Can I Use An Average-Of-An-Average?
Frequently in the business of surveying your customers, running the numbers, and reporting to supervisors and peers, time can get a little crunched. It might be tempting to take a set of aggregated averaged satisfaction scores and do a quick-and-dirty average-of-an-average. Be careful, because that road is potentially treacherous. Take the following example:
Columns A, B and C contain averages and Top 2 Box counts from a 0-10 point scale survey.
Column D shows the results that would come from doing a quick average-of-an average, for both the mean score and a Top 2 Box %. Column E shows what the actual Average and Top 2 Box % are.
In this case, the differences are not marginal and, in fact, the cocktail-napkin, average-of-an-average method gleans lower scores than the actual scores!
Of course, the reverse could have been the case, too, where the actual scores are lower than what you get from averaging averages. But mathematical outcome shouldn’t matter since, in customer satisfaction research, aren’t we trying to get the most accurate number versus the better-looking one anyway?
–John Garza
bookmark 
email 
facebook 
twitter 
linkedin 