What sorts of inferences can/can’t be drawn from the “Republican shift” (now that we have enough information to answer the question)?

Okay, so Pew, not surprisingly, happily released the partisan breakdown for all parts of its evolution question.

Pew also offered a useful explanation of what it admitted was a “puzzle” in its report–viz., how the proportion of Republicans “disbelieving” evolution could go up while the proportions of Democrats and Independents as well as the proportion of the general population “believing” in it all stayed “about the same”? Should be obvious, of course, that this was something only Pew, & not others without access to the necessary information, could do.

So now I’ll offer up some reflections on the significance of the “Republican shift”—the 9 percentage-point increase in the proportion of Republicans who indicated that they believe in the “creationist” response and the 11 percentage-point decrease in the proportion who endorsed either the “Naturalistic” or “Theistic” evolution responses to Pew’s “beliefs on evolution” item.

I’ll start with two background points on public opinion, including partisan divisions, on evolution. They are pretty critical to putting the “shift” in context.  Then I’ll offer some points that counsel against treating the “shift” as a particularly important new datum.

But to give you a sense of the theme that motivates the presentation of this information, I think the modal response to the Pew survey in the media & blogosphere was absurd.  Paul Krugman’s reaction is typical & typically devoid of reflection: “Republicans are being driven to identify in all ways with their tribe — and the tribal belief system is dominated by anti-science fundamentalists.”

He and many others leapt to a conclusion without the evidence that logic would have told them was not supplied in the original Pew summary. That’s pretty embarrassing.

And not surprisingly, the theme of their interpretation – “more evidence of Republicans being driven to anti-science extremism!” – is a testament to confirmation bias: the use of one’s existing beliefs to construe ambiguous data, which is then treated as corroborating one’s existing beliefs.

Background point 1: “Beliefs” on evolution lack a meaningful relationship to understanding evolution, to science literacy generally, or to being “pro/anti-” science.

Only aggressive disregard of empirical data—lots and lots and lots of them!—can explain why popular commentators start screaming about science illiteracy and creeping “anti-science” sensibilities in the U.S, every time a major polling outfit releases an “evolution belief” survey (about once a year).

As I’ve mentioned before, there is zero correlation between saying one “believes” in evolution and being able to give a passable (as in pass a highschool biology test) account of the modern synthesis (natural selection, random mutation, genetic variance) account of it.  Those who say they “believe” are no more likely to have even a rudimentary understanding of how Darwinian evolution works than those who say they “don’t believe” it.

In fact, neither is very likely to understand it at all.  The vast majority of those who say they “believe in evolution” believe something they don’t understand.

But that’s okay.  They’d not only be stupid—they’d be dead—if people insisted on accepting as known by science only those insights that they actually can intelligiently comprehend!  There’s way too much scientific knowledge out there, and it matters too much!

What’s not okay is to march around smugly proclaiming “my side is science literate; your’s isn’t!” because of poll results like this one.  That’s illiberal and ignorant.

It is also well established that “belief” in evolution is not a valid indicator of science literacy in general!

Answering “yes” to the simplistic “do you believe in evolution” item in the NSF’s “science indicators” battery doesn’t cohere with how one does on the rest of this science literacy test—in part because plenty of science know-nothings answer “yes” and in part because plenty of “science know a lots” answer “no.”

The item isn’t measuring the same thing as the other questions in the battery, something NSF itself has recognized.  What it is measuring is a matter I’ll address in a second.

Finally, as Pew, in one of the greatest surveys on U.S. public attitudes toward science ever has shown, “disbelieving” in evolution is not meaningfully associated with being “anti-science.”

The vast majority of people who say “I believe!” and those who say “I don’t”—“tastes great!” vs. “less filling!”—all have a super positive attitude toward science.

The U.S. is an astonishingly pro-science society. If you think otherwise, you just don’t know very much about this area.

Background point 2: “Belief”/“disblief” in evolution is a measure of identity, not a measure of science knowledge or attitudes.

As I’ve indicated, answering “I believe!” to a simple-minded “do you believe in evolution? Huh? Do you? Do you?” survey question is neither a valid measure of understanding evolution nor a valid indicator of science comprehension.

What it is is a measure of cultural identity.  People who say “yes” are expressing one sort of cultural affiliation & associated outlooks; those who say “no” are expressing another.

Religiosity is one of the main indicators of the relevant cultural styles.  The more religious a person is, the more likely he or she is to say “I don’t believe” in evolution.

Again, “belief” has nothing—zero, zilch—to do with science literacy.

Partisan self-identification—“I’m a Democrat!”; “I’m a Republican” (“tastes great! …”)—is simply another indicator of the relevant cultural styles that correspond to saying “believe” & “not believe” in evolution.

The partisan divide on evolution is old old old old news.

“MAFY” (i.e., “Making a fool of yourself based on uniformed reading of Pew poll”) point 1: Well, what do you know! Democrats don’t believe in “evolution” either!

Now that Pew has released the partisan breakdowns on its entire evolution item and not just the first half of it, it is clear, as anyone who knows anything about this area of public opinion could have told you, that the vast majority of the U.S. publicDemocrat, Republican, and Indpendentsay they “don’t believe” in evolution.

Pew initially released the breakdown only on that 1/2 of the question that asked whether respondents believed “Humans and other living things have evolved over time” or instead “Humans and other living things have existed in their present form since the beginning of time.”

The next 1/2 asks those who select “evolved” whether they believe that “Humans and other living things have evolved due to natural processes such as natural selection” or whether they believe instead that “A supreme being guided the evolution of living things for the purpose of creating humans and other life in the form it exists today.”

Get that, Paul Krugman et al?  The first position is Darwinian evolution; the second isn’t—it’s something goofy and non-scientific like “intelligent design”!

Only 37% of Democrats say they believe that humans have evolved as a result of “natural selection.”  Over 40% of the Democrats who “believe in evolution” buy either the “supreme guidance” variant or “don’t know” if evolution operates without or without God involved.

Does this mean they are “anti-science”?


What it means to say one “believes” or “disbelieves” in evolution is a complicated, subtle thing.

What groups “believe” about evolution certainly tells us something about their attitudes toward science!

But for sure what it says can’t be reduced to the simplistic (genuinely ignorant) equation “disbelieve = anti-science.”

If you would like to understand these things, rather than be a pin-up cheerleader for an embarrassingly, painfully unreflective bunch of partisan zealots– your tribe!–then you’ll have to simply accept that the world is complicated.

“MAFY” point 2: There was no meaningful “shift” in the proportion of Republicans who reject “naturalistic” or “Darwinian” evolution.

Now that Pew has released all the numbers, we know that 23% of self-identified Republicans in 2009 said they “believe” in “naturalistic” evolution—evolution via “natural selection” rather than divine “guidance”—and that 21% said that in 2013.

Not within the statistical margin of error, as far as I can tell.

And definitely not practically significant.


“MAFY” point 3: The Pew survey is really interesting but does not in itself support any inference about a significant “change” in anything since 2009.

As I indicated, the partisan division on evolution is old old old old news.  That’s because the tendency of people with culturally opposing styles to take opposing positions on it—ones that express their identity and not their knowledge of or attitudes toward science—is old old old news.

The question is whether the Pew poll—which is really an excellent piece of work, like everything else they do—justifies concluding that something material has changed in just the last four years.

I’ve thought & thought about it & concluded it really doesn’t.  Here’s why.

1st, as emphasized, the shift in the percentage of Republicans who say they believe in Darwinian or naturalistic evolution was a measly 1%.

2d, Pew has given us 2 data points.  Without knowing what the breakdown was on their question prior to 2009, it is logically fallacious to characterize the 2013 result as evidence of Republican “belief in evolution” as having “plummeted.”  For all we know, non-belief is “rebounding” to pre-2009 levels.

I don’t know if it is.  But the point is, all those asserting a shift don’t either.  They are fitting their interpretation of incomplete, ambiguous data to their preconceptions.

3rd, if something real had changed, it wouldn’t show up only in Pew’s data. Gallup has been doing polls on evolution regularly for decades.  It’s numbers show no meaningful change in the numbers, at least through 2012 (go ahead, if you are a story teller rather than a critical thinker, and invent some ad hoc account of the amazing event in 2013 that changed everything etc).

More likely, then, Pew’s result reflects just a blip.

Also supporting that view is the pretty big discrepancy between the percentage who identify as “naturalistic” as opposed to “theistic evolutionists” in Pew’s poll and those who do so in Gallup’s.  The questions are worded differently, which likely explains the discrepancy.

But that the slight word changes can generate such big effects underscores how much of a mistake it is to invest tremendous significance in a single survey item.

Good social scientists–& I’d definitely include the researchers who work for Pew in that group—know that discrepancies in the responses to individual survey items mean that individual items not a reliable basis for drawing inferences about public opinion. Because what individual items “measure” can never be determined with certainty, it is always a mistake to take any one item at face value.

Look at lots of related items, and see how they covary.  Then consider what sorts of inferences fit the overall pattern.

Here, the “overall pattern” is too indistinct, too uneven to support the inference that the 9% “shift” in the proportion of Republicans who indicated they “believe” in “creationism” in the 2009 Pew survey and the 2013 one means the world has changed in some way bearing on the relationship between beliefs in evolution and the sorts of identities indicated by partisan self-identification.

Maybe something has!

But the question is whether the survey supports that inference.  If you want to say, “Oh, I’ll construe the survey to support the conclusion that something interesting happened because I already know that’s true,” be my guest.

It’s a free country, as they say, and if you want to jump up & down excitedly & reveal to everyone in sight that you don’t know the difference between “confirmation bias” and valid causal inference, you have every right to do so!

Leave a Comment