Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Research suggests that 85 % of consumers refer to online consumer reviews (OCRs) to guide purchasing decisions (Anderson 2013), and OCRs play a pivotal role in firms’ online sales and profits (Floyd et al. 2014). We define OCRs as positive or negative comments posted on the Internet by consumers regarding goods or services sold online. Despite growing adoption, OCRs are often criticized for their credibility since review information typically is not subject to verification (Johnson and Kaye 2002); and some firms strategically manipulate OCRs (Fahey and Weinberg 2003). To distinguish truthful reviews from deceptive ones, consumers commonly use peripheral cues to assess the credibility of online information (Petty and Cacioppo 1986). Therefore, it is important for firms to understand how these cues influence trust.

Research examining OCRs and trust has been sporadic and lacks a comprehensive framework. This current study ties to a trust framework (Mayer et al. 1995) and investigates three review attributes—valence, rationality, and source—and their impact on review trustworthiness. Review valence refers to the positive or negative orientation of information about an object (Frijda 1986), and review rationality represents the degree of the argument that is relevant, objective, and verifiable (Petty and Cacioppo 1986). Review source captures the person who provides the message (e.g., reviews from social network or strangers on a retail site). We believe that examination of these attributes through the lens of trust will enrich the theoretical understanding of OCRs. In addition, we also investigate the moderating effect of one of the consumer characteristics—uncertainty avoidance. Further, given the importance of global retailing (Stock 2014) and dominant research conducted in the USA, a cross-cultural examination is particularly needed and establishes the generalizability of the framework. Therefore, our study aims to addresses three questions: (1) How do review valence, rationality, and source influence review trustworthiness? (2) How does uncertainty avoidance moderate the effect of attributes on trustworthiness? (3) Does the proposed framework hold in both China and the USA?

Theoretical Framework

In the context of OCRs, we define trustworthiness as the expectancy held by online users that written statements posted by online users can be relied on. Research in personal psychology suggests that trustworthiness is a multifaceted construct, which is comprised of three factors: benevolence, ability, and integrity (Mayer et al. 1995). Building on this three-factor trust model, we propose that the three review attributes drive the trustworthiness of OCR. Review valence ties to the benevolence driver of trust, review rationality links to the ability driver of trust, and review source speaks to the integrity driver of trust. We further propose that the effect is moderated by uncertainty avoidance. Uncertainty avoidance, the characteristic of a review reader, describes people’s risk orientation toward uncertain or unknown situations (Hofstede 2001), and in this context dealing with reviews from unknown reviewers about unknown goods/services. The degrees of uncertainty avoidance also differ across cultural stereotypes (Hofstede 2001), and thus fit with our objective of cultural validation.

The Effects of Review Attributes on Review Trustworthiness

Review valence. Review valence refers to the positive or negative affective content of a review (Sridhar and Srinivasan 2012). Benevolence is the extent to which a trustee is believed to want to do good to a trustor, aside from an egocentric profit motive (Mayer and Davis 1999). We tie review valence to the benevolence driver of trust. This can be explained through the “positive test strategy,” a type of “confirmation bias” people have (Klayman and Ha 1987, p. 212). According to this strategy, people test a hypothesis by examining instances in which the property or event is expected to occur, or by examining instances in which it is known to have occurred. When a situation is ambiguous and concrete information is lacking (e.g., unknown products and experiential services in online contexts), people rely on this strategy as a default heuristic. Customers who look for quality goods/services (e.g., a decent hotel) are more likely to attend to reviews that provide positive information and have greater chances of supporting their hypotheses (Jones et al. 2006). In addition, in an online review context, a review that describes distinguishing features or expresses a sense of constructiveness, helpfulness, and optimism projects an image of benevolence (Mayer and Davis 1999). A higher degree of perceived benevolence, or the belief of goodwill, associates with a higher degree of trust (Mayer et al. 1995). Conversely, a review that expresses negative viewpoints prohibits readers from associating reviewers with the characteristics of benevolence, and results in lower trust. Thus:

  • H1. Positive reviews are more trustworthy than negative reviews.

Review rationality. Review rationality represents the perceived degree of emotion a reviewer is exercising when creating a review message. A review message could be emotional or factual, regardless of the valence of the review (i.e., positive or negative). The rationality of the review content appeals to the ability dimension of trust. Ability refers to a group of skills, competencies, and characteristics that allow a party to have influence within some domain (Mayer and Davis 1999). People are biased regarding perceptions of intellectual ability in that emotional statements such as angry reviews often suggest that reviewers are incapable of analyzing a situation thoroughly and arriving at objective conclusions (Lerner and Tiedens 2006). Emotional evaluations are often perceived as subjective and driven by emotion rather than cognitive thinking, thereby indicating lack of intelligence (Rohrer 2010). Factual descriptions present facts, exhibit sensible reasoning, appear rational, and therefore are perceived as more intelligent (Rohrer 2010). Research suggests that competency is a driver of trust in the context of sharing knowledge (Levin et al. 2002), especially in experiential and/or credential service contexts such as hotel, medical, and legal services. Claims that are relevant, objective, and verifiable tend to be more persuasive and are perceived as more credible, thus having a positive effect on consumer attitudes and purchase intentions (Pornpitakpan 2004).

  • H2. Factual reviews are more trustworthy than emotional reviews.

Review source. Review source signals the motive of a reviewer and appeals to the third driver of trust, integrity. Integrity describes a trustor’s perceptions that a trustee adheres to principles that the trustor finds acceptable (Mayer and Davis 1999). The trustworthiness of a review is driven largely by a reader’s interpretation of a reviewer’s motive (i.e., why he/she writes a review in such a way). The strength of the ties between a reviewer and reader matters during such interpretation. For example, an inner circle is represented by expressive ties in which the closest relationships such as family members and friends are often viewed as more honest, trustworthy, and less self-interested (Granovetter 1983), while strangers who write reviews on a retailer’s site belong to the outmost circle, represented by instrumental ties, and their motives for information sharing are more questionable (Hwang 1987). Thus, friends from one’s social network are perceived with more integrity than strangers on a retailer’s site. Conversely, a retailer who controls the design and/or posting of reviews can manipulate reviews for its own interests (Fahey and Weinberg 2003). Reviews posted on a retailer’s site are viewed as possibly influenced by the retailer, are less independent, and have a higher possibility of conflicts of interests than those posted on one’s social network. This presence and/or absence of self-interest influences the integrity driver of trust.

  • H3. Reviews posted on a social network site are more trustworthy than reviews posted on a retailer’s website.

The Moderating Effect of Uncertainty Avoidance

Uncertainty avoidance (UA) signals a person’s risk orientation (Jones et al. 2006). A high-UA person avoids the unfamiliar and often follows a strict structure, while a low-UA person is more willing to take risks (Hwang and Lee 2012). Generally, positive reviews are perceived as potential gains, while negative reviews are perceived as possible losses (Tversky and Kahneman 1981). High-UA consumers who are reluctant to change and are risk averse (Lim et al. 2004) show a stronger preference for gains over losses (Tversky and Kahneman 1981), thereby trusting positive reviews more than negative reviews (Jones et al. 2006). The positive test strategy is more important to high-UA customers since they are more risk averse and show greater tendency to look for evidence to support their hypotheses (Klayman and Ha 1987).

  • H4a. UA positively moderates the effect of review valence on trustworthiness.

Likewise, risk-averse consumers demonstrate a greater need for formality and structure, and hence they rely more on factual messages that are verifiable and better structured (Dillard and Shen 2005), while emotional messages have lower argument quality due to a lack of evidence and reasoning, and have a lower ability to meet high-UA consumers’ requests for rationality (Petty and Cacioppo 1986). Therefore, high-UA consumers place more weight on factual reviews, and the difference between factual and emotional reviews becomes stronger.

  • H4b. UA positively moderates the effect of review rationality on trustworthiness.

Research classifies external information into personal and marketer-dominated information (Money and Crotts 2003). Reviews posted on a retailer’s website are considered a marketer-dominated source of information, where the posting, editing, and removing of the reviews are managed by a retailer (Money and Crotts 2003), while reviews posted on social network sites such as Facebook resemble a personal category. Marketer-dominated information is perceived riskier than personal information (Chaloupka 1999). Therefore, high-UA consumers who are risk averse place more trust on reviews posted on their social network sites than retailers’ websites to control risk and reduce uncertainty, resulting in a stronger effect of review source.

  • H4c. UA positively moderates the effect of review source on trustworthiness.

Methodology

A 2 (valence) by 2 (rationality) by 2 (source) between-subject experimental design was used to collect data across the USA and China. Hotel review is selected as the context. Subjects read hotel reviews before booking hotels, and were then assigned randomly to one of the eight conditions. Review valence was manipulated at two levels: positive reviews (with all four hotel attributes positive) and negative reviews (with all four attributes negative). We manipulated review rationality at two levels: factual description (e.g., describing hotel features carefully and justifying the evaluations) and emotional release (e.g., “Run, don’t just walk away from this hotel”). Review source was manipulated at two levels: a review appearing on an online retailing site (booktrip.com) and on a social network site (Friendbook.com). An equivalent sampling method was used to recruit subjects in both countries. US data (299 respondents) were collected through Amazon’s Mechanical Turk, while Chinese data (261 respondents) were collected using Wenjuanxing, one of the most popular online consumer research panels in China.

The measurement model fit the data adequately for both samples; and measurement equivalence was established between the two subsamples (Baggozi and Yi 1988). We conducted univariate analysis of covariance (ANCOVA) to test the hypotheses including three review attributes and the moderator of uncertainty avoidance as independent variables, review trustworthiness as the dependent variable, and age, income, education, and service quality expectation as covariates. In support of H1 and H2, we found that review valence and review rationality influenced trustworthiness in both countries (USA: F valence = 4.196, p < .05; F rationality = 76.210, p < .001; China: F valence = 28.183, p < .001; F rationality = 21.049, p < .001). Review source influenced trust in the USA (F source = 27.970, p < .001) but only marginally in China (F source = 3.297, p = .071), supporting H3 partially. Regarding moderation by UA, UA positively moderated the valence-trust link for China (F valence×ua = 6.683, p < .01), but not for the USA. A significant and positive moderation effect on the rationality-trust link (F rationality×ua = 3.900, p < .05) and marginal moderation on source-trust (F source×ua = 3.510, p = 0.062) was found for the USA but not for China.

Discussion

This research makes several theoretical contributions. First, most research has focused on examining OCR’s effect on sales. Our study departs by studying another important outcome—consumer trust toward OCRs—a factor critical to formation of purchasing intentions (Racherla et al. 2012). Second, OCR research mostly focuses on one aspect of reviews (e.g., review valence or review source), while ours adopts a multidimensional view by considering three review attributes and tying them to the drivers of trust to gain a comprehensive understanding of OCRs. Overall findings support the dominant role of review attributes on review trustworthiness in both countries. Results suggest that positive reviews, factual reviews, and reviews posted on social networks are more trustworthy. Third, empirical findings regarding OCRs are inconsistent (Chevalier and Mayzlin 2006; Kusumasondjaja et al. 2012), suggesting a need to examine contingency factors. This chapter examines the moderating role of uncertainty avoidance. Finally, cultural disparities were more intriguing regarding moderation. For Chinese consumers, review valence was of critical importance such that the advantage of positive reviews is stronger for high-UA Chinese consumers than for low-UA consumers. In contrast, for American consumers, review rationality and review source are contingent on uncertainty avoidance. For example, for American customers who are highly risk averse, the advantage of factual review over emotional review was stronger than for consumers who are risk takers, and similar effects were found regarding review source.

References available upon request