Friday, May 25, 2007

Christians and pornography

(Post 6 in a series)

In this series, I am going through various sexual behaviors and examining rates of Christians' participation in them versus those of people of other religions and no religions. Today's topic is pornography.

Before getting to the data, let me point readers to These guys do a *great* job in helping the Church deal with the issue of pornography. They also made an hilarious and interesting documentary about their getting going, called Missionary Positions. (Watch it here).

The General Social Survey asks respondents if they have watched X-rated movies in the previous year. Taking data from the last 20 years, here are rates by religion and attendance to religious services. I define active as attending on about a weekly basis.

37%, No religious affliation
33%, Non-active Black Protestants
33%, Non-Active Catholics
32%, Non-active other religions
28%, Non-active Evangelicals
26%, Jews
25%, Total sample
24%, Non-active Mainline Protestants
20%, Active Black Protestants
15%, Active Catholics
14%, Active in other religions
10%, Active Mainline Protestants
9%, Active Evangelicals

Sample size = 17,711


Corey said...

Brad... do you know how NORC deals with "social-desirability" bias? I would think that this sort of bias for these kinds of questions is not distributed randomly in the population. Someone who hits all of the indicators you are using to measure evangelical would likely experience strong cognitive dissonance in admitting they use/consume pornography. If that's true, then I would expect more lying on these items than others.

Has that been studied anywhere?

Brad Wright said...

No I don't know about NORC... certainly social desirability bias is a potential problem. I suppose it might vary by how far about the questions are and which comes first.

Nonetheless, I hear enough statistics about "Christians gone bad"--emphasizing the problems experienced by Christians, that I'm happy posting these analyses that portray Christians as less involved in some things than other people are.

It could be social desirability bias. Could reflect group differences.

In a way, "why" this difference exists is beyond what I'm trying to accomplish here. (Whether the why is methodological or substantive). I'm trying to get the basic social facts (or, at least, social statistics) straight.

Corey said...

(Whether the why is methodological or substantive). I'm trying to get the basic social facts (or, at least, social statistics) straight.

Yeah, but that's my point. You report GSS data indicating that 9% of active evangelicals have watched an X-rated film in the last year (the lowest % among the various categories). This is consistent with what you've been showing & arguing through the whole series.

However, to be 16 percentage points below the sample mean strikes me as a difference worth thinking about. If the xxxchurch people are correct, ("47% of Christians say porn is a major problem in their home") we have the problem of dueling statistics.

When it comes to the reporting of descriptive statistics, I follow Joel Best's advice and try to figure out where the number comes from. Statistics can be incredibly useful rhetorical devices for making arguments. People generally accept the statistic as a fact... as something that was picked off a tree and exists as some sort of independent object. But the reality is more complicated; these numbers are crafted (or constructed if we must use the constructivist terminology). There are known methodological niggles in getting people to accurately describe their private behavior in a self-report survey. It's one thing to report your marital history, after all much of that is public. It's quite another to admit (even to one's self) that they frequently watch pornography.

So my question is... can we really get the basic social facts straight without critical consideration of where these facts come from?

Also note, it's possible that the NORC people have come up with a way to minimize this sort of bias problem; or perhaps people are more willing to be honest with survey researchers than I give them credit for. [Having spent some time in the trenches administering & tabulating school drug surveys, you will have trouble convincing me on that score... but I'm open to discussion.]

Anyway, I'm not trying to pick a fight. I find what you're doing in this series interesting and hence worth commenting on.

sapience said...

It might also be worth noting that the type of pornography that is inquired about here might skew the results if we're interested in pornography as a whole. Viewing internet porn might have a much higher percentage across the board than viewing x-rated films. It's a heck of a lot easier (and more private) to get porn through the internet than to get x-rated films, especially for those who experience what Corey called cognitive dissonance.

Brad Wright said...

Hey Corey... great comments. You raise a lot of important issues here, and I think I'll post on them later this week.

As far as social desirability effect... absolutely it may be taking place, or it may not be. I'll look around the research and see what I can find. If it's in the GSS, which is about as good as social stats get, it would certainly be in other data sets, no? Taken to its extreme, it suggests that we really can't study people's sexual behavior with standard quant. measures, which may be the case.

As far as the social construction of stats, again I agree, and in a sense that's the subtext of what this series is about. There's been a constructed stereotype of Christians as no different in sexual behavior than others, even with behaviors deemed immoral by Christians. It's this construction that I'm addressing with this series, and it's why I'm focusing on comparing group means.

If stats can't be trusted, which may be the case, then we say we don't know if Christians are different.

If we say stats can be trusted, then we say Christians are different.

The reality is probably somewhere in between, but I don't see any empirical evidence for the Christians-same argument that Christian leaders sometimes put forth.

Corey said...

Brad, I don't want to hijack this discussion on your series. But, regarding "Taken to its extreme, it suggests that we really can't study people's sexual behavior with standard quant. measures, which may be the case."

There was a really interesting exchange in the New York Review of Books about ten years ago concerning Laumann et. al's National Health and Social Life Survey which was designed, in part, to systematically correct the biases that were produced in the Kinsey studies of human sexual behavior.

The reviewer, Richard C. Lewontin, argues that we should be suspicious of what people tell us in self-report surveys. This led to a rather interesting subsequent set of letters to the editor, including a rebuttal by Professor Laumann, a similar defense of survey research methods by Arthur Stinchecomb, and a critique of survey truths by Richard Sennett.

In all, the exchange is well worth reading. I assign this in my graduate qualitative methods seminar to examine the weaknesses of all empirically grounded truth claims. (And, believe it or not, I am an empiricist... just a cynical one).

Brad Wright said...

Great phrase... cynical empiricism. That certainly sounds like the way to go!

I just glanced at the exchange (I'm off to teach summer school), and it looks very informative. Looks like it belongs on any reading list for methods courses. Thanks!