Antecedents of information helpfulness and purchase intentions in e-retailers providing consumer reviews
Consumers are increasingly turning to online reviews to diagnose the real quality of the products and services that they plan to buy. Thus, it is very important for online retailers to understand the determinants of online reviews helpfulness and their influence on consumer behavior. However, there is a dearth of studies on the determinants of review helpfulness, especially from the consumer perspective. To fill this gap, we adopt dual-process theory and explore the influence of informational and normative cues on information diagnosticity, as well as its link with consumers’ purchase intentions. Predictions are tested using structural equation modelling with 401 users of travel reviews. Results show that information quality, overall product ranking, product popularity are strong predictor of review helpfulness and that high ranking scores together with helpful reviews provided by highly credible sources will affect consumers’ purchase intentions. This study extends the application of dual-process theory to e-word of mouth.
1. Introduction
More and more consumers are trusting online consumer reviews (OCRs) and using them to assess the quality and performance of the products and services that they plan to purchase. The importance of consumer reviews has fostered e-retailers to provide their products and services with customer reviews (Mayzlin, 2006). Scholars have provided evidence of the influence that online reviews have on product sales (e.g., Liu, 2006; Dellarocas et al., 2007; Duan, Bin, Whinston, 2008; Zhu & Zhang, 2010), information processing and adoption, and purchasing decisions (Park, Lee & Han, 2007; Zhang & Watts, 2008; Filieri & McLeay, 2014). However, little is known about what makes online reviews diagnostic by consumers using Electronic word-of mouth (e-WOM) (Pan & Zhang, 2011; King et al., 2014).
Information is diagnostic if consumers perceive it helps them to understand and evaluate the quality and performance of products sold online (Jiang & Benbasat, 2004). Diagnosticity is often conceptualised as the degree of helpfulness of information (Skowronski & Carlston, 1987; Qiu et al., 2012). Not all consumer reviews are helpful and understanding the antecedents of information helpfulness is paramount for e-retailers because the more helpful the reviews the higher will be e-retailers’ sales (Chen et al., 2008). In addition to customer reviews, e-retailers provide several cues and signals to help customers diagnose the quality and performance of products including overall ranking scores, product popularity signals, and product quality marks. In order to address gaps in the e-WOM literature, we investigate the factors which contribute the most to consumers’ perceptions of online review helpfulness.
Existing studies on review helpfulness mostly use databases of reviews from e-retailers such as Amazon using voting mechanisms which ask readers ‘was this review helpful ?’ to assess review helpfulness (e.g., Mudanbi & Schuff, 2010; Pan & Zhang, 2011; Baek et al., 2012). However, in this study we analyse review helpfulness from the consumer’s perspective for two main reasons: firstly, scholars have found that voting mechanisms can be easily manipulated (Lim et al., 2010); second, some aspects that might affect review helpfulness such as perceived source homophily cannot be assessed with textual analysis. We attempt to address gaps in the extant literature by identifying the determinants of online reviews helpfulness and its link with purchase intentions. We have used dual-process theory because it can explain the influence of social and informational factors on an individual’s psychological processes (Deutsch & Gerard, 1955).
1. E-WOM Literature
e-WOM refers to ‘any positive or negative statement made by potential, actual or former consumers about a product or company, which is made available to a multitude of people via the Internet’ (Hennig-Thurau et al., 2004, p. 39). Third-party e-retailers, namely online agencies who sell on behalf of a service provider (e.g. Booking.com), are increasingly providing customer reviews on their websites (Mayzlin, 2006) in an attempt to facilitate the consumer decision journey and increase their sales.
Online consumer reviews have attracted considerable interest from researchers who have found that OCRs directly affect sales of products (e.g., Liu, 2006; Dellarocas et al., 2007; Zhu & Zhang, 2010) and influence elements of consumer behavior including: information adoption (Cheung et al., 2008; Zhang & Watts, 2008; Filieri & McLeay, 2014); product considerations and choice (Huang & Chen, 2006); attitudes towards products (Lee et al., 2008) and purchase intentions (Park et al., 2007; Park & Lee, 2008; Lee & Lee, 2009). Despite the importance of information diagnosticity in explaining persuasion in WOM, e-WOM research on this construct is still scant (Pan & Zhang, 2011).
2. Theoretical background: Dual-process theory
Dual-process theory (DPT) was developed by social psychologists to differentiate between two types of social influences: informational and normative (Deutsch & Gerard, 1955). DPT postulates that individuals are influenced by others because they are dependent on others either for information that removes ambiguity and thus establishes subjective validity, or for reasons of social approval and social acceptance. Informational influence includes the relevant components of the information, such as the content, source, and receiver, which are considered as important sources of influence (Hovland, Janis & Kelley, 1953; Cheung et al., 2009). Normative influences is defined as ‘an influence to conform to the positive expectations of another, while informational influences is defined as an influence to accept information obtained from another as evidence of reality’ (Deutsch & Gerard, 1955; p.629).
Drawing on DPT, we argue that social influence in e-WOM communications may occur via informational influences, which include: the quality of the argument provided by others in consumer reviews, the credibility of a source, its similarity (homophily) with the reader, product quality signals provided by e-retailers; as well as via normative influence; which include consumer’s overall evaluations (ranking) of products and their popularity. In the discussion that follows, we conceptualize and discuss each of these constructs in more detail.
4. Hypotheses Development
4.1 Informational influences
4.1.1 Information quality
Information quality is defined as ‘the quality of the content of a consumer review from the perspective of information characteristics’ (Park et al., 2007, p. 128). Information quality has been shown to affect information usefulness (Cheung et al., 2008), information adoption (Filieri & McLeay, 2014), and review credibility (Cheung et al., 2009, 2012). In studies using datasets of customer reviews from e-retailers, scholars have identified that review depth and review length are information quality dimensions that affect review helpfulness (Mudanbi & Schuff, 2010; Pan & Zhang, 2011; Baek et al., 2012). However, information quality dimensions and textual analysis can only reveal the tip of the iceberg of information quality criteria that are likely to contribute to perceived information helpfulness. Therefore, we have used Churchill’s (1979) approach to identify additional information quality dimensions for perceived helpful reviews.
The new dimensions identified through interviews included: review factuality, relevance, two-sided information, and credibility. Information factuality is the degree to which a comment in a review is logical; is based on specific facts related to experiencing a product; and is free from emotional, subjective, and vacuous comments. Information relevance refers to the extent to which a review message is applicable to and helpful for the task at hand and depends on a specific customer need in a specific situation (Wang Strong, 1996). Two-sided information refers to a review message that discusses both the positive and negative sides of a product (Kamins et al., 1989). Information credibility is defined as the extent to which a user perceives a message as believable or true (Cheung et al., 2009).
In this study, we hypothesize that if a review is perceived to be of high quality, it will affect consumers’ perceptions of the level of diagnosticity of the review. The more an online review is detailed, long, based on facts, contains both positive and negative comments, and is relevant to consumer needs, the more consumers will find such information helpful.
H1: Information quality significantly and positively influences perceived info diagnosticity
In addition, scholars have suggested that information quality influences consumer purchase intentions in an e-WOM context (Park et al., 2007; Lee et al., 2008). In fact, the more informative the review is, the more favorable associations consumers may have, resulting in an increase in behavioral intention. Thus, we hypothesize:
H1a: Information quality significantly and positively influences purchase intentions.
4.1.2 Source credibility
Credible sources are among the most persuasive sources of influence (e.g., Hovland et al., 1953). e-WOM research shows that source expertise and trustworthiness do not influence perceived information usefulness (Cheung et al., 2008). Based on dual-process theory (Deutsch Gerard, 1955), we argue that credible sources are more likely to provide diagnostic information than non-credible ones. Thus, we hypothesize:
H2: Source credibility significantly and positively influences info diagnosticity.
In addition, marketing scholars have proved that source expertise and trustworthiness positively influence consumer purchase intentions and purchase behavior (Gilly, Graham, Wolfinbarger, & Yale, 1998). Zhang and Watts (2008) show that source credibility has a positive and significant influence on information adoption for online travel websites. Thus:
H2a: Source credibility significantly and positively influences purchase intentions.
4.1.3 Source homophily
Perceptual homophily represents the result of the ‘textual interaction’ between a reader and a source of communication in e-WOM. In e-WOM communications people have to retrieve profile information or read the content of reviews to make inferences about their similarity with a reviewer. Perceptual homophily concerns the similarities among people regarding their likes, dislikes, values, and experiences (Bruyn de Lilien, 2008). Research has suggested that consumers tend to have greater levels of interaction, trust and understanding with people who are similar to them (Ruef, Aldrich, & Carter, 2003). In e-WOM homophily predicts trust (Tang et al., 2013) as well as source trustworthiness and expertise (Ayeh et al., 2013). In this study, we argue that consumers will find reviews from other consumers who are similar to them in terms of their viewpoints, experiences and preferences to be more diagnostic. For example, a backpacker traveller will find the opinion and reviews of people who share the same style of travelling more useful while a young couple with kids will look for reviews from people travelling with their family members. Thus:
H3: Perceived source homophily significantly and positively influences info diagnosticity.
Additionally, scholars have attempted to prove the role of homophilous ties on consumer decisions. For instance, Brown and Reingen (1987) suggest that homophilous sources of information will be perceived as more credible than heterophilous ones, which should result in greater influence. Thus, we hypothesize:
H3a: Perceived source homophily significantly and positively influences purchase intentions.
4.1.4 Product quality marks
The technological environment limits e-retailers’ capabilities for providing specific product attributes information such as smell, taste, touch, feel and the like (Grewal et al., 2004). It follows that e-retailers must leverage signals that facilitate a consumer’s ability to make accurate quality assessments about products being sold (Pavlou et al., 2007). Marketing scholars have investigated how e-retailers use different trustmarks including third-party marks, symbols or logos such as the VeriSign logo to reduce perceptions of the potential risks involved in an online transaction (Aiken & Boush, 2006). Similarly, many e-retailers provide quality marks as signals to communicate product quality and facilitate consumers’ choices. Quality marks can be defined as any symbol, icon, signal that is presented by an e-retailer in an effort to reduce ambiguity and uncertainty about the quality of a product or service. Many third-party e-retailers provide quality marks. For example Booking.com uses an ok-hand icon to signal their preferred hotels which they believe offer the best value for money and achieve high satisfaction scores from previous customers. In this study, we hypothesize that quality marks can help consumers in assessing the quality and performance of a product that they are interested in. Additionally, we also expect that quality marks can also influence consumers’ purchase intentions. Thus, we hypothesize:
H4: Website quality marks significantly and positively influence info diagnosticity.
H4a: Website quality marks significantly and positively influence purchase intentions.
4.2 Normative influences
4.2.1 Overall Ranking
Overall ranking is a summary statistic of how all customers have rated (reviewers’ average evaluation) a product or service in a specific category, such as the ranking of hotels available in a particular destination. Overall ranking is what social psychologists refer to as base-rate information and defined as ‘general information, usually factual and statistical, about an entire class of events’ (Hogg Vaughan, 2014, p.70). For example, when using Agoda.com, every reviewer can rate the overall quality of a hotel using a scale from one to ten (superb). Such a summary statistic is a unique feature of e-WOM communications and indicates how all customers have evaluated a product or service. Research on the role of summary statistics in e-WOM is still scant. Scholars have studied the role of individual ratings on the perceived trustworthiness of retailers (Benedicktus et al., 2011) or focused on review rating consistency (Baek et al. 2012). Chevalier and Mayzlin (2006) conclude that consumers read review text rather than rely solely on summary statistics for books, while Qiu et al. (2012) focus on conflicting aggregated ratings and their influence on the diagnosticity of single reviews.
In this study, we argue that consumers benefit from access to summary statistics (rankings). By classifying the products in a category through the use of average ratings (from best to worst), a crowd of customers communicate how a product or service is performing relative to competitors. Accordingly:
H5: Overall ranking significantly and positively influences perceived info diagnosticity.
H5a: Overall ranking significantly and positively influences consumer intentions.
4.2.2 Product popularity
A product is considered popular when many people talk about it or purchase it. Online, e-retailers and online communities provide signals that communicate a product’s popularity. For example, the number of download counts indicates the quality and reliability of software products (Hanson Putler, 1996). The volume of consumer reviews is perceived by consumers as an indicator of the market performance of a product (Chevalier & Mayzlin, 2006; Huang & Chen, 2006) as it is associated to the number of consumers who have bought a product (Chatterjee, 2001). Social influence scholars observe that when individuals are uncertain about a situation they observe what other people do and imitate their behavior (e.g. Asch, 1951). Such imitative behavior can occur also in e-WOM communications. For example, when consumers are unsure about which product to buy they may look at the number of reviews per product, which communicates how many people are buying the product, to help their purchase decisions. To this extent, consumers think that the more people choose a specific product, the higher will be its quality; thus product popularity can be helpful information for consumers.