Reading, Republican Brains, and Post-Hoc Fallacies

If you read the scientific bloggysphere at all, you probably have come across discussion of Chris Mooney’s The Republican Brain. In it, he argues that there are significant cognitive differences–or points along the continuum of ‘normal’–that correspond to conservatives and liberals (in the U.S. sense*). In a recent post, Mooney describes some original data found in the book that when conservatives are given essays that, regardless of political bent in the essays, conservatives spend less time reading them (boldface mine):

“The Republican Brain” itself tries to add a little to our growing body of knowledge. The book closes with a new psychology experiment, designed by the political psychologist Everett Young and run on a group of 144 college undergraduates. We were trying to test the idea that conservatives reason more defensively than liberals – doubling down on their beliefs. We found some modest supportive evidence for this idea, at least when it comes to political topics like global warming and nuclear power (although the effect was not present on apolitical topics).

But that turned out not to be our biggest finding. The biggest finding came out of nowhere and slapped us in the face.

Here’s what our data showed: The conservatives in the study were spending a lot less time reading our essays than liberals. Unbeknownst to them, we had a timer, measuring down to the millisecond how long each student spent on each page of essay material we gave them to read (on a computer screen). And it didn’t matter whether it was an essay that supported their beliefs or one that opposed them. It didn’t matter if it was a political topic or an apolitical one. And it didn’t matter if they were an economic conservative, a social conservative, an authoritarian and so on. Across the board, all kinds of conservatives spent less time reading our essays.

Being a good liberal and appreciating scientific uncertainty and the need for nuance, I must caution that this finding needs to be replicated in subsequent research. We only found it once, in one group of undergraduates, at one university. So, one must careful.

Regular readers (hell, infrequent readers) will know that I’m not a conservative, but I call bullshit here. If I were a conservative, and one with a mean streak, I would simply argue that liberals are stupid people who read slowly. And that actually could have some factual basis. As you move up the income ladder, adults are more likely to be conservative. Both SAT verbal scores and WORDSUM in the GSS survey are correlated with income. This very well could be a case of conflating conservative ideology with reading ability. To do this properly, one would have to control for SAT scores (which should be available as college students were the subjects; my perusal of a copy of the book doesn’t suggest SAT scores were examined).

The tell to me is this:

But that turned out not to be our biggest finding. The biggest finding came out of nowhere and slapped us in the face.

Essentially, the designed study that (hopefully) has adequate controls for confounding effects gave you a weak result. So you took the data, shook it all around, and after looking at a bunch of stuff, you managed to find a significant outcome. Of course, the experiment wasn’t designed to ask this question, so any number of alternative explanations could work (I gave one example above).

And more, similar studies won’t help, unless educational performance (i.e., SAT verbal scores) is considered.

Does this sound familiar at all? If this were an ‘epidemiological’ medical research study, we would be stomping on this. Someone does some underpowered post-hoc testing after a marginal result and WHHEEEEEE!!!!

I bring this up, not to rip Mooney (I bear him no personal animus, unlike many targets of my ire), but it makes me skeptical of the other studies and how they’re interpreted. No doubt I’ll be accused of being a stereotypical liberal.

But this is a general problem I have regarding much of the human cognition work (including IQ): the results are very difficult to understand. And the societal stakes are very high. This isn’t being wishy-washy, but demanding old-school rigor. Kinda conservative, actually.

*Which, of course, in much of the world, correspond to far right and neo-liberal/moderate, but that’s a completely separate issue.

This entry was posted in Conservatives. Bookmark the permalink.

4 Responses to Reading, Republican Brains, and Post-Hoc Fallacies

  1. Yoder says:

    Haven’t read the book, so I may be bringing up something already covered, but: it seems like the obvious way to figure out what this observation means is to do some simple comprehension testing after the students are done reading. If conservatives read faster, but score about the same on comprehension, then they really are just better readers.

  2. Darkling says:

    I haven’t read the book, but from your comments sounds like he’s data dredging. It’s not horrible if you’re looking for novel hypotheses, but it should serve as the start of the experimental (re)design, rather than as the end point of the analysis.

    Reminds me of a horrible manuscript that I was asked to comment on some years back where the authors wanted to demonstrate a relationship between several variables. They did the formal statistical test (ANOVA in that case), but none of the parameters were significant. They then went on to show the biplots of their favourite variable against the response. It did show the relationship they wanted, but was somewhat meaningless since they’d just shown that when other variables were taken into account it was non-significant.

  3. wiseguy says:

    Everyone knows most conservatives still dislike reading from a computerscreen, otherwise they wouldn’t be so conservative now would they?

  4. Pingback: Agnosus » Recommended Links from the Week of April 23rd

Comments are closed.