Statistical literacy

Posted by AzBlueMeanie:

Kathleen Geier at the Political Animal blog has this important post for anyone who has ever had to suffer through analyzing the "research" (sic) produced by right-wing billionaire-funded think tanks like the Goldwater Institute. On the importance of statistical literacy:

Over at Ten Miles Square, my blogging colleague Kevin Carey recently argued that Everyone Should Learn Statistics. I heartily concur. It is my longstanding belief that becoming statistically literate is one of the most valuable skills a person could possibly acquire, especially if she is a political junkie (and I suspect that many reading this blog fall into that category). Kevin writes about his experience with jury duty and makes the argument that understanding basic statistics makes you a more competent citizen, and that is certainly true.

But statistical literacy is also a must if you want to understand public policy. After all, the way we analyze the real-world effects of various public policies is via research, and most of the best research is quantitative in nature. Unfortunately, there are vast amounts of crap social science out there that pollute our public discourse, much of it emanating from the right. And if you can’t distinguish a solid statistical argument from a slick but bogus one, you’ll be an easy mark for disingenuous ideologues and hucksterish policy entrepreneurs of all stripes.

Crap social science pollutes our body politic in two ways. The first is direct, when ideologues — and let’s be clear, 95% of the time we’re talking about right-wing ideologues here — deploy these dubious studies and statistics directly, in their own writings, speeches, and interviews. [We're looking at you, Goldwater Institute.] And that is certainly bad enough, especially when they spout off about this stuff but are not held accountable for the veracity of what they say.

What’s worse, though, is when shoddy research and dubious factoids get respectful attention from mainstream reporters and pundits. This, unfortunately, tends to happen a lot, and when these “facts” gets disseminated, poisonous “conventional wisdom” about a subject can develop, and the general public is bamboozled. Much reporting about social science is, unfortunately, quite terrible. Sensation-seeking journalists will promote the research that reports the most dramatic, headline-grabbing results; research that reaffirms pre-existing prejudices tends to get a disproportionate amount of attention (this is particularly true, I’ve notice, when it comes to research about gender differences).

All too often, the most basic questions about a piece of research are not asked. Such questions include, but are not limited to: who conducted the study? Is it a well-respected academic or some think tank hack? (Let’s not forget that the right has a small army of slick, well-funded “think tanks” which devote enormous resources to publicizing their findings and influencing policymakers and the general public). What do the experts in the field think of the study? Does it have a kosher research design? An adequate sample size? Are the findings consistent with previous research, or is the study an outlier? Sometimes the most dramatic results occur in studies whose findings are never replicated, which suggest that the results may have been a statistical fluke, or the product of unique conditions that cannot be duplicated (for example: a small pilot program that is staffed by unusually highly motivated, and highly qualified, personnel, and thus would be difficult to implement on a large scale).

Continue reading On the importance of statistical literacy.