Author ORCID Identifier
0009-0002-3972-4279
Document Type
Dissertation
Date of Award
12-31-2025
Degree Name
Doctor of Philosophy in Business Data Science - (Ph.D.)
Department
School of Management
First Advisor
Jorge Eduardo Fresneda Fernandez
Second Advisor
Chelsey Helena Hill
Third Advisor
Shanthi Gopalakrishnan
Fourth Advisor
Ellen Thomas
Fifth Advisor
Plavini Punyatoya
Sixth Advisor
Javier Cabrera
Abstract
Online product reviews have become increasingly multimodal, combining text with media-rich elements such as images. However, academic research has largely examined textual features in isolation, overlooking how visual content and its interaction with text shape perceived helpfulness. This dissertation addresses that gap by developing and empirically validating a comprehensive framework capturing how textual, visual, and contextual features collectively influence review evaluation. Grounded in the Elaboration Likelihood Model (ELM) and extended through the Text-Image Elaboration Likelihood Model (TI-ELM), the framework advances understanding of how consumers process content from both user- and business-generated sources. It also lays the foundation for examining emerging dynamics such as extremely helpful reviews—those garnering disproportionate consumer endorsement even in nonsocial review environments.
Structured in three empirical chapters, the dissertation builds a systematic multimodal perspective on review helpfulness. The first study introduces the TI-ELM framework and applies it to a large-scale dataset of user-generated online hotel reviews, custom-extracted for this research. It models helpfulness as a function of three psychological mechanisms—informativeness, persuasion, and credibility—operationalized through review quality, message neutrality, information entropy, information depth, message and author credibility. Using negative binomial regression, the study finds that multimodal reviews significantly outperform text-only reviews, with Persuasion—excluding contextual drivers—emerging as the most influential psychological mechanism. It also develops an innovative perceptual hashing algorithm to quantify new visual information in review images.
The second study applies this TI-ELM framework to Business-Generated Content (BGC). While user-generated content (UGC) has been the primary focus of prior research, firms now actively participate in review environments by posting branded content. This study compares the effectiveness of BGC and UGC in driving helpfulness and examines how aligning business-generated to user-generated visuals boosts helpfulness and strengthens persuasion and informativeness drivers. The findings suggest that BGC is significantly more credible and effective when it contextually aligns with UGC. We provide algorithmic strategies for managers to post content that maximizes helpfulness engagement and enhances business credibility.
The third study investigates what distinguishes extremely helpful reviews—those in the top percentile of helpfulness distributions—from moderately helpful ones. Using a cluster-then-predict strategy and Bayesian quantile regression (INLA), the study identifies that content and visibility features—such as reviewer and seasonality—play a more dominant role in the extreme end of the helpfulness distribution. Conversely, extremely helpful reviews of popular hotels rely more on content drivers such as entropy and text sidedness than on other contextual and visibility features. These findings deepen understanding of "virality" within nonsocial media environments, showing that online review helpfulness is highly context sensitive in nature.
Recommended Citation
Aguado Marin, Alvaro J., "Beyond words: a systematic multimodal framework for text, images, and extreme helpfulness in online reviews" (2025). Dissertations. 1863.
https://digitalcommons.njit.edu/dissertations/1863
Included in
Business Analytics Commons, Data Science Commons, Marketing Commons, Social Media Commons
