Women are from Venus, men drink scotch

Editor’s note: Robert Sims is senior market research analyst at Market Dynamics Research Group, Inc., New Orleans.

Not too many years ago the notion that “real men don’t eat quiche” gained some notoriety, and popular psychologist John Gray has made a career out of his observation that men and women are from different planets. While it is not a novel notion that men and women act differently, think differently and have different preferences, these differences do, nonetheless, create a problem to the manufacturer or distributor who seeks to market to one sex while not alienating the other.

This was the problem faced by the Sazerac Company, Inc., a New Orleans-based alcohol distributor, which was seeking to create a new, more modern, sexy image for a vanilla liqueur product, Dr. McGillicuddy’s Vanilla Schnapps. A new bottle design was developed, and several new names were tested including Kiss, Vanilla Kiss, French Kiss and French Vanilla Kiss. The product was to be designed primarily to appeal to women, but not so much so as to represent “quiche” to “real men.”

Part of the research task that we faced was to test the willingness of men to try the newly repackaged product. The research was designed to test a model developed by a member of our staff. The model was developed based on the results of a series of focus groups that we conducted and works on three basic assumptions:

Assumption 1: Though men and women likely share some preferences for alcoholic beverages, there are some drinks that are perceived to be distinctly feminine and some perceived to be distinctly masculine.

Assumption 2: Alcoholic beverages can be arrayed on a continuum ranging from most feminine to most masculine.

Assumption 3: Male consumers occasionally drink certain drinks that they consider closer to the female end of the spectrum. But there is a point at which a drink becomes too feminine for them.

Our task was to find out whether or not the client’s product fell beyond the range of drinks that men would feel comfortable ordering.

Ample basis

In addition to our own focus group findings, previous research provides an ample basis for the assumptions. For example, Traffanstedt, Rohr, and Sparks (1997) found that, though rates of alcohol consumption did not vary by gender, men were more likely than women to prefer beer as their main alcoholic drink. Ullrich and Tepper (2000) found that, due to physiological differences in taste sensitivity, men were more likely than women to prefer bourbon, gin and beer. These physiological differences may account for the psychological associations that have developed in terms of what constitutes masculine as opposed to feminine drinks.

Furthermore, the feminine-masculine dimension of alcoholic beverage preference is not a uniquely American phenomenon. Research conducted by the Social Issues Research Centre (2000) concludes that: “...association of the qualities of men’s and women’s beverages with ‘masculine’ and ‘feminine’ attributes is...a near-universal phenomenon. ‘Feminine’ drinks are often weaker, sweeter, softer or less ‘pure’ than their ‘masculine’ counterparts.”

Develop instrument

Given a reasonable basis for the assumptions of our model, the most difficult problem that the research question posed was how to develop a measurement instrument capable of validly rendering the presumed feminine-masculine dimension of alcohol preference. Data for this study were collected via intercept interviews conducted among bar customers in two states in which the product was marketed.

Two measurement possibilities immediately came to mind. First, respondents could be presented with a numbered scale. On one end of the scale would appear words to the effect of “most feminine,” and on the other the words “most masculine.” Respondents could be asked to indicate the point on the scale that served as the borderline between the point at which they would feel comfortable ordering a drink and the point at which they would not. They would then be asked to place the client’s product on this scale, and comparisons would be made to see whether or not the product fell within the respondent’s comfort zone. Though primary interest here was in male respondents, female respondents also would be asked to complete this exercise as a validity check.

Scale problematic

Though the analytic strategy implied above seemed reasonable, the scale itself seemed problematic. First, using simple masculine/feminine terminology may have been too abstract for this purpose as it assumes that respondents know what types of drinks fall into these categories. Second, men may have been hesitant to place themselves on the feminine side of the scale, especially not knowing what drinks that side represented. Conversely, men, not wanting to appear old-fashioned or to deny their “inner woman” may have been inclined to place themselves too far on the feminine side, or farther than they would actually be willing to go.

The solution to these problems was to develop a second instrument, which would be used like the first, but which would feature at each scale point the name of an actual alcoholic drink. The main problem here would be to validly array drink names along the masculine-feminine continuum, and to do so in such a way as to be understood by respondents.

Working in conjunction with the client, eight alcoholic beverages were selected for placement on what would form a seven-point scale. They were, moving from most feminine to most masculine, a cosmopolitan, a strawberry daiquiri, a white Russian, rum, vodka, beer, bourbon, and scotch. Survey respondents were shown a scale featuring these drinks and then asked to mark the point on the scale at which they would no longer feel comfortable ordering among a group of friends. (See Figure 1 for the actual question as presented to respondents.) They were then asked to place the client’s product, French Vanilla Kiss, on this scale. The scale ranged in value from 1 to 7, with 7 representing the most masculine point.

Figure 1

Given the novel nature of the measurement scheme, it was imperative to validate the measure once the data were collected. Assumption 3 implies that male drink preferences should be bounded somewhere on the feminine side of the preference scale, though not at the extreme end of the scale. Construct validation of this assumption would involve demonstration that the average boundary score for males was less than 3.5 (the scale midpoint) though greater than 1 (the extreme feminine end of the scale). In correlational terms, the assumption would be a negative correlation between gender and boundary placement.

Table 1

Table 1 presents the results of the construct validation test. A simple regression of the alcohol preference scale on gender indicates confirmation of the assumption and validation of the measure. The correlation of gender with the preference scale is, in fact, negative and significant as indicated by the coefficient for gender (coded as a dummy variable) and its associated standard error. The gender coefficient, in conjunction with the constant for the regression equation, indicates that the average male boundary score (2.86) is in fact to the feminine side of the scale, though not at the extreme (Cohen and Cohen, 1983). These tests serve to validate not only the assumption concerning male preferences, but also the ordinality of the scale.

Assured of the validity of the measurement scheme, the final step in the process was to attempt an answer to the client’s question, i.e., was French Vanilla Kiss too feminine for men? Average scores on both the alcoholic beverage preference scale and the French Vanilla Kiss placement scale were computed. A positive test from the client’s perspective would indicate a higher (more masculine) score for French Vanilla Kiss than the average male boundary score. Figure 2 indicates that, among male respondents, the average placement of French Vanilla Kiss (3.03) was within the male comfort zone. This average was close to, but significantly higher (p>.05) than the male boundary average of 2.86 as indicated above.

Figure 2

Figure 3
It should be noted that the average female product placement also fell well within the female comfort zone (see Figure 3). This result was not of particular interest to the client since it was assumed that the appeal of the product would be greatest for women. Nevertheless, the fact that the average product placement scores for men and women were equal, and the fact that the female placement score was much farther from the preference boundary than was the case among men, attests, once again, to the validity of the measurement scheme and of the analytic procedure overall.

Not unique

To note that men and women may have different product preferences is not unique, nor is it unique to take gender into consideration in analyzing research data. However, the problem faced here went beyond the mere demonstration or consideration of gender differences to the development of a measurement and analytic scheme for the purpose of developing a product primarily designed for women which would maximize general consumer interest in the face of gender differences. In so doing, it would appear that we developed a measurement that in large part avoids what Linz (1969) describes as the tautologous nature of much survey-based research. In using actual alcoholic beverage names to mark the scale, the measures were sufficiently distinct from the underlying conceptualization (the male-female continuum) to prevent false positive readings that often occur when theoretically distinct questions are worded so similarly that the wordings themselves generate positive results.

Finally, it should be stated that the general procedure developed here should be useful not only when gender-based preferences are involved but also when racial/ethnic, economic or other demographic related preferences are under study. All that is required is a product set that can be validly arrayed along the underlying demographic continuum.

References

Cohen, Jacob and Patricia Cohen. 1983. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Lawrence Erlbaum Associates, Publishers.

Linz, Juan J. 1969. Ecological Analysis and Survey Research, in Dogan, Mattei and Stein Rokkan, eds. Quantitative Ecological Analysis in the Social Sciences. MIT Press.

Social Issues Research Centre. 2000. Social and Cultural Aspects of Drinking, report to the Amsterdam Group presented to the European Commission, November 29, 2000.

Traffanstedt, Bobby K., Shelley Rohr, and Darla Sparks. 1997. Factors Related to Alcohol Preference and Consumption. Psi Chi Journal. Vol. 2 No.1.

Ullrich, N. and B. J. Tepper. 2000. Food preferences are influenced by gender and genetic taste sensitivity to 6-n-propylthiouracil (PROP), paper presented at the 2000 Annual Meeting of the Institute of Food Technologists. Dallas.