According to research, potential buyers should read reviews and consumer ratings to get a better knowledge of the products being reviewed and the opinions of the reviewers (Mudambi and Schuff 2010; Cheema and Kaikati 2010; Packard and Berger 2017). According to Chen and Lurie (2013) and Packard and Berger (2017), these reviews frequently influence review persuasiveness, which in turn influences downstream factors like sales. Perhaps businesses encourage customers to write evaluations with particular material as a result of this phenomenon. In fact, Facebook has switched from soliciting user suggestions to gathering review ratings (BrightLocal 2018).
Repurchase intentions and recommendations are just two examples of limited expressions of review persuasiveness. We specifically investigate which of these two loyalty expressions influences review persuasiveness more favorably. We also look into a boundary condition (i.e., the frequency of purchases) and the underlying mechanism.
We investigate the differential impact of these two loyalty expressions in online reviews across eight studies (a field study and seven experiments, three of which are reported in Web Appendix E) and show that repurchase intentions have a stronger positive effect on review persuasiveness than on recommendations because of reviewer credibility.
However, we also demonstrate that for regular purchases as opposed to infrequent purchases, repurchase intentions (vs. recommendations) have a stronger impact on review persuasiveness. In contrast, recommendations (as opposed to intents to make another purchase) have a higher impact on purchases. In contrast, recommendations had a higher impact on review persuasiveness for occasional purchases than for frequent purchases (vs. repurchase intentions).
As a result, we show that the boundary condition for our key proposition is the frequency of purchasing. We examine the major theoretical and managerial ramifications of new insights into the persuasive impact of repurchase intentions vs. recommendations for frequently compared to infrequently purchased products.
Conceptual Framework
Many potential consumers rely on online reviews to inform their purchase decisions, but not all reviews hold the same persuasive power. As these reviews mainly come from strangers, consumers seek diagnostic cues to assess their persuasiveness. Diagnostic cues are pieces of information that help consumers reduce uncertainty, form attitudes, and make purchase decisions. Previous research has identified various diagnostic cues in online reviews, including reviewer characteristics, variance of historical ratings, expertise, similarity to the observer, device type, review length and valence, valence intensity, review age, and review volume, as drivers of review persuasiveness.
Moreover, linguistic cues in review language also influence persuasiveness. Subtle changes in review language, such as the type of explanation or endorsement style, can significantly affect review persuasiveness.
In this study, the focus is on how prospective consumers assess online review writers' loyalty intentions through the language they use. Customer loyalty is indicated by an intention to engage in behaviors that signal a motivation to maintain a relationship with a particular brand, including positive word of mouth, repeat purchasing, and allocating a higher share of the category wallet to the specific service provider. Building on this conceptualization of customer loyalty, the researchers explore the relative persuasiveness of repurchase intentions versus recommendations, which represent two facets of loyalty expressions in online reviews.
By understanding the impact of different linguistic expressions of loyalty on review persuasiveness, businesses can gain insights into how to effectively elicit loyalty expressions in online reviews to influence consumer perceptions and decision-making.
In conclusion, online reviews play a significant role in shaping consumer purchasing behavior. By recognizing the importance of diagnostic cues, including linguistic cues related to loyalty expressions, businesses can optimize their review management strategies and enhance the persuasiveness of their online reviews.
Study 1- Testing relative effects on review persuasiveness using secondary data
Researchers used a publicly accessible review dataset from the online review site Yelp for our empirical investigation. Yelp publishes user-generated ratings about various establishments (like restaurants). As previously said, we concentrated on company reviews that were positive and published between January 2017 and December 2017. Additionally, we chose reviews that expressed repurchase intentions or recommendations. The sample included 111,728 reviews with information on the variables listed below.
Use f As of the data collection date (November 14, 2018), ovotestis is a count variable that shows the number of beneficial votes earned by review t, which was posted by customer I for business j. By clicking on the "Useful" button, Yelp users can express if the review was helpful to them.
RPItij is a variable that serves as an indicator of whether the customer indicated a desire to make another purchase or made recommendations for company j in the review t. We identify reviews that include repurchase intents or recommendations, as indicated in Web Appendix C, and set RPItij to one (1) if the review had repurchase intentions and zero (0) if it had recommendations.
Examine associated control variables.
ReviewRatingtij is an indicator variable that represents the evaluation that a customer I made of a company j in a review t. As previously said, we only counted reviews with a five (5) or four (4) star rating. ReviewRatingtij is set to one (1) if the given rating is a five (5), otherwise to zero (0). The age of a review t that a customer posted for a business j is represented by the continuous variable ReviewAgetij.
Calculating ReviewAgetij involved counting the days between the review posting date and the data collection date (November 14, 2018). FOGtij represents the readability of the review, based on the FOG Index. It indicates the years of formal education needed to understand a piece of text on the first reading (Korfiatis, GarcíA-Bariocanal and SáNchez-Alonso, 2012). We computed the FOG Index using the Textstat package in Python. FOGtij is a continuous variable that can take any positive value greater than zero (0). ReviewLengthtij is a continuous variable representing the length of review t posted by consumer I for business j. We computed review length as the number of words in the review.
Results
We used SAS' PROC MCMC approach to conclude the parameters. An MCMC chain was simulated using 15,000 samples. The first 5,000 samples were discarded as burn-in, and we chose Eve from the remaining samples. Calculating ReviewAgetij involved counting the days between the iteration and retaining a total of 2,500 samples for the posterior inference of the means and standard deviations of the parameter estimates, which we display in Table 2. We theorized that prospective consumers would find reviews with repurchase intentions (compared to recommendations) more persuasive (H1).
Study 2 - Testing relative effects on review persuasiveness using experiments
Method
In a between-subjects experiment comparing the loyalty manifestations of repurchase intentions and recommendations, 200 U.S. participants (51.5% female, mean age = 39.81 years) recruited via mTurk were randomly allocated to one of the two conditions. The participants were instructed to pretend they were buying wine online. They came upon a wine that seemed promising, but before making a purchase, they wanted to read a customer review. The review with one of the loyalty emotions was then displayed to them (the study stimuli are listed in Web Appendix H).
The review's persuasiveness was assessed after the participants gave it a grade, and their responses were based on a two-item scale.
On a seven-point scale, there was one item for review usefulness (Chen and Lurie 2013) and one item for buy intention (Packard, Gershoff, and Wooten, 2016) (measures are in Appendix D). Additionally, they gauged the familiarity of respondents by having them rate their familiarity with wine-related products on a scale of 1 to 7 (extremely unfamiliar = 1, very familiar = 7). Finally, the participants filled out demographic questions (such as their age and sex).
Results
Review persuasiveness they averaged ratings on review usefulness and purchase intention into an index of review persuasiveness (r = .64).5 As expected, an ANOVA showed that repurchase intentions had a stronger effect on review persuasiveness (Mrepurchase = 5.55, SD = 1.11 vs. Mrecommendations = 5.18, SD = 198) = 4.73, p = .031, Cohen’s d = 0.306) than recommendations (see Fig. 2). This finding supports H1. The participants in the repurchase intentions and recommendations conditions did not differ in their familiarity level with wine products (Mrepurchase = 4.11, SD = 1.72 vs. Recommendations = 4.33, SD = 1.81; F(1, 198) = 0.81, p = .369, Cohen’s d = 0.124). Also, the addition of familiarity level (p = .441), sex (p = .013), and age (p = .856) as covariates in the model did not change the significance of our results (F(1, 195) = 4.04, p = .046)1.30;
Study 3A - Testing reviewer credibility as a mediator
The objective of study 3a is to examine the role of reviewer credibility as an underlying mechanism for the observed effect of loyalty expressions (repurchase intentions vs. recommendations) on review persuasiveness.
Method
A (loyalty expressions: repurchase intentions vs. recommendations) experiment with a between-subjects design was conducted with 224 U.S. participants (50.9% female, mean age = 41.49 years), who were recruited from mTurk in exchange for a nominal payment. Participants were then randomly assigned to one of the two conditions. The participants were instructed to make up a scenario in which they would enjoy a Saturday night out with pals. Before deciding on a certain restaurant, they wanted to read the restaurant reviews that were posted on an internet review site. They noticed one of the loyalty sentiments in the review.
Results
Review persuasiveness Out of the 224 participants, five failed the attention check question and were removed from the sample, leaving 219 participants. We averaged ratings on review usefulness and purchase intention into an index of review persuasiveness (r = .78). An ANOVA showed that repurchase intentions had a stronger effect on review persuasiveness (Mrepurchase = 6.14, SD = .92 vs. Mrecommendations = 5.83, SD = 1.11; F(1, 217) = 4.97, p = .027, Cohen’s d = 0.304) than recommendations, as expected. This result lends further support to H1
Study 3B - Testing mediation through moderation-of-process analysis
Study 3b's objective is to support the moderating process (Spencer et al., 2005). If the observed results are a result of enhanced credibility being induced by repurchase intents, exogenously introducing credibility signals should provide a similar level of review persuasiveness in the recommendations condition. Additionally, this approach aids in eliminating potential explanations for the paper's main finding, which is that review persuasiveness is more strongly influenced by repurchase intentions than by recommendations based on reviewer credibility perception.
Method
275 U.S. participants from mTurk (49.8% female, mean age = 39.91 years) were randomly assigned to a between-subjects design that included 2 (loyalty expressions: repurchase intentions vs. recommendations) × 2 (credibility cue: absent vs. present). Reading a hotel review was the context. Participants were asked to make up the scenario that they wanted to reserve a hotel room for a trip to a vacation spot in the United States. As a result, people want to read the hotel reviews that are posted online. They wanted to learn more about one hotel's reviews because they were thinking about staying there.
They stumbled into a review on a website. They considered a review credible if it expressed either repurchase intent or a recommendation. Credibility indicators were both absent and present. After reading the review, researchers asked respondents about the review's persuasion, how frequently they stayed at hotels, and other personal data (such as their sex and age).
Results
Finally, loyalty expression x credibility cue interaction remained significant (F(1, 278) = 4.56, p = .033) after we accounted for frequency (p = .058), sex (p = .839), and age (p = .121) as covariates in the interaction model.
Study 4 - Examining purchase frequency as a boundary condition
Method
For a nominal fee, 416 U.S. participants (57.1% female, mean age = 38.89 years) were chosen at random from mTurk and assigned to a between-subjects design with two variables: purchase frequency (frequent vs. infrequent), and loyalty expressions (return intentions vs. recommendations). Reading a dentist's review was the context. Participants were asked to see themselves relocating to a different city. Instead of having a dental crown, they want to get their teeth cleaned. They discovered a dentist who charges a copay and takes their dental insurance. They are interested in learning more about this dentist from the opinions of his or her patients because they are unfamiliar with him or her. They stumbled into a review on a website. Method
For a nominal fee, 416 U.S. participants (57.1% female, mean age = 38.89 years) were chosen at random from mTurk and assigned to a between-subjects design with two variables: purchase frequency (frequent vs. infrequent), and loyalty expressions (return intentions vs. recommendations). Reading a dentist's review was the context. Participants were asked to see themselves relocating to a different city. Instead of having a dental crown, they want to get their teeth cleaned. They discovered a dentist who charges a copay and takes their dental insurance. They are interested in learning more about this dentist from the opinions of his or her patients because they are unfamiliar with him or her. They stumbled into a review on a website.
Results
Planned contrast results reveal that in the frequent purchase condition, revisit intentions have a stronger impact on review persuasiveness (Mrevisit = 6.09, SD = .86 vs. Mrecommendations = 5.70, SD = 1.14; F(1, 412) = 8.62, p = .004, Cohen’s d = .386) and credibility (Mrevisit = 5.51, SD = .85 vs. Mrecommendations = 5.20, SD = 1.04; F(1, 412) = 5.95, p = .015, Cohen’s d = .326) than recommendations. In contrast, recommendations have a stronger impact on review persuasiveness (M recommendations = 6.10, SD = .82 vs. Mrevisit = 5.76, SD = .96; F(1, 412) = 6.24, p = .013, Cohen’s d = . 381) and credibility (M recommendations = 5.45, SD = .89 vs. Mrevisit = 5.17, SD = .94; F(1, 412) = 4.64, p = .032, Cohen’s d = .305) than revisit intentions in the infrequent purchase condition, consistent with hypothesis 4. We present the results for review persuasiveness in Fig. 5.
General Discussion
Even though the distinction between repurchase intentions and recommendations as two distinct facets of loyalty expressions is widely acknowledged (de Matos and Rossi 2008; Lam et al., 2004; Maxham III and Netemeyer, 2002; Soderlund 2006), there is little conceptual or empirical research that systematically compares how persuasive repurchase intentions vs. recommendations are in online reviews. We first demonstrate, using a multimethod approach, that reviews with repurchase intents were more compelling than reviews with recommendations. This was done through a field study and seven tests. In the text, we discussed the findings from field research, four experiments, and three more experiments were included in Web Appendix E. We credit the reviewer's credibility for the repurchase intents' increased persuasiveness impact (in comparison to recommendations). Particularly, potential customers view declarations of repurchase plans as more reliable.
Future Research
The comprehension of the relative persuasiveness of repurchase intentions vs. recommendations in online review environments has been expanded by this study, but we are aware of its limits and the need for more empirical research. First, we have emphasized the positive aspects of loyalty statements and have shown how recommendations and repurchase intentions have varied effects on the persuasiveness of reviews. Future studies, however, can look at the negative side of displays of loyalty and demonstrate the asymmetry of any negative loyalty declarations (for example, "I do not recommend it" vs. "I will not come back again").
Finally, it might be argued that the credibility of the reviewer comes first and not the other way around. Future studies can therefore look into this reverse causality (Shi et al., 2020). We, therefore, ask for more empirical research on the significance of repurchase intentions and recommendations in online reviews as we draw to a close.
Conclusion
To expand our understanding of how repurchase intents and recommendations affect review persuasiveness in online situations, we urge additional empirical research. Future research can help firms that want to manage and use online reviews to improve consumer decision-making processes by addressing these limitations and examining fresh angles.
