See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/366578427

A moderated mediation model of situational context and brand image for

online purchases using eWOM

Article in Journal of Product & Brand Management · November 2022

DOI: 10.1108/JPBM-02-2022-3857

CITATIONS READS

23 328

2 authors:

Le Nguyen Hoang Le Thanh Tung

Ho Chi Minh University of Banking Ho Chi Minh City Open University
12 PUBLICATIONS 69 CITATIONS 109 PUBLICATIONS 713 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Le Thanh Tung on 02 April 2023.

The user has requested enhancement of the downloaded file.

 1 Citation “Hoang, L.N. and Tung, L.T. (2023), "A moderated mediation model
2 of situational context and brand image for online purchases using eWOM",
3 Journal of Product & Brand Management, Vol. 32 No. 4, pp. 661-672.
4 https://doi.org/10.1108/JPBM-02-2022-3857”

6 A moderated mediation model of situational context and brand

7 image for online purchase using eWOM

8 Structured Abstract

9 Purpose- The moderating role of situational context in the effects of electronic word-of-
10 mouth (eWOM) on online purchase intention through brand image has failed to receive much
11 empirical support. Hence, this study aims to examine whether the situational context affects
12 the direction and strength of the relationships between different aspects of eWOM and brand
13 image, which then lead to online purchase intention.

14 Design/methodology/approach- To extend the existing research in this field, the model was
15 tested using a sample of 546 online shopping customers during the fourth wave of COVID-19
16 in Ho Chi Minh City. Specifically, the measurement and structural models were used to test
17 the direct relationships and the mediating role of brand image. The moderated mediation
18 model was then created to examine the moderating role of situational context. Furthermore,
19 the authors probed the interactions by discovering how the relationships from eWOM to
20 online purchase intention through brand image change at different levels of situational
21 context.

22 Findings- Without the moderating effect of situational context, the results indicated that the
23 influence of either eWOM credibility or quantity on intention is positively partially mediated
24 through brand image. With the moderating effect of situational context, the findings indicated
25 that high versus low disease avoidance customers will go through less eWOM credibility and
26 more eWOM quantity to develop their brand images and shape their intentions.

27 Originality/value- The findings have theoretical implications for understanding the

28 customers’ online purchase intentions under the pressure of disease avoidance. For practical
29 implications, tactics for profit and non-profit purposes are suggested.

1
1

2 Keywords: Online buying behaviour, E-commerce, Electronic word-of-mouth (eWOM),

3 Consumer behaviour, Brand image, Consumer brand congruence, Purchase situations,
4 Situational context.

2
 1 1. Introduction

2 The internet has fostered the market space by becoming an essential platform where people
3 become the media for sharing information, and the customers’ comments about a product or
4 brand on the internet, which is so-called electronic word-of-mouth (eWOM), have emerged
5 as one of the key thematic evolutions of e-commerce research (Verma and Yudav, 2021). In
6 the modern world, customers no longer trust brands for their self-generated information on
7 products (Srivastava and Sivaramakrishnan, 2020). They instead tend to seek eWOM to
8 gather the testimonials of others, reduce uncertainty, and secure lower prices in their pre-
9 purchase decisions (Ngarmwongnoi et al., 2020). This source of information can have an
10 impact on customers’ views of a brand (Moran and Muzellec, 2017; Krishnamurthy and
11 Kumar, 2018) and therefore on their purchase behaviour (Nyadzayo and Khajehzadeh, 2016;
12 Erkan and Evans, 2016; Bakri et al., 2020). Because a customer’s tendency to process
13 information about a brand is determined by a match between the sense of self and the brand
14 image (Sirgy, 2018; Wallace et al., 2021), examining the self-concept requires consideration
15 of the situational context or the environment (Michel et al., 2022).
16 Recent reviews of eWOM (Donthu et al., 2021; Ismagilova et al., 2020; Verma and Yudav,
17 2021) have shown the prominent topics in this field (e.g., negative eWOM, online reviews
18 and ratings, brand loyalty, just to name a few), but none of these reviews explores the
19 situational variables that influence purchase behaviour. In the marketing and consumer
20 behaviour literature, studies related to the situational context and its effect on intention and/or
21 behaviour have been conducted since the 1980s (Sirgy, 1982; Markus and Kunda, 1986).
22 Technology advancements such as the internet and social media have enabled customers to
23 maintain their brand dialogues with or without the marketer, hence strengthening or
24 weakening the customer-brand connection (Wirtz et al., 2013). However, a customer’s
25 intention occurs not only with the brand itself and other online customers but also with the
26 situational context, which has a demonstrable and systematic effect on the current behaviour
27 under examination. The question of how consumers’ perception of brand fit as well as how
28 their participation in brand communities influences a brand is seen as one of the emerging
29 and underrepresented research areas (King, 2017).
30 Moreover, while the questions of how the effects operate (the mechanisms) and when they
31 occur (the boundary conditions) are not necessarily independent (Hayes and Rockwood,
32 2020), behavioural scientists often treat them separately. To our best knowledge, no empirical
33 research exists to examine the boundary conditions of situational context by which its effect

3
 1 is transmitted into a behavioural intention. It is therefore questionable whether the situational
2 context affects the direction and strength of the relationships between different aspects of
3 eWOM and brand image, which then lead to online purchase intention. The objectives of this
4 study are to: (1) examine the direct effects of eWOM on online purchase intention; (2) test
5 the mediating role of a brand image between eWOM and online purchase intention; (3) test
6 the moderating effect of situational context on the relationships between eWOM and brand
7 image; and (4) explore how the relationships from eWOM to online purchase intention
8 through brand image change at different levels of situational context.
9 The intricate effects of situational context examined in this study involve the COVID-19
10 pandemic. This pandemic is an abnormal crisis because public health measures and lockdown
11 have resulted in significant changes in the market dynamics, including a transformation in
12 consumer behaviour, even though how much the transformation has changed is still
13 questionable (Mehta et al., 2020). The effect of situational context on online purchase
14 intention becomes even more appropriate during the outbreak of the COVID-19 pandemic in
15 Vietnam, especially in Ho Chi Minh City. Only essential businesses (i.e., the supply of
16 essential goods) and e-commerce have been allowed to operate during that period (Tran et al.,
17 2020). It thus provides a unique context in which the authors can examine the effect of
18 situational context on the translation of actor-related variables into online purchase intention.

19 2. Theoretical background

20 2.1. The effect of situational context on online purchase behaviour

21 Situational variables can influence online purchase behaviour (Chocarro et al., 2013;
22 Grimmer et al., 2016; Dominici et al., 2021; Kvalsvik, 2022). The concept of situational
23 variables was first defined by Belk (1975) as comprising all situational variables that are
24 specific to a place and time of observation but do not result from an object or personal
25 characteristics. To prevent confusion with intrapersonal factors such as personality traits,
26 Belk (1975) also provided five taxonomies of purchase situations. In terms of physical
27 surroundings, a physical distance from a store can be an appropriate factor affecting online
28 purchase behaviour. Online shopping can solve the issues associated with the difficulties of
29 reaching offline stores located far from customers. Distance to an open-air market is thus
30 likely to affect the probability of purchasing products online (Chocarro et al., 2013; Dominici
31 et al., 2021).

4
 1 The social surroundings feature involves the social experience of shopping, the presence of
2 others, and the interaction with them. The rapid development of the internet can allow the
3 sharing of information among people worldwide without limitations (Teng et al., 2016;
4 Thomas et al., 2019). Through various communication platforms on the internet, customers
5 can engage in eWOM to share and recommend their opinions (Erkan and Evans, 2016; Bulut
6 and Karabulut, 2018; Donthu et al., 2021) and increase their purchase confidence (Chocarro
7 et al., 2013). Although the impersonal nature of online purchases allows customers (e.g.,
8 particularly those who are socially averse) to avoid the social interaction required by
9 traditional face-to-face commerce (Dominici et al., 2021), older customers who lack
10 confidence in using the Internet perceive online shopping as a loss of shopping enjoyment
11 (Kvalsvik, 2022).
12 The temporal perspective refers to time-related factors that can influence a decision to shop
13 online (Grimmer et al., 2016; Peng et al., 2019). Online shopping is considered to save time
14 because individuals can compare many different types of goods and easily buy products from
15 their preferred places (Chocarro et al., 2013). As compared to offline shopping, while
16 customers tend to be discouraged from buying online if the delivery time is more than three
17 days (Kvalsvik, 2022), online shopping can save the time and effort required to get to the
18 pickup store (Kim et al., 2017).
19 The task definition feature refers to the purpose or goal of buying a specific product. For
20 instance, customers shop at different types of stores (e.g., Internet, catalogue, and in-store
21 shopping) depending on whether they shop for a gift or themselves (Gehrt and Yan, 2004).
22 Finally, the antecedent states refer to momentary moods (e.g., anxiety, pleasantness, or
23 excitation) or momentary conditions (e.g., fatigue, illness). Situational variables such as bad
24 health status and lack of mobility are the primary triggers for buying groceries online
25 (Dominici et al., 2021), especially for older customers (Kvalsvik, 2022).
26 The narrative literature on the effect of situational context on online purchase behaviour is
27 summarised in Table I. The initial search required articles to be based on the work of Belk
28 (1975) within the time frame since 2000 (e.g., when social media was introduced) on Google
29 Scholar. The subsequent search used “situation” and “online shopping” as the keywords. The
30 chosen papers were limited to the journals listed in the Social Sciences Citation Index (SSCI).
31 Previous relevant studies that applied the interdependence technique (e.g., variables are not
32 classified as dependent or independent; see Hair et al., 2019) or considered online post-
33 purchase behaviour were not included.

5
 1 The majority of previous studies relating to this topic have been conducted using
2 inventories of situational scenarios (Grimmer et al., 2016; Kim et al., 2017) and/or choice
3 alternatives (Gehrt and Yan, 2004; Chocarro et al., 2013; Dominici et al., 2021; Kvalsvik,
4 2022). Notably, it is frequently impossible for all of the features of a situation to be utilised.
5 Rather, the most relevant features must be selected for a particular research setting. As
6 compared to prior empirical research in the online setting, it may well be argued that little
7 attention has been paid to the effect of social surroundings (e.g., eWOM) and no research has
8 been carried out to study the interactive effects of social surroundings and antecedent states
9 (e.g., disease avoidance). These considerations not only help reflect the key characteristics of
10 the online shopping context but also help capture the pattern of how customers share online
11 information during the pandemic.

12 < Insert Table I around here >

13 2.2. eWOM Credibility

14 In a simple form, eWOM is defined as customer-generated communication directed to other
15 customers via digital tools (Rosario et al., 2020). While traditional word-of-mouth is
16 restricted to a small network due to its limited reach and accessibility, eWOM is web-based
17 communication where individuals may share and look for suggestions beyond their instant
18 social sphere through the internet (Teng et al., 2016; Thomas et al., 2019; Donthu et al.,
19 2021).
20 However, a high volume of eWOM messages from strangers often poses difficulties for
21 customers in evaluating the credibility of online reviews (Tsao and Hsieh, 2015; Moran and
22 Muzellec, 2017; Ismagilova et al., 2020). This concept thus appears to be essential when
23 studying customers’ online purchase decisions (Thomas et al., 2019; Ngarmwongnoi et al.,
24 2020) because it refers to the plausibility of message content perceived by the receivers
25 (Erkan and Evans, 2016; Verma and Dewani, 2020). In brief, eWOM credibility can be
26 defined as the degree to which one sees others’ reviews or recommendations as true or factual
27 (Levy and Gvili, 2015). When assessing the credibility of online reviews, the receivers often
28 depend on the source expertise (e.g., the perceived knowledge of the source) and
29 trustworthiness (e.g., the perceived goodness of the source) (Moran and Muzellec, 2017; Yin
30 et al., 2018; Verma and Dewani, 2020).
31 Previous researchers have considered the relationship between eWOM credibility and
32 online purchase intention (Wang et al., 2015; Koo, 2016; Teng et al., 2016; Ngarmwongnoi
33 et al., 2020). For instance, Koo (2016) conducted a study based on a sample of 302 students

6
 1 in South Korea and showed that eWOM credibility significantly affects online purchase
2 intention for skin care services, meals at a restaurant, and airline tickets. Taking into
3 consideration that online information transfers occur between individuals who may not have
4 any prior connection, it is important to examine how the perceived credibility of information
5 influences a customer’s online shopping intention (Teng et al., 2016; Ismagilova et al., 2020).
6 Therefore, it is hypothesised that:
7 H1a: eWOM credibility has a positive effect on online purchase intention

8 2.3. eWOM Quantity

9 eWOM quantity refers to the volume of online reviews sent by the reviewers to express their
10 opinions (Teng et al., 2016; Srivastava and Sivaramakrishnan, 2020). When individuals look
11 for eWOM, the quantity of eWOM makes information more observable (Thomas et al., 2019;
12 Ismagilova et al., 2020) and thus leads to greater consciousness and confidence by the
13 receivers (He and Bond, 2015).
14 Previous studies have shown that eWOM messages in both credibility and quantity
15 perspectives can affect customers’ intentions (Teng et al., 2016; Bulut and Karabulut, 2018;
16 Ngarmwongnoi et al., 2020). An experimental survey conducted by He and Bond (2015)
17 showed that eWOM volume positively affects customers’ purchase intentions in different
18 contexts, including desk lamps, flash drives, paintings, and music albums. Bulut and
19 Karabulut (2018) further found that eWOM quantity positively influences customers’ trust in
20 an online seller and their online repurchase intentions. With a huge number of eWOM made
21 in the context of the COVID-19 pandemic, there are concerns about the quantity of eWOM.
22 Although fake online reviews make customers feel suspicious about a large number of online
23 reviews caused by companies’ deceptive practices (Verma and Dewani, 2020), a large
24 number of reviews contribute to the authentication of eWOM (Thomas et al., 2019).
25 Additionally, customers often become cautious when there is a social bias related to a small
26 number of reviews, leading to a need to search for more information before making their
27 online purchases (Ngarmwongnoi et al., 2020). It is thus hypothesised that:
28 H1b: eWOM quantity has a positive effect on online purchase intention

29 2.4. The mediating role of brand image

30 The frequently cited definition of brand image is Keller (1993). Recent researchers
31 (Veloutsou and Delgado-Ballester, 2018; Fetscherin et al., 2021; Parris and Guzmán, 2022),
32 however, have challenged this traditional approach of conceptualisation by arguing that what
33 was defined as brand image refers to brand reputation and that brand image is a determinant

7
 1 of brand reputation (Fetscherin et al., 2021). While a brand image is the individual-specific
2 view of the brand, a brand reputation is seen as the summation of the individual’s judgements
3 (Parris and Guzmán, 2022). To acknowledge this academic discussion, a brand image is
4 defined in this study as an individual’s perception of a brand after one real or mental
5 encounter with that brand (Veloutsou and Delgado-Ballester, 2018). Because a customer, as
6 an individual, not only responds to his/her environment or group but also interacts with each
7 other, this study seeks to understand why a customer in a group or community has a different
8 perception of a brand.
9 Before making purchases, customers normally search for information to support their
10 purchase decisions and advance their perception of a brand (Moran and Muzellec, 2017;
11 Srivastava and Sivaramakrishnan, 2020; Wallace et al., 2021). The internal source of
12 information at this stage refers to what is saved and restored from a customer’s memory,
13 while all other sources (i.e., eWOM) are seen as external ones (Krishnamurthy and Kumar,
14 2018). It is thus important to understand the impact of eWOM on online purchase decisions,
15 especially since customers use this source of information to develop their brand images
16 (Gensler et al., 2015; Ngarmwongnoi et al., 2020). The attribution theory, which was first
17 coined by Heider (1958), can be used to illustrate the mediating role of brand image in the
18 relationship between eWOM and online purchase intention. This theory deals with how
19 people make sense of others’ behaviour and communication (Ullrich and Brunner, 2015), and
20 their interpretation plays a vital role in determining their reaction to these events (Chang et
21 al., 2015). Previous research also indicated that brand image can play a mediating role in
22 explaining the effect of eWOM on purchase intention (Jalilvand and Samiei, 2012; Bakri et
23 al., 2020). Hence, it is hypothesised that:
24 H2a: The relationship between eWOM credibility and online purchase intention is positively
25 mediated through brand image

26 H2b: The relationship between eWOM quantity and online purchase intention is positively
27 mediated through brand image

28 2.5. The moderating role of situational context

29 Previous researchers (Chocarro et al., 2013; Grimmer et al., 2016; Dominici et al., 2021;
30 Kvalsvik, 2022) have used a range of variables relating to the situational context to study
31 customers’ online purchase behaviour. However, all of their findings showed inconsistent
32 results across different situational items. One reason is that the relationship between
33 independent variables and online purchase intention may be moderated and/or mediated by

8
 1 other factors not covered in their studies. Specifically, customers often prefer different brands
2 in different consumption situations (Belk, 1975; Graeff, 1997). The brand-situation congruity
3 principle (or the self-congruity theory as named by Sirgy, 1982) stated that customers prefer a
4 brand which has a greater congruity between their characteristics (or the sense of self) and the
5 brand image (Sirgy, 2018; Wallace et al., 2021; Michel et al., 2022).
6 Although the self-concept is viewed as adjustable and flexible to different social situations
7 and environments in social psychology (Markus and Kunda, 1986), the moderating role of
8 situational context in the effects of eWOM on the brand image has failed to receive much
9 empirical support from consumer researchers. In normal conditions, disease avoidance
10 behaviour is not realised by customers while they shape their intentions to purchase online.
11 Such disease avoidance behaviour has become significantly manifest during a disease
12 outbreak such as COVID-19 because the perceived scarcity of products can significantly
13 influence consumer choices (Ratner et al., 2014; Hamilton et al., 2019). The scarcity of
14 products may also lead to a decrease in the credibility of external sources of information
15 (Pantano et al., 2020). For instance, drugstores and retail stores in Vietnam ran out of hand
16 sanitisers and surgical masks in just a few days, despite the government’s announcements to
17 not panic stockpile. The authors, therefore, advance the following hypotheses:
18 H3a: The relationship between eWOM credibility and brand image is negatively moderated
19 by the situational context (disease avoidance as an antecedent state)

20 H3b: The relationship between eWOM quantity and brand image is positively moderated by
21 the situational context (disease avoidance as an antecedent state). See Figure 1 for the
22 research framework.

23 < Insert Figure 1 around here >

24 3. Method

25 3.1. Measures
26 Using a scale with less than 7 scale points can lead to probable information loss and therefore
27 may affect the discovery of a moderating effect (Memon et al., 2019). The online survey
28 hence included 23 items on a 7-point Likert-type scale, ranging from 1 (strongly disagree) to
29 7 (strongly agree). The eWOM credibility consisted of five items adapted from Yin et al.
30 (2018). Four items to measure the eWOM quantity were adapted from Thomas et al. (2019).
31 Five items for measuring the brand image were drawn from Iglesias et al. (2019). As for the
32 situational context (disease avoidance as an antecedent state), five items measuring this

9
 1 construct were based on Brown et al. (2015). The online purchase intention was measured
2 using four items developed by Thomas et al. (2019). Some descriptive questions were also
3 added to the questionnaire (i.e., gender, age, education, occupation, and monthly income).

4 3.2. Participants and procedure

5 The national lockdown started in Vietnam on August 23rd, 2021 due to the fourth wave of
6 COVID-19. Given its unique context, online shopping customers in Ho Chi Minh City were
7 seen as the target population. Because no suitable sampling frame was identified, this issue
8 made the possibility of using the probability sampling method difficult to envisage (Bell et
9 al., 2018). The quota sampling was thus used to match the target population structure in terms
10 of age and gender, and then the convenience sampling was adopted to collect data in each
11 quota group from September 15th to October 15th, 2021. Before the main study, a pilot study
12 with 10 participants was conducted, and there was no problem with understanding.
13 An online-based survey was seen as the only choice of data collection method due to the
14 strict social distancing to ensure the safety of participants. To reach the target participants, the
15 link generated from Google Forms was posted on the researchers’ social networks in Ho Chi
16 Minh City via Zalo. It was the most popular messaging application used by the Vietnamese
17 government to send daily alerts about the COVID-19 (Winkie and Nambudiri, 2022). To
18 double-check the survey quality and the relevant geolocation, the authors also requested the
19 email addresses of the participants together with obtaining their IP addresses via Pabbly, an
20 online marketing and sales software that can be integrated with Google Forms. A filtering
21 question was used to choose suitable participants, and only customers with previous
22 experience in online shopping were chosen. In total, the authors received 619 responses from
23 volunteers, of which 25 respondents were not online shopping customers, and 48 responses
24 were seen as outliers (cases with standardised values fall outside the range of ± 3; see Hair et
25 al., 2019). Hence, only 546 responses were found valid for further analysis.
26 When considering the sample size, an indirect effect in the moderated mediation model was
27 formed as a product of two effects. To detect an indirect effect of certain sizes, if bias-
28 corrected bootstrapping to correct for skew in the population is used and the magnitudes of
29 two effects that form the indirect effect are both small, the sample size of 462 or a larger
30 sample will be required for the power level at 0.8 (Fritz and MacKinnon, 2007; Hayes, 2018).
31 Thus, the sample size of 546 in this study was considered sufficient.
32 To check for a potential non-response bias to assure the data quality (Armstrong and
33 Overton, 1977), the sample was divided into thirds (with n=182 for each group) according to

10
 1 the time between the starting and closing announcement of the online survey. The results of
2 independent sample t-tests indicated that there was no statistically significant difference in
3 the mean value between the first and third groups at the 5% level of significance.
4 Since the data could be seen as self-reported, the issue of common method bias was also
5 tested using Harman’s single-factor test and the common method factor technique (Podsakoff
6 et al., 2003). All the indicators were loaded into an exploratory factor analysis with one
7 single factor. The results showed that the total variance extracted by a single factor (46.2%)
8 was less than the threshold of 50%, indicating that common method bias was not a threat to
9 the study. The authors also included a common method factor in the measurement model to
10 control the effects of an unmeasured latent method factor (Collier, 2020). The comparison
11 between the measurement model with and without the common method factor yielded Δχ2 of
12 3.729 and Δdf of 1, indicating that the common method bias was not a substantial issue.

13 3.3. Sample characteristics

14 The final sample included 207 males (37.9%) and 339 females (62.1%). Nearly half of the
15 respondents (40.5%) were below 26 years old. The majority of respondents (58.4%) had a
16 bachelor’s degree. All respondents were active online shoppers, 35.1% of them were
17 students, and 33.2% of them had a monthly income of less than 5 million VND. The
18 proportion of respondents roughly reflected the demographic composition of Ho Chi Minh
19 City’s residents (see Table II).

20 < Insert Table II around here >

21 4. Results

22 The authors used structural equation modelling (SEM) to analyse the survey data and test the
23 hypotheses in AMOS 26 and SPSS 20. The measurement and structural models were used to
24 test the direct relationships and the mediating role of brand image. The moderated mediation
25 model was then created to examine the moderating role of situational context. Additionally,
26 the authors probed the interactions by discovering how the relationships between the
27 independent variables and the dependent variable change at different levels of the moderator.
28 The results are shown as the followings.

29 4.1. Measurement model results

30 The descriptive and validity statistics for each construct are presented in Table III. Significant
31 positive correlations existed among eWOM credibility (eC), eWOM quantity (eQuan),

11
 1 situational context (SC), brand image (BM), and online purchase intention (Int). All
2 constructs had average variance extracted (AVE) values of over 0.50 and none of the inter-
3 correlations of the constructs surpassed the square root of AVE (Fornell and Larcker, 1981).
4 Thus, it is possible to assume the convergent and discriminant validity of the scales.
5 < Insert Table III around here >

6 The construct reliability was assessed by calculating Cronbach’s alpha and composite
7 reliability (Pallant, 2020). Moreover, Hair et al. (2019) suggested that individual reflective
8 item reliability is acceptable when an item has a significant factor loading on its
9 corresponding construct (e.g., should be 0.5 or higher, and ideally above 0.7). As shown in
10 Table IV, all Cronbach’s alpha and composite reliability values were above the threshold
11 values of 0.8, and items loaded strongly and significantly above the cut-off value of 0.7.
12 Thus, it is safe to confirm the reliability of all scales in this study.

13 < Insert Table IV around here >

14 4.2. Structural model results

15 The authors followed the framework suggested by Collier (2020) to test the structural model.
16 The non-mediated models examined the direct relationship between the independent variables
17 and the dependent variable without the mediator and the moderator. These models met the
18 requirements of a good overall fit (χ2=48.051, df=25, GFI=0.981, TLI=0.990, CFI=0.993,
19 RMSEA=0.041 for eWOM credibility; and χ2 =24.506, df=19, GFI=0.988, TLI=0.997,
20 CFI=0.998, RMSEA=0.023 for eWOM quantity) (Hair et al., 2019). Additionally, either
21 eWOM credibility (β=0.490, t=13.000, p<0.001) or eWOM quantity (β=0.773, t=14.600,
22 p<0.001) had a significant positive effect on online purchase intention without the mediator
23 of brand image and the moderator of situational context at 99% confidence interval (CI), thus
24 supporting H1a and H1b.
25 In the mediated models without the moderator of situational context, as presented in Table
26 V, the requirements of a good overall fit were observed. Either eWOM credibility (indirect
27 β=0.298, 95% CI [0.208, 0.418], p<0.001) or eWOM quantity (indirect β=0.300, 95% CI
28 [0.182, 0.460], p<0.001) had a significant positive indirect effect on online purchase intention
29 through the construct of brand image. In combination with the results from the non-mediated
30 models, the influence of either eWOM credibility or eWOM quantity on online purchase
31 intention was positively partially mediated through the construct of brand image, hence
32 supporting H2a and H2b.

12
 1 < Insert Table V around here >

2 4.3. Moderated mediation model results

3 The authors used the mixed model method suggested by Collier (2020) to test the moderated
4 mediation model. This method included the latent variables as the independent and dependent
5 variables but used the composite moderator variable along with the composite interaction
6 terms. The authors first attempted to mean centre all the composite variables before forming
7 the interaction variables for testing the interaction effects. The interaction effects were then
8 probed at a low or high level by taking the mean of the original moderator minus or plus one
9 standard deviation.
10 The moderated mediation model results showed a good overall fit, as presented in Table VI.
11 All the direct relationships were significant at 95% CI, except for the direct effect of eWOM
12 credibility on online purchase intention (β=-0.047, t=-0.756, p=0.45) which was non-
13 significant (or the indirect effect of eWOM credibility on online purchase intention was fully
14 mediated through the brand image with the presence of situational context). The results also
15 showed the significant moderating role of situational context on brand image, even though
16 the direct effect of the interaction of eWOM credibility and situational context on the brand
17 image was significant with a negative coefficient (β=-0.193, t=-3.732, p<0.001).
18 With the presence of situational context, the indirect effects of both eWOM credibility
19 (indirect β=0.131, 95% CI [0.059, 0.246], p<0.001) and eWOM quantity (indirect β=0.111,
20 95% CI [0.049, 0.207], p=0.002) on online purchase intention via brand image were
21 significant at 95% CI. When probing the interaction at one standard deviation below and
22 above the mean, all the moderated indirect effects were significant at 95% CI, except for the
23 moderated indirect effect of eWOM quantity on online purchase intention at the low level of
24 moderator, which was only significant at 90% CI. Furthermore, the indexes of moderated
25 mediation were all significant at 95% CI, thus supporting H3a and H3b. The indexes of
26 moderated mediation indicated that the indirect effect of eWOM credibility on online
27 purchase intention through brand image was weakened as the situational context increased
28 (Index=-0.067). Meanwhile, the indirect effect of eWOM quantity on online purchase
29 intention through brand image was strengthened as the situational context increased
30 (Index=0.058).
31 The authors also controlled the effects of gender, age, education, and monthly income on
32 the final dependent variable. These control variables were recoded into binary variables (e.g.,
33 male/female, low/high) for the regression and multi-group invariance analyses. The sample

13
 1 size of each group (n>200) followed the rule of thumb suggested by Kline (2005). The results
2 showed that only gender (β=0.283, p=0.001) and monthly income (β=-0.259, p=0.002) had
3 significant effects on online purchase intention. To test invariance, Δχ2 and Δdf between the
4 unconstrained and the constrained model were calculated (Byrne, 2004). The comparisons
5 yielded Δχ2 of 22.003 and Δdf of 11 (p=0.024) on gender-constrained equal and Δχ2 of
6 29.745 and Δdf of 11 (p=0.001) on monthly income-constrained equal. The results implied
7 that while female participants were more likely to use eWom to form their online purchase
8 intentions, high monthly income participants were less likely to consider eWom as the input
9 information for making their decisions. Further analysis was not performed due to the
10 inadequate sample size to capture the indirect relationships in each group.

11 < Insert Table VI around here >

12 5. Conclusions and discussion

13 5.1. Conclusions
14 This study aims to examine whether the situational context affects the direction and strength
15 of the relationships between different aspects of eWOM and brand image, which then lead to
16 online purchase intention. In normal conditions without the mediating and moderating
17 effects, the results showed that both eWOM credibility and quantity have positive effects on
18 online purchase intention. With the presence of brand image as the mediator, it was found
19 that the influence of either eWOM credibility or eWOM quantity on online purchase
20 intention is positively partially mediated through the construct of brand image. These results
21 are in line with previous research about the direct relationships between either eWOM
22 credibility or eWOM quantity and online purchase intention (Wang et al., 2015; Teng et al.,
23 2016; Thomas et al., 2019), and their indirect effects on online purchase intention via the
24 brand image (Jalilvand and Samiei, 2012; Krishnamurthy and Kumar, 2018).
25 Consistent with Teng et al.’s (2016) study, customers tend to assess the credibility of
26 eWOM in their decision-making processes to increase their confidence and decrease risks and
27 uncertainties. Meanwhile, the quantity of eWOM delivers a signal of product popularity with
28 a diversity of information, thus leading to an increase in online purchase intention
29 (Ngarmwongnoi et al., 2020). In terms of the mediating role of the brand image between
30 eWOM and online purchase intention, the results can be explained by the attribution theory
31 suggested by Heider (1958); and that the more a product’s reviews can be attributed to
32 stimulus-level characteristics (i.e., product quality or functionality), the more credible they

14
 1 are observed by the receivers (Moran and Muzellec, 2017), resulting in enhancing brand
2 image and online purchase intention (Jalilvand and Samiei, 2012; Bakri et al., 2020). Yet,
3 these results should be treated with caution, as previous studies have shown that situational
4 variables can moderate these effects.
5 In abnormal conditions with the moderating effect of situational context (e.g., disease
6 avoidance), the negative interactive direct effect between eWOM credibility and situational
7 context on brand image is worth noting here. Customers during the pandemic tend to depend
8 less on eWOM credibility for building their brand images when the propensity for disease
9 avoidance increases. Since these findings contradict prior research about the role of eWOM
10 credibility (Teng et al., 2016; Thomas et al., 2019), it demands further examination.
11 Moreover, high versus low disease avoidance customers tend to go through less eWOM
12 credibility and more eWOM quantity to develop their brand images and intentions. These
13 findings can be explained by the psychology of persuasion written by Cialdini (1993; 2006).
14 More specifically, product scarcity (e.g., during the pandemic) can shape customers’
15 decisions by influencing how they process information. When the incentive to process
16 information is low (e.g., in normal conditions before the pandemic), customers tend to focus
17 more on systematic processing (e.g., careful consideration of both eWOM credibility and
18 quantity). In contrast, when the incentive to process information is high (e.g., during the
19 pandemic), customers are likely to focus more on heuristic processing (e.g., focusing on
20 easily comprehended cues such as eWOM quantity).

21 5. 2. Theoretical implications
22 This study extends the marketing theory in two profound ways. First, the present study
23 enriches the self-congruity theory by integrating the self-concept (Sirgy, 1982; 2018) with the
24 concept of situational variables (Belk, 1975). Although previous researchers have extensively
25 studied the self-concept and its variations (e.g., actual self, ideal self, social self, ideal social
26 self) (Wallace et al., 2021; Michel et al., 2022), aligning the self-concept and the brand image
27 can be a complicated process. Indeed, the malleability of the self-concept (Markus and
28 Kunda, 1986) indicated that customers in different situations express different selves. The
29 meanings of the self are thus not able to be carried over from one situation to another. By
30 shifting toward the dynamic nature of the self in the social media context, the present study
31 casts more light on how consumers’ perception of a brand fit under the pressure of situational
32 context and the effect of social surroundings influences their purchase decisions.

15
 1 Another key insight involves the context of the study. With the effects of the COVID-19
2 pandemic, customers are experiencing transformations in their behaviour (Mehta et al.,
3 2020). Before the pandemic, customers preferred eWOM because it provided more reliable
4 information than companies’ self-generated information (Ngarmwongnoi et al., 2020;
5 Srivastava and Sivaramakrishnan, 2020). The increased awareness of fake online reviews
6 caused by companies’ deceptive practices even makes customers feel more concerned about
7 the credibility of eWOM (Thomas et al., 2019; Verma and Dewani, 2020). Yet during the
8 pandemic, customers tend to depend less on eWOM credibility and more on eWOM quantity
9 for building their brand images. These findings contribute to the sociological field for
10 understanding customers’ online purchase intentions under the pressure of disease avoidance
11 because the findings turn upside down the current literature that usually seeks to promote the
12 credibility of online reviews. This insight is important because it can aid in shaping
13 customers’ brand images in a time of emergency such as the COVID-19 pandemic.

14 5. 3. Practical implications
15 In terms of practical implications, marketers can use eWOM as a signal to encourage online
16 purchase intentions. However, a company with a strong brand image should not depend on
17 the benefits of a high brand image that can be found in the literature (Upamannyu and
18 Sankpal, 2014; Nyadzayo and Khajehzadeh, 2016). Instead, companies should be aware of a
19 large number of online reviews (e.g., how many reviews were made about their brands) rather
20 than the credibility of online reviews (e.g., where reviews come from), because even a high
21 brand image can be significantly weakened by a large number of negative eWOM (Jalilvand
22 and Samiei, 2012). Such effects will become even more important with the advancement of
23 network technology.
24 Another interesting practical implication of the findings relates to the use of scarcity
25 persuasion tactics (e.g., those that are often relevant during the pandemic) by online retailers
26 (Hamilton et al., 2019). For instance, online retailers can use a countdown clock to display
27 the remaining time of a promotion, show the number of online reviews jointly considering the
28 offering, and display the number of units still available for sale. For customers with a low
29 intention to purchase, such tactics are likely to increase stimulation, making customers more
30 focused on differentiating attributes of the product. Finally, regardless of whether the
31 COVID-19 crisis is damaging or aiding online retailers’ profits, the common goal of placing
32 people before profitability should be converted into actions. For instance, online retailers
33 should control the panic stockpiling of essential goods through eWOM quantity, especially

16
 1 for the spread of fake news or misinformation online. To do so, online retailers can
2 periodically ask customers to rate the credibility of a randomly selected review, thus recalling
3 customers about the credibility in an indirect way that should avoid reluctance (Pennycook et
4 al., 2020). Such an approach can potentially reduce the amount of misinformation flowing
5 online without depending on government institutions.

6 5.4. Limitations and future research

7 To examine the moderating role of situational context, the authors only made use of the
8 disease avoidance construct developed by Brown et al. (2015) to measure the online purchase
9 situation. Although this construct is relevant during the disease outbreak, using only one core
10 social goal or motivation that promotes or prevents an individual’s behaviour may affect the
11 generalisability of the findings to other situational contexts. Therefore, expanding the
12 situational variables in future research is a useful extension of this study.
13 Another limitation is the absence of eWOM valence (e.g., negative versus positive
14 information). It is unclear in this study how negative and positive reviews generate their
15 persuasive power. More experiments can be used to examine the persuasive effects of varying
16 eWOM credibility and quantity (e.g., negative versus positive) on customers’ decision-
17 making. Furthermore, while eWOM quality is similar to eWOM credibility (Teng et al.,
18 2016), considering the effect of this construct may add another insight into how relevant,
19 unbiased, and verifiable reviews influence the customers’ behavioural intentions.
20 Additionally, the control variables (i.e., gender and monthly income) were found to
21 influence the participants’ online purchase intentions, but further analysis was not performed
22 due to the insufficient sample size in each group. Hence, future research with a relatively
23 larger sample size may examine these control variables and others (e.g., product knowledge
24 or product involvement) to enhance the validity of the results.
25 Finally, while this study sheds more light on the role of situational context in the consumer
26 behaviour literature, the authors do not know how customers will behave when the pandemic
27 is over. Thus, this is another path for future research to use a longitudinal design to examine
28 the moderating role of situational context.

17
 1 References
2

3 Armstrong, J. S. and Overton, T. S. (1977), “Estimating non-response bias in mail

4 surveys”, Journal of Marketing Research, Vol. 14 No. 3, pp.396-402.

5 Bakri, M., Krisjanous, J. and Richard, J. E. (2020), “Decoding service brand image through
6 user-generated images”, Journal of Services Marketing, Vol. 34 No. 4, pp.429-442.

7 Belk, R. W. (1975) “Situational variables and consumer behavior”, Journal of Consumer

8 research, Vol. 2 No. 3, pp.157-164.

9 Bell, E., Bryman, A. and Harley, B. (2018), Business research methods, 5th ed., Oxford
10 University Press, New York, NY.

11 Brown, N. A., Neel, R. and Sherman, R. A. (2015), “Measuring the evolutionarily important
12 goals of situations: Situational affordances for adaptive problems”, Evolutionary
13 Psychology, Vol. 13 No. 3, pp.1-15.

14 Bulut, Z. A. and Karabulut, A. N. (2018), “Examining the role of two aspects of eWOM in
15 online repurchase intention: An integrated trust–loyalty perspective”, Journal of
16 Consumer Behaviour, Vol. 17 No. 4, pp.407-417.

17 Byrne, B. M. (2004), “Testing for multigroup invariance using AMOS graphics: A road less
18 travelled”, Structural equation modelling, Vol. 11 No. 2, pp.272-300.

19 Chang, H. H., Tsai, Y. C., Wong, K. H., Wang, J. W. and Cho, F. J. (2015), “The effects of
20 response strategies and severity of failure on consumer attribution with regard to
21 negative word-of-mouth”, Decision Support Systems, Vol. 71, pp.48–61.

22 Cialdini, R. B. (1993), The psychology of persuasion, William Morrow, New York, NY.

23 Cialdini, R. B. (2006), Influence: the psychology of persuasion, Revised ed., William

24 Morrow, New York, NY.

25 Chocarro, R., Cortinas, M. and Villanueva, M. L. (2013), “Situational variables in online

26 versus offline channel choice”, Electronic Commerce Research and Applications, Vol.
27 12 No. 5, pp.347-361.

28 Collier, J. E. (2020), Applied structural equation modeling using AMOS: Basic to advanced
29 techniques, Routledge, New York, NY.

18
 1 Dominici, A., Boncinelli, F., Gerini, F. and Marone, E. (2021), “Determinants of online food
2 purchasing: The impact of socio-demographic and situational factors”, Journal of
3 Retailing and Consumer Services, Vol. 60, pp.1-9.

4 Donthu, N., Kumar, S., Pandey, N., Pandey, N. and Mishra, A. (2021), “Mapping the
5 electronic word-of-mouth (eWOM) research: A systematic review and bibliometric
6 analysis”, Journal of Business Research, Vol. 135, pp.758-773.

7 Erkan, I. and Evans, C. (2016), “The influence of eWOM in social media on consumers’
8 purchase intentions: An extended approach to information adoption”, Computers in
9 Human Behavior, Vol. 61, pp.47-55.

10 Fetscherin, M., Veloutsou, C. and Guzman, F. (2021), “Models for brand relationships: guest
11 editorial”, Journal of Product & Brand Management, Vol. 30 No. 3, pp. 353-359.

12 Fornell, C. and Larcker, D. F. (1981), “Evaluating structural equation models with

13 unobservable variables and measurement error”, Journal of Marketing Research, Vol.
14 18 No. 1, pp.39-50.

15 Fritz, M. S. and MacKinnon, D. P. (2007), “Required sample size to detect the mediated
16 effect”, Psychological science, Vol. 18 No. 3, pp.233-239.

17 Gehrt, K. C. and Yan, R. N. (2004), “Situational, consumer, and retailer factors affecting
18 Internet, catalog, and store shopping”, International Journal of Retail & Distribution
19 Management, Vol. 32 No. 1, pp.5-18.

20 Gensler, S., Völckner, F., Egger, M., Fischbach, K. and Schoder, D. (2015), “Listen to your
21 customers: Insights into brand image using online consumer-generated product
22 reviews”, International Journal of Electronic Commerce, Vol. 20 No. 1, pp.112-141.

23 Graeff, T. R. (1997), “Consumption situations and the effects of brand image on consumers'
24 brand evaluations”, Psychology & Marketing, Vol.14 No.1, pp.49-70.

25 Grimmer, M., Kilburn, A. P. and Miles, M. P. (2016), “The effect of purchase situation on
26 realized pro-environmental consumer behavior”, Journal of Business Research, Vol. 69
27 No. 5, pp.1582-1586.

28 Hair, J. F., Black, W. C., Babin, B. J. and Anderson, R. E. (2019), Multivariate Data
29 Analysis, 8th ed., Cengage Learning, Hampshire, England.

19
 1 Hamilton, R., Thompson, D., Bone, S., Chaplin, L. N., Griskevicius, V., Goldsmith, K. et al.
2 (2019), “The effects of scarcity on consumer decision journeys”, Journal of the
3 Academy of Marketing Science, Vol. 47 No. 3, pp.532-550.

4 Hayes, A. F. (2018), Introduction to mediation, moderation, and conditional process

5 analysis: A regression-based approach, 2nd ed., The Guilford Press, New York, NY.

6 Hayes, A. F. and Rockwood, N. J. (2020), “Conditional process analysis: Concepts,

7 computation, and advances in the modeling of the contingencies of mechanisms”,
8 American Behavioral Scientist, Vol. 64 No. 1, pp.19-54.

9 He, S. X. and Bond, S. D. (2015), “Why is the crowd divided? Attribution for dispersion in
10 online word of mouth”, Journal of Consumer Research, Vol. 41 No. 6, pp.1509-1527.

11 Heider, F. (1958), The psychology of interpersonal relations, John Wiley & Sons, New York,
12 NY.

13 Iglesias, O., Markovic, S., Singh, J. J. and Sierra, V. (2019), “Do customer perceptions of
14 corporate services brand ethicality improve brand equity? Considering the roles of
15 brand heritage, brand image, and recognition benefits”, Journal of Business Ethics, Vol.
16 154 No. 2, pp.441-459.

17 Ismagilova, E., Slade, E. L., Rana, N. P. and Dwivedi, Y. K. (2020), “The effect of electronic
18 word of mouth communications on intention to buy: A meta-analysis”, Information
19 Systems Frontiers, Vol. 22 No. 5, pp.1203-1226.

20 Jalilvand, M. R. and Samiei, N. (2012), “The effect of electronic word of mouth on brand
21 image and purchase intention: An empirical study in the automobile industry in Iran”,
22 Marketing Intelligence & Planning, Vol. 30 No. 4, pp.460-476.

23 Keller, K. L. (1993), “Conceptualizing, measuring, and managing customer-based brand

24 equity”, Journal of Marketing, Vol. 57 No. 1, pp.1-22.

25 Kim, E., Park, M. C. and Lee, J. (2017), “Determinants of the intention to use Buy-Online,
26 Pickup In-Store (BOPS): The moderating effects of situational factors and product
27 type”, Telematics and Informatics, Vol. 34 No. 8, pp.1721–1735.

28 King, C. (2017), “Brand management - standing out from the crowd”, International Journal
29 of Contemporary Hospitality Management, Vol. 29 No. 1, pp.115–140.

20
 1 Kline, T. (2005), Psychological testing: A practical approach to design and evaluation, Sage,
2 Thousand Oaks, CA.

3 Kvalsvik, F. (2022), “Understanding the role of situational factors on online grocery

4 shopping among older adults”, Journal of Retailing and Consumer Services, Vol. 68,
5 pp.1-7.

6 Koo, D. M. (2016), “Impact of Tie Strength and Experience on the Effectiveness of Online
7 Service Recommendations”, Electronic Commerce Research and Applications, Vol. 15,
8 pp.38-51.

9 Krishnamurthy, A. and Kumar, S. R. (2018), “Electronic word-of-mouth and the brand

10 image: Exploring the moderating role of involvement through a consumer expectations
11 lens”, Journal of Retailing and Consumer Services, Vol. 43, pp.149-156.

12 Levy, S. and Gvili, Y. (2015), “How credible is e-word of mouth across digital-marketing
13 channels?: The roles of social capital, information richness, and interactivity”, Journal
14 of Advertising Research, Vol. 55 No. 1, pp.95-109.

15 Markus, H. and Kunda, Z. (1986), “Stability and malleability of the self-concept”, Journal of
16 Personality and Social Psychology, Vol. 51 No. 4, pp.858-866.

17 Mehta, S., Saxena, T. and Purohit, N. (2020), “The new consumer behaviour paradigm amid
18 COVID-19: permanent or transient?”, Journal of health management, Vol. 22 No.2,
19 pp.291-301.

20 Memon, M. A., Cheah, J. H., Ramayah, T., Ting, H., Chuah, F. and Cham, T. H. (2019),
21 “Moderation analysis: Issues and guidelines”, Journal of Applied Structural Equation
22 Modeling, Vol. 3 No. 1, pp.1-11.

23 Michel, G., Torelli, C. J., Fleck, N. and Hubert, B. (2022), “Self-brand values congruity and
24 incongruity: Their impacts on self-expansion and consumers’ responses to brands”,
25 Journal of Business Research, Vol. 142, pp. 301-316.

26 Moran, G. and Muzellec, L. (2017), “eWOM credibility on social networking sites: A

27 framework”, Journal of Marketing Communications, Vol. 23 No. 2, pp.149-161.

28 Ngarmwongnoi, C., Oliveira, J. S., AbedRabbo, M. and Mousavi, S. (2020), “The

29 implications of eWOM adoption on the customer journey”, Journal of Consumer
30 Marketing, Vol. 37 No. 7, pp.749-759.

21
 1 Nyadzayo, M. W. and Khajehzadeh, S. (2016), “The antecedents of customer loyalty: A
2 moderated mediation model of customer relationship management quality and brand
3 image”, Journal of Retailing and Consumer Services, Vol. 30, pp.262-270.

4 Pallant, J. (2020), SPSS survival manual: A step by step guide to data analysis using IBM
5 SPSS, 7th ed., Open University Press, Berkshire, England.

6 Pantano, E., Pizzi, G., Scarpi, D. and Dennis, C. (2020), “Competing during a pandemic?
7 Retailers’ ups and downs during the COVID-19 outbreak”, Journal of Business
8 Research, Vol. 116, pp.209-213.

9 Parris, D. L. and Guzmán, F. (2022), “Evolving brand boundaries and expectations: Looking
10 back on brand equity, brand loyalty, and brand image research to move forward”,
11 Journal of Product & Brand Management, Vol. ahead-of-print No. ahead-of-print.
12 https://doi.org/10.1108/JPBM-06-2021-3528

13 Peng, L., Zhang, W., Wang, X. and Liang, S. (2019), “Moderating effects of time pressure on
14 the relationship between perceived value and purchase intention in social e-commerce
15 sales promotion: Considering the impact of product involvement”, Information &
16 Management, Vol. 56 No. 2, pp.317-328.

17 Pennycook, G., Epstein, Z., Mosleh, M., Arechar, A., Eckles, D. and Rand, D. (2020),
18 “Understanding and reducing the spread of misinformation online”, ACR North
19 American Advances, Vol. 48, pp.863-867.

20 Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y. and Podsakoff, N. P. (2003), “Common

21 method biases in behavioural research: A critical review of the literature and
22 recommended remedies”, Journal of applied psychology, Vol. 88 No. 5, pp.879-903.

23 Ratner, R. K., Zhu, M., Shah, A. K., Shafir, E., Mullainathan, S., Thompson, D. V. and
24 Griskevicius, V. (2014), “Why having so little means so much: scarcity shapes
25 consumer decision making”, Advances in Consumer Research, Vol. 42, pp.230-234.

26 Rosario, A. B., de Valck, K. and Sotgiu, F. (2020), “Conceptualizing the electronic word-of-
27 mouth process: What we know and need to know about eWOM creation, exposure, and
28 evaluation”, Journal of the Academy of Marketing Science, Vol. 48 No. 3, pp.422-448.

29 Sirgy, M. J. (1982), “Self-concept in consumer behavior: A critical review”, Journal of

30 Consumer Research, Vol. 9 No. 3, pp.287-300.

22
 1 Sirgy, M. J. (2018), “Self-congruity theory in consumer behavior: A little history”, Journal of
2 Global Scholars of Marketing Science, Vol. 28 No. 2, pp. 197-207.

3 Srivastava, M. and Sivaramakrishnan, S. (2020), “The impact of eWOM on consumer brand

4 engagement”, Marketing Intelligence & Planning, Vol. 39 No. 3, pp. 469-484.

5 Teng, S., Khong, K. W., Chong, A. Y. L. and Lin, B. (2016), “Examining the impacts of
6 electronic word-of-mouth message on consumers’ attitude”, Journal of Computer
7 Information Systems, Vol. 57 No. 3, pp.238-251.

8 Thomas, M. J., Wirtz, B. W. and Weyerer, J. C. (2019), “Determinants of online review

9 credibility and its impact on consumers’ purchase intention”, Journal of Electronic
10 Commerce Research, Vol. 20 No. 1, pp.1-20.

11 Tran, T. P. T., Le, T. H., Nguyen, T. N. P. and Hoang, V. M. (2020), “Rapid response to the
12 COVID-19 pandemic: Vietnam government’s experience and preliminary success”,
13 Journal of Global health, Vol. 10 No. 2, pp.1-14.

14 Tsao, W. C. and Hsieh, M. T. (2015), “eWOM persuasiveness: do eWOM platforms and

15 product type matter?”, Electronic Commerce Research, Vol. 15 No. 4, pp.509-541.

16 Ullrich, S. and Brunner, C. B. (2015), “Negative online consumer reviews: effects of

17 different responses”, Journal of Product & Brand Management, Vol. 24 No. 1, pp.66–
18 77.

19 Upamannyu, N. K. and Sankpal, S. (2014), “Effect of brand image on customer satisfaction

20 & loyalty intention and the role of customer satisfaction between brand image and
21 loyalty intention”, Journal of Social Science Research, Vol. 3 No. 2, pp.274-285.

22 Veloutsou, C. and Delgado-Ballester, E. (2018), “New challenges in brand management”,

23 Spanish Journal of Marketing, Vol. 22 No. 3, pp.255-272.

24 Verma, D. and Dewani, P. P. (2020), “eWOM credibility: a comprehensive framework and

25 literature review”, Online Information Review, Vol. 45 No. 3, pp.481-500.

26 Verma, S. and Yadav, N. (2021), “Past, present, and future of electronic word of mouth
27 (EWOM)”, Journal of Interactive Marketing, Vol. 53, pp.111-128.

28 Wallace, E., Torres, P., Augusto, M. and Stefuryn, M. (2021), “Outcomes for self-expressive
29 brands followed on social media: Identifying different paths for inner self-expressive

23
 1 and social self-expressive brands”, Journal of Business Research, Vol. 135, pp. 519-
2 531.

3 Wang, X., Teo, H. H. and Wei, K. K. (2015), “Simultaneity and interactivity of the effects of
4 communication elements on consumers’ decision making in eWOM systems”, Journal
5 of Electronic Commerce Research, Vol. 16 No. 3, pp.153-174.

6 Winkie, M. J. and Nambudiri, V. E. (2022), “A tale of two applications: lessons learned from
7 national LMIC COVID applications”, Journal of the American Medical Informatics
8 Association, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1093/jamia/
9 ocac146

10 Wirtz, J., den Ambtman, A., Bloemer, J., Horváth, C., Ramaseshan, B., van de Klundert, J.
11 and Kandampully, J. (2013), “Managing brands and customer engagement in online
12 brand communities”, Journal of Service Management, Vol. 24 No. 3, pp.223-244.

13 Yin, C., Sun, Y., Fang, Y. and Lim, K. (2018), “Exploring the dual-role of cognitive
14 heuristics and the moderating effect of gender in microblog information credibility
15 evaluation”, Information Technology & People, Vol. 31 No. 3, pp.741-769.

24
Table I Summary of research on the effect of social surroundings (SS) and/or antecedent states (AS) on online purchase behaviour
Author(s) Study context(s) Theory/Theories Method(s) Exogenous variable(s) Endogenous Effects of SS Interactive
variable(s) effects of SS
and AS
Gehrt and Channel Literature on Multivariate Time pressure, gift or personal use Channel No No
Yan (2004) preference situational analysis of variance characteristics
(Internet, influence and retail (MANOVA)
catalogue, and attributes
store shopping)
Chocarro et Channel Literature on Multinomial Product category, distance-to-store, store Probability of Yes – Presence No
al. (2013) preference (online situational factors logistic regression tidiness, clarity of website layout, time online/offline of others and
versus offline) of day of purchase, time pressure, purchase social
presence of others, and social interaction interaction
Grimmer et Online green Theory of reasoned Structural equation Purchase situations (i.e., physical Implementation No No
al. (2016) products action and theory modelling with surroundings, temporal perspective, task intentions, pro-
consumption of planned moderation definition), intentions environmental
behaviour analysis consumer
behaviour
Kim et al. Buy online and The diffusion of Factorial design Relative advantages, compatibility, Intention to use No No
(2017) pick up in-store innovations theory, complexity, perceived risk of an online BOPS
(BOPS) literature on store, location convenience, product
situational factors involvement
and product
involvement
Dominici et Online food Literature on e- Binary logistic Socio-demographic variables, situational The likelihood of No No
al. (2021) consumption grocery purchases regression factors (i.e., physical surroundings, individuals buying
temporal perspective, antecedent states), food online
control variables
Kvalsvik Online grocery Literature on Conjoint analysis Distance to a nearby store, prices at the The probability of Yes-Social No
(2022) shopping among situational factors nearby store, social interaction, delivery online grocery interaction
older adults time, poor health, lack of mobility shopping
Current Essential goods Self-congruity Structural equation eWOM credibility, eWOM quantity, Brand image, Yes – eWOM Yes
study during the theory modelling with situational context (disease avoidance as online purchase
COVID-19 moderated an antecedent state) intention
pandemic mediation analysis

25
Table II Sample characteristics
Gender: Age:
Male 37.9% Young Adults (below 26) 40.5%
Female 62.1% Adults (26-40) 29.9%
Education: Middle Age Adults (41-60) 26.7%
High School Diploma 7.3% Older Adults (over 60) 2.9%
Bachelor’s Degree 58.4%
Master’s Degree and Above 15.8% Occupation:
Others 18.5% Student 35.1%
Monthly income (Million VND): Private employee 26.7%
Below 5 33.2% Government employee 14.5%
From 5 to below 10 22.2% Self-employee 14.1%
From 10 to below 20 29.9% Housewife 3.7%
From 20 to above 14.7% Others 5.9%

Table III Means, standard deviations, and correlations

Construct Mean SD AVE eC eQuan SC BM Int
eC 5.295 1.141 0.658 0.811
eQuan 5.661 0.997 0.666 0.789*** 0.816
SC 6.260 0.778 0.575 0.502*** 0.631*** 0.758
BM 5.530 1.085 0.605 0.725*** 0.749*** 0.618*** 0.778
Int 5.809 0.968 0.677 0.613*** 0.704*** 0.742*** 0.686*** 0.823
Note: SD=standard deviation; AVE=Average Variance Extracted; Diagonal elements (bold)
indicate the square root of AVE between the constructs and their measures; *** p<0.001.

26
Table IV Factor loadings, Cronbach’s α, and composite reliability

Construct/ α CR Factor T-value

Items Loading
eWOM Credibility (eC) (from Yin et al., 2018) 0.909 0.906
The person who writes this posting is trustworthy 0.802 N/A
The person who writes this posting is sincere 0.777*** 24.791
***
The person who writes this posting is knowledgeable on this topic 0.830 21.616
***
The person who writes this posting is an expert on this topic 0.787 20.171
The person who writes this posting is experienced in this topic 0.857*** 22.512
eWOM Quantity (eQuan) (from Thomas et al., 2019) 0.888 0.888
There is a great number of reviews from different authors about the 0.846 N/A
product of my concern
There is a variety of reviews about the product of my concern 0.828*** 23.294
***
There is a multitude of information about the product of my concern 0.812 22.606
The number of online reviews about the product of my concern is high 0.776*** 21.111
Situational Context (SC) (from Brown et al., 2015) 0.870 0.871
I did something to maintain or improve my health 0.732 N/A
I avoided something gross or disgusting 0.775*** 17.316
***
I felt physically disgusted 0.765 17.100
I avoided things in public 0.761*** 17.010
I avoided touching people 0.756*** 16.902
Brand Image (BM) (from Iglesias et al., 2019) 0.882 0.884
The brand of my concern provides good value for money 0.803 N/A
There is a reason to buy the brand of my concern instead of others 0.804*** 20.651
***
The brand of my concern has a personality 0.740 18.568
***
I have a clear impression of the type of people who consume the brand 0.822 21.240
of my concern
The brand of my concern is different from competing brands 0.715*** 17.783
Online purchase intention (Int) (from Thomas et al., 2019) 0.893 0.894
I intend to buy a product about which I have read positive online 0.846 N/A
reviews
If somebody asks me for advice on buying a product, I would 0.832*** 23.407
recommend a product about which I have read online reviews
As a matter of principle, I inform myself before buying a product with 0.776*** 21.111
the help of online reviews
In the future, I will buy a product about which I have read online 0.837*** 23.608
reviews
Note: α=Cronbach’s alpha; CR=Composite reliability; *** p<0.001; N/A=Not available due
to constrained items for identification purposes in AMOS.

27
Table V Mediated model results
Hypothesised Direct Indirect Confidence Interval P-value
Conclusion
relationship effect effect Low High
H2a: 0.199 0.298 0.208 0.418 <0.001 Partial mediation
eC→BM→Int (4.094)
H2b: 0.477 0.300 0.182 0.460 <0.001 Partial mediation
eQuan→BM→Int (6.863)
Model fit statistics:
For eC→BM→Int: χ2=192.121,df=73,GFI=0.950,TLI=0.971,CFI=0.977,RMSEA=0.055
For eQuan→BM→Int: χ2=104.444,df=62,GFI=0.971,TLI=0.988,CFI=0.991,RMSEA=0.035
Note: Unstandardised coefficients reported. Values in parentheses are t-values. Bootstrap
sample=5000 with replacement at 95% bias-corrected confidence interval.

28
Table VI Moderated mediation model results
Unstandardised
Direct relationship T-value P-value Result
coefficient
eC→BM 0.376 5.989 <0.001
BM→Int 0.348 5.937 <0.001
eC→Int (0.047) (0.756) 0.45
SC→BM 0.307 5.185 <0.001
H3a: eC X SC→BM (0.193) (3.732) <0.001 Significant*
eQuan→BM 0.319 4.347 <0.001
eQuan→Int 0.457 6.433 <0.001
H3b: eQuan X SC→BM 0.168 2.786 0.005 Significant
Model fit statistics: χ2=429.367,df=173,GFI=0.929,TLI=0.959,CFI=0.966,RMSEA=0.052
Moderated indirect relationship Indirect effect Confidence Interval P-value
Low High
eC→BM→Int** 0.131 0.059 0.246 <0.001
Probing the interaction of SC
Low level: 0.183 0.086 0.328 <0.001
High level: 0.078 0.023 0.182 0.001
Index of moderated mediation (0.067) (0.130) (0.023) <0.001
eQuan→BM→Int** 0.111 0.049 0.207 0.002
Probing the interaction of SC
Low level: 0.065 (0.007) 0.160 0.071
High level: 0.156 0.073 0.281 0.001
Index of moderated mediation 0.058 0.008 0.142 0.017
Note: Unstandardised coefficients reported. Values in parentheses are negative. Bootstrap
sample=5000 with replacement at 95% bias-corrected confidence interval; * Significant with
negative coefficient; ** The indirect effect is moderated by the construct of situational
context.

29
 eWOM Brand
Image
H2a
eWOM
Credibility
H1a
H2b Online
Purchase
Intention
eWOM
Quantity H1b

H3b H3a

Situational
Context

Figure 1 Conceptual model

Note: The dash arrows represent the moderation effects; eWOM: electronic word-of-mouth.

30

View publication stats