Thursday, August 31, 2006

If Wishes were Fishes...

by Drek

Folks who recall one or more of my previous contributions to this blog will probably know that I am an atheist. As such, I have the dubious honor of being a member of one of the least liked religious groups in the United States. Since I'm a natually disagreeable sort, this doesn't particularly bother me, but it does make my life fairly interesting from time to time.

I've been an atheist for a long, long time now and, as such, have often been on the receiving end of attempts at evangelism. These attempts usually range from well-meaning and gentle, to heavy-handed and condemning. My personal favorites, however, are the staggeringly inept and illogical as, frankly, they signal that the evangelist is going to be a fair amount of fun to mess with. What can I say? If I have to be subjected to some guy babbling on about his sky beast, I may as well find some sort of redeeming value in it.

With all of these attempts at conversion I've been exposed to an awful lot of arguments in favor of god. For the most part, however, attempts to argue me into submission seem to follow a general pattern. It usually begins with "How can you not believe in god?! He's everywhere!" Once my would-be savior has been convinced that, no, I don't see god everywhere, the next response is often, "Just try believing in god, you'll feel such peace!" On making the point that I used to be religious and feel more peace as an atheist than I ever did as a theist, I usually get the Pascal's Wager argument, "But, don't you think it's safer to believe in god just in case?" When I point out that I can't simply fake belief in front of an omniscient being* and that wagering on that kind of long-shot seems like a poor bet, it usually ends up devolving to what I like to think of as "The Final Insult." What is the final insult, you ask? It's when my erstwhile associate remarks in frustration, "Well, it's a good thing more people don't believe like you do or society would be even more screwed up than it already is!"

The final insult has always, more than nearly any other argument for god, annoyed the ever living hell out of me. The reasons are simple: atheists are, as a general rule, indistinguishable from the rest of the population. We obey the law, we pay our taxes, we work hard, have families, and participate in our communities. We blend in so well that most people are shocked to discover that an atheist has somehow crept into their midst. Given all that, what earthly reason is there to assume that a larger proportion of atheists would, somehow or other, signal the demise of western civilization?

The answer, of course, is that there's no earthly reason for such an assumption. Instead, we come back to the voyeur in the clouds himself. Since morality comes from god, the theists reason, those who do not believe in god must not, themselves, be capable of morality. Since atheists so obviously are capable of morality, it has always seemed to me that those who make this sort of argument are more than a little willfully ignorant.

Of course, pointing this out never seems to help as they often take refuge in a variant on the original argument that more or less goes as follows, "Okay, atheists can be ethical when surrounded by godly people but if they were ever left on their own, it would be chaos! Look at the Soviet Union! Communism was atheism!**" So, in other words, we're only civilized when we have lots of religious people to keep us in line. As I've said before, it's good to be loved.

All of this springs to mind not because I'm just generally a bitter asshole (although, seriously, I'm that too) but because of a study I recently came across. This research, published in the Journal of Religion and Society, does something that I have rarely seen attempted: it tries to empirically evaluate the claim that religious belief is associated with societal health and well-being. Seriously. I'll let the introduction speak for itself:

Two centuries ago there was relatively little dispute over the existence of God, or the societally beneficial effect of popular belief in a creator. In the twentieth century extensive secularization occurred in western nations, the United States being the only significant exception (Bishop; Bruce; Gill et al.; Sommerville). If religion has receded in some western nations, what is the impact of this unprecedented transformation upon their populations? Theists often assert that popular belief in a creator is instrumental towards providing the moral, ethical and other foundations necessary for a healthy, cohesive society. Many also contend that widespread acceptance of evolution, and/or denial of a creator, is contrary to these goals. But a cross-national study verifying these claims has yet to be published. That radically differing worldviews can have measurable impact upon societal conditions is plausible according to a number of mainstream researchers (Bainbridge; Barro; Barro and McCleary; Beeghley; Groeneman and Tobin; Huntington; Inglehart and Baker; Putman; Stark and Bainbridge). Agreement with the hypothesis that belief in a creator is beneficial to societies is largely based on assumption, anecdotal accounts, and on studies of limited scope and quality restricted to one population (Benson et al.; Hummer et al.; Idler and Kasl; Stark and Bainbridge). A partial exception is given by Barro and McCleary, who correlated economic growth with rates of belief in the afterlife and church attendance in numerous nations (while Kasman and Reid [2004] commented that Europe does not appear to be suffering unduly from its secularization). It is surprising that a more systematic examination of the question has not been previously executed since the factors required to do so are in place. The twentieth century acted, for the first time in human history, as a vast Darwinian global societal experiment in which a wide variety of dramatically differing social-religious-political-economic systems competed with one another, with varying degrees of success. A quantitative cross-national analysis is feasible because a large body of survey and census data on rates of religiosity, secularization, and societal indicators has become available in the prosperous developed democracies including the United States.

In summary, this gentleman hopes to determine how much of an impact religion really has on popular morality. To conduct this research he employs the International Social Survey Project (ISSP) data to compare rates of religiosity and social dysfunction cross-nationally. To measure religiosity, the author uses indicators of belief in biblical literalism, absolute belief in a creator, measures of frequency of prayer, and measures of frequency of attendance at religious services. This represents a nice cross-section of both attitudes and behaviors. For social dysfunction he uses rates of homicide, youth suicide, teenage pregnancy, and abortion, with some additional analyses of sexually transmitted diseases thrown in for good measure. While some of these might be a little debatable, I think we can all agree that a maximally healthy society probably isn't one with high murder or suicide rates.

So what does he find? Well, in a development that will shock some theists, he finds that heightened religiosity is associated with heightened social dysfunction. Again, letting the article speak for itself:

In general, higher rates of belief in and worship of a creator correlate with higher rates of homicide, juvenile and early adult mortality, STD infection rates, teen pregnancy, and abortion in the prosperous democracies (Figures 1-9). The most theistic prosperous democracy, the U.S., is exceptional, but not in the manner Franklin predicted. The United States is almost always the most dysfunctional of the developed democracies, sometimes spectacularly so, and almost always scores poorly. The view of the U.S. as a “shining city on the hill” to the rest of the world is falsified when it comes to basic measures of societal health. Youth suicide is an exception to the general trend because there is not a significant relationship between it and religious or secular factors. No democracy is known to have combined strong religiosity and popular denial of evolution with high rates of societal health. Higher rates of non-theism and acceptance of human evolution usually correlate with lower rates of dysfunction, and the least theistic nations are usually the least dysfunctional. None of the strongly secularized, pro-evolution democracies is experiencing high levels of measurable dysfunction.

And there you have it: religion isn't associated with social health. Oh, the direction of causation is in some doubt- it's possible that when things get bad, people turn to religion, rather than religion making things bad, but either way, a lack of religion isn't itself a bad thing.

So does this research excite me? Am I even now preparing leaflets to distribute in church parking lots? No. It doesn't, and I'm not.

The reason has to do with the article itself. First, there's the minor issue that makes me nervous: this article cites Robert Putnam. Don't get me wrong, I have no objection to Bob Putnam. He seems like a nice enough guy and his work is provacative. No, my problem here is that this article consistently refers to him as "Putman." That doesn't mean that the article is bad, per se, but it does raise my eyebrows a smidge.

Then we come to the next issue: the author of this paper isn't a sociologist, or an anthropologist, or a psychologist, or an economist, or any other kind of social scientist. As it turns out the author, Gregory S. Paul, is a freelance paleontologist. In perfect honesty, I never realized that there was a market for free agents in paleontology, but there you go. Again, this doesn't condemn the work, but it does make me still more nervous.

And then, ladies and gentlemen, we get to the things that really weird me out. The first is a brief paragraph that appears early in the paper:

Regression analyses were not executed because of the high variability of degree of correlation, because potential causal factors for rates of societal function are complex, and because it is not the purpose of this initial study to definitively demonstrate a causal link between religion and social conditions. Nor were multivariate analyses used because they risk manipulating the data to produce errant or desired results, and because the fairly consistent characteristics of the sample automatically minimizes the need to correct for external multiple factors (see further discussion below). Therefore correlations of raw data are used for this initial examination.

Yes, you read that right: the entire paper consists of a report on basic correlations taken out of the ISSP data. Under the best of circumstances this would worry me for the simple reason that an awful lot of heterogeneity between observations is being left uncontrolled. Sadly, however, these are not the best of circumstances. In our present circumstances we're comparing data drawn from a variety of countries whose respective natural resources, geographic locations, and histories can have profound effects on their levels of social dysfunction. Hell, just the presence or absence of a strong welfare state and socialized medicine (common and effective in many European countries, effectively absent in the U.S.) could be expected to have a profound effect on many measures of societal dysfunction. And none of that natural variation, not a scrap, is being controlled for. We have no multi-level models, no random effects, just good old correlations. Are the developed democracies included in the study relatively similar to each other? Sure. Does that mean we don't need to worry about their very real differences? Um... no. This is especially true considering the author refers to "...the high variability of degree of correlation..." Here's a thought: maybe that's because you need to add some control variables! And don't even get me started on that "further discussion below" remark. There is no further discussion of this issue anywhere in the paper.

And what of those correlations, anyway? Well, as you can see in the results section, none of them are reported. We have no correlation coefficients, no standard errors, no significance levels. The best we get are a series of rather confusing figures that cast little light on the issue.

In the end what we're left with is a paper written by someone outside of his expertise, who has well-known authors miscited, who eschews the use of any but the most basic statistical analyses, and who fails to report any of the technical details of his findings. It's the scientific equivalent of, "Hey, man, just trust me!" As a result, I simply can't believe these results, no matter how much I want to.

And I do want to, more than you might expect. To have actual empirical results to back up my contention that atheism is not the death knell of a society would be truly wonderful. Such a finding would give all atheists ammunition in the fight to be accepted and, just maybe, help us put an end to a damaging stereotype. As much as I want to run with this finding, however, I simply can't do so in good conscience.

There are two expressions, "If wishes were fishes I'd cast my net in the sea," and "If wishes were horses, beggars would ride." I often combine them,*** resulting in quite a bit of silliness but, in either case, the point is simple: if wishes meant something the world would be very different from the way it really is. However, the world isn't different, and wishes really don't change anything. We can wish all day for world hunger to end, and it will still be there. I can wish that this paper were convincing but, ultimately, it just isn't. The simple reality is that science isn't about what we want to be true, but rather about finding out what is true. It's that distinction that often makes accepting science so difficult, for academics and laypeople alike.

In the final analysis, my inability to accept this paper probably explains why I'm an atheist in the first place. The idea of a loving god is a nice one, the idea that there's someone watching out for us all every moment of our lives is very reassuring. On some level, I suppose that I wish it were true, if only because it would mean that somehow people would end up getting what they deserve. But just wishing that something were true doesn't make it so.

But wishes aren't fishes, and beggars don't ride.

* Come to think of it, I couldn't do it BEHIND an omniscient being either, but that's not the point.

** Yes, this is an actual argument I've heard used numerous times. It makes my brain melt out of my skull nearly every time.

*** i.e. "If wishes were fishes, beggars would ride." Apparently in my subconscious, beggars are aquatic.
This is as good a time as any to provide the periodic link to the Best Weblog Post, Ever.

Clearly, part of the problem is that the JRS's editor appears to have no social science training at all; by JRS standards of quantitative analysis, this may look like they snagged something from the Journal of Economic Theory.

Still, it's painful to see something showing such poor understanding of elementary univariate and multivariate statistical methods. Not so painful, though, as having one's multivariate statistical analysis evaluated by someone with this sort of command of the methods.
Atheists are typically more fervent in their belief of the Non-god than theists are in their belief of God. Do you try to convert them as well, or just annoy them?
Well, "Lord," your assertion (which strikes me as threadbare) doesn't ring true. I don't doubt that many people attend religious ceremonies without being particularly devout believers. But is the Freedom From Religion Foundation to U.S. atheism, say, what evangelical protetstantism is to mainstream U.S. Christianity? I think not.
As usual, great post, but also as usual, I have a slightly different take. On one hand, I completely agree that (a) as a social scientist, this guy is clueless, and (b) the editor and reviewers should be ashamed of themselves for letting this pass. (Not that you said this, but I will.)

On the other hand, I'm not a big fan of "fly-over" comparative analyses -- fitting a multivariate regression or its extensions (e.g., multi-level models, etc.) to cross-national data, extracting the partial correlation of interest, and calling it a day. The problem is that in country-level analyses, the unobservables multiply far faster than observables, so just adding more countries and more controls isn't necessarily going to yield any less biased estimates of the effect of interest, in this case, religiosity. I'd like to see more scholars doing careful pairwise comparisions of countries, where there is a concerted effort to match not just on observables (the implicit assumption behind regression estimates) but also on known unobservables.

It wouldn't surprise me, incidentally, if the positive relationship between religiousity and murder rates held up in a multivariate analysis. The US is such an outlier on both that it can generate this positive (partial) correlation single-handedly, much as China can single-handedly generate a global trend toward greater income equality. But, also as usual, I digress.
Tom: Maybe I missed it but... did you mean to put up a link? If so, it isn't working for me.

Lord: Actually, the answer to your question is, "it depends." I don't go out of my way to proselytize theists. I do, really and truly, think that religious freedom is the only viable solution and I'm perfectly happy to let people believe in whatever god they want, or no god at all. Once someone decides to try and convert me however, I figure they've opened themselves up to counter-arguments. To put it more succinctly, when it comes to evangelicals, they shouldn't do the crime if they can't do the time. As for your claim about the fervor of the typical atheist... I'm not even going to touch that. Tom did a plenty good job already.

Kim: First off, thanks for the compliment. Secondly, as usual you raise useful and interesting points. I also tend to start with simple correlations and then get more serious from there, and I agree that just throwing variables into a model is no panacea. Your desire for careful pairwise comparisons is, likewise, very reasonable. I think my ire was provoked at least in part by the sheer absurdity of the arguments that the author made. The justification for having NO controls was mostly non-existent and, when present, was poor. If he had made something like your argument, I might not have been so irritated. What he actually did, though, was try to sweep the issue under the rug.

I'm open to a lot of methods, but for any method to be useful it has to be used properly and its strengths and weaknesses honestly discussed.
Drek: I must have messed up the 'a' tag -- try this, or here's the cut-and-paste link for you:

The thing about throwing variables at the model is that you'd like to have some a priori reason to include what you include, otherwise the specification search does horrible things to the distribution of the estimators. Nevertheless, the general rule wrt which Paul seems to be totally oblivious is that the 'confounding factors' problem is worse for univariate methods.

Off the top of my head, I'd think that you could do something with regional variation in religiosity in the U.S., so unobservable country factors weren't an issue, and you might even be able to dredge up some panel data to deal with regional unobservables.
So you aren't with those that have to sue everyone else to feel freedom from religion? Atheism is generally just another religion, as intolerant of other religions as they are of it, as certain of their correctness as they are of the incorrectness of their opponents.
I think it's useful for the FFRF to do its part to minimize governmental expressions of piety. I subscribe to Drek's view that everyone should be free (privately) to believe whatever s/he wants -- and so I sometimes find the FFRF to be a bit overboard on the anti-religious front. In fact, since I think atheism is generally 'free-thinking' along those lines, the characterization of it as a form of religion intolerant of other religions is all but self-deconstructing.
Tom Bozzo said... "In fact, since I think atheism is generally 'free-thinking' along those lines, the characterization of it as a form of religion intolerant of other religions is all but self-deconstructing."

Free thinking it may be, but it is quite the same as Christianity in that it believes everyone else is wrong. If your idea of 'free-thinking' is allowing other people to be wrong without preaching to them then I don't think the quality of 'free-thinking' can be applied to atheism, since atheism is an abstract concept. Atheists do share their worldview, whether it be to their students, their audience, their children, or the people who read their books. This is obvious. To be fair I ask you, do atheists attempt to rescue theists from their fallacy with the same vigour that Christians attempt to rescue atheists from hell? Perhaps it isn't fair for me to disagree with you when what you really described ("intolerant religion") was not Christianity in the form I have known. I think that atheists and those who accept the bible both love the idea of truth.

I think calling atheism a religion is fun, don't you?
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?