Is the Food Really Better at Expensive Restaurants?

The next time you find yourself reaching for an expensive purse or other luxury good, bear something in mind: a little-understood concept called heteroscedasticity.

People often assume that there’s a straightforward relationship between price and quality—a low price indicates something is of low quality; a high price indicates it’s high quality. However, this relationship is more complicated and can vary hugely depending on the price of the product at issue.

Quality can also vary more at one end of the price spectrum than at the other—the phenomenon of heteroscedasticity. Laundry detergent is usually of low quality at low prices, but of unpredictable quality at higher prices. The opposite is true of wine, for which higher prices predict better quality, but the quality is less predictable at lower prices.

To gauge how consumers are affected by heteroscedasticity, Chicago Booth’s Ann L. McGill, along with University of Colorado’s Bart de Langhe, Erasmus University’s Stefano Puntoni, and Cornell University’s Stijn M. J. van Osselaer, asked participants to look at the kinds of price listings and quality ratings that you’d see on a website such as Yelp. They had people view high-priced restaurants with consistently high ratings, and midpriced restaurants with variable ratings. When then asked to predict the quality of low-priced restaurants, participants assumed all low-priced restaurants were lousy.

The opposite was true, too. When shown listings of cheap, terrible restaurants, and less terrible midpriced restaurants, participants predicted expensive restaurants would be superb. In fact some expensive restaurants can also be terrible, but participants assumed otherwise, taking for granted a pattern linking quality and cost. “What you’re doing is making the relationship much tighter than it actually is. You’re making too extreme a prediction,” says McGill.

The same phenomenon is observable when people gather data over time, as opposed to simultaneously, as on Yelp. Another experiment presented participants with the prices of different bottles of wine, one by one, and their respective qualities. Later, the participants were asked to guess the quality of new bottles they were shown, based on what they’d learned previously. Again people used their knowledge about products at one end of the price spectrum to make too-extreme predictions about products at the other end.

The authors say marketers can use this phenomenon to their advantage. “Our studies suggest that low-cost rivals can be beneficial for the perceived quality of higher-priced brands if the quality of those low-cost rivals is consistently low. In fact, this may increase the effectiveness of the price signaling strategies that lie at the heart of luxury branding,” they write.

As for consumers, McGill says they can use the knowledge of heteroscedasticity to make smarter choices. “We shouldn’t make too much of the way the world works in our own narrow experiences,” she says.