You might be a geek if you recently searched Amazon for a copy of your first book on statistics. You just might be an old geek if it was published in 1954. And only a true fossil would turn to Mark Twain for his blog title. Well, my industrial amigos, I plead guilty, guilty and guilty.
Recent advertising and government blather (There’s fine line between the two, isn’t there?) sent me looking for How to Lie with Statistics by Darrell Huff, published in 1954. I bought the Kindle version, but you can go to http://archive.org/details/HowToLieWithStatistics for a free download. I first read the book in the Junior High School library, so it’s not the usual mathematical slog. In fact, I’d call it an entertaining must-read for anyone who leaves the newspaper or TV with that hollow I-think-I’ve-just-been-lied-to feeling in his or her stomach.
Huff didn’t write to help his readers lie with statistics, but to help them identify how they were being lied to. He covers a lot of the old standard statistical shell games, like false causal relationships in simultaneous occurrences. A modern example would be the recurring statement that taxes paid by the 1% richest Americans have been reduced during the Great Recession, while the rest of us are suffering financially. Of course the taxes of wealthy business owners fell in 2008 – 2011. Their income dropped off the map!
Your CMMS isn’t immune to causal distortion, either. Did the seal fail and let coolant into the bearing, or was the seal destroyed as part of the bearing change? Only your tradesperson knows for sure. Make certain that he or she completes a description of the failure for the CMMS as part of the job closure process.
Huff also covers the dodge of the carefully (or badly) selected sample. His example was the Literary Digest’s use of phone polls and subscription postcards to predict the presidential election of 1936. The magazine forecast Roosevelt’s landslide defeat by Alf Landon. It turns out that people who were buying Literary Digest subscriptions and talking on phones in 1936 were not a random sample of US voters. They were US citizens who could afford phones and high-class subscriptions.
During the many shows and symposia that writers attend, we are told that lubrication failures cause 70 or 80% of all equipment failures, bearing failures cause 60% of all breakdowns, and electrical faults cause some other huge percentage of downtime. These might all be true, since lubrication can cause bearing failure and trip an electrical safety, but the combination is not very informative. Make sure your breakdown data keep you informed of what really needs fixing in your organization.
My current favorite from Washington is the Bureau of Labor Statistics’ assertion that the total inflation of the US dollar for the last five years is a shade over 8%, or 1.7% per year. They even let you calculate it yourself. Anyone who buys groceries for a US family knows that statement to be a load of organic fertilizer. Here’s the hitch; the Consumer Price Index, upon which the inflation statistic is based, does not include the cost of groceries or fuel, since their “volatility and seasonality” would skew the numbers. It does include real estate, though, which has dropped like a rock. Face it, we’re all getting skewed.
Modern politicians have extended the art of lying with statistics beyond anything Darrell Huff had experienced, though. Watch for the next forecast of savings or other improvement from any politician’s latest bill or initiative. Two things will be true: First, the benefits will be stated to sound as if they are annual, though the careful listener will hear that they are for a much longer period of years. Second, the year in which the grand total of results will be realized almost surely comes after the braggart will be out of office. That way, it’ll be no skin off his or her hide when the fraud comes to light.
The parallel to industrial faith healers is probably unnecessary.
You’d think that the 24/7 TV and radio news blitz that now bombards us all would result in closer examination of all this baloney, but the opposite seems to be taking place. As reporters must now fill all that time by interviewing each other, rather than the actual newsmakers, both the questions and answers are controlled by the media. This morning’s editorial guesses about the impact of events will become this afternoon’s facts to be dealt with by our leaders. “Senator, what are your plans for addressing the expected . . .?” Never mind which talking head’s expectations are being quoted.
The 2013 prize for statistical prevarication, though, has to go to the governmental and advertising crooks who make the claim that their bills or products cost “three times less” and/or leave “99% fewer cooties” than competitors’ offerings. The phrase “X times less” means absolutely nothing! What number is three times less than 100?
The next time a consultant promises you X times less cost or equipment failure, show him how the door works. He’s either dishonest or stupid. Neither is something you want from a solutions provider.
Oh, and by the way, you just might also be a geek if you’re reading about governmental and industrial statistics in Plant Services. I certainly hope so.
Read Stanton McGroarty's monthly column, Strategic Maintenance.