“Businessmen are finding that while they have more numbers than ever before, they still do not know what these numbers mean.”
“This book was written to help organizations overcome the effects of numerical naivete.”
Understanding Variation is a classic in understanding and measuring how businesses work. Its key lesson is that things vary. When we see that trade is down in February compared to last year, it’s tempting to assume that means something. The media will certainly report it as such, but we really don’t have any reason to do so. For that data to be meaningful would require that last year be ‘normal.’ I’m not sure that’s ever happened.
Variation is a significant challenge for businesses. If a business has bad results one month, should it make changes? Wait? Ignore it? There are no hard and fast rules, though entire disciplines have arisen over what type of results mean something (see: Six Sigma). Even good managers fall victim to regression to the mean: when they see results that are worse than average, they reprimand the employee, and things get better! Of course, when they see good results, they reward the employee, and things get worse. Managers learn to always be mean, rather than the true lesson that things can’t permanently stay above or below average.
Wheeler presents a few key rules, including that no data have meaning apart from their context, and that to detect a signal, filter out the noise. Much of the book, however, focuses on laying out tools that can help businesses identify meaningful variation. Wheeler swears by the control or XmR chart, a time series combined with a graph showing the size of the change each period. It means you can track both level and variability simultaneously, and look for significant changes in either.
The book is a bit outdated now, but the lessons ring true. Variation was and remains a challenge, and one it is far too easy to neglect.