Shopping online requires a lot more strategy than swinging by a department store. When the item you’re looking for has options from dozens of brands, all selling for similar prices, making an informed purchase often comes down to sifting through product ratings. If you’ve used this plan of attack before, a team of researchers has some advice for you: When it comes to reviews, trust quality over quantity. Following that rule will help you choose the better product, but according to their new study published in the journal Psychological Science, consumers tend to do just the opposite.

As Quartz reports, researchers looked at 15.6 million reviews of over 350,000 products on Amazon for the study. Plugging the data into a statistical model proved that a product with a high number of subpar reviews—say, a pair of headphones with an average rating of 2.6 and 200 reviews—is likely to be lower in quality than a different pair of headphones with the same rating and only 10 reviews. This isn’t a huge surprise—it makes sense that an item with more poor reviews has a greater chance of actually being shoddy than an item with less of them. But seeing those large numbers seems to give people a much different impression.

For one part of the study, subjects were shown two Amazon products that had both been given an average rating of 3.1 stars but had different review quantities. The first product, which had 154 reviews, had just a 40 percent chance of being superior to the second product with 29 reviews. Swayed by the power of popularity, 90 percent of participants chose the item with the most reviews. When the researchers conducted similar experiments, the would-be shoppers chose the most reviewed (and likely worse quality) products 72.3 percent of the time. “Overall, participants’ judgments suggested that they failed to make meaningful statistical inferences,” the researchers wrote in the paper.

Online customers may not always be able to spot the best purchases based on instinct, but that doesn’t mean they can’t learn to change their ways. If you’re a habitual Amazon shopper, start your searches by branching outside the featured sellers lists and looking for deals that don't necessarily benefit from the retailer's algorithms. Once you’ve found a few products you like at prices that work for you, then you can start parsing reviews—just remember to keep your popularity bias in check.

[h/t Quartz]