Stars versus Bars: How the Aesthetics of Product Ratings “Shape” Product Preference

He (Michael) Jia*, Echo Wen Wan, Wanyi Zheng

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


Websites commonly use visual formats to display numerical product ratings. Highlighting the overlooked notion of the “aesthetics” of product ratings, the current research examines how the shape of basic visual rating units (rectangular vs. non-rectangular) influences product preference. Seven experiments (and 23 supplementary experiments; N = 17,994) demonstrate a visual rounding effect. Specifically, compared to the rectangular rating format (e.g., bar ratings), the non-rectangular rating format (e.g., star ratings) increases product preference when product ratings (e.g., 3.7, 3.8, 3.9) are below the nearest integer. In contrast, the non-rectangular rating format decreases product preference when product ratings (e.g., 4.1, 4.2, 4.3) are above the nearest integer. Occurring for both the overall rating and by-attribute ratings of a product, the visual rounding effect results from a visual completeness restoration process, wherein consumers perceive non-rectangular rating units to be incomplete after vertical cutting. This research contributes to the product rating and visual marketing literatures and provides actionable implications by demonstrating what visual rating format should be adopted based on rating distribution, how the visual rounding effect can be prevented if needed, and who are even more susceptible to the visual rounding effect.
Original languageEnglish
Article numberucac043
Number of pages25
JournalJournal of Consumer Research
Publication statusE-pub ahead of print - 26 Sept 2022

User-Defined Keywords

  • product rating
  • shape
  • visual marketing
  • aesthetics
  • completeness
  • preference
  • numerical cognition


Dive into the research topics of 'Stars versus Bars: How the Aesthetics of Product Ratings “Shape” Product Preference'. Together they form a unique fingerprint.

Cite this