There are two Laws of Central Tendency -- Reality and Perception

From our very first year most market researchers are taught about the law of Central Tendency as it relates to repeated observations of similar events.  There is, however, another more discrete law of central tendency that is seldom written about.  That is the law of over and under estimates.  Actually I had forgotten about it until a couple of weeks ago when Mr. Herb Sorensen pointed out an article in the Journal of Marketing Research by Le, Ho and Toh titled "Are consumer survey results distorted? -- Systematic impact of behavioral frequency and duration on survey responses."    The article covers a case study conducted in conjunction with AT&T.  In the study they observe how actual behavioral frequency and duration systematically affect the direction of errors in the consumer responses.  Since they had the actual frequency of use and duration of each long distance call for each respondent, they could correlate the response with the actual number and duration of calls. The major discovery was that heavy users tend to underestimate both their frequency of calls and the duration while light users tend to overestimate their frequency and duration.

In most cases this phenomena may be of little importance.  However, if the study involves relationships between heavy and light users, it can be of great importance.  My first thoughts revolve around clinical research as one of the more important types of research where very accurate information is required.  Unfortunately in our world of market research we seldom have a secondary source by which we can assess each individual as in the AT&T study.  In these cases, I would frequently use diary responses in the place of a personal estimation.  Diary data permits the researcher to look at the distribution of responses and assess the nature of the distribution as well as establishing various cut-off points.  It allows the researcher to be more of an analyst.

How much, if any, is this phenomena a factor in studies forecasting purchases of a new product?  This is another reason why I prefer methods similar to the Disposable Test market where your forecasting is based on actual purchases and not stated purchase intent.  Do the over and under estimates compensate for each other?  It can be a big gamble.

Recently I came across another example of the above type of central tendency.  This case involved the estimation of knowledge among students by students.  It turned out that bright students underestimated their level of knowledge while less bright students had a tendency to overestimate their level of knowledge.

The more I thought about the student example, it reminded me of our profession.  Have you ever noticed the same phenomena among the practitioners of our trade?  That is, the more someone knows about our trade the more they are concerned with how projects are carried out.  While those with little knowledge or experience seem to be far more confident about their abilities.  I'm concerned if the less experienced individual really believes his abilities to be beyond the realm of truth or is it a sales approach?  If he really knows his limitations, I am comfortable that he will take some action to minimize the effect on the study.  However, if he truly believes his abilities to be beyond practical, I'm in trouble.  Who said "Buyer Beware"?

There are two laws of central tendency, reality and perception.  Both are important and both can have a profound effect on your research.