So I was lucky enough to have a person who was curious to know what I thought draw my attention to Gordon Gauchat “Politicization of Science in the Public Sphere: A Study of Public Trust in the United States, 1974 to 2010,” published on-line today in the American Sociological Review.
Gauchat analyzes 35 yrs of responses to the General Social Survey item that measures how much “confidence” the public has in “the scientific community” and finds that the spread between liberals and conservatives has been widening in the last 15 years or so. Indeed, before that, there really wasn’t any gap to speak of.
Gauchat had to make some judgment calls about how to carve up his data: e.g., whether & how to aggregate responses to the GSS item (which uses a crappy three-point response measure: “great deal of confidence,” “only some” or “hardly any”); how to deal with the shifting proportion of respondents identifying as “liberal” or “conservative” over the time period; whether & how to try to break the data up into discrete time periods in order to assess trends (I suspect people who do time series work might take issue with his strategy); and what variables to include as “controls” in multivariate regressions.
But I think it’s clear that the trend he points to is there. And that it’s interesting — indeed, thought provoking.
Here are some thoughts the paper has provoked in me:
1. A tale of two trends. The trend that Gauchat identifies looks pretty similar to the one that public opinion surveys identify in views on climate change. That issue started to polarize people on political/ideological lines sometime close to when conservatives and liberals started to disagree on the GSS “confidence” or “trust in science” item. Compare Gauchat’s Figure 1 (which I’ve cropped at around the point when the trend he identifies starts; the uncropped Figure is in the inset to the right) with a couple of Figures that I’ve taken from Dunlap, R.E. & McCright, A.M. A Widening Gap: Republican and Democratic Views on Climate Change. Environment 50, 26-35 (2008), who summarize Gallup polling on climate change during this period:
2. Three possible meanings. I’m conjecturing, of course, but I suspect that these two trends are in fact linked. Whether they are is something that would have to assessed with more evidence, of course. And even more important, such assessing would have to be informed by some sort of hypothesis about what the link consists in. Here are three possibilities:
a. The “confidence” item doesn’t mean what it says — it means “how do you feel about climate change?” One possibility is that the political polarization on responses to the GSS item that started in the 1990s is just an indirect measure of the politicization of climate change. That is, as climate change became more salient as a partisan issue, the question “how much confidence do you have in the scientific community” started to bear a politicized resonance that generated the same pattern of responses. On this view, “how confident are you in scientists” is essentially just an indicator of a latent attitude toward climate change. It’s also a relatively weak indicator: it doesn’t provoke as much division, in fact, as the climate change issues (in Gauchat’s Figure, the y-axis is the fraction of conservatives or liberals who selected “great deal of confidence” vs. “only some” or “hardly any” combined).
If conservatives (or a significant number of them) are translating the question “do you trust scientists” into the question “what do you think about climate change,” moreover, then the answer isn’t a very reliable indicator of how conservatives feel about scientists in general or in nonpoliticized settings.
b. The item means what it says — and measures the cost that climate change has imposed on the credibility of scientists with conservatives. Alternatively, conservatives are answering the question they are being asked — and the thing that has caused them to become less trustful generally of science is the climate change controversy. That would be very sad.
c. The item means what it says — and is the source of climate change politicization. The final possible explanation for the linked trends (or the final one I can think of right now) is that the GSS item measures a genuine and growing distrust of scientists among conservatives by conservatives and that growing distrust is itself what caused conservatives to become distrustful of climate change science in the mid to late 1990s.
That strikes me as the least plausible explanation, actually. Why did conservatives just happen to get distrustful of scientists at that very moment?
Indeed, Gauchat’s study would have lent more support to the hypothesis that some dispositional distrust of science is the cause of conservative resistance to climate-change science if he had found that conservatives distrusted scientists well before evidence of climate change started to accumulate. Because conservatives weren’t more distrustful of scientists than liberals before the mid 1990s, his data actually undercut the assertion that conservatism is associated with anti-science or closed-minded reasoning styles.
Or so it seems to me; am eager to see how others react. Particularly Chris Mooney, a thoughtful proponent of the “asymmetry thesis” (AT) (i.e., that Republicans or conservatives are more vulnerable to motivated reasoning than Democrats or liberals). Gauchat sees Mooney’s earlier “Republican War on Science” (RWoS) thesis — that Reagan & the Bush Presidencies launched partisan attacks against the scientific community — as corroborated by his data. But that actually raises the question whether RWoS and AT are consistent!
3. Some additional puzzles if one is trying to make sense of political orientations and dispositions toward science.
a. Liberals have historically “distrusted scientists” on environmental risks. It is a staple of the scientific study of public risk perceptions that “distrust” of science predicts concern over environmental risks — most prominently, as a historical matter, nuclear waste disposal. Historically, too, the left (liberals, and in cultural theory egalitarians) have been most distrustful of scientists in connection with those issues. More evidence that “distrust of scientists” is often not what it seems — a general distrust of scientists — but a (weak) indicator of some general orientation toward the risk-issue du jour.
b. Moderates distrust scientists the most! Gauchat is interested, understandably, in the growing division between conservatives and liberals in the last 15 or so years. But across the entire three-decade period of the study, the group most distrustful has been self-described moderates.
Moreover, historically, more people characterized themselves as “moderates” than as either “liberals” or “conservatives.” Conservatives, then, have historically been more trusting than most ordinary, non-partisan citizens.
Recently, conservatives have been increasing and now have basically “caught up” to moderates. Well, because moderates are the most “distrustful,” the migration of “moderates” to “conservative” could be expected to increase the proportion of “conservatives” who are “distrustful” on the GSS item.
4. What’s the story with religion? It’s got to be a different one.
Gauchat also finds that there is a parallel increase in distrust associated with religiosity (measured by church attendance). Of course, that religiosity would predict distrust (or lack of confidence) in scientists is not so surprising (not that I think this is inevitable!). But it isn’t obvious that such distrust would have increased over this period.
Gauchat’s analysis, moreover, doesn’t really make it obvious to me why it occurred. I read Gauchat himself as seeing the trend associated with religion as being of a piece with — as having the same source, essentially — as the trend associated with conservativism and distrust of science (viz., Mooney’s RWoS thesis).
But in fact, Gauchat’s statistical analysis suggests that the association between religiosity and distrust of science occurred independently of the trend involving conservatism and distrust (he doesn’t report any interactions between ideology and church attendance). That is, if one was a regular church goer, one became less trustful of scientists over the time period in question whether one was liberal, moderate, or conservative. Did Reagan and Bush cause liberal church goers to become anti-science too!?
I suppose the climate change controversy could be making even highly religious liberals and moderates more distrustful of science — although in fact, I would be super surprised if this is so, since I know from my own research that highly religious egalitarians are the most concerned of all about climate change risk!
So — I dunno what’s going on. Which I don’t mind so much; one can’t experience the pleasure of seeing a mystery solved if one is never perplexed.
(This is an aside, but treating religion and ideology as independent variables in a model like this is arguably a bad idea, since religion and conservative ideology are probably common indicators of a latent disposition that predicts science distrust and attitudes toward environmental risks more generally. If they are, the regression estimates for each influence controlling for the other will be unreliable. I will likely post something on the vice of “over-controlling” in studies that try to identify latent dispositional influences on risk perceptions sometime! In any case, it is clear from the raw data that Gauchat’s finding on conservatism is not by any means an artifact of this modeling strategy.)
* * *
As I said, thought-provoking study — one that will make people smarter as they share their reactions to it.