Was the Uber taxi ride really that incomparably good? Or do you just not want to ruin the already difficult business for the poor guy behind the wheel? The fact is: the ratings with the stars that are very common online (1 = bad, 5 = great) are getting better and better. In the USA, for example, more than 90 percent of all trips with the Uber X service are now rated with five stars. Scientists have had the problem on their radar for some time and are talking about it review inflationvaluation inflation.
Rebecca Janßen, researcher at the Leibniz Center for European Economic Research (ZEW), and Matthew Ribar from Stanford University in the US have examined this Phenomenon assumed from a special aspect: Some users are quite familiar with the things they rate, others less so. The research question was therefore, in short: Does rating inflation also exist among experts or does it only apply to laypeople? The surprising finding: Experts are even more inclined to give good star ratings than less experienced users.
The model for the study was Vivino, a company that would like to become a kind of Amazon for wine and describes itself as the largest platform in this industry. Users can use Vivino to scan labels or enter the name of a wine and then retrieve information. You can buy wines and rate wines you have already purchased.
The scientists evaluated millions of reviews using machine learning
According to the study, Vivino was interesting because the platform not only allows star ratings, but also in text form, and users can also rate the ratings themselves. The researchers evaluated and classified millions of these texts using machine learning. This produced a clear result: the quality of the wines evaluated, as was evident from the text reviews, remained essentially the same in the study period from 2014 to 2020. But the star ratings have gotten better over the years. This proves the thesis of valuation inflation.
But where does this star flation come from? Why do people give more stars even though they are about the same products? ZEW researcher Janßen can only speculate. Psychological reasons could be responsible for this, as could an increase in fake reviews, she told the SZ. There are also other theories, including manipulation that users of online platforms are exposed to. For example, when sellers explicitly ask to rate their service and make it clear that they hope for a positive rating.
The experts often have more in mind with their reviews than just making the decision easier for other potential buyers. They sometimes also try to attract users to their own sites, where they can then generate advertising revenue or sell something themselves.
The results can be summarized as follows: Users should not put too much faith in pure star ratings, even if they come from experts. Reviews provide a better guide. This has also been shown by previous studies. After all, texts allow you to go into the properties of a product in a more differentiated way than pure numerical values between one and five.