How to lie with statistics

Anyone watch this heated exchange between Wolf Blitzer and House Oversight Committee Chairman Jason Chaffetz (R-UT)? The interview concerned the committee investigation of Planned Parenthood so I will avoid commenting on this politically charged topic. I want to comment on the value of critical thinking and information literacy.

When I taught Introduction to Psychology, one of the core concepts I thought was of value to all concerned notions regarding the processes of science. What qualifies as good evidence? One of the claims I made was that you can’t lie with statistics, but you can lie or at least misinform through the use of a faulty methodology leading to flawed interpretations.

There is a classic exchange in the Blitzer interview when Wolf brings up a chart used as part of the oversight committee hearing. Wolf kept saying the chart was misleading and Chaffetz kept saying he stood behind the statistics. The chart was drawn without a numerical marking of the Y axis and showed two crossing lines with money for abortions increasing and money for other services to women decreasing. The misleading thing about the chart was that most would assume the descending and ascending lines that actually crossed on the chart were based on the same scale. This was not the case. When drawn using the same scale, the abortion line was far below the line representing other statistics. While it did show a small increase and the line showing other spending a decrease, the flawed representation of scale gave a very different impression about how funds were allocated and how allocations may have been changing. The link I provide above contains the video, but also shows the chart. Take a look and see what you conclude.

Here is a more appropriate representation

I wish I could use this example in class. If done purposefully to misinform, it clearly is a lie. The lie kind of involves statistics, but it is really an example of misinformation based on the trust of those who assume you would not attempt to deceive them. Was Chaffetz lying with statistics? I guess it depends on how you define lying. Whatever the political position one supports, there should at least be honesty in how data are presented.