Female Winemakers In California Are Generally Better Than Their Male Counterparts --A Rejoinder
June 29, 2015
Our research focuses on illuminating the presence of California women winemakers and their contributions to the wine industry (see www.womenwinemakers.com).
Our first study assessed the perception that women winemakers have shattered the glass ceiling. Results showed that only 9.8% of California wineries have a woman as the lead winemaker, illustrating a discrepancy between perception and fact. Our second study investigated whether winery acclaim was associated with this discrepancy. Coding of winery data in Opus Vino (2010) provided support for the hypothesis of proportionally greater acclaim for wineries having women as their lead winemakers.
Imagine our surprise to see a blog post highlighted by Wine Industry Insight on June 25th (see reference below) stating that our studies “purported to show that women are more successful winemakers than are male winemakers” and using the results of our studies in making a causal claim that “female winemakers in California are generally better than their male counterparts.” The research methods, data, and conclusions made in the blog post were so flawed that we as educators are obligated to make a response that we hope others will find instructive.
The post reported on three “methods.” The first “metric and measure” involved using our data on the percentage of wineries, by CA wine region, with lead women winemakers, somehow adjusting these percentages without access to the original data. The blogger then looked at how these percentages fit with some numbers, themselves never revealed, that were distributed “against four tiers of regions.” Based on this magical method, the writer then declares that the score is “Gals 1, Guys 0.”
Sorry folks, but one needs to (1) use accurate and clearly defined data in calculating correlations; (2) employ the appropriate statistical analyses to calculate a correlation coefficient; and (3) then use a probability table to decide if the relationship between the two variables is statistically significant.
The author’s “second metric and method” employed the 2014 Wine Spectator Top 100 wines. The blogger identified 20 CA wines in the set of 100 wines and then coded those wines by the sex of the winemaker. From this coding, the author claimed that 3 of the 20 wines (15%) were made by women winemakers.
The codings of winemaker sex are not correct, however. One of the wines coded “F” (for female), Turley Zinfandel, was not crafted at Marcassin by Helen Turley but rather at Turley Wine Cellars where Larry Turley is the Proprietor. Another of the wines coded “F” is Ponzi Pinot Noir Willamette Valley, which obviously is not a California wine.
Correcting that error drops the number to 1 for wines crafted by a female winemaker out of the now-reduced total of 19 wines, or 5.3%, rather than the 15% computed by the blogger. This makes the conclusion reached by the blogger that there “is a huge (50%) over-representation [of women]” erroneous. (In addition, here again, the appropriate statistical test should have been calculated before reaching any conclusion.)
We could continue our critique with other examples but will instead end our lesson on research methods by reminding readers that research is a serious enterprise. It is not a game in which one makes up the rules and then counts to see who wins. We were particularly troubled that the blogger misrepresented our research and its purposes. Our research is about illuminating the presence of women winemakers in their roles as lead winemakers and recognizing their significant contributions in a field that remains male dominated.
One last educational point. A correlation describes the relationship between two variables, such as the amount of chocolate consumed by a country and the number of its Noble Prize laureates. There is no causal relationship here. One can feast on chocolate for every meal but that will do nothing to increase the chances of being awarded a Nobel Prize.
The same principle applies to the correlates of winemaking: A correlation between, say, the cost of wine and the percentage of women who are the winery owners or winemakers only tells you that there is an association between the two variables; one does not cause the other. All to say, one cannot conclude that women or men are better at winemaking from looking at the association between their percentage as lead winemakers in a wine area and the cost of the wine or its ratings.