A friend emailed me and suggested I look into the GSS and figure out which groups are smarter in terms of political & religious combinations. So again, my methodology is simple. I took the WORDSUM variable, which is the # correct out of 10 on a vocabulary test as a proxy for intelligence. Sliced & diced the survey sample into “Secular” vs. “Religious” (i.e., those whose confidence in the existence of god is 0 to sometimes vs. those who are mostly to totally sure), and “Right” vs. “Left” ((aggregating from “slightly” to “extremely” liberal or conservative). I also split the results including non-whites, and only with whites. Here are the means for the classes:
As you can see, the rank order is Secular Left > Secular Right > Religious Right > Religious Left. If you look at the original data, you will note interestingly that the most and least intelligence classes on a more granular level seem to be the extreme liberals who are non-religious (the smartest) and religious (dumbest). This applies both when you have a racially diverse, or whites only, sample. Extreme conservatives who are non-religious don’t seem very bright, but the N is small (19), so I don’t know what to say about that. If I had to hazard a hypothesis, it might be that very smart people know what society expects of their worldviews. So secular people should be on the Left, and the bright ones know this and conform their disparate Weltanschauung appropriately. The rather dull on the other hand don’t realize that society expects the religious to be on the Right, and so avow their extreme liberalism without any doubt or self-consciousness. Perhaps the same could be said of non-religious individuals who assert their extreme conservatism? (Or perhaps they don’t even know what “extreme conservatism” entails)
But you want to know the distributions, right?
Obviously you can conclude from these data one definitive point: secular liberals are smart. On the other hand, only ~30% of liberals are secular, vs. ~15% of conservatives. The self perception of “liberals” that they are much smarter than “conservatives” probably has to do with the fact that the intelligentsia is dominated by secular liberals.
These results are meaningless in the absence of a statistical confidence test. The same holds true of your other posts using this dataset.
No they’re not. You don’t demand confidence tests for verbal arguments. This isn’t being submitted into peer review, but you do have confidence tests in the GSS. But p-value 0.05 isn’t magical, it’s a useful convention. If you want all the statistics, do it yourself.
I am confused by the vertical axis. What is the scale? what are the units?
% of each class. Add the Y’s for each class going along the X axis and you get ~ 100% (some would not respond, etc., so it might not be perfect). Some raw data:
WORDSUM Secular Right
0 1.2
1 1.4
2 2.1
3 3.1
4 7
5 17.8
6 21.6
7 12.8
8 15.3
9 9
10 8.7
So 8.7% of those who were “Secular Right” (God(1-4), Polviews(5-7)) got a 10 out of 10 on the vocab test.
In terms of information, note the mode for the Secular Left. They’re obviously pretty smart if the mode is so high.
10 vocabulary words as a arbiter of intelligence?
Maybe you ought to start off explaining your concept of intelligence that a simple vocabulary test, let alone one of such a puny scope, could ever equate to intellignce.
Education maybe, but then it would still have to be a far larger scope for evaluation. It’d also be good to know what the vocabulary words were, who chose them, and why.
You’ve also failed to plot any standard deviation. I wager that if you do, you’ll find that there’re no significant differences in your samples.
Yuo’ve also not compared your distributions to the groups as a whole by percentage. Not that it matters–thre doesn’t appear that there’d be any significant deviations there either.
Additionaly, why on earth break out white respondents without breaking out any other group (including, say, “non-white”) for comparison. Further meaninglessness.
Maybe you ought to start off explaining your concept of intelligence that a simple vocabulary test, let alone one of such a puny scope, could ever equate to intellignce.
GSS Wordsum proxy for IQ, not definitive measure of it.
Education maybe, but then it would still have to be a far larger scope for evaluation.
Already in some of my previous posts.
You’ve also failed to plot any standard deviation. I wager that if you do, you’ll find that there’re no significant differences in your samples.
DO NOT WAGER IF YOU WON’T DO IT YOURSELF! I made my methodology transparent so that people could actually replicate and do it themselves, INSTEAD OF SPECULATING. Is that clear? I tire of people who criticize me for not investing the time to explore & transcribe all the statistics WHO REFUSE TO THEMSELVES EVEN BEGIN WHAT I STARTED.
(caps lock on again, to indicate my genuine fury)
Additionaly, why on earth break out white respondents without breaking out any other group (including, say, “non-white”) for comparison. Further meaninglessness.
Most of the sample is white. The N’s get small for secular conservative non-whites. Though they do exist….
@David Hume
Yet you offered the white breakout as though it were meaningful? You intended it to relate to a larger sample that also includes it? What are you trying to show?
Hoping not to extend Razib’s fury (do I have to call you David Hume on this site? I don’t mind but there was already another bloke named that…), but I need to know more about RTO Trainer’s question:
“You’ve also failed to plot any standard deviation. I wager that if you do, you’ll find that there’re no significant differences in your samples.”
Is this to imply that the differences in these means are meaningless if applying 1 SD in the proper direction causes them to reverse themselves? I.e., if you subtract 1 SD from the secular right’s mean, and/or add an SD to the religious left’s mean, and the latter comes out ahead of the former, then Razib’s conclusion is wrong?
Because if that is the implication I don’t agree. I don’t see why the standard deviations would make Razib’s information meaningless. Means are means.
@RTO Trainer
Black people are way more liberal than white people according to surveys, but they are also more religiously fundamentalist. Controlling for race eliminates the exogenous impact of the Southern Strategy.
Normally, one would determine teh central tendency in the data sets by mean, mode, or median and then calculate the standard deviation. The SD is then added and subtracted from the center. If the value of given sets does not exceed the + or – SD, then the difference between them is considered statistically immaterial.
No matter, the sample and the measurements are fatally flawed.
No matter, the sample and the measurements are fatally flawed.
Bullshitter. You actually didn’t check, you’re just making stuff up. Even separating on a more granular level many of these classes are outside each other’s 95% intervals.
I’m letting this bullshitter’s comments through as an example. You don’t even have to do your own calculations when you use the GSS interface, it generates many statistics for you (e.g., std).
@RTO Trainer
Hmm … I’ve never seen std used that way. If I’m reading you right, a difference of 1 to 14 IQ points is “considered statistically immaterial”? (I’m using IQ only because large samples should have a 15-point std.) Do you have a source for this?
RTO is wrong. If you repeat Hume’s analysis and get the ANOVA stats there’s a significant (p = .005) interaction effect. However, the definition of secular is important. If you restrict secular to GOD = DONT BELIEVE + NO WAY TO FIND OUT, the the liberal/conservative difference is not significant.
RTO is also wrong about wordsum. It’s a well studied IQ measure (strongly correlated with full scale IQ), and vocabulary tests in general are the single best IQ tests.
Here’s how wordsum means break down by right/left and three groups of religiosity
Atheist + Agnostic:
left = 7.41
right = 7.51
Some Higher Power + Believes Sometimes:
left = 7.11
right = 6.39
(there’s your difference)
Believes in God (w/ and w/o doubts):
left = 6.31
right = 6.38
Bravo ‘zib.
What if…we overlay political affiliation on the bellcurve of IQ? Presuming we had some valid metric of IQ, of course. Would we have seen…..20%democrats/50%republicans/30%democrats in 2004?
What would we see now?
30%/40%/30%? 20%/40%/40%?
And what if we regress political affiliation on Gini?
These results don’t surprise me. Everyone knows that the atheist left tends to be smart. Intelligence is part of their problem. They think they’re smart enough to tell everyone else how to live and can even out-think the market. Besides, many of them are Jews.
I have often said that stupid people make the best leaders. President Reagan was semi-senile and probably only slightly more intelligent than average when younger. Excellent writer though, because he wrote so simply.
The definitions of right and left are becoming more and more blurred. I consider myself a secular rightist, yet I score slightly left libertarian on an online political test. I am both an atheist and pro-abortion, not just pro-choice, and reasonably intelligent (some grad school). The SPLC claims me as the leader of four right-wing hate groups in Augusta GA, surely a record for anyone. I am also a 25 year member of my Rotary Club and past president. Interesting combination.
Would being pro-abortion mark me as a leftist? Would being a race realist mark me as a rightist? David, where would I fit in on your graphics?
Lol.
I didn’t see the Church of Atheism pouring 20+ million congregation dollars into tellin’ Mormon citizens they couldn’t marry.
Not to mention that well-intentioned evangelical bumbler that just trashed our economy in his eightyear stewardship and spent 700 billion dollars and 4000 lives from the finest military on the planet to midwife the birth of another Islamic state.
Perhaps if GW had been a smart atheist we would be in a different world today.
Somehow I don’t think smart atheists are our problem.
This is a pretty scary comment.
Obviously you can conclude from these data one definitive point: secular liberals are smart. On the other hand, only ~30% of liberals are secular, vs. ~15% of conservatives. The self perception of “liberals” that they are much smarter than “conservatives” probably has to do with the fact that the intelligentsia is dominated by secular liberals.
Could I not conclude that smart people figure out that the way to get ahead today is to be a secular leftist? Hence, the incentives are in place to be a secular leftist, so we get more…secular leftists?
AC, there was a lot of discussion at Volokh a while ago about a study showing evangelical christians are the religious group most thoroughly disliked by academics, with mormons second.
In general discussion people theorized that evangelicals and mormons were the groups most strongly percieved to be political agitators, and also the groups that proselytized the most heavily.
It would be interesting to repeat the study and see if post Prop 8 activism Mormon dislike has moved above evangelical dislike.
Yeah, I was rather amused by that. Fits into all the stereotypes about the right, just or otherwise.=.
@”David Hume”
With regards to the earlier defense of using wordsum as a means to gauge IQ:
You link – one assumes as justification – to a blog attempting to defend use of the wordsum as proxy. Unfortunately, the blog, rather amusingly, argues that the best way to correct for inherent errors in the test itself is to parse the test into finer and finer grains with more and more correlative factors.
Kind of like adding that deferant to explain retrograde motion.
That’s not legitimate or rigorous, and perhaps you did not mean it as justification.
If so I continue to wait for some explanation of why you feel so certain wordsum GSS data can be reliably used to draw conclusions. Despite voluminous peer reviewed research suggesting that while it can suggest trends it absolutely cannot be used for conclusive statements.