation of extremely negative and positive ratings using latent semantic analytics and regression analysis. Comput. Human Behav. 114:106527. doi: 10.1016/j.chb.2020.106527

Finn, A. (2012). Customer delight: distinct construct or zone of nonlinear response to customer satisfaction? J. Serv. Res. 15, 99–110. doi: 10.1177/1094670511425698

Fornell, C., and Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Market. Res. 18, 39–50. doi: 10.1177/002224378101800104

Goyette, I., Ricard, L., Bergeron, J., and Marticotte, F. (2010). e-WOM Scale: word-of-mouth measurement scale for e-services context. Can. J. Admin. Sci. 27, 5–23. doi: 10.1002/cjas.129

Gozukara, E., Ozyer, Y., and Kocoglu, I. (2014). The moderating effects of perceived use and perceived risk in online shopping. J. Glob. Strateg. Manage. 16, 67–81. doi: 10.20460/JGSM.2014815643

Grange, C., Benbasat, I., and Burton-Jones, A. (2019). With a little help from my friends: Cultivating serendipity in online shopping environments. Inf. Manage . 56, 225–235.

Grewal, D., Levy, M., and Kumar, V. (2009). Customer experience management in retailing: An organizing framework. J. Retail. 85, 1–14. doi: 10.1016/j.jretai.2009.01.001

Guzel, M., Sezen, B., and Alniacik, U. (2020). Drivers and consequences of customer participation into value co-creation: a field experiment. J. Product Brand Manage. doi: 10.1108/JPBM-04-2020-2847

Hair, J. F., Anderson, R. E., Tatham, R. L., and Black, W. C. (1998). Multivariate Data Analysis . 5th ed., Hoboken, NJ: Prentice-Hall.

Hair, J. F., Celsi, M., Ortinau, D. J., and Bush, R. P. (2010). Essentials of Marketing Research , Vol. 2. New York, NY: McGraw-Hill/Irwin.

Hair, J. F., Ringle, C. M., and Sarstedt, M. (2011). PLS-SEM: indeed a silver bullet. J. Market. Theor. Pract. 19, 139–152. doi: 10.2753/MTP1069-6679190202

Hassan, S., Shamsudin, M. F., and Mustapha, I. (2019). The effect of service quality and corporate image on student satisfaction and loyalty in TVET higher learning institutes (HLIs). J. Tech. Educ. Train. 11:4.

Hechler, S., and Kessler, T. (2018). On the difference between moral outrage and empathic anger: anger about wrongful deeds or harmful consequences. J. Exp. Soc. Psychol. 76, 270–282. doi: 10.1016/j.jesp.2018.03.005

Hosany, S. (2012). Appraisal determinants of tourist emotional responses. J. Trav. Res. 51, 303–314. doi: 10.1177/0047287511410320

Ishfaq, M., Nazir, M. S., Qamar, M. A. J., and Usman, M. (2020). Cognitive bias and the Extraversion personality shaping the behavior of investors. Front. Psychol. 11:556506. doi: 10.3389/fpsyg.2020.556506

Jain, S. (2021). Examining the moderating role of perceived risk and web atmospherics in online luxury purchase intention. J. Fash. Market. Manage. Int. J. 05:89. doi: 10.1108/JFMM-05-2020-0089

Jalonen, H., and Jussila, J. (2016). “Developing a conceptual model for the relationship between social media behavior, negative consumer emotions and brand disloyalty,” in Conference on e-Business, e-Services and e-Society (Cham: Springer), 134–145. doi: 10.1007/978-3-319-45234-0_13

Jin, N., Line, N. D., and Merkebu, J. (2016). The impact of brand prestige on trust, perceived risk, satisfaction, and loyalty in upscale restaurants. J. Hospital. Market. Manage. 25, 523–546. doi: 10.1080/19368623.2015.1063469

Johnson, M. S., Sivadas, E., and Garbarino, E. (2008). Customer satisfaction, perceived risk and affective commitment: an investigation of directions of influence. J. Serv. Market. 5:120. doi: 10.1108/08876040810889120

Kähr, A., Nyffenegger, B., Krohmer, H., and Hoyer, W. D. (2016). When hostile consumers wreak havoc on your brand: the phenomenon of consumer brand sabotage. J. Mark. 80, 25–41. doi: 10.1509/jm.15.0006

Kim, D. J., Ferrin, D. L., and Rao, H. R. (2008). A trust-based consumer decision-making model in electronic commerce: the role of trust, perceived risk, and their antecedents. Decis. Support Syst. 44, 544–564. doi: 10.1016/j.dss.2007.07.001

Kim, M., Vogt, C. A., and Knutson, B. J. (2015). Relationships among customer satisfaction, delight, and loyalty in the hospitality industry. J. Hospital. Tourism Res. 39, 170–197. doi: 10.1177/1096348012471376

Klaus, P. P., and Maklan, S. (2013). Towards a better measure of customer experience. Int. J. Market Res. 55, 227–246. doi: 10.2501/IJMR-2013-021

Kline, R. B. (2005). Principles and Practice of Structural Equation Modeling , 2nd Edn. New York, NY: Guilford Press.

Kumar, A., Chaudhuri, D., Bhardwaj, D., and Mishra, P. (2020). Impulse buying and post-purchase regret: a study of shopping behaviour for the purchase of grocery products. Int. J. Manage. 11:57. doi: 10.34218/IJM.11.12.2020.057

Kumar, A., and Kim, Y. K. (2014). The store-as-a-brand strategy: the effect of store environment on customer responses. J. Retail. Consum. Serv. 21, 685–695. doi: 10.1016/j.jretconser.2014.04.008

Lee, S. H., and Cotte, J. (2009). Post-purchase Consumer Regret: Conceptualization and Development of the PPCR Scale . ACR North American Advances.

Lemon, K. N., and Verhoef, P. C. (2016). Understanding customer experience throughout the customer journey. J. Market. 80, 69–96. doi: 10.1509/jm.15.0420

Li, S., Zhou, K., Sun, Y., Rao, L. L., Zheng, R., and Liang, Z. Y. (2010). Anticipated regret, risk perception, or both: which is most likely responsible for our intention to gamble? J. Gambl. Stud. 26, 105–116. doi: 10.1007/s10899-009-9149-5

Liao, C., Lin, H. N., Luo, M. M., and Chea, S. (2017). Factors influencing online shoppers' repurchase intentions: the roles of satisfaction and regret. Inform. Manage. 54, 651–668. doi: 10.1016/j.im.2016.12.005

Lindenmeier, J., Schleer, C., and Pricl, D. (2012). Consumer outrage: emotional reactions to unethical corporate behavior. J. Bus. Res. 65, 1364–1373. doi: 10.1016/j.jbusres.2011.09.022

Liu, M. W., and Keh, H. T. (2015). Consumer delight and outrage: scale development and validation. J. Serv. Theory Pract . 25, 680–699. doi: 10.1108/JSTP-08-2014-0178

Loureiro, S. M. C., and Kastenholz, E. (2011). Corporate reputation, satisfaction, delight, and loyalty towards rural lodging units in Portugal. Int. J. Hospital. Manage. 30, 575–583. doi: 10.1016/j.ijhm.2010.10.007

Ludwig, N. L., Heidenreich, S., Kraemer, T., and Gouthier, M. (2017). Customer delight: universal remedy or a double-edged sword? J. Serv. Theory Pract . 8:197. doi: 10.1108/JSTP-08-2015-0197

Ma, J., Gao, J., Scott, N., and Ding, P. (2013). Customer delight from theme park experiences: the antecedents of delight based on cognitive appraisal theory. Ann. Tour. Res. 42, 359–381. doi: 10.1016/j.annals.2013.02.018

Malhotra, N., Hall, J., Shaw, M., and Oppenheim, P. (2006). Marketing Research: An Applied Orientation . Melbourne, VC: Pearson Education Australia.

Mamuaya, N. C., and Pandowo, A. (2020). Determinants of customer satisfaction and its implications on word of mouth in e-commerce industry: case study in Indonesia. Asia Pacific J. Manage. Educ. 3, 16–27. doi: 10.32535/apjme.v3i1.740

Mattila, A. S., and Ro, H. (2008). Discrete negative emotions and customer dissatisfaction responses in a casual restaurant setting. J. Hosp. Tour. Res. 32, 89–107. doi: 10.1177/1096348007309570

Mendez, J. L., Oubina, J., and Rubio, N. (2008). Expert quality evaluation and price of store vs. manufacturer brands: an analysis of the Spanish mass market. J. Retail. Consum. Serv. 15, 144–155. doi: 10.1016/j.jretconser.2007.11.003

Mikulić, J., Kreši,ć, D., and Šerić, M. (2021). The factor structure of medical tourist satisfaction: exploring key drivers of choice, delight, and frustration. J. Hospital. Tour. Res. 1177:1096348020987273. doi: 10.1177/1096348020987273

Moors, A., Boddez, Y., and De Houwer, J. (2017). The power of goal-directed processes in the causation of emotional and other actions. Emot. Rev. 9, 310–318. doi: 10.1177/1754073916669595

Nassauer, S. (2020). Walmart sales surge as Coronavirus drives Americans to stockpile. Wall Street J . Availale online at: https://www.wsj.com/articles/walmart-sales-surge-as-coronavirus-drivesamericans-to-stockpile-11589888464?mod=hp_lead_pos5 (accessed on May 18, 2020).

Oliver, R. L. (1989). Processing of the satisfaction response in consumption. J. Consum. Satisfact. Dissatisfact. Complain. Behav. 2, 1–26.

Oliver, R. L., and DeSarbo, W. S. (1988). Response determinants in satisfaction judgments. J. Consum. Res. 14, 495–507. doi: 10.1086/209131

Oliver, R. L., Rust, R. T., and Varki, S. (1997). Customer delight: foundations, findings, and managerial insight. J. Retail. 73:311. doi: 10.1016/S0022-4359(97)90021-X

Olotewo, J. (2017). Examining the antecedents of in-store and online purchasing behavior: a case of Nigeria. J. Market. Res. Case Stud. 15, 1–16. doi: 10.5171/2017.668316

Pandey, N., Tripathi, A., Jain, D., and Roy, S. (2020). Does price tolerance depend upon the type of product in e-retailing? role of customer satisfaction, trust, loyalty, and perceived value. J. Strateg. Market. 28, 522–541. doi: 10.1080/0965254X.2019.1569109

Parasuraman, A., Ball, J., Aksoy, L., Keiningham, T. L., and Zaki, M. (2020). More than a feeling? toward a theory of customer delight. J. Serv. Manage . 3:34. doi: 10.1108/JOSM-03-2019-0094

Park, E. O., Chung, K. H., and Shin, J. I. (2010). The relationship among internal marketing, internal customer satisfaction, organizational commitment and performance. Product. Rev. 24, 199–232. doi: 10.15843/kpapr.24.2.201006.199

CrossRef Full Text

Pavlou, P. A., Liang, H., and Xue, Y. (2007). Understanding and mitigating uncertainty in online exchange relationships: a principal-agent perspective. MIS Q. 105–136. doi: 10.2307/25148783

Rahmadini, Y., and Halim, R. E. (2018). The “Influence of social media towards emotions, brand relationship quality, and word of Mouth (WOM) on Concert's Attendees in Indonesia,” in MATEC Web of Conferences (EDP Sciences) , 05058.

Redman, R. (2020). “Online grocery sales to grow 40% in 2020,” in Supermarket News . Available online at: https://www.supermarketnews.com/onlineretail/online-grocery-sales-grow-40-2020 (accessed on May 21, 2020).

Rizal, H., Yussof, S., Amin, H., and Chen-Jung, K. (2018). EWOM towards homestays lodging: extending the information system success model. J. Hosp. Tour. Technol. doi: 10.1108/JHTT-12-2016-0084

Roseman, I. J. (1984). Cognitive determinants of emotion: a structural theory. Rev. Person. Soc. Psychol. 5, 11–36.

Roseman, I. J. (2013). Author reply: on the frontiers of appraisal theory. Emot. Rev. 5, 187–188. doi: 10.1177/1754073912469592

Rossolov, A., Rossolova, H., and Holguín-Veras, J. (2021). Online and in-store purchase behavior: shopping channel choice in a developing economy. Transportation 20, 1–37. doi: 10.1007/s11116-020-10163-3

Salancik, G. R., and Pfeffer, J. (1978). Uncertainty, secrecy, and the choice of similar others. Soc. Psychol. 23, 246–255. doi: 10.2307/3033561

Saleh, M. A. H. (2016). Website design, technological expertise, demographics, and consumer's e-purchase transactions. Int. J. Market. Stud. 8, 125–138. doi: 10.5539/ijms.v8n1p125

Scherer, K. R. (1997). The role of culture in emotion-antecedent appraisal. J. Pers. Soc. Psychol. 73:902. doi: 10.1037/0022-3514.73.5.902

Schneider, B., and Bowen, D. E. (1999). Understanding customer delight and outrage. Sloan Manage. Rev. 41, 35–45. doi: 10.1016/S0022-4359(01)00035–45

Shamsudin, M. F., Razak, A. A., and Salem, M. A. (2018). The role of customer interactions towards customer satisfaction in theme parks experience. Opcion 34, 546–558.

Shim, S., Eastlick, M. A., Lotz, S. L., and Warrington, P. (2001). An online prepurchase intentions model: the role of intention to search: best overall paper award—the Sixth Triennial AMS/ACRA Retailing Conference, 2000? J. Retail. 77, 397–416. doi: 10.1016/S0022-4359(01)00051-3

Smith, C. A., and Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. J. Pers. Soc. Psychol . 48:813.

Tandon, A., Aakash, A., and Aggarwal, A. G. (2020). Impact of EWOM, website quality, and product satisfaction on customer satisfaction and repurchase intention: moderating role of shipping and handling. Int. J. Syst. Assur. Eng. Manage. 54, 1–8. doi: 10.1007/s13198-020-00954-3

Tandon, U. (2021). Predictors of online shopping in India: an empirical investigation. J. Market. Anal. 9, 65–79. doi: 10.1057/s41270-020-00084-6

Tandon, U., Kiran, R., and Sah, A. N. (2016). Understanding online shopping adoption in India: unified theory of acceptance and use of technology 2 (UTAUT2) with perceived risk application. Serv. Sci. 8, 420–437. doi: 10.1287/serv.2016.0154

Tarhini, A., Alalwan, A. A., Al-Qirim, N., and Algharabat, R. (2021). “An analysis of the factors influencing the adoption of online shopping,” in Research Anthology on E-Commerce Adoption, Models, and Applications for Modern Business (Pennsylvania: IGI Global), 363–384.

Tarofder, A. K., Nikhashemi, S. R., Azam, S. F., Selvantharan, P., and Haque, A. (2016). The mediating influence of service failure explanation on customer repurchase intention through customers satisfaction. Int. J. Qual. Serv. Sci. 4:44. doi: 10.1108/IJQSS-04-2015-0044

Taylor, S. A., and Baker, T. L. (1994). An assessment of the relationship between service quality and customer satisfaction in the formation of consumers' purchase intentions. J. Retail. 70, 163–178. doi: 10.1016/0022-4359(94)90013-2

Teo, T., Lee, C. B., Chai, C. S., and Wong, S. L. (2009). Assessing the intention to use technology among preservice teachers in Singapore and Malaysia: a multigroup invariance analysis of the technology acceptance model (TAM). Comput. Educ. 53, 1000–1009. doi: 10.1016/j.compedu.2009.05.017

Thibaut, J. W., and Kelley, H. H. (2017). The Social Psychology of Groups . Routledge. doi: 10.4324/9781315135007

Torres, E. N., Milman, A., and Park, S. (2019). Customer delight and outrage in theme parks: a roller coaster of emotions. Int. J. Hospital. Tour. Administr. 16, 1–23. doi: 10.1080/15256480.2019.1641455

Tsiros, M., and Mittal, V. (2000). Regret: a model of its antecedents and consequences in consumer decision making. J. Consum. Res. 26, 401–417. doi: 10.1086/209571

Tzeng, S. Y., Ertz, M., Jo, M. S., and Sarigöll,ü, E. (2021). Factors affecting customer satisfaction on online shopping holiday. Market. Intell. Plann. 8:346. doi: 10.1108/MIP-08-2020-0346

Ul Haq, J., and Bonn, M. A. (2018). Understanding millennial perceptions of human and nonhuman brands. Int. Hospital. Rev . 9:14. doi: 10.1108/IHR-09-2018-0014

Unal, S., and Aydin, H. (2016). Evaluation of consumer regret in terms of perceived risk and repurchase intention. J. Glob. Strateg. Manage. 2, 31–31. doi: 10.20460/JGSM.20161024354

Venkatesh, V., Thong, J. Y., and Xu, X. (2012). Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Q. 157–178. doi: 10.2307/41410412

Wai, K., Dastane, O., Johari, Z., and Ismail, N. B. (2019). Perceived risk factors affecting consumers' online shopping behaviour. J. Asian Financ. Econ. Bus. 6, 246–260. doi: 10.13106/jafeb.2019.vol6.no4.249

Wang, C. Y., and Mattila, A. S. (2011). A cross-cultural comparison of perceived informational fairness with service failure explanations. J. Serv. Market . 25, 429–439. doi: 10.1108/08876041111161023

Wang, X. (2011). The effect of unrelated supporting service quality on consumer delight, satisfaction, and repurchase intentions. J. Serv. Res. 14, 149–163. doi: 10.1177/1094670511400722

Westbrook, R. A., and Oliver, R. L. (1991). The dimensionality of consumption emotion patterns and consumer satisfaction. J. Consum. Res. 18, 84–91. doi: 10.1086/209243

Whelan, J., and Dawar, N. (2014). Attributions of blame following a product-harm crisis depend on consumers' attachment styles. Mark. Lett. 27, 285–294. doi: 10.1007/s11002-014-9340-z

Woodside, A. G., Frey, L. L., and Daly, R. T. (1989). Linking service quality, customer satisfaction, and behavio. Mark. Health Serv. 9:5. doi: 10.1016/S0022-4359(01)0009-5

Wu, R., and Wang, C. L. (2017). The asymmetric impact of other-blame regret versus self-blame regret on negative word of mouth: empirical evidence from China. Eur. J. Market. doi: 10.1108/EJM-06-2015-0322

Yang, Y., Gong, Y., Land, L. P. W., and Chesney, T. (2020). Understanding the effects of physical experience and information integration on consumer use of online to offline commerce. Int. J. Inf. Manage. 51:102046. doi: 10.1016/j.ijinfomgt.2019.102046

Yap, B. W., and Khong, K. W. (2006). Examining the effects of customer service management (CSM) on perceived business performance via structural equation modelling. Appl. Stochast. Models Bus. Indus. 22, 587–605. doi: 10.1002/asmb.648

Zeelenberg, M., and Pieters, R. (2007). A theory of regret regulation 1.0. J. Consum. Psychol. 17, 3–18. doi: 10.1207/s15327663jcp1701_3

Zeithaml, V. A., Berry, L. L., and Parasuraman, A. (1996). The behavioral consequences of service quality. J. Market. 60, 31–46. doi: 10.1177/002224299606000203

Zhang, H. (2017). Understanding the Consumption Experience of Chinese Tourists: Assessing the Effect of Audience Involvement, Flow and Delight on Electronic Word-of-mouth (eWOM) (Doctoral dissertation).

Zhang, Z., Ye, Q., Song, H., and Liu, T. (2015). The structure of customer satisfaction with cruise-line services: an empirical investigation based on online word of mouth. Curr. Issues Tourism . 18, 450–464.

Keywords: consumer perception, online shopping, actual experiences, customer satisfaction, direct shopping, perceived risk, delight, outrage

Citation: Rao YH, Saleem A, Saeed W and Ul Haq J (2021) Online Consumer Satisfaction During COVID-19: Perspective of a Developing Country. Front. Psychol. 12:751854. doi: 10.3389/fpsyg.2021.751854

Received: 02 August 2021; Accepted: 30 August 2021; Published: 01 October 2021.

Reviewed by:

Copyright © 2021 Rao, Saleem, Saeed and Ul Haq. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Junaid Ul Haq, junaid041@yahoo.com

Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

A theoretical model of factors influencing online consumer purchasing behavior through electronic word of mouth data mining and analysis

Roles Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing

Affiliation School of Economics and Management, Zhengzhou University of Light Industry, High-tech District, Zhengzhou City, Henan Province, China

Roles Conceptualization, Funding acquisition, Project administration, Supervision

* E-mail: [email protected]

Affiliation School of Politics and Public Administration, Soochow University, Gusu District, Suzhou City, Jiangsu Province, China

Roles Data curation, Funding acquisition, Project administration

Roles Formal analysis, Funding acquisition, Project administration

• Qiwei Wang, 
• Xiaoya Zhu, 
• Manman Wang, 
• Fuli Zhou, 
• Shuang Cheng

• Published: May 18, 2023
• https://doi.org/10.1371/journal.pone.0286034
• Peer Review
• Reader Comments

The coronavirus disease 2019 pandemic has impacted and changed consumer behavior because of a prolonged quarantine and lockdown. This study proposed a theoretical framework to explore and define the influencing factors of online consumer purchasing behavior (OCPB) based on electronic word-of-mouth (e-WOM) data mining and analysis. Data pertaining to e-WOM were crawled from smartphone product reviews from the two most popular online shopping platforms in China, Jingdong.com and Taobao.com . Data processing aimed to filter noise and translate unstructured data from complex text reviews into structured data. The machine learning based K-means clustering method was utilized to cluster the influencing factors of OCPB. Comparing the clustering results and Kotler’s five products level, the influencing factors of OCPB were clustered around four categories: perceived emergency context, product, innovation, and function attributes. This study contributes to OCPB research by data mining and analysis that can adequately identify the influencing factors based on e-WOM. The definition and explanation of these categories may have important implications for both OCPB and e-commerce.

Citation: Wang Q, Zhu X, Wang M, Zhou F, Cheng S (2023) A theoretical model of factors influencing online consumer purchasing behavior through electronic word of mouth data mining and analysis. PLoS ONE 18(5): e0286034. https://doi.org/10.1371/journal.pone.0286034

Editor: Ahmad Samed Al-Adwan, Al-Ahliyya Amman University, JORDAN

Received: April 19, 2023; Accepted: May 5, 2023; Published: May 18, 2023

Copyright: © 2023 Wang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the manuscript and its Supporting Information files.

Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Henan Province Philosophy and Social Science Planning Project (grant number. 2020CZH012), the Henan Key Research and Development and Promotion Special (Soft Science Research) (grant number. 222400410126), the Jiangsu Province Social Science Foundation Youth Project (grant number. 21GLC012) and the Doctor Fund of Zhengzhou University of Light Industry (grant number. 2020BSJJ022, 2019BSJJ017). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

1. Introduction

A prolonged quarantine and lockdown imposed by the coronavirus disease 2019 (COVID-19) pandemic has changed the human lifestyle worldwide. The COVID-19 pandemic has negatively impacted various sectors such as manufacturing, import and export trade, tourism, catering, transportation, entertainment, especially retail and hence the global economy. Consumer behavior has gradually shifted toward contactless services and e-commerce activities owing to the COVID-19 [ 1 ].

Consumers are relying on e-commerce more than ever to protect their health. Recent advances in information technology, digital transformation, and the Internet helped consumers to encounter the COVID-19 to meet the needs of the daily lives, which led to an increase in the importance of e-commerce and changes in consumers’ online purchasing patterns [ 2 ]. When consumers shop online, their behavior is considered non-traditional, and is illustrated by a new trend and current environment. To analyze the influencing factors of online consumer purchasing behavior (OCPB), it is necessary to consider several factors, such as the price and quality of a product, consumers’ preferences, website design, function, security, search, and electronic word-of-mouth (e-WOM) [ 3 ]. As the current website design and payment security have become a user-friendly and guaranteed system compared with a decade ago, some factors are no longer considered as essential. By contrast, greater diversity and complexity have become the main characteristics of the influencing factors. Furthermore, under the traditional sales model, consumers’ purchase decisions were simple, while online consumers have more options in terms of shopping channels and decision choices. Meanwhile, in recent years, consumers’ preferences have gradually shifted from standardized products to customized and personalized. In line with these changes, information technology and data science, such as big data analytics, data mining from e-WOM, and machine learning (ML), adaptively analyze data regarding online consumers’ needs to obtain more accurate data.

Since the concept of big data was proposed in 2008, it has been applied and developed lasting 14 years, emerging as a valuable tool for global e-commerce recently. However, most enterprises have failed to seize the benefits generated from big data. In the context of big data, a huge number of comments were posted regarding e-malls (Amazon, Taobao, etc.) and online social media (blogs, Bulletin Board System, etc.). For instance, Amazon was the first e-commerce company to establish an e-WOM system in 1995, which provided the company with valuable suggestions from online consumers. E-WOM has greater credibility and persuasiveness, compared with traditional word of mouth (WOM), which is limited by various subjective factors. Moreover, e-WOM has the advantage of containing not only structured data (e.g., ratings) but also unstructured data (e.g., the specific content of consumer reviews). However, e-WOM provides product-related information that cannot be directly transformed to a research objective. Thus, an innovative method of big data analytics needs to be utilized to explore the influencing factors of OCPB, which shows the advantage of interdisciplinary applications.

The research problems are to explore the factors influencing OCPB through e-WOM data mining and analysis and explain the most important influencing factors for online consumers that are likely to exist in the future within the context of the COVID-19. The study fulfills the literature gaps on exploring influencing factors of OCPB from the perspective of e-WOM. The study makes a significant contribution to the consumer study because its findings can adequately identify the influencing factors of OCPB. It also provides the theoretical and managerial implications of its findings including how e-commerce platforms can use such data to adapt their platforms and marketing strategies to diverse situations.

The remainder of this is organized as follows. Section 1 presents the introduction. Section 2 discusses the literature review and hypotheses. Section 3 provides the methodology, including data mining and analysis. Section 4 describes the results, including K-means results, performance metrics, hypotheses results, and a theoretical model. Sections 5 and 6 provide discussion and conclusion, respectively.

2. Literature review and hypotheses

2.1 influencing factors of ocpb.

Online shopping has an increasing sales volume each year, which has become huge challenges for offline retailers. Venkatesh et al. [ 4 ] found that culture, demographics, economics, technology, and personal psychology were the main antecedents of online shopping, and the main drivers of online shopping were congruence, impulse buying behavior, value consciousness, risk, local shopping, shopping enjoyment, and browsing enjoyment by a comprehensive model of consumers online purchasing behavior. Within the context of COVID-19, OCPB is positively impacted by attitude toward online shopping [ 5 ]. Melović et al. [ 6 ] focused on millennials’ online shopping behavior and noted that the demographic characteristics, the affirmative characteristics, risks and barriers of online shopping were the key influencing factors. Based on the stimulus-organism-response (SOR) theory model, consumers’ actual impulsive shopping behavior is impacted by arousal and pleasure [ 7 ]. Furthermore, the influencing factors of consumers’ purchase behavior toward green brands are green perceived quality, green perceived value, green perceived risk, information costs saved, and purchase intentions by perceived risk theory [ 8 ]. The positive and negative effects of corporate social responsibility practices on consumers’ pro-social behavior are moderated by consumer-brand social distance, although it also impacts consumer behavior beyond the consumer-brand dyadic relationship [ 9 ]. Green perceived value, functional value, conditional value, social value, and emotional value may impact green energy consumers’ purchase behavior [ 10 ]. Recipients’ behavior and WOM predict distant consumers’ behavior [ 11 ]. Moreover, consumer behavior is significantly impacted by financial rewards, perceived intrusiveness, attitudes toward e-mail advertising, and intentions toward the senders [ 12 ]. Store brand consumer purchase behavior is positively impacted by store image perceptions, store brand price-image, value consciousness, and store brand attitude [ 13 ]. A meta-analysis summarizes the influencing factors of consumer behavior, household size, store brands, store loyalty, innovativeness, familiarity with store brands, brand loyalty to national brands, price consciousness, value consciousness, perceived quality of store brands, perceived value for money of store brands, and search versus experience positively impact consumer behavior, whereas price–quality consciousness, quality consciousness, price of store brands, and the consequences of making a mistake in a purchase negatively impact consumer behavior [ 14 ].

Based on protection motivation theory and theory of planned behavior (TPB), consumers are more likely to use online shopping channels than offline channels during the COVID-19 pandemic [ 15 ]. The TPB is also adapted to explain the influencing factors of consumers’ behavior in different areas. For instance, the attitude, perceived behavioral control, policy information campaigns, and past-purchase experiences significantly impact consumers’ purchase intention, whereas subjective and moral norms show no significant relationship based on the extended TPB [ 16 ]. Although green purchase behavior has different antecedents, only personal norms and value for money have fully significant relationships with green purchase behavior, environmental concern, materialism, creativity, and green practices. Functional value positively influences purchase satisfaction, physical unavailability, materialism, creativity, and green practices, and negatively influences the frequency of green product purchase by extending the TPB [ 17 ]. Meanwhile, Nimri et al. [ 18 ] utilized the TPB in green hotels and showed that knowledge and attitudes, as well as subjective injunctive norms, positively impacted consumers’ purchase intention. Yi [ 19 ] observed that attitude, social norm, and perceived behavioral control positively impacted consumers’ purchase intention based on the TPB. The factors of supportive behaviors for environmental organizations, subjective norms, consumer attitude toward sustainable purchasing, perceived marketplace influence, consumers’ knowledge regarding sustainability-related issues, and environmental concern are the influencing factors of consumers sustainable purchase behavior [ 20 ]. Consumers’ green purchase behavior is impacted by the intention through support of the TPB [ 21 ].

2.2 Influencing factors of emergency context attribute

Consumers exhibited panic purchase behavior during the COVID-19, which might have been caused by psychological factors such as uncertainty, perceptions of severity, perceptions of scarcity, and anxiety [ 22 ]. In the reacting phase, consumers responded to the perceived unexpected threat of the COVID-19 and intended to regain control of lost freedoms; in the coping phase, they addressed this issue by adopting new behaviors and exerting control in other areas, and in the adapting phase, they became less reactive and accommodated their consumption habits to the new normal [ 23 ]. The positive and negative e-WOMs may have significant influence on online consumers’ psychology. Specifically, e-WOM that conveys positive emotions (pride, surprise) tends to have a greater impact on male readers’ perception of the reviewer’s cognitive effort than female readers, whereas e-WOM that conveys negative emotions (anger, fear) has a greater impact on cognitive effort of female readers than male readers [ 24 ]. When online consumers believe their behavioral effect is feasible and positive, while their behavioral decision is related to the behavioral outcome [ 25 ]. Traditionally, there are five stages of consumer behavior that include demand identification, information search, evaluation of selection, purchase, and post-purchase evaluation. In addition, online purchase behavior involved in the various stages can be categorized into: attitude formation, intention, adoption, and continuation. Most of the important factors that influence online purchasing behavior are attitude, motivation, trust, risk, demographics, website, etc. “Internet Adoption” is widely used as a basic framework for studying “online buying adoption”. Psychological and economic structures associated with the IT adoption model can be used as the online consumer’s behavior models for innovative marketers. The adoption of online purchasing behavior is explained by different classic models of attitude behavior [ 26 ]. Consumer behaviors represented by customer trust and customer satisfaction, influence repurchase and positive WOM intentions [ 27 ]. Return policy leniency, cash on delivery, and social commerce constructs were significant facilitators of customer trust [ 28 ]. Meanwhile, seller uncertainty was negatively influenced by return policy leniency, information quality, number of positive comments, seller reputation, and seller popularity [ 29 ]. Social commerce components were a necessity in complementing the quality dimensions of e-service in the environment of e-commerce [ 30 ]. Perceived security, perceived privacy and perceived information quality were all significant facilitators of online customer trust and satisfaction [ 31 ].

E-service quality, consumer social responsibility, green trust and green perceived value have a significant positive impact on green purchase intention, whereas greenwashing has a significant negative impact on green purchase intention. In addition, consumer social responsibility, green WOM, green trust and green perceived value positively moderated the relationship between e-service quality and green purchase intention, while greenwashing and green participation negatively moderated the relationships [ 32 ]. Large-scale online promotions provide mobile users with a new shopping environment in which contextual variables simultaneously influence consumer behavior. There is ample evidence suggesting that mobile phone users are more impulsive during large-scale online promotion campaigns, which are the important contextual drivers that lead to the occurrence of mobile users’ impulse buying behavior in the “Double 11” promotion. The results show that promotion, impulse buying tendency, social environment, aesthetics, and interactivity of mobile platforms, and available time are the key influencing factors of impulse buying by mobile users [ 33 ]. Environmental responsibility, spirituality, and perceived consumer effectiveness are the key psychological influencing factors of consumers’ sustainable purchase decisions, whereas commercial campaigns encourage young consumers to make sustainable purchases [ 34 ]. The main psychological factors affecting consumers’ green housing purchase intention include the attitude, perceived moral obligation, perceived environmental concern, perceived value, perceived self-identity, and financial risk. Subjective norms, perceived behavioral control, performance risk, and psychological risk are not included. Meanwhile, the purchase intention is an important predictor of consumers’ willingness to buy [ 35 ]. The perceived control of flow and focus will positively affect the utilitarian value of consumers, while focus and cognitive enjoyment will positively impact the hedonic value. Moreover, utilitarian value has a greater impact on satisfaction than hedonic value. Finally, hedonic value positively impacts unplanned purchasing behavior [ 36 ]. Utilitarian and hedonic features achieve high purchase and WOM intentions through social media platforms and also depend on gender and consumption history [ 37 ].

Therefore, we present the following hypothesis:

• Hypothesis 1 (H1): Perceived emergency context attribute is the influencing factor of OCPB.

2.3 Influencing factors of perceived product attribute

Product quality and preferential prices are the major factors considered by online consumers, especially within the context of the COVID-19. Specifically, online shopping offers lower price, more choices for better quality products, and comparison between them [ 1 ]. Under the circumstance of online reviews, an original equipment manufacturer (OEM) selling a new product carefully decides whether to adopt the first phase remanufacturing entry strategy or to adopt the phase 2 remanufacturing entry strategy under certain conditions. Meanwhile, the OEM adopts penetration pricing for new and remanufactured products, when the actual quality of the product is high. Otherwise, it adopts a skimming pricing strategy, which is different from uniform pricing when there are no online reviews. Online reviews significantly impact OEM’s product profits and consumer surplus. Especially when the actual quality of the product is high enough, the OEM and the consumer will be also reciprocal [ 38 ]. Online reviews reduce consumers’ product uncertainty and improve the effect of consumer purchase decisions [ 39 , 40 ]. Uzir et al. [ 41 ] utilized the expectancy disconfirmation theory to prove that product quality positively impacts customer satisfaction, while product quality and customer satisfaction are mediated by customer’s perceived value. Product quality and customer’s perceived value will have greater influence with higher frequency of social media use. Nguyen et al. [ 42 ] studies consumer behavior from a cognitive perspective, and theoretically develops and tests two key moderators that influence the relationship between green consumption intention and behavior, namely the availability of green products and perceived consumer effectiveness.

Both sustainability-related and product-related texts positively influence consumer behavior on social media [ 43 ]. Online environment, price, and quality of the products are significantly impacted by OCPB. Godey et al. [ 44 ] explained the connections between social media marketing efforts and brand preference, price premium, and loyalty. Brand love positively impacts brand loyalty, and both positively impact WOM and purchase intention [ 45 ]. Brand names have a systematic influence on consumer’s product choice, which is moderated by consumer’s cognitive needs, availability of product attribute information, and classification of brand names. In the same choice set, the share of product choices with a higher brand name will increase and be preferred even if it is objectively inferior to other choices. Consumers with low cognitive needs use the heuristic of “higher is better” to select options labeled with brand names and choose brands with higher numerical proportions [ 46 ].

• Hypothesis 2 (H2): Perceived product attribute is the influencing factor of OCPB.

2.4 Influencing factors of perceived innovation attribute

Product innovation increases company’s competitive advantage by attracting consumers, whereas the enhancement of innovative design according to consumer behavior accelerates the development of sustainable product [ 47 , 48 ]. The innovation, WOM intentions and product evaluation can be improved positively by emotional brand attachment and decreased by perceived risk [ 49 ]. Based on the perspective of evolutionary, certain consumer characteristics, such as buyer sophistication, creativity, global identity, and local identity, influence firms’ product innovation performance, which can increase the success rate of product innovation, and enhance firms’ research and development performance [ 50 ]. However, technological innovation faces greater risk as it depends on market acceptance [ 51 ]. Moreover, electronic products rely more on technological innovation compared with other products, which maintain the profit and market [ 52 ]. The technological innovation needs to apply logical plans and profitable marketing strategies to reduce consumer resistance to innovation. Thus, Sun [ 53 ] explains the relationship between consumer resistance to innovation and customer churn based on configurational perspective, whereas the results show that response and functioning effect are significant but cognitive evaluation is not.

Based on the perspective of incremental product innovation, aesthetic and functional dimensions positively impact perceived quality, purchase intention, and WOM, whereas symbolic dimension only positively impacts purchase intention and WOM. By contrast, aesthetic and functional dimensions only positively impact perceived quality, whereas symbolic dimension positively impacts purchase intention and WOM. Furthermore, perceived quality partially mediates the relationship between aesthetic and functional dimensions and purchase intention and WOM by incremental product innovation, whereas perceived quality fully mediates the relationship between aesthetic and functional dimensions and purchase intention and WOM by radical product innovation [ 54 ]. Contextual factors, such as size of organizations and engagement in research and development activity, moderate the relationship between design and product innovation outcomes [ 55 ]. For radical innovations, low level of product innovation leads to more positive reviews and less inference of learning costs. As the functional attribute of radical innovations is not consistent with existing products, it is difficult for consumers to access relevant product category patterns and thus transfer knowledge to new products. The product innovation of aesthetics, functionality, and symbolism positively impact willingness to pay, purchase intention, and WOM through brand attitude [ 56 ]. This poor knowledge transfer results in consumers feeling incapable of effectively utilizing radical innovations, resulting in greater learning costs. In this case, product designs with low design novelty can provide a frame of reference for consumers to understand radical innovations. However, incremental product innovation shows no significant difference between a low and high level of design newness [ 57 ].

• Hypothesis 3 (H3): Perceived innovation attribute is the influencing factor of OCPB.

2.5 Influencing factors of perceived motivation attribute

The research has proven that almost all consumers’ purchases are motivated by emotion. Under this circumstance, an increase in online consumers’ positive emotions increases, their purchase frequency, whereas an increase in online consumers’ negative emotions reduces their purchase frequency. Additionally, user interface quality, product information quality, service information quality, site awareness, safety perception, information satisfaction, relationship benefits and related benefit factors have negative impacts on consumers’ online shopping emotionally. Nevertheless, only product information quality, user interface quality, and safety perception factors have positive effects on online consumer sentiment [ 58 ]. E-WOM carries emotional expressions, which can help consumers express the emotions timely. Pappas et al. [ 59 ] divides consumers’ motivation into four factors, namely entertainment, information, social-psychological, and convenience, while emotions into two factors, namely positive and negative. Specifically, according to complexity and configuration theories, a conceptual model by a fuzzy-set qualitative comparative analysis examines the relationship between a combination of motivations, emotions, and satisfaction, while results indicate that both positive and negative emotions can lead to high satisfaction when combing motivations.

From the perspective of SOR theory, consumers’ motivation is greatly influenced by self-consciousness, while conscious cognition plays the role of intermediary. First, after being stimulated by the external environment, online consumers will form “cognitive structure” depending on their subjectivity. Instead of taking direct action, they deliberately and actively obtain valid information from the stimulus process, considering whether to choose the product, and then react. Second, the stimulation stage in the retail environment can often attract the attention of consumers and cause the change of their psychological feelings. This stimulation is usually through external environmental factors, including marketing strategies and other objective influences. Third, organism stage is the internal process of an individual. It is a consumers’ cognitive process about themselves, their money, and risks after receiving the information they have seen or heard. Reaction includes psychological response and behavioral response, which is the decision made by the consumer after processing the information [ 60 ]. Based on literature review, 10 utilitarian motivation factors, such as desire for control, autonomy, convenience, assortment, economy, availability of information, adaptability/customization, payment services, absence of social interaction, and anonymity and 11 hedonic motivation factors, such as visual appeal, sensation seeking/entertainment, exploration/curiosity, escape, intrinsic enjoyment, relaxation, pass time, socialize, self-expression, role shopping, and enduring involvement with a product or service, are refined [ 61 ]. Consumers’ incidental moods can improve online shopping decisions impulsivity, while decision making process can be divided into orientation and evaluation [ 62 ]. Sarabia‐Sanchez et al. [ 63 ] combine K-means cluster and ANOVA analyses to explore the 11 motivational types of consumer values, which are achievement, tradition, inner space, universalism, hedonism, ecology, self-direction (reinforcement, creativity, harmony, and independence), and conformity.

• Hypothesis 4 (H4): Perceived motivation attribute is the influencing factor of OCPB.

3. Materials and methods

3.1 research design.

Given the present study’s objective to identify the influencing factors of OCPB, we analyzed e-WOM using big data analysis. To obtain accurate data of the influencing factors on OCPB, smartphones were the main object of data crawling. The rationale behind this choice is as follows. First, the time people spend using their smartphones is gradually increasing. Nowadays, smart phones are not only used for telephone calls or text messages, but also for taking photographs, recording video, surfing the web, online chatting, online shopping, and other such uses [ 64 ]. Second, smartphones have become a symbol of personal identification, as users’ using fingerprint or facial scans are frequently used to unlock devices, conduct online transactions, and make reservations, etc. Finally, smartphones’ software and hardware are updated frequently, so they may be considered high-tech products. Therefore, smartphones were chosen as the research object to determine which influencing factors affect OCPB.

Fig 1 shows the e-WOM data mining process and methods used. A dataset obtained from Taobao.com and Jingdong.com was collected by utilizing a Python crawling code, additional details of which are provided in Section 2.3. Section 2.4 addresses issues regarding language complexity. Moreover, Section 2.5 refers to the clustering of the influencing factors of OCPB through the K-means method of ML.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pone.0286034.g001

3.2 Data collection

The data were crawled from the e-commerce platforms Jingdong.com and Taobao.com by utilizing Python software. Jingdong and Taobao are the most powerful and popular platforms in China having professional e-WOM and user-friendly review systems. Specifically, the smartphone brands selected for analysis were Apple, Samsung, and Huawei because these three smartphone companies occupy the largest percentage of the smartphone market.

The authors determined that the analysis of the influencing factors of OCPB would be more persuasive and realistic by choosing smartphone models with high usage rate and liquidity. Thus, products reviews were crawled for the purchase of newly launched smartphones from Apple, Samsung, and Huawei in 2022. Specifically, to guarantee high-quality data, reviews from Taobao flagship stores and Jingdong directly operated stores were selected. However, we only collected reviews’ text content instead of images, videos, ratings, or rankings, the rationale was to ensure the reliability of data and meet research objectives. For instance, some e-commerce sellers attempt to increase their sales volume through deceitful methods, such as by faking ratings, rankings, and positive comments. Furthermore, online sellers and e-commerce companies (rather than consumers) often decide which smartphones are highest-rated and highest-selling. Finally, nowadays, the content of online reviews is not limited to text, as they also involve pictures, videos, and ratings, which have limited contribution in analyzing influencing factors of OCPB. Thus, the analyzed data regarding e-WOM in reviews was limited to text content.

In addition, to accurately reflect the real characteristics of OCPB during the COVID-19 pandemic, the study period ranged between February and May, 2022 (4 months). During that 4-month period, consumers exhibited a preference for buying products from e-commerce platforms. Specifically, the number of text reviews for the aforementioned types of smartphones was 51,2613 and 44,3678 in Taobao and Jingdong, respectively, for a total of 956,291 reviews.

3.3 Textual review processing method

As the crawled data exhibited noise, several data cleaning methods were adopted to filter noise and transform unstructured data of complex contextual review into structured data. Fig 1 shows the main procedures of the reviews’ pre-processing and the details are as follows.

First, to identify the range of sentences and for further data processing, sentences were apportioned using Python’s tokenizer package.

Second, this study employed Python’s Jieba package to perform word segmentation. The Jieba package is the Python’s best Chinese word segmentation module, comprising three modes. The exact mode was used to segment the sentences as accurately as possible, so they may be suitable for textual context analysis. The full mode was used to scan and process all words in each sentence, although it had a relatively high speed, it had a low capacity to resolve ambiguity. Additionally, the search engine mode segmented long words a second time, which allowed for the improvement of the recall rate, and was suitable for engine segmentation based on Jieba’s exact mode.

Third, stop words were deleted by referring to a stop words list. These included conjunctions, interjections, determiners, and meaningless words, among others. Finally, Python’s Word-to-vector (Word2vec) package was imported in the next step. Word2vec is an efficient training word vector model proposed by Mikolov [ 65 , 66 ]. The basic starting point was to match pairs of similar words. For instance, when “like” and “satisfy” appeared in a same context, they showed a similar vector, as both words had a similar meaning. Kim et al. [ 67 ] stated that a word could be considered a single vector and real numbers in the Word2vec model. In fact, most supervised ML models could be summarized as f ( x )−> y . Moreover, x could be considered a word in a sentence, while y could be considered this word in the context. Word2vec aimed to decide whether the sample of ( x , y ) could match the laws of natural language. Namely, after the process of Word2vec, the combination of word x and word y could be reasonable and logical or not. Table 1 shows the results of text processing.

https://doi.org/10.1371/journal.pone.0286034.t001

3.4 Influencing factors analysis by K-means

ML styles are divided into supervised and unsupervised algorithms. This study mainly utilized unsupervised algorithms to analyze the clusters of influencing factors of OCPB. Unsupervised algorithms consist in the clustering of unknown or unmarked objects without a trained sample [ 68 ]. This study utilized K-means to cluster the influencing factors.

For a given sample set, the K-means algorithm divides the sample set into k clusters according to the distance between samples. The main algorithm’s logic is to make the points in the cluster as close as possible, and to make the distance between the clusters as large as possible. Assuming that clusters can be divided into ( C 1 , C 2 ,…, C k ), the Euclidean distance of E is shown in Eq 1 .

The main procedures of K-means were the following.

Step 1 consisted of inputting the samples D = { x 1 , x 2 ,… x m }, K is the number of clusters, and appears as C = { C 1 , C 2 ,… C k }.

In Step 2, K samples were randomly selected from data set D as the initial K centroid vectors: { μ 1 , μ 2 ,… μ k }.

For Step5, it was necessary to repeat Steps 3 and 4, until all the centers μ remained steady. The final clustering result can be shown as C = { C 1 , C 2 ,… C k }.

The main procedures of K-means, according to Jain [ 69 ], are shown in Table 2 .

https://doi.org/10.1371/journal.pone.0286034.t002

4.1 K-means results

Based on the main procedures of K-means ( Table 2 ), the results are presented in Figs 2 – 4 .

https://doi.org/10.1371/journal.pone.0286034.g002

https://doi.org/10.1371/journal.pone.0286034.g003

https://doi.org/10.1371/journal.pone.0286034.g004

Four clusters of influencing factors of OCPB can be clearly identified in the analyses of the Jingdong dataset, Taobao dataset, and combined Jingdong and Taobao dataset. After checking the context of four clusters, even though small differences were found, their influence was found to be negligible for our analyses. Thus, Fig 4 was chosen as the benchmark of influencing factors of OCPB. In Section 4.3, the explanation and analysis of influencing factors of OCPB will be presented.

4.2 Performance metrics

First, performance metrics of sum of the square errors (SSE) and silhouette coefficient were adapted to verify the clustering results of K-means.

When the number of clusters does not reach the optimal numbers K, SSE decreases rapidly with the increase of the number of clusters, while SSE decreases slowly after reaching the optimal numbers, and the maximum slope is the optimal numbers K.

Where C i is the i th cluster, p is the sample point in C i (the mean value of all samples in C i ), and SSE is the clustering error of all samples, which represents the quality of clustering effect.

Fig 5 indicates that the SSE decreases rapidly when K equals the number of four.

https://doi.org/10.1371/journal.pone.0286034.g005

The range of sc i is between -1 and 1, the clustering effect is bad when sc i is below zero, whereas the clustering effect is good when sc i is near 1 conversely.

Based on Fig 6 , it is obviously to show that the silhouette coefficient reaches highest when K equals the number of four. Therefore, the results of the SSE and the silhouette coefficient jointly prove the number of K is four.

https://doi.org/10.1371/journal.pone.0286034.g006

4.3 Hypotheses results

Based on the K-means analysis, this section presents the influencing factors identified in the data from Jingdong and Taobao, which indicate the influencing factors influencing OCPB.

The first cluster comprises the perceived emergency context attribute, such as logistics, expressage, delivery, customer service, promotion, and reputation.

The second cluster comprises the perceived product attribute, such as appearance, brand, hand feeling, color, cost-performance ratio, price, design, and usability.

The third cluster comprises the perceived innovation attribute, such as photograph, quality and effects, screen quality, audio and video quality, pixel density, image resolution, earphone capabilities, and camera specifications.

The fourth cluster comprises the influencing factors, such as processing speed, operation, standby time, battery, system, internal storage, chip, performance, and fingerprint and face recognition, which cannot represent the perceived motivation attribute.

The results match the findings of Zhang et al. [ 70 ] to some extent, who identified 11 smartphone attributes based on online reviews: performance, appearance, battery, system, screen, user experience, photograph, price, quality, audio and video, and after-sale service. In addition, other scholars have explained the relationship between feature preferences and customer satisfaction [ 71 , 72 ], usage behavior and purchase [ 73 , 74 ], importance and costs of smartphones’ features and services [ 75 ], brand effects [ 76 ], and purchase behavior of people of different ages and gender groups [ 77 – 79 ]. Thus, H1, H2 and H3 are supported, while H4 is not supported according to the results of the K-means analysis.

4.4 Theoretical framework and validity of OCPB influencing factors

Kotler’s five product level model states that consumers have five levels of need comprising the core level, generic level, expected level, augmented level, and potential level. First, the core benefit is the fundamental need or want that consumers satisfy by consuming a product or service. Second, the generic level is a basic version of a product made up of only those features necessary for it to function. Third, the expected level includes additional features that the consumer might expect. Fourth, the augmented level refers to any product variations or extra features that might help differentiate a product from its competitors and make the brand a preferred choice amongst its competitors. Finally, a potential product includes all augmentations and improvements that a product might experience in the future [ 80 ].

In contrast with these levels, this study proposed the four influencing factors of OCPB. Based on Table 3 , first, the perceived emergency context H1 is not included in Kotler’s five products level, while the influencing factor expresses the significant characteristics of OCPB compared with Kotler’s model. Second, the perceived product attribute H2 could be considered the core and generic level. Third, the perceived innovation attribute H3 could be considered the potential level. Fourth, the results of H4 mainly reflects additional or special function of product, which meets the definition of the expected and augmented level. To refine the theoretical framework, H4 changes to the perceived functionality attribute by combing the explanation of the expected and augmented level, instead of the perceived motivation attribute. The details are shown in Fig 7 .

https://doi.org/10.1371/journal.pone.0286034.g007

https://doi.org/10.1371/journal.pone.0286034.t003

Fig 7 shows the four influencing factors of the theoretical framework of OCPB. Specifically, according to Kotler’s five products level, the perceived product attribute is the necessary influencing factor of OCPB, which meets the core drive and basic requirement. For instance, the core drive of purchasing of a smartphone is the core function of communication, and then the appearance, brand, color, etc. The perceived functionality attribute is the additional influencing factor of OCPB, which meets the expected and augmented requirement. For instance, when smartphones are in the same price range, consumers prefer to choose a smartphone belonging to better quality, smarter design, or better functionality. Moreover, the perceived innovation attribute is the attractive influencing factor of OCPB, which reflects the potential level. For instance, most consumers are the Apple fans mostly because the Apple products offer innovative usage experience and different technology elements yearly. Finally, the perceived emergency context attribute is the adaptive influencing factor of OCPB, which shows the main distinction with Kotler’s five products level. Further, because of the COVID-19, consumers only have online channel to purchase product under a prolonged quarantine and lockdown. Thus, in the emergency context, consumers primarily consider whether the product can be purchased in the e-commerce platform, whether the product can be delivered normally, or whether the packaged has been disinfected fully.

5. Discussion

Traditional consumer behavior is mainly affected by psychological, social, cultural, economic, and personal factors [ 81 , 82 ]. Park and Kim [ 83 ] conducted an empirical study to identify the key influencing factors that impact OCPB, which include service information quality, user interface quality, security perception, information satisfaction, and relational benefit. Further, Sata [ 84 ] conducted an empirical study and found that price, social group, product features, brand name, durability and after-sales services were important to consumers’ buying behavior when choosing a smartphone for purchase. Simultaneously, some studies have utilized big data technology to explore OCPB, exploring online consumers’ attitude toward products in different countries, and identified product features. However, these studies do not identify the influencing factors of OCPB and ignore e-WOM. To better explain OCPB influencing factors, e-WOM should be integrated into the theoretical framework and used in practical applications. Thus, this study contributes to OCPB research by data mining and analysis that can adequately identify the influencing factors based on e-WOM.

5.1 Theoretical implications

First, perceived emergency context attribute is the influencing factor of OCPB. Because of the COVID-19, e-commerce is the priority choice for consumers under circumstances of prolonged quarantine and lockdown, and then considering logistics and delivery. Furthermore, customer service, packaging, promotion, and reputation are critical to online consumers.

Second, perceived product attribute is the influencing factors of OCPB. The basic features of product, such as appearance, brand, hand feeling, price, and design, positively attract online consumers. Elegant appearance, famous brand, better hand feeling, lower price, and better design would be more impactful to OCPB.

Third, perceived innovation attribute is the influencing factor of OCPB. For smartphone, online consumers would show more interest in the innovation of speed, operation, standby time, chip, etc. Scientific and technological innovation for most products could improve the level of OCPB. Thus, the guarantee and improvement of functionality of a product could create more opportunities for online consumers to make purchasing decisions.

Fourth, according to Kotler’s five products level, perceived product attribute satisfies the characteristics of core drive and basic, while the perceived innovation attribute satisfies the characteristics of the potential level. Because hypothesis of perceived motivation attribute is not supported. Based on the analyzing results, the perceived functionality attribute is refined instead of the perceived motivation functionality attribute, which satisfies the expected and augmented. Meanwhile, the perceived emergency context attribute is not included, which shows the main difference with Kotler’s five products level.

5.2 Managerial implications

The influencing factors of OCPB were clustered into four categories: perceived emergency context, product, innovation, and function attributes. The definition and explanation of these categories may have important managerial implications for both OCPB and e-commerce. First, the findings of this study suggest that e-commerce enterprises should pay more attention to improving the quality, user experience, and additional design features of their products to arouse the interest of OCPB. However, this may be difficult for e-commerce enterprises because achieving these goals requires updating the software and hardware constantly, which involves significant investment. For most scientific and technical corporations, making heavy investments is not particularly difficult, however, service-type enterprises and small and medium enterprises may have insufficient funds to afford such heavy investments. This is the main reason that most online consumers buy products from famous brands instead of small and medium enterprises. Therefore, to improve their situation, both types of companies could jointly develop products or services, for instance, small and medium enterprises may purchase patents from large enterprises, jointly researching and developing products, or large enterprises could share their achievements at a price.

Second, the pandemic has accelerated the spread of e-commerce considerably, changing consumers’ shopping style in the process. Accordingly, e-commerce enterprises should adapt their marketing strategies, especially as the COVID-19 pandemic is still ongoing, due to the rapid development of the economy and its dynamic environment. For instance, e-commerce platforms should realize that changes in OCPB will continue to contribute to the growth of the e-commerce market. Moreover, e-commerce enterprises should combine their online presence with brick-and-mortar stores. Even more importantly, e-commerce enterprises should successfully operate their supply chain to adapt to the implementation of lockdown measures and the closing of manufacturing factories. Consumers should exercise caution when facing e-commerce enterprises’ adaptive financial policy, such as interest-free rates, which may cause financial burden.

Third, e-commerce enterprises should offer a simple and smooth shopping experience, clearly display practical information, increase the value of goods (by improving the quality, design, and performance of products or services) and improve their brand image for online consumers. However, e-commerce enterprises sometimes rely on certain fraudulent methods to increase their sales volume, such as falsifying positive e-WOM and deleting negative feedback, as was identified during the data processing stage. Therefore, online consumers should select online stores cautiously to avoid buying products of poor quality or performance.

Fourth, nowadays, technology is constantly evolving at an accelerated rate, particularly in the smartphone industry, as companies launch new products with innovative functions each year. Thus, e-commerce enterprises should strive to innovate to secure their position in the market. In addition, consumers should reconsider the need to experience the state-of-the-art products because these may have high prices.

6. Conclusion and limitations

In conclusion, during the COVID-19 pandemic, consumers highly preferred to buy products online, because most brick-and-mortar stores were closed due to lockdowns and social distancing measures. Additionally, with the rapid development of e-commerce, online shopping has become the most popular shopping style because it allows consumers to not only save time and money, but also review e-WOM before purchasing a product. Moreover, e-WOM is much more reliable compared with traditional WOM. Thus, this study proposed a theoretical framework to explore and define the influencing factors of OCPB based on e-WOM data mining and analyzing. The data were crawled from Jingdong and Taobao, while the data process was also fully demonstrated. Comparing the results, the influencing factors of OCPB were clustered around four categories: perceived emergency context, product, innovation, and function attributes. Moreover, perceived emergency context attribute is the main difference compared with Kotler’s five products level, while perceived product attribute meets the core and generic level, perceived functionality attribute meets the expected and augmented level, and perceived innovation attribute meets the potential level.

However, this study still has certain limitations. First, the data were crawled from Chinese e-commerce websites, hence, they may not be generalized in contexts where the influencing factors and dimensions may vary compared with other countries or regions. Second, this study only explored and defined the antecedents of OCPB. Data should be added from Western e-commerce websites. Moreover, the present study’s results should be compared with Western studies to generate a more comprehensive view of the antecedents of OCPB. Future studies should explore the underlying mechanisms influencing OCPB.

Supporting information

S1 dataset..

https://doi.org/10.1371/journal.pone.0286034.s001

• View Article
• Google Scholar
• PubMed/NCBI
• 25. Krishna A, Strack F. Reflection and impulse as determinants of human behavior. Knowledge and Action,Springer, Cham; 2017. p. 145–67.
• 37. Mikalef P, Pappas IO, Giannakos M. Consumer intentions on social media: a fsQCA analysis of motivations. Conference on e-Business, e-Services and e-Society: Springer, Cham; 2016. p. 371–86.
• 58. Cinar D. The effect of consumer emotions on online purchasing behavior. Tools and Techniques for Implementing International E-Trading Tactics for Competitive Advantage. USA: IGI Global; 2020. p. 221–41.
• 61. Martínez-López F. J. P-G, C., Gázquez-Abad JC, Rodríguez-Ardura I. Online consumption motivations: an integrated theoretical delimitation and refinement based on qualitative analyses. Strategic e-Business Management. Berlin, Heidelberg: Springer; 2014. p. 347–70.
• 80. Kotler P. Principles of marketing. Boston: Pearson; 2016.

Log in using your username and password

• Search More Search for this keyword Advanced search
• Latest content
• For authors
• Browse by collection
• BMJ Journals More You are viewing from: Google Indexer

You are here

• Volume 11, Issue 2
• The pandemic of online research in times of COVID-19
• Article Text
• Article info
• Citation Tools
• Rapid Responses
• Article metrics

• http://orcid.org/0000-0002-3196-8757 Jeroen De Man 1 ,
• Linda Campbell 2 ,
• Hanani Tabana 3 ,
• http://orcid.org/0000-0003-2268-3829 Edwin Wouters 2 , 4
• 1 Department of Family Medicine and Population Health , University of Antwerp , Antwerp , Belgium
• 2 Centre for Population, Family and Health, Department of Sociology , University of Antwerp , Antwerp , Belgium
• 3 School of Public Health, Faculty of Community and Health , University of the Western Cape , Cape town , South Africa
• 4 Centre for Health Systems Research & Development , University of the Free State , Bloemfontein , South Africa
• Correspondence to Dr Jeroen De Man; jeroen.deman{at}uantwerpen.be

The COVID-19 pandemic has led to an explosion of online research using rating scales. While this approach can be useful, two of the major challenges affecting the quality of this type of research include selection bias and the use of non-validated scales. Online research is prone to various forms of selection bias, including self-selection bias, non-response bias or only reaching specific subgroups. The use of rating scales requires contextually validated scales that meet psychometrical properties such as validity, reliability and—for cross-country comparisons—invariance across settings. We discuss options to prevent or tackle these challenges. Researchers, readers, editors and reviewers need to take a critical stance towards research using this type of methodology.

• statistics & research methods
• social medicine
• public health
• mental health

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:  http://creativecommons.org/licenses/by-nc/4.0/ .

https://doi.org/10.1136/bmjopen-2020-043866

Statistics from Altmetric.com

Request permissions.

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Assessment of the biopsychosocial impact of the COVID‐19 pandemic has been identified as a pressing need. 1 For a variety of reasons, including pragmatic ones, the COVID-19 pandemic has pushed many researchers to measure these biopsychosocial consequences using a particular methodological approach: the assessment of latent variables measured with rating scales in online questionnaires. This approach may indeed circumvent barriers in data collection if based on rigorous methods and a clear and appropriate research question. However, while using this approach initially seems straightforward, it has a long-standing track record of challenges and uncertainties, from setting up the research to the analyses and reporting of results. 2 , 3 The pressure to use online surveys in a short space of time may therefore be leading to another pandemic: one of studies with poorly constructed scales and non-representative sample sizes. In this commentary, we highlight possible drawbacks of the use of online surveys and rating scales, and we offer potential solutions for those who endeavour to use them.

Selection bias in online research

Sample validity is an essential requirement in survey research and means that each participant of the study population has the same chance to participate. Potential threats to sample validity are well known when using online studies. 2 First, where and how surveys will be made available will strongly determine their participants. Studies using distribution channels that only reach a subgroup of the target population suffer from selection bias. 4 Differences in health literacy and online access may strongly skew participation, especially in low-income and middle-income countries or in societies with large differences in educational and socioeconomic levels. Second, online surveys will typically attract participants who have a special interest or a close relationship with the topic (ie, self-selection bias). 4 On the other hand, specific subgroups may be less inclined to respond or complete the survey (ie, non-response bias). 4 When only a subgroup is being reached, selection bias will typically grow with an increasing diversity of the target population. This implies that extra caution is needed when studying more general populations.

Even though online surveys may obtain large sample sizes, this does not necessarily compensate for selection bias and may even make it worse. 5 Correction of such bias is often a daunting task, if possible at all. Suggestions to prevent it include the following:

Balance information in the introduction of the survey to sufficiently inform potential participants and to avoid eliciting interest from a particular subgroup. For example, mentioning that the survey relates to COVID-19 may be needed to attract sufficient participants, while introducing the survey as ‘COVID-19 and your mental health status’ may only attract a specific subgroup.

Include a broad array of items to measure sociodemographic and other characteristics that may potentially determine participation. Reporting these characteristics will help authors and readers to appraise sample validity and recognise the study’s limitations. Using the same questions as in large surveys, such as a population census, may allow better assessment of the sample’s representativeness and may allow application of sample weighting. A recent study by De Man et al could serve as an example. 6 This study investigated associations between COVID-19-related stressors and depression in Belgian students attending higher education. When comparing their study sample with governmental data on higher education students enrolled in the previous year, they found a higher proportion of women (±20%), while other sociodemographic characteristics were comparable. This higher proportion of women needs to be taken into account when interpreting the findings of the study, especially since depression is more common among women.

How and to whom the survey is distributed is crucial. Potential distribution channels for online surveys include social media, news outlets, phone messages, email lists and quick response (QR) codes on printed material. Pursuing sample validity requires a tailored approach that facilitates equal participation of all relevant subgroups of the target population. For example, distributing the survey through academic networks will likely result in a very skewed image of the general national population. A recent study on the use of Facebook as a recruitment strategy tested an intervention to improve sample validity: the implementation of male-only advertisement increased the proportion of male participants. 7 However, obtaining a representative sample may sometimes not be possible, and researchers may need to rethink whether launching an online questionnaire is warranted. If researchers choose to go ahead, robust reporting of the study procedures in the methods and discussion section is essential. The Checklist for Reporting Results of Internet E-Surveys (CHERRIES) checklist may serve as a useful guide for this matter. 2

Given the current mushrooming of online initiatives, respondents may feel overloaded by the sheer number of questionnaires they are presented with, thereby reducing their interest. Avoiding lengthy questionnaires can help. Joining forces with other research groups can reduce the number of duplicate initiatives and increase access to different distribution channels. In particular, for surveys on COVID-19, early registration of projects may facilitate such collaboration. A global COVID-19 research registry for public health and social sciences can be found here ( https://converge.colorado.edu/resources/covid-19/public-health-social-sciences-registry ).

Keeping participants informed about the results (eg, through the press and individual base) and presenting them as coproducers of knowledge may encourage participation in future initiatives. In the previously mentioned study by De Man et al , authors may consider giving feedback to their study participants through student associations, fraternities or the university communication.

The use of rating scales

If and how rating scales should be used in the assessment of latent variables has been subject of an ongoing debate. 8 A theoretical discussion on the type of data that rating scales represent (eg, continuous, ordinal and interval) is beyond the focus of this commentary. However, if one decides to use rating scales, current consensus converges on the need to meet specific psychometrical properties, especially if the results are used as unit-weighted composite scores (ie, the summation of scores per participant). In the following paragraph, we highlight some essential properties that are often neglected. For more details and psychometrical estimation methods, we refer to relevant textbooks or online resources such as the ones developed by Revelle. 9

To draw meaningful conclusions, a scale needs to be valid for the studies’ target population.

Validity can refer to a scale’s proven ability to predict a certain outcome (ie, predictive criterion validity) or a scale’s relationship with a well-established measure or gold standard (ie, concurrent criterion validity). Contextual similarity between the actual study and the validation setting is essential and often overlooked.

Content validity corresponds to the scale measuring all facets of a given construct. A common and well-intended practice to reduce the length of the survey is to use a selection of the items of an existing scale. However, this shortening may also reduce construct coverage and consequently affect a scale’s content validity.

For cross-validation purposes, researchers may also consider using scales that have been used in large representative surveys in their study setting (eg, the Demographic and Health Survey and the European Social Survey).

The use of unit-weighted composite scores is often justified by a coefficient alpha estimate above a defined cut-off depending on the purpose and field of research. First of all, it is important to note that a high value of alpha is not a sufficient criterion to use composite scores. Often overlooked is the participants’ conceptual understanding of the items as has previously been noted. Theoretically, items that measure completely different concepts could correlate with each other and result in an acceptable alpha. In addition, testing reliability of a composite score is not straightforward as the validity of alpha depends on rather strict conditions. 10 One such condition is unidimensional data (ie, the scale measures only one concept). A scale that measures various subfactors besides an overarching ‘general’ factor may therefore overestimate coefficient alpha. In addition to alpha, the use of model-based estimators such as total omega and hierarchical omega is now recommended in order to arrive at a more nuanced estimation of the reliability of a scale. 10 In particular, if a scale shows deviations of unidimensionality (ie, the scale does not measure only one concept), reliability assessment becomes complex, and we would not do justice to formulate a simple guideline for its estimation. More details on the recommended procedures to test reliability of a scale can be found in textbooks or online. 9 Finally, it is important to mention that, even though the use of composite scores may be justified based on reliability estimates, techniques such as factor analysis and structural equation modelling are usually preferred when it comes to accuracy of measuring relationships with one or more latent variables.

Comparison over time, across different settings (eg, different countries) and between subgroups, may be of special interest when studying latent constructs in participants exposed to rapidly changing environments because of COVID-19 or related preventive measures (eg, lockdown). While often overlooked, comparison of subgroups, be it through factor analysis or through composite scores, requires measurement equivalence, which can be defined as ‘whether or not, under different conditions of observing and studying phenomena, measurement operations yield measures of the same attribute’. 11 Impaired measurement equivalence precludes a meaningful interpretation of measurement data and can be due to various reasons. 12 For instance, different subgroups may attribute a different meaning to certain words of an item because of a different socioeconomic background or because of a different language use: the interpretation of ‘feeling stressed’ may substantially differ among countries or even between men and women.

Depending on the purpose of the study, measurement equivalence can be tested through different levels of measurement invariance. For instance, if the relationship between different constructs is being studied, equivalence of factor loadings (ie, metric invariance) is required. However, if the purpose is to compare subgroup means of a certain construct, be it through factor analysis or composite scores, additional equivalence of intercepts is required (ie, scalar invariance). Invariance is typically assessed based on a structural equation modelling (SEM) framework, but it can also be tested using an item response theory framework or a combination of both approaches. 13 A recent compilation of recommendations by Putnick et al provides more detail on the SEM framework approach. 13

In conclusion, lack of in-person access to participants and timeliness may have pushed researchers to use online surveys and rating scales in particular. When researchers consider using this approach, they need to balance the added value of their research against the potential drawbacks such as selection bias and the use of non-validated or poorly validated scales. Moreover, state-of-the-art analysis of latent variables often requires tedious and advanced modelling techniques. While using these methods can be particularly useful during the current pandemic, authors, readers and reviewers should take a critical stance towards the results of such studies, even when sample sizes are large.

• O’Connor DB ,
• Aggleton JP ,
• Chakrabarti B , et al
• Eysenbach G
• Hobart JC ,
• Zajicek JP , et al
• Eysenbach G ,
• Kaplan RM ,
• Chambers DA ,
• van de Velde S , et al
• Foreman J ,
• Capasso A , et al
• Zinbarg RE ,
• Revelle W ,
• Yovel I , et al
• Vandenberg RJ ,
• Putnick DL ,
• Bornstein MH

Twitter @JeroenDeMan1

Contributors JDM, LC, HT and EW contributed to the conception of the study. JDM drafted the manuscript. LC, HT and EW critically revised the work and read and approved the final manuscript.

Funding This work was supported by the Faculty of Medicine and Health Sciences of the University of Antwerp, grant number 37025.

Competing interests None declared.

Patient consent for publication Not required.

Provenance and peer review Not commissioned; externally peer reviewed.

Read the full text or download the PDF:

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 28 January 2021

The effect of the definition of ‘pandemic’ on quantitative assessments of infectious disease outbreak risk

• Benjamin J. Singer 1 ,
• Robin N. Thompson 2 , 3 &
• Michael B. Bonsall 1  

Scientific Reports volume  11 , Article number:  2547 ( 2021 ) Cite this article

33k Accesses

19 Citations

200 Altmetric

Metrics details

• Applied mathematics
• Epidemiology

In the early stages of an outbreak, the term ‘pandemic’ can be used to communicate about infectious disease risk, particularly by those who wish to encourage a large-scale public health response. However, the term lacks a widely accepted quantitative definition. We show that, under alternate quantitative definitions of ‘pandemic’, an epidemiological metapopulation model produces different estimates of the probability of a pandemic. Critically, we show that using different definitions alters the projected effects of key parameters—such as inter-regional travel rates, degree of pre-existing immunity, and heterogeneity in transmission rates between regions—on the risk of a pandemic. Our analysis provides a foundation for understanding the scientific importance of precise language when discussing pandemic risk, illustrating how alternative definitions affect the conclusions of modelling studies. This serves to highlight that those working on pandemic preparedness must remain alert to the variability in the use of the term ‘pandemic’, and provide specific quantitative definitions when undertaking one of the types of analysis that we show to be sensitive to the pandemic definition.

Similar content being viewed by others

Interplay of social distancing and border restrictions for pandemics via the epidemic renormalisation group framework

Giacomo Cacciapaglia & Francesco Sannino

Second wave COVID-19 pandemics in Europe: a temporal playbook

Giacomo Cacciapaglia, Corentin Cot & Francesco Sannino

Regional excess mortality during the 2020 COVID-19 pandemic in five European countries

Garyfallos Konstantinoudis, Michela Cameletti, … Marta Blangiardo

Introduction

In the early stages of an infectious disease outbreak, it is important to determine whether the pathogen responsible may go on to cause an epidemic or a pandemic 1 , 2 , 3 , 4 , 5 . There is extensive literature on determining the probability of a major epidemic given a small population of initial infected hosts 6 , 7 , 8 , 9 . This research has produced a natural mathematical definition of an epidemic, based on the bimodal distribution of outbreak sizes given by simple stochastic epidemiological models when \(R_0\) is larger than but not close to one 10 . The term ‘pandemic’ has no corresponding theoretical definition, and there is no consensus mathematical approach to determining the probability of a pandemic. In this study, we examine how alternative definitions of ‘pandemic’ affect quantitative estimates of pandemic risk assessed early in an infectious disease outbreak.

The term ‘pandemic’ is used extensively, appearing in phrases such as ‘pandemic preparedness’ 11 , 12 , 13 , ‘pandemic influenza’ 14 , 15 , 16 , and ‘pandemic potential’ 17 , 18 , 19 . A Google Scholar search returns 25,800 results using the term ‘pandemic’ for 2019 alone.

The International Epidemiology Association’s Dictionary of Epidemiology defines a pandemic as “an epidemic occurring worldwide, or over a very wide area, crossing international boundaries and usually affecting a large number of people” 20 . Notably this definition makes an explicit reference to national borders. Contrastingly, a World Health Organization (WHO) source makes reference to a pandemic as “the worldwide spread of a new disease” 21 .The use of the word ‘new’ here is ambiguous in the context of infectious diseases. HIV/AIDS is often referred to as a global pandemic, but is certainly not new on the timescale of, say, the emergence of influenza strains 22 , 23 . A study by Morens et al. in 2009 finds that there is little in common between all disease outbreaks that have been referred to as pandemics, except that they have a wide geographical extension 24 .

These kinds of differences between pandemic definitions can often go unnoticed, but in certain circumstances they can cause confusion between different stakeholders (e.g. between scientists and governments, or governments and the public), who may not have a shared background understanding of the term. In 2009, the WHO declared a pandemic of H1N1 influenza, using criteria related to the incidence and spread of the virus in different WHO regions 25 . The criteria did not include reference to morbidity or mortality 26 . This fact led to some controversy over whether the declaration of a pandemic was appropriate, as the declaration prompted some governments to mount an intensive response to an outbreak that resulted in fewer yearly deaths than a typical strain of seasonal flu 27 , 28 , 29 , 30 .

International health organisations such as the WHO have not provided any formal definitions of the term ‘pandemic’, and the WHO no longer uses it as an official status of any outbreak 25 , 31 . It would, however, be hasty to dismiss the importance of the term on these grounds. Although the WHO no longer uses the term ‘pandemic’ officially, the WHO Director-General drew attention to their use of the term as recently as March 2020, to describe the status of the COVID-19 outbreak 32 . The Director-General cited “alarming levels of inaction” as one of the reasons to use the term, along with the caveat that “describing the situation as a pandemic does not change WHO’s assessment of the threat posed by this virus”. The WHO’s use of the term was of interest to the public, receiving extensive press coverage 33 , 34 , 35 . The term ‘pandemic’ clearly continues to be important to indicate serious risk during disease outbreaks.

Regardless of the extent to which the pandemic definitions currently in use do or do not agree, they are all qualitative in nature, using descriptions such as “very wide area” and “large number of people”. Perhaps as a result of this, many quantitative studies on pandemics do not make use of a quantitative definition of a pandemic, but instead focus on causally related concepts, such as sustained transmission 19 , or emergence of novel viruses 36 . Others treat the spread of a pathogen at a pandemic level as a context in which to study transmission dynamics, without paying special attention to how those dynamics might lead to a pandemic as distinct from an epidemic or a more limited outbreak 37 , 38 , 39 . In this paper, we examine the effects of alternative pandemic definitions on the analysis of key epidemiological questions. The results provide a foundation for deciding the appropriate quantitative definition of ‘pandemic’ in a given context.

We use a metapopulation model to investigate the effects of pandemic definition on the results of a quantitative assessment of the probability of a pandemic. Metapopulation models are commonly applied to pathogens that spread between regions of the world, and so are appropriate for modelling pandemics 40 , 41 , 42 , 43 , 44 , 45 . We represent states of our metapopulation model as states of a Markov chain, allowing us to calculate the probability of a pandemic directly, as opposed to simulating many stochastic outbreaks and recording the proportion which result in pandemics. We explore two different kinds of pandemic definition, following Morens et al. 2009 24 , specifically:

the family of transregional definitions, where a pandemic is defined as an outbreak in which the number of regions experiencing epidemics meets or exceeds some threshold number n . We refer to specific transregional definitions as n -region transregional definitions, e.g. a 3-region transregional definition.

the interregional definition, where a pandemic is defined as an outbreak in which two or more non-adjacent regions experience epidemics.

Note that these definitions require a specific sense of ‘region’. These regions could be countries, or they could be larger or smaller than individual countries—from counties to health zones to WHO regions. Our metapopulation model (detailed in the Methods section below) can be used to model regions of any size. We have chosen not to include definitions with criteria relating to the number of people infected or killed, instead of, or in addition to, geographical extension. Extension is the only universal factor in pandemic definitions, and so is the focus of the current study 24 .

Three questions that help form public health policy at the beginning of an outbreak are:

Would interventions restricting travel reduce the risk of a pandemic?

Does a portion of the population have pre-existing immunity, and does this affect the risk of a pandemic?

How is the risk of a pandemic affected by regional differences in transmission?

Using our metapopulation model, we explore how changing the pandemic definition does or does not affect our answers to these questions. We show that the precise definition of a pandemic used in modelling studies can (but does not always) affect the inferred risk. The predicted effects of travel restrictions, the influence of pre-existing immunity, and the impact of regional differences in transmission can all vary when alternative definitions of ‘pandemic’ are used. This demonstrates clearly the need to consider carefully the pandemic definition used to assess the risk from an invading pathogen. This is necessary for clear communication of public health risk.

Travel rates

One important question about pandemic risk is what effect inter-regional travel rates have on the probability of a pandemic occurring 16 , 17 , 46 , 47 . Here we model epidemics occurring in regions connected on a network in which the connections and their weighting can be set at fixed values representing the rates of travel between regions. We consider simple networks that can illustrate the effects of our different pandemic definitions—namely, the star network, in which one central region is connected to all others with equal weighting and the non-central regions lack any other connections, and the fully connected network, in which each region is connected to every other with equal weighting. Figure  1 illustrates that the connectivity of the full network is much higher than that of the star network. Using the star network allows us to make the distinction between adjacent and non-adjacent regions, thus allowing us to distinguish between transregional and interregional pandemic definitions.

Illustrations of ( a ) a star network and ( b ) a full network, each with ten regions. Circles represent regions, and straight lines represent travel routes between regions.

Unless otherwise stated, all figures in the current study are generated with a transmission rate of \(\beta = 0.28\) per day, a recovery rate of \(\mu = 0.14\) per day, and an inter-regional travel rate of \(2\times 10^{-4}\) per day. This corresponds to a within-region basic reproduction number ( \(R_0\) ) of 2. These values are within the plausible range for both seasonal and pandemic influenza, and as such they can be used to simulate a plausible pathogen of pandemic potential 38 . We further assume an initial population of 1000 susceptible individuals in each region, and that the outbreak is seeded by a single infectious individual in one region. In the full network, all regions are equivalent, so we seed the outbreak in a single arbitrary region. In the star network, we take the average probability of a pandemic over outbreaks seeded in each region.

Using a model with ten regions allows us to test a range of different transregional definitions of a pandemic. The pandemic probability under a range of n -region transregional definitions for a 10-region network with a variety of travel rates is shown in Fig.  2 . An n -region transregional definition effectively provides a threshold number n —if more than n regions experience epidemics, the outbreak is counted as a pandemic, and otherwise it is not. Thus we indicate the different possible n -region definitions through their threshold numbers in Figs.  2 , 5 , and  6 .

Pandemic probability for a range of between-region travel rates and a range of transregional pandemic definitions. The “pandemic threshold number” refers to the minimum number of regions that must experience epidemics before a pandemic is declared. The pandemic probability is, in general, sensitive to the pandemic definition used, but the degree of sensitivity depends on network structure and travel rates. ( a ) Pandemic probability for a star network. The pandemic probability is, in general, highly sensitive to the pandemic definition used. ( b ) Pandemic probability for a fully connected network. The sensitivity of the pandemic probability to the pandemic definition used is significantly reduced at high travel rates.

The 1-region transregional definition merges the definitions of ‘pandemic’ and ‘epidemic’ in an implausible way, but it is included in these figures for comparison. The comparison between the pandemic probability according to the 2-region definition and according to the 10-region definition shows the difference between pandemic definitions that are satisfied by any transregional transmission and definitions that are satisfied only by truly global spread. For the star network, or for the fully connected network with low travel rates, there is a marked difference between the probability of either of these definitions being satisfied. However, for the fully connected network at medium or high rates of travel, if the pathogen invades the initial region successfully, then it will go on to spread globally. As such, the probability of a pandemic is nears the maximum of 0.5 (i.e. \(1-1/R_0\) ) at all thresholds. For any definition, the probability of a pandemic increases with the connectivity of the network, and with travel rates across the network.

We can also explore the difference in pandemic probability between the transregional and interregional definitions, which make use of a distinction between adjacent and non-adjacent regions. This is shown for a 10-region star network in Fig.  3 a, in which we consider the 2-region transregional and 2-region interregional definitions. We choose a star network as it is one of the simplest network types in which there are adjacent and non-adjacent regions. There is a difference between the 2-region interregional and transregional definitions, but the difference is much smaller than that between the 2-region interregional and 10-region (global) definition, and reduces as travel rates increase. In the case of a fully connected network, all regions are essentially adjacent to each other, so we compare only the 2-region transregional and global definitions. We find that the definitions are clearly distinct for low travel rates, but as the travel rate increases the difference between the likelihood of a pathogen causing an epidemic in one region and the likelihood of it causing epidemics in all regions disappears. This is due to the fact that the pathogen can be introduced into any population from any other.

Plots of pandemic probability against between-region travel rate for a range of pandemic definitions. The difference in probability for different pandemic definitions changes as travel rates increase. ( a ) Plot of pandemic probability for a star network. ( b ) Plot of pandemic probability for a full network. For a fully connected network all regions are adjacent, so no line is shown for the interregional definition, which requires non-adjacent regions to experience epidemics.

In this section we have shown that, when a pandemic is defined in terms of which regions experience epidemics of a disease, different definitions can produce very different estimates of the pandemic probability at low connectivity or travel rates, but have a much smaller effect at high connectivity and travel rates. In the supplementary information, we illustrate that effects due to network structure are mostly due to the difference in motility between the full network and the star network, although topology still plays an important role.

Cross-immunity

Some pathogens with pandemic potential have a prior history of infecting humans, such as pandemic influenza. Newly emerged pathogens with no history of infecting humans are less likely than these established pathogens to encounter regions where susceptible individuals have partial immunity to infection. Established pathogens may encounter individuals with partial immunity acquired from infections with previously circulating strains—i.e. cross-immunity 48 , 49 . It can be important in responding to an outbreak to consider whether any individuals might have existing immunity. We can therefore investigate the interaction between immunity generated by prior exposure and pandemic definition by examining how cross-immunity affects our calculation of the pandemic probability on a network.

We modelled the spread of a pathogen over a ten-region network with no cross-immunity initially, where the initial infected individual could originate in any region. We only included cases where at least one region experienced an epidemic of this initial pathogen. To simulate the emergence of a strain with higher pandemic potential, we then introduced a second pathogen with a higher transmission rate of \(\beta = 0.42\) (corresponding to a basic reproduction number of 3), to which infection with the initial pathogen conferred some degree of partial immunity to infection. The strength of this immunity is written as \(\alpha\) . See the Methods section for details of how cross-immunity is incorporated into our modelling framework. We defined a pandemic as occurring when all ten regions experienced epidemics of the second pathogen, and repeated the model for two values of the level of cross-immunity at a variety of between-region travel rates. The results are presented in Fig.  4 .

Plots of pandemic probability against travel rate for high and low levels of cross-immunity ( \(\alpha\) ) on ten-region networks. A pandemic is defined here as all ten regions experiencing epidemics, i.e. the 10-region transregional definition. The plots show a large relative difference both in the probability of pandemics and in how that probability scales with travel rates for different levels of cross-immunity. The initial infected individual for each outbreak originates in a randomly chosen region. ( a ) Plot of pandemic probability for a star network. ( b ) Plot of pandemic probability for a full network.

First, increasing cross-immunity decreases the probability of a pandemic. Second, the presence of cross-immunity changes how pandemic probability scales with travel rates. In general, the pandemic probability increases faster with travel when the level of cross-immunity is low, except when it reaches a point of saturation as in Fig.  4 b.

Figure  5 shows the simultaneous effects of different n -region transregional pandemic definitions and the degree of cross-immunity in determining the pandemic probability. Here we fix the travel rate at \(2.0 \times 10^{-4}\) per day. In the full network there is a distinct transition from higher risk to lower risk, as cross-immunity approaches one. However, in the star network there is, on average, less circulation of the initial pathogen, so the effect of cross-immunity is less dramatic. Increased cross-immunity can also increase the difference in risk for different pandemic definitions—for the fully connected network, when cross-immunity exceeds \(\alpha = 0.5\) , differences in probability between different thresholds become visible that are much smaller at lower values. This suggests that the probability that an outbreak will develop into a pandemic may be more sensitive to the exact pandemic definition for outbreaks of pathogens that encounter pre-existing immunity than for pathogens which encounter only fully susceptible populations. However, this effect is not seen for the star network, in which the low connectivity of the network results in larger differences in probability between different thresholds even at low levels of cross-immunity.

Pandemic probability for various levels of cross-immunity ( \(\alpha\) ) and a range of transregional pandemic definitions, on a ten-region network. ( a ) Pandemic probability for a star network. ( b ) Pandemic probability for a fully connected network. Here the sensitivity of the pandemic probability to the pandemic definition used increases with cross-immunity, until the probability of any epidemic becomes very low.

Heterogeneous transmission

A topic of great concern during a pandemic is heterogeneity in risk between different countries or regions 50 , 51 . Cross-immunity can create one kind of heterogeneity, since it is common for previous exposure to a pathogen to differ between regions 52 . Another kind of heterogeneity is that due to different public health interventions. Here we ignore cross-immunity and instead examine a heterogeneous fully connected network of ten regions, five of which have a higher rate of transmission of the pathogen than the other five. This can be thought of as an approximation to the difference between poor regions with a relative lack of public health interventions, and wealthy regions with well-funded public health organisations and increased access to healthcare.

The level of heterogeneity was defined as the ratio of the transmission rate in the higher-transmission regions to the transmission rate in the lower-transmission regions. The average transmission rate across all regions was kept fixed at \({\bar{\beta }} = 0.28\) per day, corresponding to a basic reproduction number of 2. The simultaneous effects of heterogeneity and the pandemic definition in determining the pandemic probability are illustrated in Fig.  6 .

Pandemic probability for various degrees of heterogeneity of transmission rates and a range of transregional pandemic definitions, on a fully connected ten-region network where five regions are classed as higher-transmission and the other five regions are classed as lower-transmission. Note that the colour scales differ between the two plots, in order to make the variation in plot ( a ) clearer. ( a ) Pandemic probability for a pathogen emerging in a higher-transmission region. For low thresholds heterogeneity increases the pandemic probability, but at the 10-region threshold the pandemic probability grows and then decreases with increasing heterogeneity. ( b ) Pandemic probability for a pathogen emerging in a lower-transmission region. At all thresholds increasing heterogeneity decreases the pandemic probability.

The row for the 1-region definition shows how the risk of any outbreak varies with the changing basic reproduction number of the pathogen in the region in which it emerges. More complex effects can be seen for higher n -region definitions, especially the 10-region definition, where, at high levels of heterogeneity, even pathogens emerging in higher-transmission regions are prevented from spreading globally due to the low chance of epidemics in lower-transmission regions. Thus the probability of a pandemic under a 10-region definition increases and then decreases with increasing heterogeneity. In the supplementary information, we show that this increasing-decreasing effect exists in networks of different sizes and structures. It appears at different thresholds in different networks. No corresponding effect exists for a pathogen emerging in a lower-transmission region, where increasing heterogeneity always decreases the chance of a pandemic, however it is defined.

In this study, we have developed a theoretical framework to estimate the probability of a pandemic, as detailed in the Methods section below. We use a Markov chain technique based on SIR dynamics to model the spread of a pathogen. The results of this modelling framework reveal in which situations the definition of ‘pandemic’ has a strong effect on the calculated pandemic risk and in which situations it does not. The models also illustrate the effects of differing epidemiological parameters on the pandemic risk under different definitions, and how these effects interact with each other.

Returning to the three epidemiological questions introduced in the introduction, we can see that our results show how the answers can depend on our definition of a pandemic, and on key population and pathogen parameters. The first question was “Would interventions restricting travel reduce the risk of a pandemic?” In Fig.  2 , we see that reductions in travel rates always reduce risk in a network with low connectivity, where travel occurs mainly through a central hub. However, in a highly connected network with high travel rates, travel would have to be extremely highly suppressed to change the probability of a pandemic substantially, under most definitions. This accords with previous findings regarding the effectiveness of restricting travel 53 . Additionally, in the highly connected network, changing the definition of a pandemic makes little difference to the pandemic probability, for high enough values of the travel rate.

Figure  3 further illustrates the effects of different definitions. Changing the pandemic definition can sometimes greatly alter the estimated probability of a pandemic, as seen in Fig.  3 a between the yellow line, representing the 2-region transregional definition, and the purple line, representing the 10-region transregional definition. The effect on the pandemic risk of reducing travel rates also differs substantially between these two definitions. However, there are situations where changing the definition does not significantly change the pandemic probability, as seen in the same figure between the yellow line and the dashed green line, representing the 2-region interregional definition. Both the estimated risk and the effect of reducing travel are very similar in these two cases. So, while some changes in definition do not cause a large change in quantitative analyses of the risk of a pandemic, others may significantly alter both our point estimates and the predicted effects of key parameters. Figure  3 b shows that this may depend on the values of those key parameters themselves. For low travel rates, the pandemic probability is very different for the two illustrated definitions, but at high travel rates the pandemic probabilities for the two definitions converge.

The second question was “Does a portion of the population have pre-existing immunity, and does this affect the risk of a pandemic?” The presence of immunity can significantly alter the results discussed in the paragraphs above. In Fig.  5 b, the leftmost column is equivalent to the column from Fig.  2 b in which \(\lambda = 2.0 \times 10^{-4}\) per day, but with a higher transmission rate of \(\beta =0.42\) . However, as cross-immunity increases, a marked difference in the pandemic probability between different definitions becomes visible. This shows that the conclusion that precise pandemic definitions are of reduced importance in a highly connected network with high travel rates is context sensitive—if the population has high immunity, differences between definitions re-emerge.

The third question was “How is the risk of a pandemic affected by differences between regions?” In Fig.  6 , we examined how heterogeneous transmission rates in different regions affect the pandemic probability. Many pathogens have higher transmission rates in lower income countries, and novel pathogens are more likely to emerge in low income countries 50 , 51 , 54 . Putting these two facts together, we see that pathogens are most likely to emerge in countries in which they have higher transmission rates. Motivated by this, we compared the scenarios of emergence in a higher-transmission and lower-transmission region, finding that pandemic definition makes a larger difference for diseases emerging in a higher-transmission region. In particular, when the pandemic definition requires many countries to experience epidemics to qualify an outbreak as a pandemic, including countries with lower transmission rates, we see striking non-linearity in the relationship between heterogeneity and the pandemic probability. For these definitions, as the difference in transmission rates between higher- and lower-transmission regions increases, the pandemic probability increases initially, before decreasing. This initial rise is due to the enhanced spread between high-transmission regions increasing the importation rate to low-transmission regions. This result implies that, when the mean value of the transmission rate is fixed, a small gap in the effectiveness of public health infrastructure between wealthy and poor regions puts all regions at greater risk, while a larger gap protects wealthier regions while the risk for poor regions continues to increase.

To illustrate this concept, consider the contrasting examples of Ebola and COVID-19. The 2014 outbreak of Ebola virus followed the pattern of high incidence in low income countries but low incidence in high income countries. The virus spread through several low-income African countries but was effectively contained when introduced to high-income countries 55 , 56 , 57 . In this case, high-income countries had the capacity to prevent a pandemic from taking hold, being able to quickly isolate and treat symptomatic individuals. This generated high heterogeneity in transmission, corresponding to the right side of Fig.  6 a, with low-income countries at high risk and high-income countries at low risk. In contrast, high-income countries have not been able to escape the pandemic of COVID-19, in part due to asymptomatic and presymptomatic transmission of SARS-CoV-2 allowing it to evade surveillance and public health measures 58 , 59 . This has led to more similar transmission rates between different countries, corresponding to the left side of Fig.  6 a, where risk is more uniform between regions and therefore between pandemic definitions.

In our analyses, we use a metapopulation modelling framework. Metapopulation models are widely used in pandemic modelling 40 , 41 , 42 , 43 , 44 , 45 . Our novel Markov chain approach allows us to calculate pandemic probabilities directly, without requiring large numbers of simulations to generate an approximation. We expect our overall conclusion, that the effects of key parameters on pandemic risk depend on the pandemic definition, to hold irrespective of the underlying modelling framework. Future studies could replicate our analyses using different models and modelling approaches, such as metapopulation models with additional epidemiological complexity 43 , 45 , 60 , 61 or the widely used global epidemic and mobility (GLEaM) model 62 , 63 , 64 . Exploring how our quantitative results vary for different modelling frameworks in the field of mathematical epidemiology 14 , 16 , 65 , 66 , 67 is a target for further investigation.

Other future work using our modelling framework could address the role of pandemic definitions in quantifying the effects of additional epidemiological parameters on pandemic risk, such as use of different types of travel (e.g. within-country transport or international flights) 45 , 68 , 69 , the rate of nosocomial infections 70 , or age structure 71 . Our metapopulation modelling framework is generally applicable, and this framework could be extended to represent outbreaks of many different specific pathogens emerging in various locations. An important factor for response planning is the timescale over which outbreaks develop into pandemics. The duration of the initial phase of outbreaks has been a subject of previous study 72 , as has the overall duration of outbreaks 10 , 73 , 74 , 75 , 76 . In theory, Markov chain models could be used to assess the time for a local epidemic to develop into a pandemic, and we leave this as an avenue for further work.

In summary, we have developed a novel modelling framework for estimating the pandemic risk. We have applied this framework to assess the pandemic risk in a range of different scenarios, and have interpreted the results under a variety of pandemic definitions. We have found that certain relationships, such as the effect of heterogeneity in transmission between regions on the risk of a pandemic, are highly dependent on the definition of ‘pandemic’ used, while others, such as the effect of high travel rates on pandemic risk in a highly connected network, are not. This work provides a foundation for improved communication about pandemic risk, by highlighting the contexts in which pandemic definitions need to be provided in quantitative detail. In general, we contend that, when assessing the risk that an outbreak will develop into a pandemic, the precise pandemic definition used for a given analysis should be considered and stated clearly. Future work could investigate the effects of alternative definitions in more detailed epidemiological models, and extend this framework to investigate different dynamical features of pandemics.

We have combined standard epidemiological modelling techniques with a novel Markov chain treatment of metapopulation dynamics to produce a method for calculating the probabilities of epidemics and pandemics in a network of population regions. At each step of this chain, we resolve information about which regions may experience epidemics. The order in which the status of any given region is resolved does not necessarily match the order in which the given epidemics occur in calendar time. A benefit of our model is that we can calculate the probabilities of different final outcomes directly, without requiring large numbers of stochastic simulations to estimate these values. This comes at the cost that temporal information is not represented explicitly in our model: we focus on the pandemic probability, accounting for all possible ways that a pandemic could occur, rather than estimating the possible times at which epidemics could occur in different regions or the timescale over which an outbreak will develop into a pandemic (see Discussion).

We model the transmission of a pathogen through n regions labelled \(P_1, P_2, P_3, \ldots , P_n\) . Each region \(P_j\) has associated with it some intra-region pathogen transmissibility \(\beta _j\) , disease recovery rate \(\mu _j\) , and population size \(N_j\) . From these quantities it is possible to calculate a region-specific basic reproduction number \(R_{0,j}\) . This can be fixed across all regions for a particular pathogen, or allowed to vary from region to region to reflect local epidemiological differences.

First let us consider the spread of the pathogen in a single region, using well-established results of stochastic Susceptible-Infected-Recovered (SIR) models. If a region \(P_j\) contains an initial number of infected individuals \(I_j(0)\) , then in the stochastic SIR model, the probability that these individuals do not cause an epidemic in \(P_j\) is \((1/R_{0,j})^{I_j(0)}\) when \(R_{0,j}\ge 1\) , and 1 otherwise 17 . We also define the final size of an epidemic \(R_{j}(\infty )\) (not to be confused with \(R_{0,j}\) ) as the number of recovered individuals in \(P_j\) at the end of the epidemic. This equals the total number of individuals in \(P_j\) who become infected at any time, and is given by the solution of the following equation 77 .

Infected individuals are assumed to travel from region \(P_j\) to region \(P_m\) at a rate \(\lambda _{jm}\) . We seek the probability that infected individuals travelling from \(P_j\) will not cause an epidemic in \(P_m\) , in the case where initially infected individuals in \(P_m\) do not cause an epidemic in \(P_m\) (including the case where there are no initially infected individuals in \(P_m\) ). This is equal to the probability that i infected individuals migrate from \(P_j\) to \(P_m\) , multiplied by the probability that this number of individuals fails to cause a major epidemic, summed over possible values of i . The minimum value of i is the case where no infected individuals migrate, and the maximum value is the case where all individuals in \(P_j\) that become infected at any point migrate. This gives us an expression for \(q_{jm}\) , the conditional probability that, if \(P_j\) experiences an epidemic and \(P_m\) does not experience an epidemic due to a source of infected individuals other than \(P_j\) , \(P_m\) does not experience an epidemic.

This approximation is valid when the number of infected individuals that travel between regions is much smaller than the size of the regions.

We assume that infected individuals travelling from a region \(P_j\) cannot cause an epidemic in a neighbouring region \(P_m\) if \(P_j\) does not itself experience an epidemic. Then computing the value of \(q_{jm}\) for every pair of populations \(P_j\) and \(P_m\) gives us sufficient information to determine the probability of any particular set of regions connected on a network experiencing epidemics so long as there are no interactions between different groups of migrants arriving in a region, and the total numbers of migrants in any region remains very small relative to the region’s size. If these assumptions hold, we can imagine the regions on a network with weighted directed edges, where the weight of the edge directed from region \(P_j\) to region \(P_m\) is \(q_{jm}\) .

To determine how the final probabilities of epidemics depend on the pairwise probabilities \(q_{jm}\) , we use a Markov chain. The states of this Markov chain assign one of three states to each region— N (for neutral), where it is not yet determined whether the region will experience an epidemic, E (for epidemic), where it is determined that the region will experience an epidemic but it is not yet determined in which further regions it will cause epidemics, and T (for terminal), where it is determined that the region will experience an epidemic and in which further regions it will cause epidemics due to onward transmission. As our model does not explicitly represent dynamical processes occurring over time, these states should not be interpreted as actual states of infection and recovery within regions, but rather as bookkeeping devices for the role of various regions in determining the spread of the pathogen through the network.

Suppose we have a network connecting n regions. In the initial state, each region where the initially infected individuals have caused an epidemic is in state E , and all the other regions are in state N . The global state of the network is simply the product of the states of each system. We can then define a transition matrix \(\mathbf{T }\) that acts on the global state. The elements of this matrix are denoted \(t_{x_1x_2\ldots x_n \rightarrow y_1y_2\ldots y_n}\) .

\(x_j\) is the state ( N , E , or T ) of region \(P_j\) before the transition, and \(y_j\) is its state afterwards. The expression inside the first set of square brackets ensures that the only acceptable transitions for any given region are \(N \rightarrow E\) and \(E \rightarrow T\) , and requires that all epidemic regions in the initial state must be terminated in the transition (this prevents double-counting of possible transmission paths). The expression inside the second set of square brackets gives the probability of each \(N \rightarrow E\) transition, and the expression inside the final set of square brackets gives the probability of each \(N \rightarrow N\) transition, given the regions that are in state E before the transition.

Note that these transitions do not represent a dynamical process—the order of transitions in this model does not necessarily correspond to the order in which regions experience epidemics. Instead, the transitions are simply stages along the exploration of different routes and outcomes from the disease spreading process.

The initial probability of each global state \(z_1z_2\ldots z_n\) (where \(z_i \in \{N, E, T\}\) ) is given by:

where \(Q_j = \min ((1/R_{0,j})^{I_j(0)},1)\) is the probability that the initial population of infective individuals does not cause an epidemic in region \(P_j\) . Essentially, no region can begin in state T , and the probability of each initial global state is given by the product of the probabilities of each region being in the corresponding initial regional state.

In this system, all states in which no region is epidemic are absorbing, and in each transition at least one epidemic state must become terminal. This means that the system must reach an absorbing state in at most n transitions, since at least one region becomes terminal in each transition, and a fully terminal state is absorbing. So the final probability vector \(p_{\mathrm {final}}\) is given by

with \({\mathbf{T}}\) as the transition matrix and \(p_{\mathrm {initial}}\) as the vector whose elements defined by Eq. ( 5 ). This final vector gives the probabilities of each configuration of the metapopulation, with populations in state N never experiencing an epidemic, and regions in state T experiencing an epidemic at some point.

Cross-Immunity

The model described above can incorporate certain epidemiological details, such as heterogeneity of population parameters, but is restricted to treating quite simple disease dynamics. In this section we expand the model to treat pathogens that give those who overcome infection cross-protection against future strains of that pathogen. This is necessary to be able to investigate how pre-existing immunity changes how pandemic definitions affect the results of our model.

We first describe the spread of a pathogen strain X using the methods above, introducing a superscript X to the relevant parameters to mark the strain, e.g. \(R_0^X\) , \(R^X(\infty )\) , and \(p^X_{\mathrm {final}}\) . We assume that infection with pathogen X confers cross-immunity \(\alpha\) to a second strain of the pathogen, which we call Y . In each population \(P_j\) we can define an effective basic reproductive number for Y in the case that \(P_j\) has experienced an epidemic of X , which we call \(R^Y_{e,j}\) .

This expression simply multiplies the basic reproductive number by the effective number of susceptible individuals given the prevalence of cross-immunity in the population. It is through this expression that cross-immunity enters the model—the parameter \(\alpha\) does not otherwise appear in what follows.

We can write down an equation for the expected total number of individuals in \(P_j\) infected in an epidemic of Y in analogy to Eq. ( 1 ). In the case where there has been no previous epidemic of X in \(P_j\) , the expected epidemic size is the solution \(R_{j,\mathrm{no}X}^Y(\infty )\) of

In the case where there has been a previous epidemic of X in \(P_j\) , the expected epidemic size is the solution \(R_{j,X}^Y(\infty )\) of

We assume that individuals infected with Y travel at the same rate as individuals infected with X . We then define the pairwise probabilities of transmission of Y between populations in analogy to Eq. ( 2 ). That is,

where \(R^Y_{c,m} = R^Y_{0,m}\) when \(P_m\) has not experienced a previous epidemic of X , \(R^Y_{c,m} = R^Y_{e,m}\) when \(P_m\) has experienced a previous epidemic of X , \(R^Y_{j,b}(\infty ) = R^Y_{j,\mathrm{no}X}(\infty )\) when \(P_j\) has not experienced a previous epidemic of X , and \(R^Y_{j,b}(\infty ) = R^Y_{j,X}(\infty )\) when \(P_j\) has experienced a previous epidemic of X .

These expressions for \(q^Y_{jm}\) can be substituted for \(q_{jm}\) in Eq. ( 3 ) to yield a transition matrix for modelling the spread of Y , which we will call \({\mathbf{T}}^Y(s_1s_2\ldots s_n)\) , where \(s_j\) is the final state (either N or T ) of the X outbreak in \(P_j\) . We find the initial probabilities of each state with regards to Y , \(p^Y_{\mathrm {initial}}\) , in analogy to Eq. ( 5 ), given an initial number of individuals infected with Y in each population \(I^Y_j(0)\) .

where \(Q^Y_j = \min [(1/R^Y_{0,j})^{I^Y_j(0)},1]\) when \(P_j\) has not experienced a previous epidemic of X (i.e. \(s_j=N\) ), and \(Q^Y_j = \min [(1/R^Y_{e,j})^{I^Y_j(0)},1]\) when \(P_j\) has experienced a previous epidemic of X (i.e \(s_j = T\) ). We can then write the final probabilities of each combination of possible epidemics of Y , for a given set of previous epidemics of X , as

To find the overall probability of each combination of epidemics of Y in various populations given a prior probability of each combination of epidemics of X (given by \(p^X_{\mathrm {final}}(s_1s_2\ldots s_n)\) defined in Eq. ( 6 )), we sum over the possible values of \((s_1s_2\ldots s_n)\) , weighted by their probability.

Code availability

Code is available on the Open Science Framework at https://osf.io/z52te/ .

Craft, M. E., Beyer, H. L. & Haydon, D. T. Estimating the probability of a major outbreak from the timing of early cases: an indeterminate problem?. PLoS ONE 8 , 1–7. https://doi.org/10.1371/journal.pone.0057878 (2013).

Article   CAS   Google Scholar  

Thompson, R. N., Gilligan, C. A. & Cunniffe, N. J. Detecting presymptomatic infection is necessary to forecast major epidemics in the earliest stages of infectious disease outbreaks. PLoS Comput. Biol. 12 , 1–18. https://doi.org/10.1371/journal.pcbi.1004836 (2016).

Jamison, D. T. et al. (eds.) Disease Control Priorities: Improving Health and Reducing Poverty , vol. 9 (World Bank, Washington (DC), 2018), 3rd edn. https://doi.org/10.1016/S0140-6736(15)60097-6 .

Adalja, A. A., Watson, M., Toner, E. S., Cicero, A. & Inglesby, T. V. Characteristics of microbes most likely to cause pandemics and global catastrophes. Curr. Top. Microbiol. Immunol. https://doi.org/10.1007/82_2019_176 (2019).

Article   PubMed   PubMed Central   Google Scholar  

Thompson, R. N. Novel coronavirus outbreak in Wuhan, China, 2020: intense surveillance is vital for preventing sustained transmission in new locations. J. Clin. Med. 9 , 498. https://doi.org/10.3390/jcm9020498 (2020).

Article   PubMed Central   Google Scholar  

Tarwater, P. M. & Martin, C. F. Effects of population density on the spread of disease. Complexity 6 , 29–36. https://doi.org/10.1002/cplx.10003 (2001).

Article   MathSciNet   Google Scholar  

Miller, J. C. Spread of infectious disease through clustered populations. J. R. Soc. Interface 6 , 1121–34 (2009).

Article   Google Scholar  

House, T., Ross, J. V. & Sirl, D. How big is an outbreak likely to be? Methods for epidemic final-size calculation. Proc. R. Soc. A Math. Phys. Eng. Sci. 469 , 20120436–20120436 (2012).

Ball, F., Sirl, D. & Trapman, P. Threshold behaviour and final outcome of an epidemic on a random network with household structure. Adv. Appl. Probability 41 , 765–796 (2009).

Thompson, R., Gilligan, C. & Cunniffe, N. Will an outbreak exceed available resources for control? Estimating the risk from invading pathogens using practical definitions of a severe epidemic. J. R. Soc. Interface https://doi.org/10.1098/rsif.2020.0690 (2020).

Kyoon-achan, G. & Wright, L. Community-based pandemic preparedness: COVID-19 procedures of a Manitoba First Nation community. J. Community Saf. Well-Being 5 , 45–50 (2020).

Marston, H. D., Paules, C. I. & Fauci, A. S. The critical role of biomedical research in pandemic preparedness. JAMA 318 , 1757 (2017).

Monto, A. S., Comanor, L., Shay, D. K. & Thompson, W. W. Epidemiology of pandemic influenza: use of surveillance and modeling for pandemic preparedness. J. Infect. Dis. 194 , 92–97 (2006).

Lunelli, A., Pugliese, A. & Rizzo, C. Epidemic patch models applied to pandemic influenza: contact matrix, stochasticity, robustness of predictions. Math. Biosci. 220 , 24–33. https://doi.org/10.1016/j.mbs.2009.03.008 (2009).

Article   MathSciNet   PubMed   MATH   Google Scholar  

Chowell, G., Nishiura, H. & Bettencourt, L. M. Comparative estimation of the reproduction number for pandemic influenza from daily case notification data. J. R. Soc. Interface 4 , 154–166. https://doi.org/10.1098/rsif.2006.0161 (2007).

Colizza, V., Barrat, A., Barthelemy, M., Valleron, A. J. & Vespignani, A. Modeling the worldwide spread of pandemic influenza: baseline case and containment interventions. PLoS Med. 4 , 95–110. https://doi.org/10.1371/journal.pmed.0040013 (2007).

Thompson, R. N., Thompson, C. P., Pelerman, O., Gupta, S. & Obolski, U. Increased frequency of travel in the presence of cross-immunity may act to decrease the chance of a global pandemic. Philos. Trans. R. Soc. B Biol. Sci. https://doi.org/10.1098/rstb.2018.0274 (2019).

Thompson, R. Pandemic potential of 2019-nCoV. Lancet Infect. Dis. 3099 , 30068. https://doi.org/10.1016/S1473-3099(20)30068-2 (2020).

Breban, R., Riou, J. & Fontanet, A. Interhuman transmissibility of Middle East respiratory syndrome coronavirus: estimation of pandemic risk. The Lancet 382 , 694–699. https://doi.org/10.1016/S0140-6736(13)61492-0 (2013).

Porta, M. A Dictionary of Epidemiology (Oxford University Press, USA, 2008). https://www.oxfordreference.com/view/10.1093/acref/9780195314496.001.0001/acref-9780195314496 .

World Health Organization. What is a pandemic? (2010). https://www.who.int/csr/disease/swineflu/frequently_asked_questions/pandemic/en/ .

Osman, A. S. HIV / AIDS in the last 10 years. Eastern Mediterranean Health Journal 14 , 90–96 (2007). http://www.emro.who.int/emhj-volume-14-2008/volume-14-supplement/hivaids-in-the-last-10-years.html .

Cohen, M. S. et al. The spread, treatment, and prevention of HIV-1: evolution of a global pandemic. J. Clin. Investig. 118 , 1244–1254. https://doi.org/10.1172/JCI34706.1244 (2008).

Article   CAS   PubMed   Google Scholar  

Morens, D., Folkers, G. & Fauci, A. S. What is a pandemic?. JAMA J. Am. Med. Assoc. 321 , 910. https://doi.org/10.1001/jama.2019.0700 (2009).

Doshi, P. The elusive definition of pandemic influenza. Bull. World Health Organ. 89 , 532–538. https://doi.org/10.2471/BLT.11.086173 (2011).

Coninx, J. K. et al. (eds.) Pandemic influenza preparedness and response: a WHO guidance document (WHO Press, 2009). https://www.who.int/influenza/resources/documents/pandemic_guidance_04_2009/en/ .

Bonneux, L. & Van Damme, W. Health is more than influenza. Bull. World Health Organ. 89 , 539–540. https://doi.org/10.2471/BLT.11.089078 (2011).

Simonsen, L. et al. Global mortality estimates for the 2009 influenza pandemic from the GLaMOR project: a modeling study. PLoS Med. https://doi.org/10.1371/journal.pmed.1001558 (2009).

Coronavirus Disease (COVID-19) Press Conference 17 February 2020 (2020). https://www.who.int/emergencies/diseases/novel-coronavirus-2019/media-resources/press-briefings .

Thompson, R. N., Gilligan, C. A. & Cunniffe, N. J. Control fast or control smart: when should invading pathogens be controlled?. PLoS Comput. Biol. 14 , 1–21. https://doi.org/10.1371/journal.pcbi.1006014 (2018).

Nebehay, S. WHO says it no longer uses ’pandemic’ category, but virus still emergency (2020). https://www.reuters.com/article/uk-china-health-who/who-says-it-no-longer-uses-pandemic-category-but-virus-still-emergency-idUKKCN20I0PD .

World Health Organization. WHO Director-General’s opening remarks at the media briefing on COVID-19 - 11 March 2020 (2020). https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-media-briefing-on-covid-19---11-march-2020 .

Kopecki, D., Lovelace Jr., B., Feuer, W. & Higgins-Dunn, N. WHO declares the coronavirus outbreak a global pandemic (2020). https://www.cnbc.com/2020/03/11/who-declares-the-coronavirus-outbreak-a-global-pandemic.html .

Boseley, S. WHO declares coronavirus pandemic (2020). https://www.theguardian.com/world/2020/mar/11/who-declares-coronavirus-pandemic .

Wan, W. What is pandemic? Why did WHO just declare one? (2020). https://www.washingtonpost.com/health/2020/03/11/who-declares-pandemic-coronavirus-disease-covid-19/ .

Webster, R. G. Predictions for future human influenza pandemics. J. Infect. Dis. 176 , S14–S19. https://doi.org/10.1086/514168 (1997).

Article   PubMed   Google Scholar  

Towers, S. & Feng, Z. Pandemic H1N1 influenza: predicting the course of a pandemic and assessing the efficacy of the planned vaccination programme in the United States. Eurosurveillance 14 , 1–3 (2009).

Coburn, B. J., Wagner, B. G. & Blower, S. Modeling influenza epidemics and pandemics: insights into the future of swine flu (H1N1). BMC medicine 7 (2009). http://www.ncbi.nlm.nih.gov/pubmed/19545404 .

Chowell, G., Bettencourt, L. M., Johnson, N., Alonso, W. J. & Viboud, C. The 1918–1919 influenza pandemic in England and Wales: spatial patterns in transmissibility and mortality impact. Proc. R. Soc. B Biol. Sci. 275 , 501–509. https://doi.org/10.1098/rspb.2007.1477 (2008).

Apolloni, A., Poletto, C., Ramasco, J. J., Jensen, P. & Colizza, V. Metapopulation epidemic models with heterogeneous mixing and travel behaviour. Theor. Biol. Med. Model. 11 , 1–26. https://doi.org/10.1186/1742-4682-11-3 (2014).

Hufnagel, L., Brockmann, D. & Geisel, T. Forecast and control of epidemics in a globalized world. Proc. Natl. Acad. Sci. 101 , 15124–15129. https://doi.org/10.1073/pnas.0308344101 (2004).

Article   ADS   CAS   PubMed   Google Scholar  

Watts, D. J., Muhamad, R., Medina, D. C. & Dodds, P. S. Multiscale, resurgent epidemics in a hierarchical metapopulation model. Proc. Natl. Acad. Sci. 102 , 11157–11162. https://doi.org/10.1073/pnas.0501226102 (2005).

Colizza, V., Barrat, A., Barthelemy, M. & Vespignani, A. The role of the airline transportation network in the prediction and predictability of global epidemics. Proc. Natl. Acad. Sci. 103 , 2015–2020. https://doi.org/10.1073/pnas.0510525103 (2006).

Article   ADS   CAS   PubMed   MATH   Google Scholar  

Ball, F. et al. Seven challenges for metapopulation models of epidemics, including households models. Epidemics 10 , 63–67 (2015).

Pei, S., Kandula, S., Yang, W. & Shaman, J. Forecasting the spatial transmission of influenza in the United States. Proc. Natl. Acad. Sci. U. S. A. 115 , 2752–2757. https://doi.org/10.1073/pnas.1708856115 (2018).

Article   CAS   PubMed   PubMed Central   Google Scholar  

Brownstein, J. S., Wolfe, C. J. & Mandl, K. D. Empirical evidence for the effect of airline travel on inter-regional influenza spread in the United States. PLoS Med. 3 , 1826–1835. https://doi.org/10.1371/journal.pmed.0030401 (2006).

Mateus, A. L., Otete, H. E., Beck, C. R., Dolan, G. P. & Nguyen-Van-Tam, J. S. Effectiveness of travel restrictions in the rapid containment of human influenza: a systematic review. Bull. World Health Organ. https://doi.org/10.2471/BLT.14.135590 (2014).

Andreasen, V. Dynamics of annual influenza A epidemics with immuno-selection. J. Math. Biol. 46 , 504–536. https://doi.org/10.1007/s00285-002-0186-2 (2003).

Valkenburg, S. A. et al. Immunity to seasonal and pandemic influenza A viruses. Microbes Infect. 13 , 489–501. https://doi.org/10.1016/j.micinf.2011.01.007.Immunity (2013).

Walker, P. G. T. et al. The impact of COVID-19 and strategies for mitigation and suppression in low- and middle-income countries. Science 422 , 413–422. https://doi.org/10.1126/science.abc0035 (2020).

Article   ADS   CAS   Google Scholar  

Bonds, M. H., Keenan, D. C., Rohani, P. & Sachs, J. D. Poverty trap formed by the ecology of infectious diseases. Proc. R. Soc. B Biol. Sci. 277 , 1185–1192. https://doi.org/10.1098/rspb.2009.1778 (2010).

Penman, B. S., Gupta, S. & Shanks, G. D. Rapid mortality transition of Pacific Islands in the 19th century. Epidemiol. Infect. 145 , 1–11. https://doi.org/10.1017/S0950268816001989 (2017).

Hossain, M. P. et al. The effects of border control and quarantine measures on the spread of COVID-19. Epidemics. https://doi.org/10.1016/j.epidem.2020.100397 (2020).

Jones, K. E. et al. Global trends in emerging infectious diseases HHS Public Access. Nature 451 , 990–993. https://doi.org/10.1038/nature06536 (2008).

Article   ADS   CAS   PubMed   PubMed Central   Google Scholar  

Merino, J. G. Response to Ebola in the US: misinformation, fear, and new opportunities. BMJ (Online) 349 , 1–2. https://doi.org/10.1136/bmj.g6712 (2014).

World Health Organization. Ebola virus disease - United Kingdom (2014). https://www.who.int/csr/don/30-december-2014-ebola/en/ .

Gomes, M. F. C. et al. Assessing the international spreading risk associated with the 2014 West African ebola outbreak. PLoS Curr. Outbreaks https://doi.org/10.1371/currents.outbreaks.cd818f63d40e24aef769dda7df9e0da5 (2014).

Gandhi, M., Yokoe, D. S. & Havlir, D. V. Asymptomatic transmission, the achilles’ heel of current strategies to control Covid-19. N. Engl. J. Med. 382 , 2158–2160. https://doi.org/10.1056/NEJMe2009758 (2020).

Lai, C. C., Shih, T. P., Ko, W. C., Tang, H. J. & Hsueh, P. R. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and coronavirus disease-2019 (COVID-19): the epidemic and the challenges. Int. J. Antimicrob. Agents 55 , 105924. https://doi.org/10.1016/j.ijantimicag.2020.105924 (2020).

Meakin, S. R., Tildesley, M. J., Davis, E. & Keeling, M. J. A metapopulation model for the 2018 Ebola outbreak in Equateur province in the Democratic Republic of the Congo. bioRxiv 1–30 (2018). https://doi.org/10.1101/465062 .

Baguelin, M. et al. Control of equine influenza: Scenario testing using a realistic metapopulation model of spread. J. R. Soc. Interface 7 , 67–79. https://doi.org/10.1098/rsif.2009.0030 (2009).

Balcan, D. et al. Multiscale mobility networks and the spatial spreading of infectious diseases. Proc. Natl. Acad. Sci. U. S. A. 106 , 21484–21489. https://doi.org/10.1073/pnas.0906910106 (2009).

Article   ADS   PubMed   PubMed Central   Google Scholar  

Bajardi, P. et al. Human mobility networks, travel restrictions, and the global spread of 2009 H1N1 pandemic. PLoS ONE https://doi.org/10.1371/journal.pone.0016591 (2011).

Gonçalves, B., Balcan, D. & Vespignani, A. Human mobility and the worldwide impact of intentional localized highly pathogenic virus release. Sci. Rep. 3 , 1–7. https://doi.org/10.1038/srep00810 (2013).

Bosch, F. V. D., Metz, J. A. J. & Zadoks, J. C. Pandemics of focal plant disease, a model. Anal. Theor. Plant Pathol. https://doi.org/10.1094/PHYTO.1999.89.6.495 (1999).

Chao, D. L., Halloran, M. E., Obenchain, V. J. & Longini, I. M. FluTE, a publicly available stochastic influenza epidemic simulation model. PLoS Comput. Biol. https://doi.org/10.1371/journal.pcbi.1000656 (2010).

Article   MathSciNet   PubMed   PubMed Central   Google Scholar  

Thompson, R. N. & Brooks-Pollock, E. Detection, forecasting and control of infectious disease epidemics: modelling outbreaks in humans, animals and plants (Trans. R. Soc. B Biol. Sci, Philos, 2019).

Google Scholar  

Khan, K. et al. Infectious disease surveillance and modelling across geographic frontiers and scientific specialties. Lancet Infect. Dis. 12 , 222–230. https://doi.org/10.1016/S1473-3099(11)70313-9 (2012).

Poletto, C., Tizzoni, M. & Colizza, V. Human mobility and time spent at destination: impact on spatial epidemic spreading. J. Theor. Biol. 338 , 41–58. https://doi.org/10.1016/j.jtbi.2013.08.032 (2013).

Barbarossa, M. V. et al. Transmission dynamics and final epidemic size of ebola virus disease outbreaks with varying interventions. PLoS ONE 10 , 1–21. https://doi.org/10.1371/journal.pone.0131398 (2015).

Nishiura, H., Chowell, G., Safan, M. & Castillo-Chavez, C. Pros and cons of estimating the reproduction number from early epidemic growth rate of influenza A (H1N1) 2009. Theor. Biol. Med. Model. 7 , 1–13. https://doi.org/10.1186/1742-4682-7-1 (2010).

Miller, J. C., Davoudi, B., Meza, R., Slim, A. C. & Pourbohloul, B. Epidemics with general generation interval distributions. J. Theor. Biol. 262 , 107–115. https://doi.org/10.1016/j.jtbi.2009.08.007 (2010).

van Herwaarden, O. A. & Grasman, J. Stochastic epidemics: major outbreaks and the duration of the endemic period. J. Math. Biol. 33 , 581–601. https://doi.org/10.1007/BF00298644 (1995).

Article   PubMed   MATH   Google Scholar  

Grasman, J. Stochastic epidemics: the expected duration of the endemic period in higher dimensional models. Math. Biosci. 152 , 13–27. https://doi.org/10.1016/S0025-5564(98)10020-2 (1998).

Article   MathSciNet   CAS   PubMed   MATH   Google Scholar  

Barbour, A. D. The duration of the closed stochastic epidemic. Biometrika 62 , 477–482. https://doi.org/10.1093/biomet/62.2.477 (1975).

Article   MathSciNet   MATH   Google Scholar  

Allen, L. J. & Allen, E. J. A comparison of three different stochastic population models with regard to persistence time. Theor. Popul. Biol. 64 , 439–449. https://doi.org/10.1016/S0040-5809(03)00104-7 (2003).

Miller, J. C. A note on the derivation of epidemic final sizes. Bull. Math. Biol. 74 , 1. https://doi.org/10.1007/s11538-012-9749-6 (2012).

Download references

Acknowledgements

This work was supported by funding from the Biotechnology and Biological Sciences Research Council (BBSRC) [grant number BB/M011224/1]. This research was funded by Christ Church, Oxford, via a Junior Research Fellowship (RNT).

Author information

Authors and affiliations.

Department of Zoology, University of Oxford, Oxford, UK

Benjamin J. Singer & Michael B. Bonsall

Christ Church, University of Oxford, Oxford, UK

Robin N. Thompson

Mathematical Institute, University of Oxford, Oxford, UK

You can also search for this author in PubMed   Google Scholar

Contributions

B.J.S., R.N.T., and M.B.B. conceived the study. B.J.S. carried out the analysis, wrote the manuscript, and prepared the figures. R.N.T. and M.B.B. supervised the research. All authors revised the manuscript and gave final approval for submission.

Corresponding author

Correspondence to Benjamin J. Singer .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary figures., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Singer, B.J., Thompson, R.N. & Bonsall, M.B. The effect of the definition of ‘pandemic’ on quantitative assessments of infectious disease outbreak risk. Sci Rep 11 , 2547 (2021). https://doi.org/10.1038/s41598-021-81814-3

Download citation

Received : 04 September 2020

Accepted : 29 December 2020

Published : 28 January 2021

DOI : https://doi.org/10.1038/s41598-021-81814-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

The incidence of psychosocial disturbances during the coronavirus disease-19 pandemic in an iranian sample.

• Azam Farmani
• Mojtaba Rahimian Bougar
• Hooman Farahmand

Current Psychology (2023)

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.