Damned Lies, Damned Statisticians

Joel Best has written an excellent book examining the misuse and abuse of statistics, especially those asserted in the public forum and used for social and political decision-making. The book is a great source of lecture ideas and demonstrations; an anecdote from the introduction will illustrate the kind of material you might draw from.

Best served on the dissertation committee of a student who asserted the following claim in the first sentence of her or his dissertation prospectus: Every year since 1950, the number of American children gunned down has doubled. This dramatic statistic certainly attracts attention and would seem to call for strong, unambiguous, immediate legislation of all sorts. But as Best points out, theres a certain stink hanging over this claim. Lets say, for example, that in 1950 only one child was gunned down in America. That would mean that in 1951 two children were gunned down, in 1952 four children were gunned down, in 1953 eight children were gunned down, and so on. If this statistic were accurate, by 1965 there would have been 32,768 children gunned down (Best notes that FBI statistics for 1965 revealed only 9,960 criminal homicides of any kind in the entire country). By 1970 the number of deaths would have passed 1 million, and by 1980 it would have passed
1 billion. By 1983 there would have been 8.6 billion gunned down children (more than twice the population of the planet at that time), and by 1995, when this student made this assertion, the number of American children gunned down would have been 35 trilliona staggering statistic indeed, but for a very different reason!

A little digging by Best revealed the error of the students ways. The statement was harvested verbatim from a published article in a journal in the students field, but the original statement was made by the Childrens Defense Fund. However, the original statement read, The number of American children killed each year by guns has doubled since 1950. Notice that this is a very different statement with a very different meaning: In 1994 the number of children gunned down was twice what it was in 1950. Some creative license on the part of the articles authors (and the students lifting of it) led to the combinatorial confusion revealed by Best.

But theres more to the story. As Best points out, the population of the United States also rose between 1950 and 1994, by about 73 percent. We might therefore expect all kinds of events to increase, including the number of childhood fatalities. Because the population had nearly doubled, the number of childhood shootings (and number of cars purchased, and children born, and television sets bought, and books written, and any number of things) might indeed have seemed to increase just because there were more people. Moreover, theres some fuzziness about the claim itself. Child is a little sticky, given that some Childrens Defense Fund statistics include anyone under the age of 25. Also, died by gunshot could include suicides and accidents as well as homicides. Finally, its not clear who has compiled the information on these childhood deaths or how the counting was done.

Unfortunately, there are more than enough of these types of statistical missteps. Write a short summary of the article of your choice and share your thoughts.