Crunching and Massaging the Data

Any student of statistical analysis quickly discovers that the presentation of research results can be very a very subjective enterprise. presents an excellent example today. A Honolulu Advertiser article on Sunday contained the following graph showing CO2 concentration at the Mauna Loa Observatory over the last several decades: The trend line reflects the “CO2 rises sharply” theme of the article. You can just imagine someone reading this article with their morning coffee and thinking, “Wow, this sounds pretty scary, and there’s no end in sight. I’m officially alarmed.” But Junkscience’s Steven Milloy helpfully puts this chart in perspective by simply adding the zero point to the y-axis and reducing the time interval to 2 years down from 10. The result presents the data in a much different light. Doesn’t seem like that scary of a trend line anymore, huh? Of course this is nothing unusual for those of us with an understanding of how the statistics game works, but I think it’s always useful to point this stuff out from time to time. I remember my “Aha!” moment in grad school when I realized how powerful the shaping and presentation of data really is, and my skepticism level rose tremendously after that, particularly as it relates to the media. People routinely make judgments and decisions on issues based on charts and graphs all the time. We see them in the news, we see them on financial statements, we see them everywhere. So much so that people are often inclined to trust what they’re presented with, rather than approaching new information with a healthy skepticism. Not that one should doubt every graph they see, just that they should look deeper to see the underlying structure of the data and the logical implications that flow from that, rather than keeping their focus fixed on the surface.

Leonard Gilroy is vice president of government reform at Reason Foundation and senior managing director of Reason's Pension Integrity Project.