Track 9: Social Influence
Aner Sela (University of Florida)

9A. Online Reviews

Saturday, March 5
9:30am – 11:00am EST
Discussant: Cait Lamberton (University of Pennsylvania)
MC: Jared Watson (New York University)
Calendar Invite: Add to calendar
Student Coordinator: Elina Hur (Cornell University) (

Competitive Papers

Should Reviewers and Website Moderators Censor Swearwords in Reviews?
Authors: Katherine Lafreniere (University of Lethbridge - Calgary Campus), Sarah Moore (University of Alberta)
Presenting Author: Katherine Lafreniere (University of Lethbridge - Calgary Campus)
Swearing in product reviews is offensive but helpful to readers. Can reviewers and website moderators use censored swearwords to convey similar information without causing offense? Results show that a censored (vs. uncensored) swearword in a review conveyed a similar level of the reviewer’s feelings but a weaker level of the product’s attribute. Thus, censored swearwords in reviews were less helpful and impactful. Still, the impact of censored swearwords improved when readers attribute censorship to the website as opposed to the reviewer. Thus, website moderators may still benefit from censorship if readers can infer the reviewer’s original intentions from swearword use.
The Influence of Mean Product Ratings on Review Judgments and Search
Authors: Daniel Katz (University of Chicago Booth School of Business), Dan Bartels (The University of Chicago Booth School of Business)
Presenting Author: Daniel Katz (University of Chicago Booth School of Business)
When searching for information, normative models suggest one should acquire maximally informative information. However, confirmation biases can lead people to seek redundant information or interpret information in ways that support preexisting beliefs. We examine how people search for and integrate information in product reviews because of their importance for understanding consumer behavior. We manipulated preexisting beliefs by varying the mean product rating. Reviews that were most consistent with (i.e., were closer to) the mean were rated as more helpful, lead to more extreme belief updating, and were more likely to be searched, relative to reviews that deviated further from the mean. Our results suggest people may be subject to a kind of confirmation bias when searching for, reacting to, and learning from product reviews.
I Want You to Like Me, so I’ll Wait to Share the Bad - The Influence of Self-Presentation Concerns on Consumer Ratings
Authors: Elisa Solinas (University of Southern California), Francesca Valsesia (University of Washington), Joseph Nunes (University of Southern California), Andrea Ordanini (Bocconi University)
Presenting Author: Elisa Solinas (University of Southern California)
Evidence of a declining trend in the valence of consumer reviews (i.e., ratings) online is abundant. Previous work focused on the evolution of ratings at the product level across reviewers. In contrast, using real-world Yelp data since its inception we document a negative trend in ratings at the reviewer level across products. Experimental data supports a self-presentation explanation; reviewers concerned with being perceived negatively are reluctant to post negative reviews early in their review history. Further, we identify a potential platform feature managers could employ to help mitigate reviewers’ reluctance to post negative reviews early in their review history.
The Persuasive Present (Tense)
Authors: Grant Packard (York University), Jonah Berger (University of Pennsylvania)
Presenting Author: Grant Packard (York University)
Consumers often use either past or present tense when sharing opinions (e.g., “That album was (is) great”). Does verb tense shape word of mouth’s impact, and if so, how? A multimethod investigation demonstrates present tense language increases impact, in part because it suggests the communicator is more certain. Analysis of over 500,000 online reviews reveals that reviews that use more present tense are seen as more helpful. Three experiments further demonstrate that shifting from past to present tense increases persuasion and document the underlying role of certainty. These findings highlight how a subtle, yet central language feature shapes communication’s impact.

Flash Talks

When Appealing to Agency Backfires: Evidence from a Multinational Field Experiment and the Lab
Authors: Joseph Reiff (UCLA Anderson School of Management), Hengchen Dai (UCLA Anderson School of Management), Jana Gallus (UCLA Anderson School of Management), Anita McClough (InMoment), Steve Eitniear (InMoment), Michelle Slick (NA), Charlotte Blank (Maritz)
Presenting Author: Joseph Reiff (UCLA Anderson School of Management)
Contrary to prior research about agency and to expert predictions, a field experiment across seven countries (N=430,666) revealed that attempting to make consumers feel their feedback was consequential backfired: it decreased consumers’ propensity to give companies feedback and it even increased unsubscribing from future emails. The agency appeals did particularly poorly in countries with lower trust in business (e.g., Japan) and performed better in countries with higher trust (e.g., China). Pre-registered lab evidence shows that people who do not trust business perceive these appeals as inauthentic, which helps explain when and why the appeals can reduce compliance with companies’ requests.
How Joint-Singular Evaluations and Different Reference Points Make Stars Bad
Authors: Matt Meister (University of Colorado Boulder), Nicholas S. Reinholtz (University of Colorado Boulder)
Presenting Author: Matt Meister (University of Colorado Boulder)
In a two-phase, preregistered experiment, the authors demonstrate a simple case where the presence of real star ratings leads consumers to make objectively worse choices. Specifically, when raters make singular evaluations (as is most common), but users of ratings make joint evaluations (as is also most common), star ratings may suggest that an objectively worse choice is superior. Because users do not intuit this joint-singular difference in evaluation, they are swayed by star ratings to make poor choices. This influence of star ratings persists even when presented alongside objective information.


Psychological Contracts in the Sharing Economy: The Role of Platform Design Policies when Consumers Get Reviewed
Authors: Laura Rifkin (Brooklyn College), Colleen Kirk (New York Institute of Technology), Canan Corus (Pace University)
Presenting Author: Laura Rifkin (Brooklyn College)
The Peer-to-Peer sector of the sharing economy relies on reputation systems in which peers (consumers and providers) review each other. We examine, for the first time, a turn of the tables in which consumers are being evaluated. Across six studies using multiple actual behaviors and sharing contexts (multi-product, home, ride, and car sharing), we demonstrate that a negative review of the consumer from the peer provider leads to negative WOM about the platform. Rather than overgeneralization of responses from  hosts to the platform, this effect is mediated by consumers’ perceived betrayal by the platform. Consistent with psychological contract theory, effects are exacerbated when platform promises are strengthened by through provider endorsement, and attenuated when reviews are kept private. Further, by increasing perceptions of fairness, both providing a revenge opportunity (changing the consumer’s review of the provider) and a reparation opportunity (responding on the platform) independently and together diminish the effect.
“I Like It” vs. “You’ll Like It”: The Use of Personal Pronouns in Online Reviews
Authors: Wenyan Yin (Drexel University), Yanliu Huang (Drexel University), Jonah Berger (University of Pennsylvania)
Presenting Author: Wenyan Yin (Drexel University)
This research examines the relationship between certainty and the use of personal pronouns in online reviews. We propose that review writers feeling uncertain (vs. certain) about their opinions are more likely to use “you” as the grammatical subject in their review. On the other hand, review readers tend to perceive the reviewer using “I” to be more certain about their attitudes and thus more persuasive than the reviewers using “you”. We have conducted four studies to test our propositions with study 1 from the review reader’s perspective and studies 2(a)(b) and 3 from the review writer’s perspective.
The Impact of Online Review Linguistic Features on Reviewers and Readers
Authors: Alisa Wu (Columbia University, Columbia Business School), Vicki Morwitz (Columbia University, Columbia Business School)
Presenting Author: Alisa Wu (Columbia University, Columbia Business School)
By extracting attitude bases (i.e., emotion versus cognition) and valence (i.e., positivity) from review text data, we find that these linguistic features exert distinct effects on consumer evaluations and reactions. Review writers who recount their consumption experience primarily based on their emotions (vs. cognitions) and/or on positivity feel better about their consumption experiences. Review readers rate emotion-based (cognition-based) online reviews more (less) positively, but also rate reviews that are high on positivity more negatively. We show these effects of linguistic features using actual review data and confirm their causal role by independently manipulating emotion, cognition, and valence in laboratory experiments.
Return to schedule