Statistical Lies and The Liars Who Use Them

August 16, 2013

In his book, Stat-Spotting, Joel Best gives a good example of how politicians can lie with statistics. He explains how President Bush's Secretary of the Interior Gale Norton went before reporters and announced that the United States had had its first net gain in wetlands since the Fish and Wildlife Service started tracking wetland acreage in 1954. That certainly makes it sound like the administration and its agencies are doing a good job protecting the environment, doesn't it? But it turns out that the only reason there was a supposed increase in wetlands acreage is because the term was defined in a new way, and now included water hazards on golf courses and other man-made water areas. She did not mention that the report she was referring to in the "good news" acknowledged that the acreage of natural wetland areas had declined once again.

Nice trick. We could redefine "lake" in such a way that thousands of ponds are classified as lakes, and then claim to have "created" thousands of new natural lakes. The bottom line is that when comparing current data with that from the past, you have to be comparing the same thing to be accurate and honest. It might make sense to include man-made wetlands in the definition, but then the older data should be adjusted (if possible) to include these as well, if there are comparisons to be made.

Best gives another good example of questionable use of statistics in his book. There was a report done by the Cable News Network (CNN) in 2005, in which they looked at American wealth. They reported that "the average U.S. household has a net worth of greater than $400,000." But there is more than one way to arrive at an average. In this case they decided to use the mean rather than the median. In other words they added up the net worth of all households and divided by the number of households. Why this can distort the truth of the matter is clear if we imagine a simple example.

Suppose there are a thousand people living on an island, which we'll call Statland. One of them has a billion dollars and the others all have nothing. The total of all assets is a billion dollars, so if we divide that by the thousand residents we arrive at a mean average net worth of one million dollars. Now, if we report that "the average resident of Statland is worth a million dollars," that might be accurate in one sense, but it does give an inaccurate impression of a place where 99.9% of the population is poor.

The median is arrived at by finding the measure where half fall below and half above. At the time when CNN reported the average net worth of U.S. households to be over $400,000, the median net worth of households (including home equity) was about $86,000. In other words, half of all households fell below this figure and half above. While not as extreme a difference as in Statland, the two figures do provide quite a differing view of the wealth of families in The United States, don't they?

For more on the misuse of statistics, see our page on lying with statistics. I detail several methods there which are used to distort the truth for political purposes. For example, I briefly address the way polls are manipulated through the wording of the questions on them. You can easily get very different results if you change how you ask a question.

I also have a page on my brainpower website that looks at how to lie with statistics. It covers issues like non-random sampling and biased selection of data. I also look at how the phrase "up to" is used to distort what the statistics say.

We should not be too surprised by the many statistical lies we see. Any tool can be used for various purposes, both good and bad, honest and dishonest. This is certainly true of the powerful tool that we call statistics.

Copyright - Terms / Privacy / Contact

Was this page interesting or useful? Please let others know with one of these...