Wednesday, December 23, 2015

Parameters

A parameter is a central underlying amount, figure, number such as the normal body temperature of a hummingbird, which is 107° F.  Often, parameters are a sort of human mathematical myth.  We measure the body temperature of 10 or 20 hummingbirds and take the average.  That average is our estimate of the "true" or "typical" or "normal" hummingbird body temperature.  It may well be that none of the birds had that average temperature.


Besides that, if you have read "How to Lie with Statistics" by Darrell Huff, you know that we might figure the average in several ways.  The figure that occurs most often (mode), the figure that is below half of the readings and above half (median) or the figure obtained by summing all the readings and dividing by the number of readings taken (mean).  In the end, we THINK that a hummingbird has something like an appropriate body temperature or a day has an appropriate high temperature.


One of my heroes, W. E. Deming, the man who put new emphasis on better quality in goods and services, had a rule to watch out for parameters, for numbers, especially in goals, and plans.  Maybe you have heard about the body mass index or the correct number of hours of sleep per night.  Deming cautioned to look behind such numbers.  Ask why that number?  What happens if the number is higher, lower, much higher or much lower?  What research, what reasons, what experience supports the number given?  How long has that number been the figure given and have conditions changed?  Has our thinking changed to where we should use a higher or lower number?


Back in June, I wrote a post about a Brookings Institute study that attempted to put a dollar figure on how much attending various colleges could be expected to add to a person's lifetime earnings.  As I read the study, I had to keep reminding myself that it was based on AVERAGES.  I kept thinking that there is variation around any average and that some of the figures being averaged were much higher or much lower.  I had to keep reminding myself that the averages concerned were, of course, related to what happened in the past.  As some markets emphasize, past performance is no guarantee of future success.


Make a note to yourself: when you spot a number being touted, question its accuracy and its justification.  You may have heard the saying: 98% of statistics are made up on the spot.




--
Bill
Main blog: Fear, Fun and Filoz
Main web site: Kirbyvariety

Twitter: @olderkirby

Popular Posts

Follow @olderkirby