Journal of Retailing and Consumer Services 68 (2022) 103066
Available online 9 July 2022
0969-6989/© 2022 Elsevier Ltd. All rights reserved.
Measuring the impact of online reviews on consumer purchase decisions – A scale development study
Semila Fernandes
a,b, Rajesh Panda
c, V.G. Venkatesh
d,*, Biranchi Narayan Swar
e, Yangyan Shi
f,g,**aSymbiosis Institute of Business Management, Bengaluru, India
bSymbiosis International (Deemed University), Pune, India
cXavier Institute of Management, XIM University, Bhubaneswar, India
dEM Normandie Business School, METIS Lab, France
eManagement Development Institute, Murshidabad, India
fSchool of Economics and Management, China University of Geosciences, Wuhan, China
gMacquarie Business School, Macquarie University, Sydney, Australia
A R T I C L E I N F O Keywords:
Consumer psychology Retailing
Scale development
Purchase decision. online reviews Word-of-mouth.
A B S T R A C T
Consumers’ exposure to online reviews influences their online retail shopping behavior. They search for reviews while evaluating products for purchase decisions. Past studies have indicated that online reviews affect the credibility and trust of the sellers and the products they sell on online platforms. Keeping this in view, the current paper aims to develop and validate a scale to understand the impact of online reviews on consumer purchase decisions. Data were collected from 431 young online shoppers for this research. The initial exploratory factor analysis (EFA) results helped identify four factors, viz. source credibility, volume, language and comprehension, and relevance which constitute the scale. The scale was validated by confirmatory factor analysis (CFA). The study’s findings fill the gap of having a standardized scale that online retailers can use as indicators to assist consumers in their online decision-making. The discussions and implications support consumers’ susceptibility to online reviews, an essential source for product and brand information in facilitating online consumers’ purchase decisions.
1. Introduction
In October 2020, research by Wall Street Journal revealed surprising factual statistics every business would want to know and the importance of online reviews (The Wall Street Journal, 2020). Firms need to capi- talize on their understanding of online reviews as online shoppers consider online reviews as channels of getting product information while making purchase decisions (Fu et al., 2020). Newspaper articles viz. business news daily in November 2021 also guided firms towards the dynamics of responding to an online review (Business News Daily, 2021). Unlike offline purchases, customers cannot touch or feel the product online and depend on consumer reviews (Schneider and Zielke, 2020). A Google study based on 57 million online customer reviews indicated that these reviews influenced consumers’ purchase choices
(Morrison, 2015). A ‘sales assistant’ assists a consumer by searching for product-attribute information; similarly, online customer reviews help consumers identify information about a product that best matches their needs (Schneider and Zielke, 2020). Most consumers like to go through other consumers’ opinions about retailers’ products or services before finalizing their purchase decisions.
The study is motivated by a specific behavior of consumers in the online environment - consumers spend more on products with excellent online reviews indicating the economic value of online reviews (Thakur, 2018). Extant literature underpins the importance of online reviews and identifies a few variables that consumers rely on while making online purchases (Li et al., 2020). Additionally, the context of this study is India which is an emerging market. India is the second-largest online market globally, with 622 million internet users, and it is projected that this
* Corresponding author.
** Corresponding author. School of Economics and Management, China University of Geosciences, Wuhan, China.
E-mail addresses: [email protected] (S. Fernandes), [email protected] (R. Panda), [email protected] (V.G. Venkatesh), drbiranchi.marketing@
gmail.com (B.N. Swar), [email protected] (Y. Shi).
Contents lists available at ScienceDirect
Journal of Retailing and Consumer Services
journal homepage: www.elsevier.com/locate/jretconser
https://doi.org/10.1016/j.jretconser.2022.103066
Received 26 February 2022; Received in revised form 22 May 2022; Accepted 23 June 2022
number will increase to 900 million by 2025 (IAMAI-Kantar, 2021). This shows that India will be a lucrative market for online retailers to sell their products and services with the growing online population.
Besides, 70 percent of customers refer to online reviews before finalizing their purchase decisions, and 63 per cent of customers are more likely to buy the product if it has higher product ratings and positive reviews (Rauschnabel et al., 2019; MacDonald, 2018). Young shoppers are the predominant group who buy online (Lissitsa and Kol, 2016; Singh et al., 2018). A recent study posited that a younger group of consumers looked for online reviews and agreed that online shopping was more convenient. Customers considered ‘reviews’ and ‘ratings’ as two essential sources of information (Shin and Darpy, 2020). Such re- views and opinions on the web and e-commerce portals have been referred to as e-WOM (electronic Word of Mouth).
Online reviews support consumers in gaining product-related infor- mation, thereby increasing their confidence during online purchasing.
These reviews are offered as an alternative by e-commerce firms to physically or visually interact with the product (Changchit et al., 2022).
Extant literature reveals that although users have been accustomed to online reviews, it remains far from deliberating how users perceive online reviews (Changchit et al., 2022). The motivation to share and post reviews is dependent on the consumer behavior and psychological aspects of an individual, which is more challenging to measure and observe (Iacobucci, 2010). Extant literature has attempted to identify individual constructs and variables that enhance e-WOM (Augusto and Torres, 2018). Researchers have separately analyzed the variables that explain the usefulness of online reviews, such as. quality, quantity, star ratings, credibility, reviewer identity (Filieri et al., 2020; Nieto-Garcia et al., 2019). However, studies suggest that individual attributes such as star ratings, review language, positive and negative reviews, recently received reviews, and style of writing reviews are not ideal indicators for customers’ perceived satisfaction and quality. At a macro level, the key challenge for retailers is to examine these attributes together in purchase decisions. Among the various research papers that have worked on on- line reviews, the methods adopted for measuring these individual items of online reviews are correlation (Racherla et al., 2013), Factor analysis, multiple regression & ANOVA (Schuckert et al., 2015), machine learning techniques (Leung et al., 2013) and questionnaire-based studies (Iaco- bucci, 2010). However, literature has recommended validating these attributes through a measurement model to build the credibility of on- line reviews (Schuckert et al., 2015). To the best of our knowledge, there is no such scale that comprehensively measures online reviews through a standardized scale (Schuckert et al., 2015; Kim and Song, 2018; Kyr- iazos, 2018). To study this gap, the current study endeavors to explore the following research question: Which reliable and valid indicators highlight the influence of online customer reviews? This research answers the question in three steps: identifying the research gaps and employing focus group discussions (FGDs) to generate new items and validate the items identified through literature, exploratory factor analysis (EFA) followed by confirmatory factor analysis (CFA) to systematically purify and validate the factor structure therebyproposing a new standard measurement scale.
The work contributes to the retailing domain in multiple ways.
Firstly, it develops and proposes a comprehensive measurement scale to measure online reviews. Secondly, this pioneering work in consumer retailing proposes new items such as ‘trusting the reviews from a verified customer’ and ‘the style of writing reviews’ that impact customers’
product purchase journeys. This will improve the measurement of online reviews in marketing. This robust, validated scale helps analyze the broad items and fine-tune them to reflect a few specific dimensions. This scale will enable new theoretical underpinnings and help researchers compare the scale across cultures and markets.
The paper has six sections. Section 1 includes the introduction;
Section 2 includes the theoretical and literature background. Section 3 elucidates the methodology and the scale development procedure. Our empirical scale development and validation process comprise three
studies that confirm the reliability and validity of the instrument.
Finally, we present how marketers can apply for online reviews in our relevant discussions in Section 4. Section 5 contributes to the study, including theoretical and managerial implications. Section 6 concludes with limitations and future research agenda.
2. Theoretical and literature background 2.1. Electronic-word of mouth and online reviews
Word-of-Mouth (WOM) is a dominant force in influencing con- sumers’ decisions. Electronic-WOM (e-WOM) is gaining prominence and is recognized as the new communication medium (Choi and Maasberg, 2021). e-WOM refers to informal means of communication directed to- wards consumers adopting internet-based technology, which compre- hends various websites and media forms, including online customer reviews (Gottschalk and Mafael, 2017). These consumer-generated on- line reviews for products and services are increasing in importance and popularity (Lee and Choeh, 2018). Research has indicated that e-WOM is frequently generated through social media and online shopping plat- forms. Research scholars have confirmed that these platforms of e-WOM influence purchase decisions (Duarte et al., 2018). e-WOM in online shopping platforms is generated through online reviews by existing buyers. These online customer reviews are recognized as one of the pivotal forms of e-WOM. It has also been seen that when consumers are making product decisions, they rely more on online reviews (Choi and Maasberg, 2021). Moreover, consumers look for the total number of online reviews as qualifying items, reflecting upon the popularity and awareness of that product or service (Chen et al., 2017).
Consumer-generated online reviews gauge the persuasion of e-WOM with a specific focus on the perceived effectiveness and trustworthiness of reviews (Srivastava and Kalro, 2019).
2.2. Attributes of online reviews and consumer purchase decisions Online content in the form of online reviews is becoming increas- ingly popular in Internet-Based Marketing (Choi and Maasberg, 2021).
Online reviews significantly influence consumers’ purchase intention (Huang et al., 2019). Reviews usually reflect upon arguments that the consumers categorize into pros and cons or positive arguments and opposing arguments (Risselada et al., 2018). Negative reviews discon- firmation has more significant and stronger effects than positive reviews disconfirmation (Li et al., 2020). In addition to the number of online reviews, the type of reviews also impacts f consumers, potentially yielding better purchase intention results. Studies showed that online reviews generated through internet forums are perceived to be much more credible and trustworthy than the corporate websites generated by the marketer. Consumers go through online reviews across multiple sites and internet forums (Thakur, 2018). These number/quantity of reviews and their types/quality positively influence the credibility and trust of the seller/product (Ismagilova et al., 2020). As the number of reviews increases in the online environment, obtaining specific information becomes difficult, and hence consumers look out for heuristic cues like the star ratings to simplify their search and evaluation process (Yi and Oh, 2022; El-Said, 2020).
Recent/current reviews reflect upon the evidence about the products and services, thereby showcasing higher credibility (Shareef et al., 2019). The study conducted by Shaheen et al. (2019) found that the usefulness of online customer reviews and credibility induces the adoption of reviews and the propensity to trust online retailers. Simi- larly, consumers believe negative reviews to be much more authentic than positive reviews, even in online buying behavior. Negative reviews have a more substantial influence on consumers’ assessment of services or products and purchase intention against a positive message (Weis- stein et al., 2017). Research has also indicated that the attributes of online reviews such as the richness of the review, review ratings, their
relevance (Nieto-Garcia et al., 2019), and the reviewers’ identity dis- closures and their level of expertise (Munzel, 2016; Chen et al., 2017) depict positive influence towards an online purchase decision. Re- searchers have also discovered that consumers value the online gener- ated reviews only when the reviewer has an experience of using the product or service (Stein and Ramaseshan, 2016). Empirical studies have also posited how online reviews are written, which subtly affects consumers’ purchase decisions (Dixit et al., 2019). The semantic con- tent, language, and style of writing consumer reviews influence online consumer sales. Linguistic style and Content are inseparable and rein- force the impact of online reviews, thereby making the review clearer and more unambiguous, and appealing to the reader (Stein and Ram- aseshan, 2016).
The literature supported the relationship between pre-purchase in- formation (online reviews) and purchase decisions (Saha and Sahney, 2022). Furthermore, the data collected from inbound tourists in China also reflected the influence of online reviews on the sale of consumer products (Siddiqi et al., 2020). The studies also recognized the role of online reviews, both valence and volume, in creating trust among con- sumers while shopping. The study revealed that reviews enhance the effect of a positive summary review on trust (Sebastianelli and Tamimi, 2018). In the hotel industry, the positive valence rate of reviews and the number of reviews significantly impact online hotel booking (Fu et al., 2021). Studies have also demonstrated the impact of online reviews on travelers’ online hotel booking intentions. Researchers identified mul- tiple attributes of online reviews, such as usefulness, reviewer expertise, timeliness, volume, and valence (negative and positive) (Zhao et al., 2015). The results revealed a significantly negative relation between negative online reviews and online booking intentions. However, posi- tive online reviews upon booking intentions in isolation might not in- fluence purchase, and hence there is a need to explore additional attributes that drives purchase decision (Zhao et al., 2015). Moreover, studies posited that an increase in online reviews alone has no signifi- cant impact on selling online (Davis and Khazanchi, 2008). In the case of online shopping, negative online reviews impact consumers’ purchase decisions more than positive reviews (Jin, 2007).
Consumers can evaluate better when reviewers indicate their per- sonal identity/real name with a photo (Kim, 2020; Mariani and Pre- dvoditeleva, 2019). A positive relationship between reviewer expertise and people’s purchase intentions was observed in extant literature (Tan et al., 2008). It has thus been important for reviewers to fully identify their individuality while posting online content to qualify the source as credible, although it is not mandatory (Ismagilova et al., 2020). Profile photos of the reviewers also help improve the credibility of their reviews (Kim et al., 2020). Thus, consumers’ purchase decisions are influenced when there is an interaction between the facial expression of the reviewer and the content of online reviews, which provides a strength of intention to purchase the product (Heng et al., 2018). Table 1 provides a summary of recent notable works.
Overall, the above review establishes that online reviews (a promi- nent form of e-WOM) are an essential force that influences consumers’
online purchases. Consumers frequently refer to online reviews and are primarily dependent on the type of responses shared by the reviewers.
The researchers indicated that the reviewer’s expertise, star ratings, length of reviews, the credibility of information, reviewers’ behavior, and the reviewer’s identity are the key factors that impact potential evaluations of consumers’ purchase journey. However, such variables have not yet been grouped to constitute a scale for measuring online reviews. This study identifies relevant variables and develops a mea- surement model to propose a scale for online reviews based on the literature review.
3. Research methodology 3.1. Scale development procedure
The research article adopts recognized procedures of Churchill (1979) and Brod et al. (2009) to construct the measurement scale: a qualitative and quantitative study involving a consumer survey followed by a purification study and data validation. Fig. 1 below depicts the research scheme adopted for the scale development procedure.
Table 1
Summary of recent notable works.
Sl. No Authors/Year Concept Key Findings
1 Choi & Maasberg
(2021) Effects of product
reviews Online reviews had a
persuasive impact on consumers’ processing of information.
2 Hsieh and Li
(2020) Susceptibility towards
Source Credibility Trustworthiness, fairness, and justifiability of the source were significant.
3 Mariani. and Predvoditeleva (2019)
Online reviewers’
behavior Review ratings, review comments, helpful votes, and length of the textual review positively impact review continuity and frequency.
4 Mariani and Predvoditeleva (2019)
Type of Reviewers Helpful reviewers are those who post reviews and actively travel more.
5 Nieto-Garcia et al.
(2019) Reputation and
expertise of the Reviewer, star ratings, and length of reviews
Message and messenger determine review usefulness.
6 Heng et al. (2018) Social media platforms as a source of knowledge on brands
Results indicated that review quality and source were essential items influencing the credibility of the consumer towards assessment of online review.
7 Hajli (2018) Information credibility Online WOM was found to be a credible and helpful piece of information.
8 Chen et al. (2017) Identity of reviewers on the social media platform
Quantitative variables of reviews, viz. overall review ratings, hotel stars, reviewer identity, etc., are helpful on social media.
9 Chen et al. (2017) e-WOM and association-based strategy for corporate posting on social media
When e-WOM valence was positive, corporate posting on consumers’ CSR associations was significant.
10 Hsu et al. (2017) Online reviews on
purchase intention Lower purchase intention was observed when subjects were exposed to harmful online customer reviews.
11 Zhao et al. (2019) Online textual reviews Length of the reviews reduces customer satisfaction levels.
12 Weisstein et al.
(2017) Negative online WOM
communication Consensus in online negative WOM communication impacts potential evaluations of the firm.
13 El-Said (2020) Responding to single
star reviews A complaint framework was developed for lower- star rating reviews.
Several inquiry types are imperative to develop generalizability and improve the scale’s validity (Spake et al., 2003). On that recommen- dation, the present study focuses on three phases as below:
The first phase – Qutative Study: Variables to be part of the study were identified through literature. This was followed by Four mixed-sex Focus Group Discussions (FGDs) with 10–12 shoppers each, which helped explore variables and check the applicability of variables identified through literature.
The second phase – Purification: Exploratory Factor Analysis (EFA), was conducted with 380 respondents, which resulted in four- dimensional factors – Source Credibility, Volume, Language Comprehen- sion, and Relevance. The same factor structure was confirmed through Confirmatory Factor Analysis (CFA).
The third phase– Validation: CFA was administered with a new set of 278 participants, and the 17-item scale termed ‘Online Review’ was validated.
3.2. Qualitative inquiry approach
The qualitative inquiry approach was followed by reviewing extant literature and involving focus group discussions (FGDs). This approach helped in item generation. The literature-generated item pools are re- flected in the statements below: Studies have indicated that consumers pay attention to online customer reviews (Kim and Song, 2018). These online generated reviews reflect upon the arguments that shoppers categorize into positive and negative arguments based on the pros and cons of the reviews (Risselada H. et al., 2018). Additionally, the number of reviews is considered anchored by consumers who provide aggregate review content rather than individual reviews (Grewal et al., 2020). It is not only about the number of reviews, but consumers search for heu- ristic cues in the form of star ratings that will simplify their search and evaluation process (El-Said, 2020; Herhausen et al., 2015).
On another side, extreme views depicted through ratings (5-star or 1- star) matter to the shopper when evaluating their choices (El-Said, 2020). In comparison, recent or current reviews are more credible in providing evidence about the products they search for (Shareef et al., 2019). Research has posited that consumers value those online-generated reviews reviewed based on their product use and experience (Ahani et al., 2019). Studies have reported that when re- viewers reveal their identity or their real name and photo, consumers are better positioned to evaluate their product and service choice as they get connected and relate to the overall experience (Munzel, 2016).
Consumers get positively influenced by these attributes of online customer reviews where reviewers’ identity disclosures and level of expertise are revealed (Mariani and Predvoditeleva, 2019). Literature has depicted that consumer weighs negative messages much more than positive messages in their evaluation criteria (Weisstein et al., 2017).
Hence, the sellers and managers need to respond immediately, effec- tively, and authenticate these negative messages to generate positive
customer loyalty (El-Adly, 2019). Researchers have argued that in addition to the semantic content of the messages, linguistic styles also shape shoppers’ choice decisions (Tran and Strutton, 2020).
Further, clear and unambiguous reviews appeal to the readers, reinforcing the influence of the reviews (Stein and Ramaseshan, 2016).
Finally, shoppers go through online customer reviews across multiple websites, chats, and internet forums to support their purchase decisions (Goes et al., 2014). As prompted by extant studies, the factors that in- fluence consumers’ online shopping environment are stated below:
1. I look for reviews that describe the benefit/problem of the product (Kawaf and Istanbulluoglu, 2019).
2. I look for reviews at multiple sites to confirm the review scores (Goes et al., 2014).
3. The average score/star rating is essential for a product (El-Said, 2020).
4. I prefer that the reviews included are appealing and straightfor- ward (Tran and Strutton, 2020).
5. I prefer that reviews supported by relevant arguments are critical (Risselada H. et al., 2018).
6. If the number of reviews are less, I hesitate to consider the re- views while buying the product (Grewal et al., 2020).
7. Negative reviews influence more than positive reviews (Weisstein et al., 2017).
8. I trust the information more when I believe the reviewer has adequate knowledge/expertise of the product (Vermeulen and Seegers, 2009).
9. I look for extreme views (5-star or 1-star) while evaluating a product (Sen and Lerman, 2007).
10. The reviews are more credible if the name/image of the reviewer is available (Munzel, 2016).
11. I believe reviews impact my decision to purchase or not to pur- chase a product (Kim and Song, 2018; Hsu et al., 2017).
12. I look for the seller’s response instead of the customers’ reviews (Munzel, 2016).
13. The review is more helpful if the message is clear and unambig- uous (Lu et al., 2018).
14. Recent reviews are more important than older ones (Lee et al., 2017).
15. I believe the review is more credible when the reviewer is the product’s user (gender/age-specific products) (Mariani and Pre- dvoditeleva, 2019).
3.3. Assessment of face validity
Face validity is recognized as a practice of editing and generating items by involving an expert panel. It enhances the correctness/appro- priateness of each item before getting into the content validity stage (Hardesty and Bearden, 2004). Face validity step of 15 items involved a Fig. 1.Research Scheme for Scale development.
panel of experts comprising twelve industry experts (having experience from 10 to 22 years) and three marketing researchers. It involved a Delphi process to help establish the relevance of items. All the experts agreed on the identified variables. Eighty percent of them confirmed the items and the impact of online recommendations on the consumer purchase journey. They indicated the need to classify items to measure online reviews’ overall significance in product sales. Further, experts recommended focused group discussions (FGDs) to gain more insights into variables and examine the content validity.
3.4. Focus group discussions (FGDs) – overview, characteristics of participants, and methodology
Four mixed-sex FGDs were held with 10–12 shoppers each. These FGDs helped to have a deeper understanding on the research topic and identify variables not addressed in the literature. The members for the FGDs were selected using the convenience sampling method by visits to universities, shopping complexes, and housing associations after taking appropriate consent from the establishments of these sites. Each focus group had the basic parameters defined based on gender, education, marital status, and occupation covering a good mix of students, business professionals, and those in service (blue-collar and white-collar jobs) aged 18 and 45 years. This young segment was chosen because it is the critical audience for e-tailers to understand online shopping behavior (Arul Rajan, 2020). The total participants were 44, of which 25 were female and 19 were male.
Discussions covered the following main topics regarding e-shopping and online reviews:
•Factors that help influence online shopping
•The convenience of using online reviews/online texts
•Impact of reviews on actual product purchases
•Details on how consumers evaluate online reviews
•Sentiments of online reviews
In the open discussion, the group members provided broad concerns over the underlying online reviews, their desires and interests, and sug- gested strategies they would recommend or avoid when purchasing goods and services in the online shopping channel. The participants also deliberated on additional insights on consumers’ behavioral mindset toward referring to online reviews during e-shopping beyond the topics mentioned above. More than half of the participants (59 percent) would prefer to trust online reviews if a credible source has contributed to them. Around 68 percent of the participants expressed their belief in reviews written unbiasedly. Regarding the type of reviews, 73 percent of participants reported that the top positive and negative reviews would help them evaluate their online product. Additionally, half of the par- ticipants (50 percent) were inclined towards an online product purchase if the language and style of writing reviews were precise. Such feedback guided the authors to a reality of the consumer perspectives and iden- tified four specific items. Through this exploratory phase, the following four items/variables were recognized and were appended to the list of 15 items identified through literature:
16. I believe the style of writing reviews should be unbiased or impartial.
17. The information or message in the topmost negative reviews and topmost positive reviews are helpful in my purchase decision.
18. I trust the review only if it comes from a verified customer (The online retailer mentions he/she is a verified customer).
19. The language style of reviews must be precise rather than metaphorical.
In the next stage, these 19 items generated through extant literature and substantiated by FGDs were administered using a survey instrument involving a 5-point agree/disagree Likert scale among the study
respondents to assess the dimensionality of the items identified.
3.5. Scale purification approach
A scale-purification framework is a widely accepted approach in empirical consumer behavior studies that distinguishes the dimension- ality of the constructs and item levels using statistical criteria. Those criteria assess the quantitative data involving standardized techniques, including exploratory factor analysis and comparing the cut-off values of alpha coefficient; and confirmatory factor analysis (Wieland et al., 2017).
3.6. Stage 1: Exploratory Factor Analysis (EFA)
EFA data was collected from 431 consumers aged 18–45 years in Bangalore, Karnataka. Bangalore was chosen because of the rapid growth of online shopping (Joshi et al., 2021). Out of 431 responses, 380 were retained for the final analysis; 51 missing and inconsistent re- sponses were eliminated. The sample size of 380 was appropriate, especially for EFA with factor loadings greater than equal to 0.50 and no cross-loadings and reliability of greater than equals 0.70. Convenience samples were identified through housing complexes, shopping malls, and educational campuses. The qualifying question to the respondents was: “What forms of online customer reviews do consumers rely on during product purchase decisions?”. This confirmed that the sample has knowledge and awareness of online reviews and identifies the variables related to online customer reviews in the consumer decision-making journey. The same approach was adopted for selecting and approach- ing respondents online. A self-administered questionnaire collected the data for both offline (310) and online participants (121). The ques- tionnaire contained 19 items, and the participants appraised the items on a 5-point Likert-type, i.e., agree/disagree response scale, which was deemed suitable for self-administered online questionnaire studies (Hair et al., 2010). This Likert-scale design ranges from agreement or disagreement with level of quality 5 =Strongly Agree, 4 =Agree, 3 = Neutral, 2 =Disagree and 1 =Strongly Disagree.
The 5-point response scale is the simplest of all, which allows the interviewer to complete reading the statements or collecting informa- tion on all the scale descriptors in a short period and provides flexibility to respondents for the midpoint (option 3) if they are unsure of the Table 2
Demographic profile of the respondents.
Sl. No Description Frequency %
1 Gender Male 241 63.42
Female 139 36.57
2 Marital Status Married 248 65.2
Unmarried 132 34.7
3 Age (in years) 18–22 63 16.58
23–27 135 35.53
28–32 32–36
>36a
87 61 34
22.90 16.05 4 bMonthly Income (USD = 8.9
United States dollars) USD, 334 88 23.2
USD, 669 111 29.2
USD, 1,338 130 34.2
>USD, 1,338 51 13.4
5 Occupation Student 53 13.94
Business 27 7.10
Service 259 68.15
Others 41 10.79
6 Education Graduate 112 29.47
Post-Graduate 234 61.58
Postgraduate &
above 34 8.95
a Less than or equal to 45 years.
b USD equivalent to Indian Rupee.
response. A 5-point Likert-scale design is also a good measurement model with significantly higher reliability (Adelson and McCoach, 2010). The demographic profile of the respondents is mentioned in Table 2.
The data received on the 19 items from the participants were analyzed through exploratory factor analysis (EFA). They were verified for reliability using the established principal axis factoring procedure with varimax rotation and the eigenvalue method, which helped deter- mine the list of factors (Hair et al., 2010). Varimax rotation, a common form of Principal component analysis or factor analysis, resulted in four factors with factor loadings of 0.5 and higher being considered practical and acceptable for the analysis (Hair et al., 2010). Upon evaluation, the items with loadings less than 0.5 and did not adequately load on either of the factors were removed. Two items, the information or message in the top positive and top negative reviews are helpful in my purchase decision (Item 17), and the language style of reviews must be precise rather than metaphorical (Item 19), were dropped from the study because of low factor loading in EFA (with factor loadings 0.5 and lower). The results of the EFA analysis involving the 17 items led to the identification of a four-factor structure. The four extracted factors can be named Source Credibility, Volume, Language Comprehension, and Relevance. The item-factor relationship resulted in forming factor 1 having items viz. I trust the information more when I believe the reviewer has adequate knowl- edge/expertise of the product (Item 8); I believe the review is more credible when the reviewer is the user of the product (gender/age-specific products) (Item 15); I trust the review only if it comes from a verified customer (The online retailer mentions he/she is a verified customer) (Item 18); The reviews are more credible if the name/image of the reviewer is available (Item 10); I look for reviews at multiple sites to confirm the review scores (Item 2); I believe the reviews that are supported by relevant arguments are critical (Item 5) and I look for reviews those describe the benefit/problem of the product (Item 1).
Further, the above items indicate that online customer reviews are perceived to be much more credible and trustworthy; hence the construct is Source Credibility. The item-factor relationship resulted in factor 2 having items viz. If the numbers of reviews are less, I hesitate to consider the reviews while buying the product (Item 6); I look for extreme views (5-star or 1-star) while evaluating a product (Item 9), and the average score/star rating is essential for a product (Item 3). As these items validate that the numbers of customer reviews/star ratings of online reviews are the qualifying criteria that simplify consumer’s search and evaluation process, the authors chose to identify this factor as Volume. The item- factor relationship resulted in containing factor 3 having items viz.
The review is more useful if the language of reviews is simple and appealing (Item 4); The review is more helpful if the message is clear and unambiguous (Item 13), and I believe the style of writing reviews should be unbiased or impartial (Item 16). These items provided evidence that the language and the semantic content of reviews appeal to the user towards online con- sumer sales, and hence we chose to recognize the factor as Language and Comprehension. The item-factor relationship resulted in forming factor 4 having items viz. I look for the response of the seller in lieu of the customers’
reviews (Item 12); Negative reviews influence more than positive reviews (Item 7); I believe reviews impact my decision to purchase or not to purchase a product (Item 11) and Recent reviews are more important than older ones (Item 14). These items indicate that consumers relied on the positive/
negative type of customer reviews to reinforce their online purchase choice, and therefore the authors decided to term this factor as Relevance.
3.7. Stage 2: confirmatory factor analysis (CFA)
In Stage 2, the empirical results of EFA were confirmed using an iterative process involving Confirmatory Factor Analysis (CFA) fit indices (Anderson and Gerbing, 1988) and chi-square testing (Voss et al., 2003). CFA for the 17 items was conducted using the IBM SPSS AMOS program which tests the factorial validity of the measuring instrument
(Gallagher et al., 2008; Kline, 2015). Empirical measures of the model estimation were evaluated through internal consistency reliability, convergent and discriminant validity as proposed by Hair et al. (2010) in their book Multivariate Data Analysis. The factor loadings having values greater than 0.5 are considered practical and acceptable for the analysis (Hair et al., 2010). The results of CFA and reliability values are depicted in Table 3. The Cronbach’s α(alpha) value is 0.962. The three items loading on factor 2 (Volume) have a Cronbach’s α (alpha) value of 0.959.
The three items loading on factor 3 (Language and Comprehension) have a Cronbach’s α (alpha) value of 0.977, and the four items loading on factor 4 (Relevance) have a Cronbach’s α value of 0.959 (additional informa- tion is illustrated in Table 3), which were considered acceptable (α >
0.70) (Nunnally, 1978; Kline, 2011) and hence confirming the reliability of scale dimensions.
The validity on the construct level was statistically examined by computing discriminant validity and convergent validity (Fornell and Larcker, 1981). Convergent validity was assessed, and the reliability at a level greater than 0.7 was identified as significant (as depicted in Table 4) for all items of the measurement model. Model evaluation was also conducted through discriminant validity, where the correlation between the factors was measured, and all the conditions were met as per the recommendations for use by Fornell and Larcker criteria, 1981 (as depicted in Table 5). To support the Fornell and Larcker criterion for discriminant validity, the heterotrait-monotrait ratio of correlations (HTMT) was computed. HTMT is a measure of similarity between latent variables (Henseler et al., 2015). If the HTMT is less than 0.9, then discriminant validity can be regarded as established (Horstmann, 2017), which confirms the authenticity and validity of the measurement model (Quoquab and Mohammad, 2020) (Refer to Table 6).
The measurement model consisting of the 17 variables/items iden- tified through the item-generation stage offers an acceptable model-fit index as recommended by the thresholds proposed by Hair et al.
(2010) and Byrne (2010), 2013; Hu and Bentler (1999) (as depicted in Table 7). Thus, the authors finalized and proposed the Online Customer Review (OCR) Scale, as elucidated in Table 3.
3.8. Scale validation approach
The measurement model was validated using the same procedure involving the CFA technique on an independent and new data set comprising 278 respondents. The loadings of the factors for the purifi- cation and validation phase are summarized in Table 8. The scale’s reliability was assessed through the coefficients of Cronbach’s alpha.
The coefficients of the measurement scale were computed as 0.951 for factor 1 (Source Credibility), 0.953 for factor 2 (Volume), 0.955 for factor 3 (Language and Comprehension), and 0.936 for factor 4 (Relevance) (as indicated in Table 9), which met the acceptable level of being more significant than 0.70. This established the reliability of the scale di- mensions. The measurement scale also showed acceptable convergent validity (see Table 9) and discriminant validity threshold values (see Table 10). The HTMT values also supported the establishment of discriminant validity (see Table 11). The empirical findings of CFA confirmed the four-factor structure validation of the proposed. The goodness of fit statistic fit provided a good fit (see Table 12). The results of the validation structure conducted using the CFA technique indicated that the model fit was adequate and has achieved a satisfactory model fit with field data (Somers et al., 2003).
4. Discussion
The present study aims to propose a structured scale for online re- views. Such a scale is necessary as online reviews have significantly influenced product and brand sales (Li et al., 2020). With the advent of online retailing and various shopping platforms, consumers are now shopping online and looking for opinions of other shoppers posted on online shopping sites, platforms, and blogs (Fernandes et al., 2021; Micu
Table 3
Confirmatory factor analysis (CFA).
Sl. No. Items/Variables Sources* Mean
score Standard
Deviation (SD) Factor
Loadings Reliability (Cronbach alpha)
Factor 1: 0.962
Source Credibility
1 I trust the information more where I believe the reviewer has
adequate knowledge/expertise for the product. Hsieh & Li (2020) 3.2895 1.00678 0.846 2 The review is more credible when the reviewer is a product user
(gender/age-specific products). Mariani & Predvoditeleva
(2019) 3.3974 1.10527 0.788
3 I trust the review only if it comes from a verified customer (The
online retailer mentions he/she is a verified customer). Exploratory Study 3.3211 0.96212 0.964 4 The reviews are more credible if the name/image of the reviewer is
available. Munzel (2016) 3.3237 1.01854 0.879
5 I look for reviews at multiple sites to confirm the review scores. Goes et al. (2014) 3.2763 1.03244 0.882 6 I believe the reviews that are supported by relevant arguments are
important. Risselada H. et al., 2018 3.2289 1.06650 0.888
7 I look for reviews that describe the benefit/problem of the product. Munzel (2016) 3.3500 0.96979 0.976
Factor 2: 0.959
Volume
8 If the reviews are fewer, I hesitate to consider the reviews while
buying the product. Grewal et al. (2020) 3.5842 0.89606 0.907
9 I look for extreme views (5-star or 1-star) while evaluating a
product. Sen & Lerman (2007) 3.5474 0.92766 0.933
10 The average score/star rating is important for a product. El-Said, 2020 3.6000 0.93481 0.986
Factor 3: 0.977
Language and Comprehension
11 The review is more useful if the language of reviews is simple and
appealing. Tran and Strutton, 2020 3.3263 1.09874 0.952
12 The review is more helpful if the message is unambiguous. Stein and Ramaseshan,
2016 3.3447 1.11104 0.969
13 I believe the style of writing reviews should be unbiased or
impartial. Exploratory Study 3.3500 1.11886 0.980
Factor 4: 0.959
Relevance
14 I look for the seller’s response in lieu of the customers’ reviews. Munzel (2016) 3.4816 0.92584 0.798 15 Negative reviews influence more than positive reviews. Weisstein et al. (2017) 3.5184 0.93152 0.962 16 I believe reviews impact my decision to purchase or not to purchase
a product. Kim and Song, 2018; Hsu
et al., 2017 3.5816 0.92241 0.958
17 Recent reviews are more important than older ones. Lee et al. (2017) 3.5289 0.93972 0.979
(*The items/variables have been aptly modified based on a qualitative study, which involved discussions with the industry/subject matter experts and participants of the study).
Source: Contribution by authors.
Table 4
Convergent Validity Estimates (Purification phase: 1st stage data collection).
Constructs Composite Reliability
(CR) AVE Reliability Convergent
Validity
SC 0.962 0.794 ✓ ✓
V 0.959 0.888 ✓ ✓
LC 0.977 0.935 ✓ ✓
R 0.959 0.860 ✓ ✓
Note: As per the recommended value for convergent validity, CR must be 0.70 or higher, the Average Variance Explained (AVE) must have the cut-off value of 0.50 or higher, and the CR value must be greater than AVE (Hair et al., 2010).
Table 5
Discriminant Validity Estimates (Purification Phase: 1st stage data collection).
Latent Constructs AVE SC V LC R
SC 0.794 0.891
V 0.888 0.186 0.943
LC 0.935 0.309 0.351 0.967
R 0.860 0.205 0.303 0.765 0.927
Note: The values/elements across the diagonal, indicated in bold, are the √AVE, and the off-diagonals are the correlations between constructs.
Table- 6
HTMT computations.
HTMT Values
V 0.169
LC 0.317 0.334
R 0.219 0.295 0.770
SC V LC
Table 7
CFA results: Measuring model fit (purification phase).
Indicators aCFA
Results Threshold
Value Source
CMIN/df 2.181 Three or lower Hair et al. (2012); Byrne (2010) GFI 0.934 0.90 or higher Somers et al. (2003)
NFI 0.972 0.90 or higher Hair et al. (2012); Hu and Bentler (1999)
SRMR 0.0259 0.08 or lower Kline (2011) RMSEA 0.056 0.08 or lower Brannick (2003) CFI 0.985 0.90 or higher Hu and Bentler (1999) NNFI (TLI) 0.982 0.90 or higher Hu and Bentler (1999) AGFI 0.910 0.80 or higher Byrne (2010, 2013) a The model fit indicators are within the acceptable threshold values.
et al., 2019). These reviews’ exposure and subsequent influence have also been referred to as e-WOM (Shankar et al., 2020). In the absence of face-to-face interaction, customers evaluate the reviews based on their perceived credibility (Heng et al., 2018). Literature suggests that consumer-generated online reviews are gaining importance and popu- larity (Lee and Choeh, 2018) and influencing purchase decisions (Duarte, e Silva & Ferreira, 2018). Consumer endorsements on online platforms are trustworthy compared to seller-generated information (Hajli, 2018). However, there are many factors relating to online re- views, like contributing to writing online reviews (Thakur, 2018), several reviews (Grewal et al., 2020), star ratings (Li et al., 2020), reviewer expertise (Vermeulen and Seegers, 2009), quality of reviews (Risselada et al., 2018) etc. that impact the review credibility. Hence the current study establishes that customer reviews are pivotal in online purchases by studying various variables that contribute to the reliability of the reviews and proposes a scale to describe the elements of online reviews and their impact on the consumer purchase decision. The research adopted the scientific and systematic approach to developing the scale using 17 items. These 17 items were identified through liter- ature and refined through Focus Group Discussions and Factor Analysis.
Based on the CFA output and subsequent validation, the study proposes an Online Review Scale corroborating four constructs: Source Credibility with seven items, Volume with three items, Language and Comprehension with three items, and Relevance with four items.
4.1. Source credibility
Source Credibility refers to the trust customers attach to the review’s writer (Hsieh and Li, 2020). Online platforms have multiple sellers of the same product at different prices. Such platforms allow customers to share feedback, grievances, and appreciation of the products and ser- vices that customers purchase online. Online shoppers trust a review if they believe the author of the review is an actual user of the product and if she/he has mentioned the benefits or the problems of the product (Ismagilova et al., 2020). The reliability is higher if the reviewer is a user of the product rather than merely the buyer (Thakur, 2018). This sug- gests that if a product is meant for women, a review by a man will seem less reliable to the prospective customers even if he is the actual pur- chaser of the product. Customers also check the reliability of the re- viewers, whether she/he is verified customers by the e-retailer or the platform (Mariani and Predvoditeleva, 2019). This reduces the possi- bility of robots or fake customers providing reviews of the products where they are not the actual buyers. Consumers’ reliance on online reviews is also based on the reviewer’s reputation and expertise. If the review provides valid information and product benefits and reflects the expertise or knowledge of the reviewer, then customers trust such re- views (Li et al., 2020). The construct Source credibility with seven items forms an essential part of the OCR Scale.
4.2. Volume
Though unique review content is helpful, the total number of reviews works as an anchor by providing aggregate review content (Grewal et al., 2020). Hence, in addition to reliable reviews, online shoppers also look for the Volume (number) of reviews for a particular product on the site. Though the review length may or may not facilitate consumers in their decision-making journey (Li et al., 2020), the total number of re- views helps them evaluate the product and trust the overall score if the number of reviews are more. Moreover, positive or negative reviews do influence the conversion to product sales. Positive reviews evoke active consumer purchase decisions (Weisstein et al., 2017). Even if the review is positive or has emotional content (Guo et al., 2020), if the number of reviews is perceived to be less, customers are averse to accepting it (Yi Table 8
A comparison of Factor Loadings in Purification and Validation Phases.
Purification phase Validation phase
SC1←SC 0.846 0.828
SC2←SC 0.788 0.735
SC3←SC 0.964 0.950
SC4←SC 0.879 0.836
SC5←SC 0.882 0.872
SC6←SC 0.888 0.856
SC7←SC 0.976 0.966
V1←V 0.907 0.911
V2←V 0.933 0.903
V3←V 0.986 0.984
LC1←LC 0.952 0.954
LC2 ← LC 0.969 0.939
LC3 ← LC 0.980 0.914
R1 ← R 0.798 0.701
R2 ← R 0.962 0.946
R3 ← R 0.958 0.939
R4 ← R 0.979 0.970
Table 9
Convergent Validity Estimates (Validation Phase: 2nd stage data collection).
Latent
Constructs Composite
Reliability (CR) AVE Reliability Convergent Validity
SC 0.951 0.750 ✓ ✓
V 0.953 0.871 ✓ ✓
LC 0.955 0.876 ✓ ✓
R 0.936 0.802 ✓ ✓
Table 10
Discriminant validity estimates (validation phase).
Latent Constructs AVE SC V LC R
SC 0.750 0.866
V 0.871 0.216 0.933
LC 0.876 0.295 0.369 0.936
R 0.802 0.222 0.348 0.733 0.896
Note: The values indicated across the diagonals (identified in bold) are the
√AVE, and the off-diagonals are the correlations between the constructs.
Table 11
HTMT values (validation phase).
HTMT Values
V 0.196
LC 0.295 0.355
R 0.240 0.346 0.737
SC V LC
Table 12
CFA results: Measuring model fit (validation phase).
Indicators *CFA Results Cut off Criteria
AGFI 0.886 0.80 or higher
CMIN/df 2.096 Three or lower
NFI 0.957 0.90 or higher
GFI 0.916 0.90 or higher
RMSEA 0.063 0.08 or lower
NNFI (TLI) 0.972 0.90 or higher
SRMR .0335 0.08 or lower
CFI 0.977 0.93 or higher
(* All the model fit indicators are as per the acceptable threshold/cut off values).
and Oh, 2022). This imposes that a few good reviews may not be suf- ficient to influence customer decisions. If the number of reviews is more, customers tend to associate more trust. While evaluating reviews, it may be easy to look at the average score/rating to make an overall impression of the product (Rauschnabel et al., 2019). However, customers are not persuaded by the average score only, and they look for extreme positive and negative views (Cao et al., 2011). A few extreme negative views can dissuade a customer from buying a product with an acceptable average rating (Li et al., 2020). Consumers consider review ratings as anchors to minimize the risks associated with choosing the right product (Yi and Oh, 2022). Consumers also find the number of reviews to be ‘helpful’ as they tend to derive aggregate information from all the reviews (Grewal et al., 2020; Yi and Oh, 2022).
4.3. Language and comprehension
In the absence of face-to-face contact with a reviewer, the language and writing style influence the review’s trustworthiness (Stein and Ramaseshan, 2016). The third construct identified in the OCR Scale is Language and Comprehension. Customers rely more on the reviews where the language is simple, appealing, and easy to understand. Customers tend to disbelieve a review if they feel the customer is biased by the language used in the review. Hence, customers can make out disgruntled and unjust customers if they have used improper language and such reviews tend to be less authentic.
Moreover, if the rating by the reviewer does not support the description, customers tend to rely less on such reviews. Clear and un- ambiguous reviews are perceived to be more reliable and positively reinforce buying decisions (Lu et al., 2018). So this suggests that if the reviewer uses simple language, uses a tone that does not reflect biases against the product, and the message is simple to understand, customers tend to trust such reviews.
4.4. Relevance
Online reviews may contain information about product descriptions, service experience, grievances, complaints, or appreciations as felt by the reviewer. However, all this information may not be pertinent to all the customers. Hence, apart from Source Credibility, Volume of reviews, and the Language of the reviews, what matters is the Relevance of the review to the customer and the response of the seller/retailer to negative reviews. So, a seller’s response to negative reviews helps customers make an informed decision by reducing the adverse effect if the response is acceptable. Customers rely more on recent reviews than older reviews (Kawaf and Istanbulluoglu, 2019). Some e-retailers also display the recent reviews at the top or give options to customers to select a time period to access the reviews. Like negative word of mouth in the physical place, negative online reviews dissuade a customer more than a positive review can influence (Li et al., 2020). Hence negative reviews are more relevant to customers than positive ones. Moreover, deceptive or fake reviews may also lead into negative WOM and dissuade consumers from purchasing (Munzel, 2016). This indicates that e-retailers and online platforms must strategize their response to negative reviews and solve customer complaints to build the trust and confidence of both existing and prospective customers.
In summary, customers check the credibility of the source of the reviews and look for a higher number of reviews to trust the rating/score during the online purchase. Simultaneously, the language and relevance of the review also influence the consumer purchase decision in enhancing the reliability of the reviews.
5. Contributions
5.1. Theoretical contributions
The study has notable theoretical contributions. First, it is the
pioneering work that measures the effectiveness of online consumer reviews. This has contributed to developing and validating an Online Review scale with 17 items and investigated the impact of online customer reviews on the consumer purchase decision. The scale mea- sures the helpfulness of online reviews in consumer purchase behavior.
The study’s empirical results indicated that the developed scale has a reliable measurement model (Hair et al., 2012).
Second, with the growing use of online platforms for purchase and the spread of the internet, shoppers are using various forms of infor- mation channels to make the best shopping decision. The study shows that while making a purchase decision, consumers give importance to online reviews, reducing their search efforts and minimizing the risk while choosing a product or service. In this context, the study’s out- comes contribute to developing a theoretical understanding of con- sumers’ purchase decisions by using the scale of the online review with few new dimensions. It focuses on the importance of four factors or di- mensions such as source credibility, relevance, volume, and language in analyzing consumer purchase decisions.
Third, the study provides new perspectives and directions to online reviewers by adding new variables. The findings of the study added two new variables to the online review scale: ‘the style of writing reviews should be unbiased or impartial’ (for ‘the language and comprehension’
dimension) and ‘customers trust the reviews if it comes from a verified customer’ (for “source credibility’ dimension).
Finally, this adds value to the existing or traditional consumer decision-making literature by providing further scope to study the interrelationship and influence among these constructs. Subsequent studies can also explore the consequences of using these identified constructs to manage reviews. The EFA, CFA, reliability, and validity tests specified that the scale developed is a sound and dependable empirical model (Byrne, 2013). The study provided a comprehensive theoretical understanding of the usage and impacts of online customer reviews and is a base for quantitative and experimental studies.
5.2. Managerial implications
Firms should actively weave their products, services, and brands into this online review discourse. This study has various implications for managers. First, the literature observes that the attributes of online customer reviews such as review quality (richness), review ratings, the relevance of the reviews (Nieto-Garcia et al., 2019), and the reviewers’
identity disclosures and their level of expertise (Mariani and Pre- dvoditeleva, 2019) depict positive influence towards consumer online purchase decision. This helps both consumers in their decision-making and online retailers regarding product design, display, and managing relationships. Secondly, online retailers and marketplaces can use the
‘Online Review’ scale to get more customer insights into how customers gather information, arrive at purchase decisions, and retailers/brands can manage online services accordingly. The scale can segment con- sumers who are most likely to get influenced by online reviews. Thirdly, it is observed that consumers decide their buying decisions based on various types of online information, such as positive or negative types of online reviews and the number of online reviews. Before making their purchase decisions, consumers check the information about the com- pany and ratings about the products on their respective websites (Stuppy et al., 2020). Our result also showed that consumers look for extreme reviews (5-star or 1-star) while evaluating a product or brand, consistent with the findings of Sen and Lerman (2007). Hence, the proposed scale will help retailers understand consumer tendencies toward the suscep- tibility of online reviews, and thereby their behavior can be monitored and evaluated to devise appropriate programs. Fourthly, consumers are generally influenced by the primacy effect, which might influence them if the online information is negative. This research will enable online retailers to target the segments that value negative reviews more than positive reviews in their purchase decision.