(Update on 23 June 2008: The final paper is now published. A summary is available here. Our summary below is based on the working draft - some aspects possibly changed prior to final publication.)

The last year or so has seen an awful lot of interesting research being done on the consumer’s preferences for wine.

The most recent entry into this field of research, by Goldstein, et al, is a fascinating study of the relationship between consumer preferences and price. The conclusion from this study was that less experienced drinkers prefer cheaper wines.In the study, tasters were given different wines to try but were not given any information about price. A book is due out soon that is supposed to give a more readable version of the study.

Earlier in the year there was a study that showed that when someone thought a wine was expensive they preferred it a cheaper wine. In this study subjects were given a wine to taste and told a price.The price they were given was not necessarily real, and in some cases the same wine was given multiple times but with different prices being stated. Subjects, in general, preferred the wine when they thought it was more expensive! This implied that people extracted pleasure (utility) from believing that they were enjoying something exclusive, or at least expensive.

Let’s take a moment to deconstruct the most recent paper – that inexperienced wine drinkers prefer cheaper wine compared to experienced drinkers.

• 506 people participated in 17 separate tastings, for a total of 6,175 observations.
• 532 different wines were tasted
• Each person rated each wine that they tried on a scale from 1(bad) to 4 (great)
• Ordinary least sum of squares (OLS) regression is used to determine coefficients relating score to price for each wine. Ordered probit was also used, but not discussed in depth other than to say that the results from ordered probit yield the same interpretation as from OLS.

The author’s conclude that for people without any formal training in wine (about 83% of the tasters), their opinion of any particular wine was negatively correlated with the average retail price of the wine. The trend was opposite for those with some formal wine training (17% of the tasters).That is, people experienced with wine generally gave higher ratings to more expensive wine.The author’s provide a simple equation relating score to the natural log of the average retail price in US dollars:

Score = -0.048 x ln(price) + 0.138 x ln(price) x Expert – 0.448 + 2.337

where “Expert” is equal to 1 if the taster has had any wine training, and 0 if they have not.

That equation tells us some interesting things. The plot below shows the calculated score based on price for the non-expert (shown in blue) and the expert wine drinker (shown in red). Two things jump out – first, the trends go opposite each other, consistent with the authors’ conclusions about wine-training, price, and score. Second, the dynamic range of calculated scores for the non-experts is much narrower than for experts.

What does this mean? I’m not sure. But I find it very interesting that over a $100 range of prices, the “expert” scores change from 1.95 to 2.30 (a range of 0.35), whereas the “non-expert” scores range between 2.30 to 2.11 (a range of 0.19). This fact, by the way, is also readily apparent by the differences in the effective coefficients applied to the LN(price) (-0.048 for the “non-expert” and 0.09 [or -0.048+0.138~0.09] for the “expert”). In other words, the size of the coefficient for the expert is roughly twice that of the non-expert. Keep in mind that the non-expert coefficient was more significant from a statistical point of view, even if it is smaller in magnitude.

This article has certainly brought no shortage of interest on-line. Everyone from Vinography to the NY Times to the NY Times (again) to Newsweek to Portfolio.com has discussed the article. And judging by the comments at Eric Asimov’s column at the NY Times, some wine folk are non-to-happy about the conclusions.

We are not terribly surprised. One article shows that people prefer wine that they think costs more, and another shows that what they actually like actually is cheaper on average. It is fun to try and guess why this might be.

Could it be that wine that large numbers of people like is made in great quantity. Vineyards that make this wine therefore look for more efficient ways to make it. This added efficiency, in turn, results in lower prices.

Or maybe the less-educated wine crowd just do not like the taste of the unique aspects of terroir, or tannins, or whatever.

Or maybe they like what they are familiar with. They drink cheaper wine because they can afford cheaper wine. They like cheaper wine because it is what they are used to.

Perhaps between the two studies this shows that vanity is the strongest driver - we want to impress with our wine. Our drive to impress outweighs our desire to have the thing we actually prefer more. I can believe that.