Economics 101: how can we improve online reviews’ accuracy?

False reviews leave consumers lost, and causes serious damage to honest sellers who are trying to earn a living

FILE PHOTO - Expedia CEO Dara Khosrowshahi poses for a portrait during the 2010 Reuters Travel and Leisure Summit in New York, U.S. February 22, 2010.  REUTERS/Lucas Jackson/File Photo
Powered by automated translation

The volume of e-commerce continues to expand rapidly across the entire world.

The ease of such transactions comes at a cost, however, which is the inability to physically inspect merchandise prior to purchase, as well as decreased convenience of returns of defective or unsatisfactory products. This is why online reviews have become such a central part of the e-shopping process - many consumers refuse to make a purchase unless the product has received a high rating. But can we trust the reviews that we read?

Online reviews are an unusual service from an economics perspective for several reasons. Many of the traditional economic principles do not apply, meaning that we have to look for unconventional ways of improving quality.

The first issue with online reviews is that it is impossible to give people an incentive to deliver accurate ones. Ultimately, how you feel about the quality of a product is your personal opinion, formed within your mind. Nobody can demonstrate that you are lying; even if you give inconsistent opinions over time, you can justifiably claim that you have changed your mind. Thus, even if someone paid you to express your opinion, they have no way of ensuring that what you outwardly express is the truth.

The second issue is that people have intrinsic incentives to provide online reviews for free. A minority of consumers will take the time to express their opinion, and these are the opinions that you ultimately benefit from when you make your own purchases. The problem is that the intrinsic incentives to provide online reviews are not distributed evenly across all opinions. Instead, people with extreme opinions have a strong intrinsic motivation to express it.

This is especially true in the negative domain. Humans have a natural tendency to react much more adversely to negative stimuli than they are to react favourably toward positive stimuli, including the sometimes cathartic experience of expressing your disgust for a bad product over the internet. This means that voluntary reviews are not representative of the underlying reality of opinions.

You probably already realise this, and make an adjustment in your mind, such as reducing the weight you place on bad reviews. However, this process is undermined by the fact that various groups have an incentive to articulate inaccurate opinions. The vendors may fabricate positive reviews, and their competitors may fabricate negative reviews.


Read more:

Walmart in talks to buy more than 40% of India's Flipkart - sources

PayPal slumps as former parent Ebay chooses new partner


A recent study by Dina Mayzlin (University of Southern California), Yaniv Dover (Hebrew University), and Judith Chevalier (Yale University) demonstrated this subversion of the review process very clearly. Travel website TripAdvisor allows travellers to review hotels even without a booking, while Expedia requires a booking via the website. The authors found that hotels that had a neighbouring competitor were much more likely to have a 1 or 2-star review on TripAdvisor than on Expedia, most likely as a result of efforts at reputational sabotage.

False reviews leave consumers lost, and causes serious damage to honest sellers who are trying to earn a living. What can be done to tackle these problems?

Online exchanges have made significant progress in addressing the problem. They use techniques such as rater reputations (Amazon), requiring purchases (Expedia), and monitoring the activity of accounts suspected of false reviews (eBay). While not perfect, on the whole, such methods have allowed consumers to feel confident that online reviews reflect real opinions, especially when shopping at major retailers such as Amazon.

But what about the issue of ensuring that online reviews are representative of all purchasers, and not just ones who were extremely happy or unhappy? A study by Ioana Marinescu (University of Pennsylvania), Nadav Klein (University of Chicago), Andrew Chamberlain (Glassdoor - an online recruiter), and Morgan Smart (also Glassdoor) sheds light on the issue.

The authors conduct a series of experiments to compare a baseline situation where online reviews are voluntary, with alternative situations where an incentive is given to users. They work Glassdoor, which allows employees to review employers. They initially hypothesise that reviews are biased toward the negative component of the spectrum, because disgruntled employees are more likely to take the time to volunteer a review than ones who feel satisfied.

The study found that when employees were given a small financial incentive to deliver a review, participation in the review process was significantly higher. Moreover, they found that the reviews were, on average, more positive than the reviews that would emerge if the website relied exclusively on voluntary reviews. They found a similar effect of a non-monetary incentive, which was a short message urging employees to review employers on pro-social grounds (helping to identify good employers and punish bad ones).

Notably, the financial incentives necessary to induce reviews were not particularly large. For this reason, if this result is replicated by other teams of researchers, or by companies conducting their own field experiments, you should expect to start being offered financial incentives to review products that you have purchased, or possibly even messages reminding you of the benefits to other users of your providing an accurate appraisal of your purchases.

Omar Al-Ubaydli (@omareconomics) is a researcher at Derasat, Bahrain.