Averages or Top Boxes?
May 17, 2005
- by Robert E. Stevens, GENESIS II(The Second Beginning) E-Mail: firstname.lastname@example.org
One of the many controversies surrounding data analysis is how we view product or concept ratings. Some, and I should say most analysts, prefer to use averages. There are others, however, that swear by top box analysis.
Why not look at both? Each approach yields a picture of the market potential from a different perspective. Each approach utilizes different assumptions. Each requires a different type of statistical analysis, parametric vs. non-parametric.
I have a habit of initially looking at the distribution of scores for each product or idea. Of prime importance is determining if we are operating in a “Ruler” or “Pie” market model. That is, could I determine if there is a single standard of excellence among the participants or are there disagreements pertaining to excellence between panelists. To do this, I look at the distribution of votes for each product to determine if the variance of each is normally distributed around the mean. If they are normally distributed, I can utilize the various statistical analyses such as the “t test” and the “Analysis of Variance.” I can still remember, after a half century, the standard phrase of statistics 101, “the errors must be normally distributed with a mean of zero and a standard deviation of one. “The classical statistical tests require that the errors be normally distributed.
If there are disagreements among the participants, that is the distribution of scores are not normally distributed, we are probably operating in a “Pie” market. In a “Pie” market, there is disagreement as to what is best. For example, the ice cream market is a prime example of a “Pie” market. There is a disagreement as to which is the best flavor. The score distributions will generally appear as a bi-modal distribution. Whereas, a ruler market is usually dominated by the end result criteria such as fewer cavities, cleaner clothes, more absorbent, etc. where there is a universal agreement among the participants concerning the standard of excellence.
When performing up-stream research, we should always look for ideas that appeal to a strong minority segment. My thinking is that a product of this type could be the next Tide or Pampers. Both of these brands had trouble initially when paired with their current competition.
The introductions of brands similar to Pampers (disposable diaper) and Tide (synthetic detergent as it was called in those days) brought paradigm shifts to their category. Each in their initial introduction was not preferred among the alternatives by a majority of the consumers. New technology that replaces current technology takes time for acceptance among a majority of the consumers, while new technology that adds to and does not replace current technology has a faster rate of acceptance, for instance, compact disks and cellular phones. This may be easier seen in the association with evolutionary and revolutionary new products. Evolutionary products usually bring together new technology with an already acceptable product execution, while revolutionary new products require the acceptance of a completely new bundle of features and benefits.
In evaluating the potential of revolutionary positioning, the important reads are: strength of acceptance among a distinct population segment, size of the segment, frequency of need, number of competitors, and how well the product delivers on its promise. In the case of a revolutionary product, the ability to win a paired comparison blind test has little to do with its potential market success.
Sponsor: Sorensen Associates Inc
~ ~ Oregon: 800.542.0123 ~ ~ Minnesota: 888.616.0123 ~ ~ Pennsylvania: 866.993.0123 ~ ~
the in-store research company™
-- Dedicated to the relentless pursuit of WHY?