Views from the Hills by R. E. Stevens, GENESIS II (The Second Beginning) E-Mail views@aol.com

Ruler vs. Pie Market Models

Recently in a discussion on appropriate analysis of pair testing, the concept of market models came up. I was surprised that the concept of Ruler vs. Pie Models was alien to those present.

The concept of the Ruler/Pie models was popular in the 1950's and 1960's. Basically the model separates the consumer's choices into two important groups, those where there is a common ideal vs. the case where different consumer groups have different ideals.

As the title, "Ruler" implies, these are situations where the consumers agree on the ideal. "Ruler" situations are usually present in "needs" categories such as less dandruff, fewer cavities, shinier floors/windows, cleaner clothes, etc.

"Pie" situations are best described as preferences dealing with "wants" such as flavors of ice cream, types of music, types of food, odor characteristics, colors, etc.

The use of averages and significance testing is quite appropriate in the "Ruler" situations. However, averages and significance testing in the "Pie" situation can be seriously misleading. In the "Pie" model there is no real "average" consumer. The data must be viewed in the appropriate segments.

For example, a 40/60 result in a "Ruler" situation may indicate a serious negative in the 40% product or an important benefit in the preferred product. In either case, the negative or advantage is seen by a minority segment of the panel. However, in a "Pie" model, the 40% may be an indication of a strong candidate in a segmented market (reference Views, March 3, 1996, " Republican Primary and Product Research").

Significance testing must be used with good judgment. If it is routinely used to test all pair tests against the 50/50 hypothesis, it may actually contribute to a misinterpretation.

When interpreting data, I lean heavily on individual ratings of products and their distributions along with the voluntary comments or reasons for preference.


[Back][Index][Forward]