related to link in previous post

[somedaybitch]

anybody with a bigger brain into numbers and analysis care to take a crack at this?

i'm not all that brilliant, but it seems to me that you can't make the causality argument here. i can see correlation, but how can one seriously argue cause? isn't there a vested interest bias in the outcome for the author?

anyone? bueller?

Comments

[fialka.livejournal.com]

Um...considering the percentages in the first box add up to more than 100% and all the rest considerably less, I'd say bollocks.

Not that I don't think you *couldn't* prove that as a generalised argument, but I reserve the right to claim exception for a few friends of mine who are both religious and intelligent.

Mind you, I also don't think any of them believe the bible should be interpreted literally, so maybe the intelligence correlation is related to that? :)

[fialka.livejournal.com]

feh, I just had to look again. Yeah, okay, I see what he's doing now and why the percentages don't add up. But still (and particularly considering his views on race and intelligence) I'd cite this as an excellent example of how statistics can be manipulated to say pretty much whatever you want depending on how you want to massage your data and what underlying factors you choose to ignore (in his case, all of them).

Yes, anecdotally, the idea of a correlation between intelligence and certain forms of religiosity certainly does seem possible (so too does the idea that there's no correlation). It's an interesting question to ask. But, as someone pointed out in the comments, WORDSUM is a measure of verbal aptitude for standardised English, not intelligence. So with one thing and another, this manipulation of the GSS data isn't going to answer the question. It will, unfortunately, get a lot of people to point a finger and say "see, I told you them folks in Marietta is dumb!"

It kind of reminds me of something my father used to say, which was, "A man with a little knowledge and the will to use it is a very dangerous thing."

[somedaybitch.livejournal.com]

okay, i feel better. and it's not that one can't make one argument or the other, it just seemed, i dunno, kind of impossible to say one caused the other. i know plenty of really religious people that are exceptionally intelligent and plenty of really stupid people that aren't in any way religious.

[fialka.livejournal.com]

Exactly. Although the teaching of creation as science may indeed be *causing* a good deal of dumbing down in Marietta, GA :)

[eve11]

But, as someone pointed out in the comments, WORDSUM is a measure of verbal aptitude for standardised English, not intelligenc

Exactly. There are a lot of underlying factors that could explain this correlation -- education, race, socioeconomic status, just for starters. On the other hand, I don't see anywhere in the article where they say that religiosity causes a lack of intelligence. They are pointing out the correlation. People with higher verbal scores tend to be more liberal and to not believe in God.

[fialka.livejournal.com]

I don't see anywhere in the article where they say that religiosity causes a lack of intelligence

Agreed, but I think the problem is he's flirting with the other paradigm, ie, that those who are less intelligent are more likely to be more religious. IOW, ranging dangerously close to 'only an idiot would believe in God'.

I'm not religious myself, but I can see how that is not only offensive, but counter-productive to trying to understand why fundamentalism is on the rise across the globe atm. His science is as sloppy as any creationist's, and evidently just as powerful in the hands of someone with an agenda.

::shudders::

[eve11]

Oh, agreed. Yeah, the "only an idiot would believe in God" argument is just ridiculous.

[boofadil.livejournal.com]

This dumb ass says "ugh".

[vee-fic.livejournal.com]

To elaborate on Fi's notes above:

1. It's just as fair to say "less intelligent people tend to be more religious." Correlations work both ways. But the evidence used seems to show, rather, that "more intelligent people tend to be less religious," which is not the same statement.

1a. And yes, the first rule of statistics is Do NOT talk about statistics Correlation is not causation.

2. Verbal IQ is a very narrow measure of intelligence. In fact, it's a shitty one, as anybody who has taken the SAT will tell you: to get into college, you have to match verbal IQ with mathematical IQ, and colleges also like things like emotional/interpersonal skills and kinesthetic skills. No college wants to turn out a bunch of maladjusted geniuses without also turning out at least a few future CEOs and Congresspeople. (Note: Assume that CEOs and Congresspeople don't tend to be maladjusted geniuses, nor geniuses of any kind.)

3. Verbal IQ is an intelligence directly related to education. How do you learn words? You are exposed to them. So based on the evidence shown, it is no less fair to say "people with more education tend to be less religious." Introducing this distinction also introduces all of the factors involved in education: income, race, childhood nutrition, family traditions/reading exposure, class size, school resources, neighborhood safety, etc. etc. That's a lot of counfounding factors! Automatically, causation must be ruled out until every one of these factors can be accounted for in the data.

3a. (The King James Bible is famed in linguistic circles for having a vocabulary of about 8,000 words -- as a foundational document, that's pretty vocabulary-poor. Bible-lovers who don't also read, e.g., Milton, would necessarily score lower on a verbal IQ test, despite deep and thoughtful analytical skills, because the Bible doesn't include words like parallelogram.)

4. Showing that the smartest segment tends to believe less is not the same as showing anything about the middle segments which mst people occupy. (13.8% has to be at least 1 standard deviation away from the middle.) This person is using the outliers to make statements about the whole sample, which is disastrously wrongheaded.

4a. I can think of a lot of reasons for outliers to hold atypical opinions, on religion as well as on other topics. They're outliers. That's kind of the point. Possibly a small percentage of the smartest people don't believe in god because they believe they themselves are god.

5. In sum, don't let people with no statistics training at a set of data. It's like letting amateurs design and build a house: you might get something pretty that doesn't have any support beams, and ten minutes after you move in, it'll collapse on your head.

[fialka.livejournal.com]

you might get something pretty that doesn't have any support beams, and ten minutes after you move in, it'll collapse on your head.

I really should say something more intelligent to all that but my WORDSUM is -3 this morning and all I can do is ROFL.

It's all putting me in mind of Penn and Teller, which I happened to watch awhile back. Explaining the creationism/evolution debate Penn reads (longly and somewhat tiringly) from Darwin. He then turns to Teller. "And your side of the argument?" Teller picks up a very large, very handsome bible. He considers it thoughtfully. And then uses it to smash Penn in the face.

[somedaybitch.livejournal.com]

*snort*

[eve11]

Showing that the smartest segment tends to believe less is not the same as showing anything about the middle segments which mst people occupy. (13.8% has to be at least 1 standard deviation away from the middle.) This person is using the outliers to make statements about the whole sample, which is disastrously wrongheaded.

I wouldn't say it's wrongheaded; they're making a statement of conditional probability. Ie, given that you're "smart" -- (to them it's top 13% of the Wordsum scale), you're more likely to not believe in God. There's really no other way you can do that aside from calculating the conditional percentage. If you have a large enough sample size, you're bound to get extreme values but those aren't necessarily the same as outliers. As long as you've got enough people in that group your results should be sound. The problem would be; if the group that they saw had only, say, 5 people in it, then they wouldn't be able to have much confidence in any of their estimates. Experimentally speaking, an outlier is a point that you believe is extreme because it didn't come from the distribution you think it did -- there's a confounding variable, there was experimental error, etc.

Ah, I just broke rule number 1, didn't I?

[somedaybitch.livejournal.com]

1. ...But the evidence used seems to show, rather, that "more intelligent people tend to be less religious," which is not the same statement.

see, that's what *i* was getting from what i read, and i wasn't sure if i was skewing what i was reading because, from anecdotal evidence that's what i've come to believe, OR, the writer was skewing the interpretation because that's what the writer believes.


1a. And yes, the first rule of statistics is Do NOT talk about statistics Correlation is not causation.

heh. and, yes, how, without examining the influence of all the other factors that go in to measuring intelligence, can you make a causal statement, when you can't even make a definitive correlative statement without at least including qualifiers?

and great additional points, especially the middle folks.

i'd really love to see someone(s) tackle this, because the topic is fascinating, but i think it would have to be from someone who has no vested interest in it's outcome other than curiosity.

[thevaliumsofalj.livejournal.com]

2. Verbal IQ is a very narrow measure of intelligence. In fact, it's a shitty one, as anybody who has taken the SAT will tell you: to get into college, you have to match verbal IQ with mathematical IQ, and colleges also like things like emotional/interpersonal skills and kinesthetic skills. No college wants to turn out a bunch of maladjusted geniuses without also turning out at least a few future CEOs and Congresspeople. (Note: Assume that CEOs and Congresspeople don't tend to be maladjusted geniuses, nor geniuses of any kind.)

3. Verbal IQ is an intelligence directly related to education. How do you learn words? You are exposed to them. So based on the evidence shown, it is no less fair to say "people with more education tend to be less religious." Introducing this distinction also introduces all of the factors involved in education: income, race, childhood nutrition, family traditions/reading exposure, class size, school resources, neighborhood safety, etc. etc. That's a lot of counfounding factors! Automatically, causation must be ruled out until every one of these factors can be accounted for in the data.

and, the SAT is not really a good measure of IQ in general. It's only really meant for forecasting the GPA of the student as a first year freshman. When we do tests to measure IQ, the mathematics/arithmatics is part of the verbal subtests, not part of the performance/processing tests that we do - like Block Design and Symbol Search - that test for the non-verbal, non-education/experience based learning.

[arthurfrdent.livejournal.com]

heh, sorry I missed this one ;) esp. with the tasty thought processes :D