Watch out for averages

We're inundated from numbers and alleged data from so many sources. I get especially annoyed when I see purportedly meaningful analysis based on the average of some parameter, or the average of one parameter compared to another. Let's be honest with ourselves: in most cases, average as a data point is at best meaningless, at worst misleading.

It's not just in the political or economic sphere, either. Consider two power supply designs, one for a relatively constant load and one that delivers power to a camera flash. Their average power and energy demand may be the approximately the same, but the requisite design challenges are substantially different in almost every respect, including the sourcing battery capacity, IR drop, thermal issues, and component and connector sizing.

Why am I so annoyed about this use and misuse of “average”? It's for several reasons. I see people try to put a quantitative veneer on their opinions and analysis, to enhance their apparent credibility (sorry to say, journalists are major offenders here); I see quoting of average values being conveniently used to show that some things are roughly equivalent when they are not (or the reverse—that they are very different, when they are surprisingly similar); and I see an oversimplification of analysis which is often the result of shallow or lazy thinking.

As engineers, we know how to use numbers and do fairly decent analyses when called needed. But we should also recognize that many of the analyses we see, and which are used to set guidelines and direction, are often flawed. I'd rather hear someone say “this number is an average, but it doesn't really prove much–and there is large uncertainty in the underlying data” than try to infuse their speculation and personal belief with numbers.

0 comments on “Watch out for averages

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.