ps418

For the benefit of anyone who wanted to see it, here is the data on religiosity from Bouchard et al's (1990) paper. Let me emphasize again that I am referring to religiosity, not religion. If anyone would like to explain to me, in a calm and dispassionate way, how the results can plausibly be reconciled with the position that all or even most of the variation in 'religiosity' is the result of environmental factors, please do.

Religiosity was assessed using two different scales. I have not seen the scales that were used. I will look at them next time I go to the library. The scales were administered to 31 pairs of identical twins reared apart ('MZA'), all of which were adopted in infancy, and grew up in different homes. The scales were also administered to 458 pairs of identical twins reared together-- in the same home at the same time with the same parents ('MZT').

If parenting and the rearing environment plays a large role in adult religiosity, than the correlation between MZTs for religiosity should be substantially greater than the correlation between MZAs. But that's not what Bouchard et al found. Their results indicate a correlation coefficient of 0.51 for the MZTs, and 0.49 for MZAs. Thus, taken at face value, parenting and the rearing environment account for ~0.02 of the variance in religiosity.

These results are consistent with an earlier finding by the same group (Waller et al., 1990), using a smaller sample, but more measures of reliosity (5 measures). <a href="http://www.apa.org/journals/psp/psp806845.html" target="_blank">Olsen et al. (2001)</a> state that substantial heritability was found on all five measures:

Broad sense heritabilty, that is. Olsen et al. again confirm these results. Their article is available online:

<a href="http://www.apa.org/journals/psp/psp806845.html" target="_blank">Olson et al., 2001. The Heritability of Attitudes: A Study of Twins. Journal of Personality and Social Psychology 80, No. 6, 845-860.</a>

Patrick

[ December 10, 2002: Message edited by: ps418 ]</p>

pz

You don't really believe that, do you? Tell me you're just being facetious here.
Whoa there. That's a pretty far out statement -- does the paper actually say that, or is that your interpretation? It's obviously false, for a number of reasons.

Consider just the MZTs for a moment. You've just said that the correlation coefficient for MZTs is 0.51. That's rather far away from a perfect correlation of 1.0, but it does say that there is a broad trend. An individual who scores as highly religious is likely to have a twin who is also religious, while any atheist is likely to have a twin who is not particularly religious. If we saw this plotted as a scatter plot, a nice positive correlation would be obvious, but we'd also see quite a bit of variation -- some individuals with high scores would have a twin with a middling score, for instance. In light of all this variation within a group of genetically identical individuals, how do you make the argument that the environment contributes to only 2% of the variation? Where does all that variation within one group come from? Just random, indeterminate noise?

So you've got a measure that shows that twins raised together have similar religious attitudes, but that there is some variation. You've got another group of twins raised apart, that also show a roughly equivalent degree of similarity, but the correlation is a little weaker, a little noisier. Now look at the ns: 458 to 31. Am I surprised that the smaller group shows a noisier signal? How can you possibly draw the conclusion you have from that comparison?

I have a much more plausible explanation for you. Raising twins together or raising twins apart in this study made essentially no difference in the correlation of their religious attitudes because they were still being raised in similar environments. You didn't have any twins where one was being raised by wealthy Episcopalians and the other was living with a vagrant mob of Holy Rollers. In fact, I'll predict that the majority of those separated twins were either living with relatives, or were both subject to the chaos of institutionalized foster parentage and the adoption system.

I'll mention other obvious flaws in the study. The subjects are not selected by random sampling, but by self-recruitment. They stepped forward as a result of Bouchard's publicizing, which made the 'eerie similarity' of twins the focus of interest. This flaw was magnified by the nature of the study, which was by self-report on questionnaires. Remember, these twins raised 'apart' are not studied in isolation -- they know of each other, and have been comparing notes. Who knows? If they gave their dogs the same name when they were 12 years old, they might get to go on Oprah.

What a ghastly paper. Let's throw a jumble of questions at people, and then fish among the results for correlations that we like. It just confirms my impression of this field that no one in it is capable of focusing on simple, clean experimental design that addresses a discrete issue, and then following through with in-depth analysis. That this paper is reviewing a host of other papers that follow the same crippled, hopeless strategy is just a further indictment of the field.
This paper starts out better. At least they admit that they are "investigating the issue of genetic determination of the differences between individuals in their expressed attitudes", and they spend a fair amount of time paying lip service to the subtleties and ambiguities of what they are studying. But then they plunge into madness. It's the same ol' 'fling more questionnaires at people, and then overanalyze the results'. I mean, seriously...they're looking at the heritability of doing crossword puzzles, capitalism, and their stance on abortion. They spit up numbers. They generate correlation coefficients and chi square values. It's all very busy and all very superficial. I've seen the same approach in Gauqelin's study that 'proved' the position of Mars in the sky at the time of birth was a robust indicator of whether someone was going to be a great athlete or not, and he had a much bigger data set to work with.

Why should I believe any of this? Given the deep, deep limitations in any such study that must work with humans ("natural experiments", right...), I find the conclusions laughable. If they are trying to support some kind of genetic determinism, they have failed. If you're going to argue that genetic determinism is a strawman, that they're looking at something more complex and subtle, then they've also failed, and worse, what they're looking for is so ill-defined and nebulous that I don't know why they wasted their time with it.

ps418

Pz,

Thank you for your reply. Now I have a much better idea about what exactly you think the flaws are. I disagree with your assessment, and think the evidence clearly supports a genetic influence on religiosity, but I will make a point to get each of these papers and fairly consider each point you have made.

However, you still have failed to clear the air regarding something you stated in the beginning of this thread. Specifically, regarding twin studies of religiosity, you stated:

Now, I am 99.9% certain that you fabricated this fact to support your prior beliefs. I regard this as a significant matter. I have all but begged you to either provide the references for this claim, so that I can verify it and accept your reductio, or to retract the claim and admit your mistake. Now I ask you for the 4th time to either provide support for that claim, or to retract it.

Thank you,

Patrick

pz

That's a rather offensive accusation you've made. No, I didn't fabricate it. I read it in one of these twin studies some time ago, and it was a rather jarring observation, which was why I noted it. I have not been able to find the original source yet, however. If I do, I'll let you know.

I really do not understand why you find this so unlikely. These papers you've cited report correlations for fondness for doing crossword puzzles, castration as a punishment for sex crimes, playing bingo, and liking roller coaster rides...and you find it so unbelievable that one of these endless studies would find a correlation in specific religious belief that you'll accuse me of lying about it?

I'll tell you what. You can go back to my reductio and substitute "beliefs about crossword puzzles and abortion rights" for sect choices, and tell me if it is no less absurd.

ps418

I disagree that you have provided any reasons for thinking this is "obviously false." But to answer your question, here is what Bouchard et al say:

I agree wholeheartedly that it is "far out" that rearing environment would account for so little. Yet this has been found over and over again, not simply with regard to religiosity, but for many other traits as well (e.g. <a href="http://www.amazon.com/exec/obidos/tg/detail/-/0684844095/qid=1039901225/sr=8-1/ref=sr_8_1/103-7389065-1655809?v=glance&s=books&n=507846" target="_blank">Harris, 1998;</a> <a href="http://www.amazon.com/exec/obidos/tg/detail/-/0898621321/qid=1039901272/sr=1-1/ref=sr_1_1/103-7389065-1655809?v=glance&s=books" target="_blank">Rowe, 1993: </a><a href="http://www.amazon.com/exec/obidos/tg/detail/-/0471319228/ref=pd_sim_books_2/103-7389065-1655809?v=glance&s=books" target="_blank">Cohen, 1999;</a> <a href="http://www.amazon.com/exec/obidos/tg/detail/-/0465076882/qid=1039902086/sr=8-1/ref=sr_8_1/103-7389065-1655809?v=glance&s=books&n=507846" target="_blank">Dunn and Plomin, 1990</a>).

Yet again you trot out the utterly irrelevant strawman of genetic determinism, despite the fact that not a single study I cited argued for the genetic determination of any trait, and when in fact they all conclude something quite the contrary (that nongenetic factors account for much if not most of interindividual differences).

[ December 14, 2002: Message edited by: ps418 ]</p>

ps418

Well, I've now reviewed many of these twin studies myself, and have yet to find the one you apparently read. But yes, please do let me know when you find the one you read.

Huh? We're not simply talking about correlations. Correlations in themselves are not evidence of heritability. For instance, there is a very good correlation between parents and offspring with regards to specific religious beliefs (e.g. belief in transubstantiation), but that data alone does not even imply heritability. On the other hand, if you took a large sample of children of Catholic parents, and reared them in protestant homes, and a large proportion of the children became Catholics rather than Protestants, then you'd have some presumptive evidence of heritability. Thats the kind of evidence Im looking for.

Like I said, if you happen to recall which twin study you were referring to, please let me know so I can verify your claim. I'm currently reviewing twin studies, and have yet to see the one you referred to.

Patrick

[ December 14, 2002: Message edited by: ps418 ]

[ December 14, 2002: Message edited by: ps418 ]</p>

ps418

This is a fundamental misunderstanding of what I said. I did not say that environment (i.e. all non-genetic factors) contributes only 2% to the variation (which would mean that genes account for 98%)! That would not only be absurd, it would be empirically falsified by the very evidence I cited. What I specifically said is that the rearing enviroment accounts for very little of the correlation. In behavior genetic jargon, this is referred to as the common or shared environment, as opposed to the unique or nonshared environment. Virtually all of the nongenetic variation can be attributed to the nonshared environment, and shared environment accounts for surprisingly little (see the sources I cited above).

No, that explanation is not at all plausible. The similarity of the rearing environments clearly does not account for the correlation. Maybe the reason you thought this was plausible was because I left out an important fact from my argument, which severely weakened it. The fact is this: the religiosity scale correlations for both MZTs and MZAs are much greater than for both DZTs and DZAs (dizygotic twins reared together and apart, respectively. In other words, even monozygotic twins reared apart show a greater concordance on measures of religiosity than dizygotic twins reared together! And I'm still looking for specific data, but I would wager than adoptive siblings reared together show even less correlation on measures of religiosity than dizygotic twins whether reared apart or together. This differential concordance along a genetic continuum from MZ twins to DZ twins to adoptive siblings (so-called virtual twins) is expected when a trait is genetically influenced.

Here is a quote from twin researcher Nancy Segal, from her book <a href="http://www.amazon.com/exec/obidos/tg/detail/-/0452280575/qid=1039969436/sr=8-1/ref=sr_8_1/103-7389065-1655809?v=glance&s=books&n=507846" target="_blank">Entwined Lives:</a>

It is the differential correlations that suggest a genetic influence. The magnitude of the influence is proportional to the difference between the MZ and DZ concordance. Note also that Segal specifically repudiates the idea that specific religious beliefs are genetically influenced.

I found another quote from prominent twin researcher and behavior geneticist David Lykken, also repudiating the notion that specific religious beliefs are genetically influenced, even though measures of religiosity are genetically influenced. The quote comes from an article titled <a href="http://www.edge.org/3rd_culture/lykken/" target="_blank">How can educated people continue to be radical environmentalists?</a> (where radical environmentalism is more or less the position you have advocated elsewhere, that personality differences are largely or solely the result of environmental differences). On <a href="http://www.edge.org/3rd_culture/lykken/lykken_p3.html" target="_blank">page 3</a> of that article, Lykken states:

Also I should point out that Lykken, along with Bouchard and McGue, has published a more recent paper on the heritability of religiosity, examining the correlations for both MZAs and DZAs (35 and 37 twin pairs, respectively), and again found significant heritabilities, comparable in magnitude to those found in the previous studies. The library I use does not have a subscription to the journal Twin Research, so it will be a while before I can get a copy of this paper. But I'll include the abstract:

Bouchard TJ Jr, McGue M, Lykken D, Tellegen A., Intrinsic and extrinsic religiousness: genetic and environmental influences and personality correlates. Twin Res 1999 Jun;2(2):88-98.

I know from past experience that you are a smart guy. Therefore, I do not believe that you are so obtuse as to not see the difference between finding a heritability for, say, authoritarian attitudes (e.g. "criminals should be punished harshly") and attitudes about thrill-seeking behavior (e.g. "roller coasters are fun") on the one hand, and highly specific religious beliefs on the other (e.g. Catholicism vrs. Protestantism). The former does not seem a priori unlikely to me, but the latter does seem unlikely, and if true, would cause me to seriously reconsider the validity of the twin studies.

And I didn't accuse you lightly. After all, fabricating data is the worst thing that someone can be found guilty of, intellectually speaking. I accused you because: 1) you repeatedly ignored my requests for references or a retraction, 2) after a near-exhaustive literature search of twin studies on religiosity, I could not find a source supporting your claim, and 3) at least two twin researchers are on record specifically denying the claim, saying that specific religious orientation/affiliation does not show heritability. I have some more to say on the matter, but I'd like to wait until I have finished my review of the literature. I thank you for pressing me to clarify my thinking on this matter.

Patrick

scigirl

My take on this topic is that our brains have certain features which allow us to believe in the supernatural (i.e. make faulty connections between temporal and spacial events). Therefore, any genes which enhance or take away this feature may play a role in determining the types of religious experiences a person can undergo.

A problem I have with all these studies we are debating here is the way in which people are being categorized. If religion is a manifestation of brain function, than we should group people into "types of religious experiences" instead of "jewish" or "catholic."

Figure out how the religious belief manifests itself into the life and beliefs of the person. Are the fanatical? Are they skeptical? Do they appeal to authority easy (take every word by the preacher/rabbi/sheik on faith) or do they question? Are they more inspired by visual experiences or by written texts?

My guess is that if we categorized people this way, we would find certain brain patterns, some of which might be due to heredity.

I found some articles and posted them <a href="http://iidb.org/cgi-bin/ultimatebb.cgi?ubb=get_topic&f=57&t=000767" target="_blank">in science and skepticism</a>. Enjoy!

scigirl