Lies, damned lies and statistics: Selective use of data can lead to faulty conclusions
In this year of census forms and elections, it seems appropriate to marvel at one of the understated wonders of our free society: the gathering of objective data.
Shortly after the Berlin Wall fell in 1989, I had the chance to live and work for a year in Hungary. I worked as a volunteer on an energy efficiency project which sought to assist Hungarian municipalities as they transitioned from planned economies to a free market system.
During a year of Kafkaesque experiences, what particularly stood out was the selective and biased statistical record compiled by the former communist regime. As we worked to gather a historical record of municipal energy usage it became apparent that those statistics that made officials look good were maintained, and those which presumably did not were somehow lost to posterity.
I remember thinking at the time how lucky we are in the United States to have myriad sources of generally complete and reliable data. We take this for granted, yet it is crucial to the effective functioning of our democracy and the efficient performance of the economy.
Recently I visited the National Science Foundation’s Division of Science Resources Statistics in Washington, D.C. This office is just one of a host of government statistics offices providing data on topics ranging from “accidents and fatalities” to “zinc production” (I pulled these topics from the index of the “Statistical Abstract of the United States” sitting on my shelf).
At the NSF we were seeking data to help us benchmark the research performance of the University of Michigan versus peer institutions. This task would be impossible without the data that NSF gathers on a regular basis.
Now the term “lies, damned lies and statistics” (popularized by Mark Twain and often credited to British Prime Minister Benjamin Disraeli) is not really an indictment of statistics themselves. Numbers don’t lie. But the selective use of data and a general mathematical illiteracy can lead to faulty conclusions.
As mathematician John Allen Paulos says in his book “Innumeracy,”: “Politicians are seldom a help in this regard since they deal with public opinion and are therefore loath to clarify the likely hazards and trade-offs associated with almost any policy.”
That's something to think about during these final weeks of campaign ads and speeches.
Daryl Weinert is the Executive Director of the University of Michigan’s Business Engagement Center. He can be reached at email@example.com.