Evangelicals and the nonreligious do differ in science literacy

Ear­li­er this week, I quot­ed this from an op-ed in the LA Times:

I recent­ly con­duct­ed sur­vey research com­par­ing the most con­ser­v­a­tive of Protes­tants — those who iden­ti­fy with a con­ser­v­a­tive Protes­tant denom­i­na­tion, attend church reg­u­lar­ly and take the Bible lit­er­al­ly, or about 11% of the pop­u­la­tion in my analy­sis — with those who do not par­tic­i­pate in any reli­gion. The con­ser­v­a­tive Protes­tants are equal­ly like­ly to under­stand sci­en­tif­ic meth­ods, to know sci­en­tif­ic facts and to claim knowl­edge of sci­ence. They are as like­ly as the non­re­li­gious to have majored in sci­ence or to have a sci­en­tif­ic occupation.

The paper report­ing this research is not in press yet, but I thought I’d try repli­cat­ing the results. After all, the basic infor­ma­tion need­ed to eval­u­ate these claims is avail­able through the Gen­er­al Social Sur­vey, a mas­sive sur­vey that has been con­duct­ed by the Nation­al Opin­ion Research Cen­ter since 1972. The GSS has a lot of ques­tions about peo­ple’s reli­gious views and back­ground, and since 2006, has includ­ed a sci­ence lit­er­a­cy sur­vey on behalf of the Nation­al Sci­ence Foun­da­tion (pre­vi­ous­ly dis­cussed here).

The data are there to eval­u­ate the claims about sci­ence lit­er­a­cy, about sci­ence meth­ods, about major­ing in sci­ence, and sci­en­tif­ic occu­pa­tions, but let’s start with the claim about sci­ence lit­er­a­cy I made in the pre­vi­ous post. To test Evan­s’s claim about sci­ence lit­er­a­cy, I used the NSF’s sci­ence lit­er­a­cy ques­tions. I defined the “non­re­li­gious” as those who either rarely attend church (less than once a year) or who answer “none” when asked what reli­gion they belong to. It should be not­ed that these “nones” are a grow­ing group in the US, and are not nec­es­sar­i­ly athe­ists: they just don’t asso­ciate them­selves with tra­di­tion­al reli­gion. There’s a ques­tion where folks are asked if they con­sid­er them­selves fun­da­men­tal­ists; that tends to over­es­ti­mate the fun­da­men­tal­ist frac­tion of the coun­try, so I instead tried to repli­cate Evan­s’s pro­to­col: I sep­a­rat­ed out peo­ple who believe the bible is the word of God (not inspired by God or a book of fables), and who attend church at least once a month. I did­n’t try to iso­late fun­da­men­tal­ist denom­i­na­tions, because those denom­i­na­tion­al ties are get­ting weak­er in many cas­es. In any event, this screen is clos­er to the stan­dard set of screens used to iden­ti­fy evan­gel­i­cals, rather than fun­da­men­tal­ists per se. If any­thing, these sam­ples are over-inclu­sive, which would tend to dimin­ish any dif­fer­ences, bias­ing us toward Evan­s’s result.

The graph above shows what per­cent­age of mem­bers of each group answered a giv­en num­ber of sci­ence lit­er­a­cy ques­tions right. As you can see, the curve for self-described fun­da­men­tal­ists match­es the curve for evan­gel­i­cals nice­ly: many peo­ple treat the terms as syn­ony­mous, so it’s not unrea­son­able that these might reflect very sim­i­lar pop­u­la­tions. Note also that these groups both have medi­an sci­ence lit­er­a­cy sub­stan­tial­ly low­er than that of the gen­er­al US, and much low­er than that of the self-described­ly non­re­li­gious. The medi­an score for the non­re­li­gious was 8, for evan­gel­i­cals and self-described fun­da­men­tal­ists it was 5, and for the gen­er­al US, it was 6. 

This stan­dard set of ques­tions does include a ques­tion about evo­lu­tion and anoth­er about the Big Bang, but those two ques­tions can­not account for a dif­fer­ence of 3 right answers between evan­gel­i­cals and the nonreligious!

Of course, reli­gious atti­tudes are hard­ly the most impor­tant fac­tors dri­ving sci­ence lit­er­a­cy, and some of those oth­er fac­tors cor­re­late with reli­gious views. For instance, there’s a cor­re­la­tion of ‑0.5 between sci­ence lit­er­a­cy and evan­gel­i­cal belief and cor­re­la­tion of 0.25 between sci­ence lit­er­a­cy and non­re­li­gion. Sci­ence lit­er­a­cy has cor­re­la­tion of 0.48 with edu­ca­tion­al attain­ment (high­est degree achieved), and 0.47 with abil­i­ty on a vocab­u­lary test. Both of those mea­sures of aca­d­e­m­ic abil­i­ty and achieve­ment cor­re­late with reli­gious atti­tudes as well. They also cor­re­late with eco­nom­ic class, both per­son­al income and parental income, as well as race and oth­er class markers.

So I ran a mul­ti­ple regres­sion, incor­po­rat­ing non­re­li­gion, evan­gel­i­cal belief, the vocab­u­lary test, edu­ca­tion­al attain­ment, race. In this mod­el, non­re­li­gion was asso­ci­at­ed with get­ting about an extra half a ques­tion right, being evan­gel­i­cal with get­ting an extra ques­tion and quar­ter wrong. Hav­ing a col­lege degree ver­sus just grad­u­at­ing high school cor­re­lates to an extra ques­tion or so right, being white is asso­ci­at­ed with about an extra right answer, and every addi­tion­al vocab­u­lary word some­one got right was asso­ci­at­ed with an extra half ques­tion right on the sci­ence survey.

It’s pos­si­ble to get a bet­ter mod­el by incor­po­rat­ing inter­ac­tions between vari­ables, but the basic gist does­n’t change (though the result for par­tic­u­lar peo­ple might be). Evan­gel­i­cal iden­ti­ty and non­re­li­gious iden­ti­ty inter­act strong­ly with vocab­u­lary score and with race. You can see the large effects of ver­bal knowl­edge, of edu­ca­tion, and of reli­gious views in the graph above.
There’s a lot going on in that graph, so let’s pick it apart. On the top and right side, there’s a his­togram of answers on the sci­ence lit­er­a­cy and vocab­u­lary tests, to help make sense of the scat­ter­plot under­neath the lines. Because it’s impos­si­ble to get frac­tion­al answers right, I shift­ed the indi­vid­ual points around in the scat­ter plot, mak­ing all the data more vis­i­ble, and show­ing the nice bivari­ate normal(ish) dis­tri­b­u­tion of the data.

I plot­ted regres­sion lines from each of four pop­u­la­tions: evan­gel­i­cals who stopped school after high school, the non­re­li­gious who stopped after high school, evan­gel­i­cals who stopped school after col­lege, and the non­re­li­gious who stopped after col­lege. The bold lines are the regres­sion mod­els, and the lighter lines of the same col­or have their para­me­ters drawn from the under­ly­ing dis­tri­b­u­tions of the data, help­ing illus­trate the inter­nal vari­abil­i­ty of these indi­vid­ual mod­els. To sim­pli­fy the graph, I left out oth­er vari­a­tion in edu­ca­tion­al attain­ment (from not com­plet­ing high school, not com­plet­ing col­lege, com­plet­ing junior col­lege, and com­plet­ing grad­u­ate school); you should be able to eye­ball where those lines would be. I did­n’t show vari­a­tion relat­ed to race because evan­gel­i­cals and the non­re­li­gious both tend to be white, and fac­tors affect­ing sci­ence lit­er­a­cy in dif­fer­ent races are a whole dif­fer­ent topic. 

The top­most line, the group with the high­est lev­el of sci­ence lit­er­a­cy, includes the most-edu­cat­ed non­re­li­gious folks, but note that increased lit­er­a­cy has less impact on this group than it does for the oth­er groups. The oth­er three lines are basi­cal­ly par­al­lel, with the regres­sion lines for non­re­li­gious groups always above those from evan­gel­i­cals (though that’d be dif­fer­ent if I had includ­ed high­er and low­er edu­ca­tion lev­els on the graph).

There’s a lot of scat­ter in the data, so the fact that the mod­els fol­low this strict order­ing by reli­gious view should not be tak­en to imply that no evan­gel­i­cals are sci­ence lit­er­ate, or that the non­re­li­gious have uni­form­ly high sci­ence lit­er­a­cy. These are just gen­er­al sta­tis­ti­cal ten­den­cies, and should­n’t be tak­en to show cau­sa­tion, either. The forces that dri­ve peo­ple into evan­gel­i­cal tra­di­tions are com­plex, as are the forces dri­ving sci­ence lit­er­a­cy and edu­ca­tion­al attain­ment. Those caus­es are inter­wo­ven, and while there may be direct causal­i­ty, it’s far more like­ly that shared cau­sa­tion is a more impor­tant fac­tor over all.

I look for­ward to see­ing Evan­s’s paper, since it looks like he used a dif­fer­ent set of sci­ence lit­er­a­cy ques­tions, and his sam­pling was sure­ly dif­fer­ent than the broad sam­pling avail­able from the GSS. If, for instance, he was look­ing at a sam­ple of col­lege stu­dents, he may have missed the shift in sci­ence lit­er­a­cy that we see between high school grads and col­lege grads, which could bring the sci­ence lit­er­a­cy dif­fer­ences to with­in a mar­gin that might be explained by the han­dling of ques­tions on evo­lu­tion and the ori­gin of the universe.

But this is also a cau­tion­ary tale. I quot­ed lots of his essay, with only mod­est caveats, because it tends to sup­port by own sus­pi­cions. When I first tweet­ed the sto­ry, Daniel Lox­ton retweet­ed it, say­ing “Skep­tics need to know this.” He then expressed some cau­tion, not­ing: “It’s been point­ed out to me that I don’t know the research @JoshRosenau describes, and can’t prop­er­ly eval­u­ate it. Fair enough,” explain­ing, “It agrees with what I already think: that peo­ple of faith are no less bright or curi­ous or imper­fect­ly sci­ence lit­er­ate than non-believ­ers.” That’s my belief, too, and the analy­sis above need not shake us from that basic prin­ci­ple. It does com­pli­cate that prin­ci­ple, because while I firm­ly believe reli­gious and non­re­li­gious folks are equal­ly capa­ble of sci­ence lit­er­a­cy, for some rea­son they don’t achieve sci­ence lit­er­a­cy at the same rate.

These data don’t tell us enough to unearth the caus­es of the dif­fer­ences we see, but I can spec­u­late based on what I’ve read else­where. Fun­da­men­tal­ism and evan­gel­i­cal­ism (to a less­er degree) are asso­ci­at­ed with author­i­tar­i­an mind­sets that seek hier­ar­chy, and (some­what para­dox­i­cal­ly) with a desire for indi­vid­u­al­ism. That’s not a mind­set that mesh­es well with sci­ence, where author­i­ty can be over­thrown rel­a­tive­ly eas­i­ly, which tends to resist imposed hier­ar­chy, and which favors com­mu­nal effort over a self­ish indi­vid­u­al­ism (though, of course, sci­ence con­tains plen­ty of hier­ar­chy and plen­ty of self­ish indi­vid­u­al­is­tic non­sense). This, I’d argue, explains the mas­sive polit­i­cal divide between sci­en­tists and the gen­er­al pub­lic. Only 6% of sci­en­tists are Repub­li­cans and 9% who called them­selves con­ser­v­a­tives (52% of sci­en­tists con­sid­er them­selves lib­er­al). Com­pare that to the 23% of the pub­lic who called them­selves Repub­li­cans in a con­tem­po­ra­ne­ous sur­vey of the US pub­lic, 37% who called them­selves con­ser­v­a­tives, and 20% who called them­selves liberals.

That same hierarchal/individualist com­plex of traits tends to asso­ciate itself with efforts to impose anti­science poli­cies (whether or not one thinks anti­science beliefs show a par­ti­san divide, it’s hard to point to anti­science poli­cies that the left has imposed or even made a con­cert­ed effort to impose), and also with reli­gious beliefs that val­i­date hierarchal/individualistic beliefs. Evan­gel­i­cal and espe­cial­ly fun­da­men­tal­ist church­es do that in spades, and are thus like­ly to attract peo­ple with a pre­dis­po­si­tion against sci­ence. That pre­dis­po­si­tion is like­ly to be deep­ened by ser­mons and by inter­ac­tions with oth­er anti­science folks in those reli­gious com­mu­ni­ties, and in some cas­es (evo­lu­tion most promi­nent­ly) that anti­science posi­tion becomes not just an indi­vid­ual assess­ment of the sci­ence, but a cul­tur­al mark­er of a per­son as a mem­ber of a par­tic­u­lar tribe. 

Evo­lu­tion became such a mark­er of fun­da­men­tal­ist iden­ti­ty in the 1920s, and glob­al warm­ing denial seems like it may well be head­ed that way as a mark­er of polit­i­cal con­ser­vatism today in the US, just as glob­al warm­ing accep­tance and con­cern have become mark­ers of polit­i­cal lib­er­al­ism in the US. The beliefs of one’s par­ents and cul­tur­al heroes will have a seri­ous impact from a young age, lay­ing a ground­work of reli­gious views, of views on sci­ence, of sci­ence knowl­edge, and of atti­tudes and affect. These issues are cor­re­lat­ed on gen­er­a­tional scales, mak­ing causal­i­ty real­ly hard to tease part.

To be clear, the folks I’m label­ing “anti­science” above would not call them­selves anti­science, and would insist that they like sci­ence a lot. But, as I said in the pre­vi­ous post on Evan­s’s work, they’d mean some­thing dif­fer­ent when they talk about sci­ence. They tend to con­flate sci­ence and tech­nol­o­gy, and praise prac­ti­cal, applied sci­ence that lets them do new things, and oppose it when it places lim­its on them, or chal­lenges beliefs they hold dear. In some cas­es, that will give the appear­ance of being pro-sci­ence (as when Newt Gin­grich stands up for NIH fund­ing), but it isn’t moti­vat­ed by an under­stand­ing of what sci­ence is, let alone an appre­ci­a­tion of sci­ence for sci­ence’s sake. As a reminder, then-Speak­er Gin­grich hap­pi­ly fund­ed NASA and the NIH, but famous­ly killed the Office of Tech­nol­o­gy Assess­ment, Con­gress’s inde­pen­dent advi­sor on sci­ence and technology. 

Evangelicals and the nonreligious don’t differ in science literacy

Soci­ol­o­gist John Evans talks about his research on evan­gel­i­cal atti­tudes toward sci­ence. Writ­ing for the LA Times, he says:

I recent­ly con­duct­ed sur­vey research com­par­ing the most con­ser­v­a­tive of Protes­tants — those who iden­ti­fy with a con­ser­v­a­tive Protes­tant denom­i­na­tion, attend church reg­u­lar­ly and take the Bible lit­er­al­ly, or about 11% of the pop­u­la­tion in my analy­sis — with those who do not par­tic­i­pate in any reli­gion. The con­ser­v­a­tive Protes­tants are equal­ly like­ly to under­stand sci­en­tif­ic meth­ods, to know sci­en­tif­ic facts and to claim knowl­edge of sci­ence. They are as like­ly as the non­re­li­gious to have majored in sci­ence or to have a sci­en­tif­ic occu­pa­tion. While oth­er stud­ies have shown that the elite sci­en­tists who work at the 20 top research uni­ver­si­ties are less reli­gious than the pub­lic, it appears that the vast major­i­ty of peo­ple with worka­day sci­en­tif­ic occu­pa­tions are like their neigh­bors, reli­gious­ly speak­ing â€¦

There are, of course, a few fact claims in which con­ser­v­a­tive Protes­tant the­ol­o­gy and sci­ence dif­fer, such as the ori­gins of humans and the uni­verse. Here we find that typ­i­cal con­ser­v­a­tive Protes­tants are like­ly to believe the teach­ing of their reli­gion on the issue and not the sci­en­tif­ic claim.

We could com­plain that they are being incon­sis­tent in believ­ing the sci­en­tif­ic method some of the time but not always. Yet social sci­ence research has long shown that peo­ple typ­i­cal­ly are not very con­sis­tent. The peo­ple who are more con­sis­tent are those who are pun­ished for incon­sis­ten­cy: philoso­phers, media pun­dits, polit­i­cal activists and politi­cians. [Note, though, that media pun­dits are not pun­ished for incon­sis­ten­cy. ‑JR] â€¦

The great­est con­flict between fun­da­men­tal­ists, evan­gel­i­cals and sci­ence is not over facts but over val­ues. While sci­en­tists like to say that their work is val­ue-free, that is not how the pub­lic views it, and con­ser­v­a­tive Protes­tants espe­cial­ly have homed in on the moral mes­sage of sci­ence. William Jen­nings Bryan, famed defend­er of the cre­ation­ist per­spec­tive at the Scopes “mon­key tri­al,” was not just opposed to evo­lu­tion for con­tra­dict­ing the Bible but also con­cerned that the under­ly­ing phi­los­o­phy of Dar­win­ism had ruined the morals of Ger­man youth and had caused World War I.

The sit­u­a­tion today is not that dif­fer­ent: Con­tem­po­rary “intel­li­gent design” advo­cates, for exam­ple, are deeply wor­ried that the teach­ing of evo­lu­tion has a neg­a­tive effect on chil­dren’s values.

The same research that shows fun­da­men­tal­ists gen­er­al­ly believe in sci­ence’s abil­i­ty to gath­er facts about the world also shows that they do not want sci­en­tists to lead the pub­lic debate on issues con­cern­ing morality.

Is it still futile then? Can the two “sides” nev­er agree?

No, it isn’t futile. Under­stand­ing what con­cerns the “oth­er side” would help. â€¦

To move for­ward, we, as a coun­try, need to low­er the polit­i­cal con­flict. Yes, the views found in fun­da­men­tal­ist church­es are not exact­ly the same as those at the Nation­al Sci­ence Foun­da­tion. But we would see less of the polar­iz­ing “we real Amer­i­cans” rhetoric from the reli­gious right if its mem­bers were not ridiculed as know-noth­ings. Con­ser­v­a­tive Protes­tants are not fun­da­men­tal­ly opposed to all science.

These points large­ly echo those I’ve made on this blog for years, though there are places I’d quib­ble. Evan­gel­i­cals val­ue sci­ence and tech­nol­o­gy, sure­ly, but their view of sci­ence is a cod-Bacon­ian sci­ence, not so much 19th cen­tu­ry as 17th, and wrong even from Bacon’s per­spec­tive. A love of tech­nol­o­gy and adher­ence to an out­dat­ed and restric­tive induc­tive mode of inquiry is not love of sci­ence, and it is in that gap which one can find the “fun­da­men­tal oppo­si­tion to science.”

But Evans is right that this sort of research can guide us towards bet­ter ways to engage peo­ple around sci­ence. Know­ing that evan­gel­i­cals treat sci­ence as what we can touch and influ­ence direct­ly does indeed sug­gest that cli­mate change com­mu­ni­ca­tion should focus on direct mea­sure­ments, not on mod­els, and the same for evo­lu­tion. (The trick, of course, is that the real­ly inter­est­ing parts of cli­mate sci­ence or evo­lu­tion­ary sci­ences involve tak­ing direct mea­sure­ments and mak­ing mod­el-based, testable extrapolations.)

I’m espe­cial­ly intrigued by the find­ing that evan­gel­i­cals are as like­ly as non­re­li­gious folks to major in sci­ence and that both groups per­form com­pa­ra­bly on sci­ence lit­er­a­cy tests. Elaine Howard Eck­lund’s sur­vey of sci­en­tists at elite uni­ver­si­ties found almost no evan­gel­i­cals, so clear­ly there’s some­thing keep­ing evan­gel­i­cals from con­tin­u­ing in sci­ence past col­lege, and that’s worth explor­ing in more detail. Do they apply for grad school in equal num­bers? If not, why not, and if so, why do they so rarely become professors? 

It’s con­ceiv­able that the flawed con­cept of sci­ence which evan­gel­i­cals often hold becomes an obsta­cle to advanced sci­en­tif­ic study. But I’d like to think that a sci­ence major will have had to con­front that issue before grad­u­at­ing. If not, it sug­gests a flaw in col­lege-lev­el sci­ence education.

Part of the dropoff is sure­ly explained by the cor­re­la­tion of evan­gel­i­cal belief and con­ser­v­a­tive polit­i­cal ide­ol­o­gy. For var­i­ous social rea­sons, con­ser­v­a­tives don’t choose to pur­sue advanced degrees as often as lib­er­als do, and evan­gel­i­cals are often con­ser­v­a­tives, so the same forces will apply. But that’s not a big enough effect to explain the mag­ni­tude of this shift.