Shoppers warned to not fall for star ratings after UBC research finds reviews can be misleading

Star ratings vs numerical ratings can influence online shoppers in different ways, says UBC study. (Courtesy: UBC)

Online shoppers might change the way they regard purchase reviews following new research from UBC that says the ratings could be misleading consumers.

The study, carried out by UBC’s Sauder School of Business and published earlier this month, found shoppers browsing online typically overestimate star ratings but underestimate those given in a number format.

A popcorn machine with a 3.5 star rating would typically be rounded up to a four, for example, while a 3.5 numerical rating would be rounded down to a three.

“If there is a rating using stars, people tend to complete the incomplete picture, that’s why it feels more than what it is,” said the study’s lead author Deepak Sirwani.

“When the same rating is presented numerically, people focus on the first digit, and that’s why it seems slightly less than what it is. That changes how people think about the product, which drives perception about the product quality and whether they want to buy it or not.”

Sirwani, who began the research with colleagues from New York’s Cornell University before publishing the study as a marketing and behavioural science professor at UBC, said the study involved around 15,000 paid-survey participants from across the U.S.

The project recruited a further 40,000 people from Meta, via ads sent out onto social media platforms Facebook and Instagram. A/B testing was used to ensure the same ad with two different rating styles could be delivered to social media users.

Half of those surveyed were delivered the ad with star ratings, while the other half would have seen the ad with numerical ratings.

“We observed which ad was more likely to get more clicks,” said Sirwani.

Consumers faced with a three-and-a-half or four-and-a-half star rating likely wouldn’t realize they are rounding up as the quick assumption, the same “filling in” process that helps identify a friend behind a mask or a partially hidden logo, is made within the first 10-20 nanoseconds of looking at a rating, said Sirwani.

“It is extremely automatic what we do. It’s why I tell people that, whenever they see a star rating, try to convert it to a number first before thinking about the rating and making a purchase.”

Researchers are hoping the findings from the study will help push for a regulatory body that can intervene and set standards for the consumers’ rating system.

“If companies have an unconscious process by which they can manipulate people, then there needs to be a body to intervene and stop this,” said Sirwani.

“Let’s not fool consumers by showing these stars.”