Outsourcing my critical reading (i.e., just plain reading!) of this article worked really well. Given the great points that came out in the comments, I don’t think there’s any value added in my offering a full assessment of the paper, which reported the results of a study that analyzed less than ideal data with a questionably specified statistical model from which the authors seemed to draw very debatable inferences.
But I do think it might be interesting to explore, at least to a degree, what one might learn about the authors’ research questions if one applied a valid statistical model to data that could support some reasonable inferences.
Basically, the authors purport to find that “religious variables,” but not “partisan identification,” predict a general hostility to science as manifested by attitudes toward climate change, evolution, and stem-cell research. They treat this finding as suggesting reason to doubt claims that partisan political predispositions (e.g., Gauchat 2012), shaped by elite discourse (e.g., Mooney 2005; Brulle et al. 2012), account for public conflict over science issues.
Because the authors believe that religiosity bears a greater share of the responsibility for such conflicts than is normally appreciated, let’s call this the “Religion-science Antipathy thesis” (RAT).
The main “political” competitor the authors advert to is one that attributes such conflict to antipathy—perhaps psychologically grounded (Mooney 2012), perhaps economically (Brulle et al., 2012), or maybe both—between a conservative political orientation and science. Let’s call this the Conservativism-science Antipathy Thesis” (CAT).
So here’s what I’m going to do. I’ve compiled a bunch of observational (i.e., survey!) data and modeled them in a way that I think arguably bears on the relative plausibility of RAT and CAT.
But exercising a credulity-defying degree self-restraint, I am going to refrain, for at least 24 hrs, from telling you what sorts of conclusions I think these data support.
In that period, you, the 14 billion readers of this blog, will be afforded the exclusive opportunity to specify and defend your own inferences!
How many of the myriad other “cultural cognition blogs” out there do you think would have the necessary levels of self-confidence and respect for their readers to surrender their “first word” prerogative? That’s right—not a one!
Okay, then.
Let’s start with global warming.
As you can see, this figure plots the probability of agreeing that there is “solid evidence” for human-caused global warming in relation to right-left political outlooks conditional on one’s level of “religiosity.” Political outlooks and religiosity are measured with separate multi-item scales. (Because partisan self-identification and liberal-conservative ideology are both indicators of the same thing—an unobserved or latent political disposition—it is really not a good idea to treat them as “independent” right-hand side variables in a multivariate regression.) The colored hashmarks are the 0.95 confidence intervals for the predicted probability at the indicated point on the left-right political outlook scale.
If you want to “see” the regression model or the “raw data,” then click on the specified thumbnails in the margin.
As you can tell, there’s an interaction between religiosity and political outlooks: the contribution that moving left in outlook makes to acceptance of climate change is bigger the less religious one is.
If you get only this, demand a refund!But I won’t say anything more than that! What this signifies in the battle between RAT and CAT is your call to make!
Next, let’s look at belief in evolution. Same model, used now to examine the impact of political outlooks on belief in evolution conditional on religiosity.
Significance? You tell me!
Next, support for stem-cell research.
Actually, I’ve never collected data on this topic. So I popped open a canned dataset that has such data: the super great 2009 Pew public attitudes toward science survey.
I again constructed a political outlook scale by aggregating response to partisan self-identification and liberal-conservative ideology items.
Pew didn’t have all the same items from which I constructed the religiosity scale in the previous models. So I constructed one using self-reported church attendance (one of the items that I did have in my CCP data set), self-identified “born again” evangelical status, and a “non-religious” self-identification variable that separated out persons who self-identified as agnostics or atheists from those who reported affiliation with any religious denomination.
How good a measure is this? I wasn’t sure, so I came up with a method to externally validate it.
It turns out the Pew survey also has measures for global-warming acceptance and belief in evolution (the authors of the study that inspired this exercise should have used the Pew dataset rather than the 2006 GSS dataset, which lacked a genuine measure of global-warming acceptance). When I used the Pew religiosity scale and the right-left political outlook measures as predictors of these beliefs, the Pew religiosity scale behaved very comparably to the CCP-dataset religiosity scale in the modelsreported above. That struck me as pretty good evidence that the Pew scale is tapping into pretty much the same unobserved or latent disposition being tapped into by the CCP religiosity scale.
Here’s the result for stem-cell funding:
What do you think?
These are the three issues—global-warming acceptance, belief in evolution, and support for stem-cell research—that the article we read used to test RAT vs. CAT.
But some of you pointed out that disbelief in evolution and opposition to stem-cell research are arguably the sorts of positions one might expect highly religious individuals to form independent of any sort of general hostility to science. For that reason, one might conclude they don’t supply as clean a test of RAT vs. CAT as, say, climate-change acceptance, where generalized “science hostility” is less likely to be confounded with issue-specific religious concerns.
So, I decided to add one more issue to try to make the fight more fair: nuclear power!
The great Pew study had two nuclear-power items, positions on which I also modeled in relation to political outlooks conditional on religiosity:
So there you go.
Infer away!
Refs
Brulle, R.J., Carmichael, J. & Jenkins, J.C. Shifting public opinion on climate change: an empirical assessment of factors influencing concern over climate change in the US, 2002–2010. Climatic Change 114, 169-188 (2012).
Gauchat, G. Politicization of Science in the Public Sphere. American Sociological Review 77, 167-187 (2012).
Mooney, C. The Republican Brain: The Science of Why They Deny Science–And Reality (John Wiley & Sons, Hoboken, NJ, 2012).
Mooney, C. The Republican war on science (Fine Communications/MJF Books, New York, NY, 2009).
So now one of the 14 billion readers of this blog, Chris Mooney, has offered his views on RAT vs. CAT in his own writeup on wonkblog.
Others might want to consult what he has to say as they continue to formulate their own interpretations.
You know, it’s not too late for me to turn this into one of the cool contests that we regularly run here on CCP blog.
Submitter of “best interpretation” (as determined by “scientific consensus” survey) will get his or her choice of any of these prizes (subject to availability):