How Much Testing is Too Much?

The article actually had nothing at all to do with Market Research… at least not on the surface. It was a discussion of a debate stirred up by Mark Cuban (owner of the Dallas Mavericks and entrepreneur).

Nearly all men can stand adversity, but if you want to test a man's character, give him power.

Abraham Lincoln

Mark Cuban feels having regular (say, quarterly) blood tests to “establish a baseline of your own personal health.” this is a great way to stay on top of our health situation, allowing us to better understand when something is out of line and abnormal. A number of “experts” though, including many in the medical profession, claim that this is a bad idea. Their argument revolves mostly on two things: First, that frequent testing will result in more cases of “false positives”, thus leading to unnecessary treatment and any potential side effects. The second major argument is that one’s health is better when it is not being obsessed over. They claim that a big part of being healthy is just taking good care of yourself and not worrying about what “might” happen. There are “experts” on both sides of the debate and it is no doubt an interesting discussion.

I see a distinct parallel though, with some of our work as marketing researchers. There are two things that immediately come to mind. The first involves Statistical Significance Testing. When a study is complete, one of the first questions that come from a client is always, “Are there any significant findings?” Often, we (in part) answer that question by looking over a mountain of data tables for those little letters that denote “statistical significance”. These come about as every single piece of data is essentially compared to every other single piece of data to determine where the “significant” differences exist. Every time I am a part of this exercise I literally hear my graduate advisor’s voice in my head saying, “Now remember, one out of every 10 of those significant findings are just noise”. I’ll spare you the lesson here on hypothesis testing. Suffice it to say, every time I engage in this process I get a strong check in my statistical spirit. I know that we are finding things that aren’t real just because of the sheer number of things we are testing. And if decisions to change things are being made on the noise and not on the signal, then we are headed for trouble. Both the companies we advise and we as researchers, need to be looking at more than just “statistical significance”.

Another, parallel I see with the medical testing discussion involves tracking studies. This one I think is a little trickier. In many ways, the analogy to medical testing is much more direct. We track, for example, our brands “health”. We do so that we have a baseline of understanding of what “normal” is. This is almost the exact language that Cuban uses in his argument. Tracking gives us the means to determine when something changes and alerts us that we may need to do something to address or otherwise remedy the situation. In a marketing situation, how would we know to do this without the tracking component? How would we know when we needed to make changes? Should we just do the best job we can with all aspects of the business and trust that it will all just work out? But once we do learn of a problem, will it be too late?

I’m interested in your thoughts on this one… Please comment below with your take, either on the medical testing issue, or even better on the application it has on Marketing Research. I look forward to seeing everyone’s thoughts!

~ Bud Sanders


Leave a comment


  • No comments found