I just saw a fascinating study on the subject,
Eisenberg et al., Who is the 40-Year-Old Virgin and Where Did He/She Come From? Data from the National Survey of Family Growth.
Unfortunately, the New Scientist blog post about the study erred in an important way: It reported that among men and women “aged 25 to 45,” “13.9 per cent of men and 8.9 per cent of women said they have never had sex.” But this is true only among the single men and women (the study explicitly mentioned this limitation, but the New Scientist item didn’t).
Furthermore, the study refers only to absence of heterosexual sex, and I take it that most homosexuals wouldn’t really be thought of as “virgins” juts because they hadn’t had heterosexual sex. Limiting the results to virginity among single non-homosexuals (i.e., people who say they are heterosexuals, bisexuals, or “something else”) age 25 to 45 would yield 11.1% virginity among men and 7.7% among women. And the study also appears to count only genital sex, and not, for instance, oral sex; I know there’s disagreement among people about whether someone who has had oral sex qualifies as a virgin. (See the description for the EVERSEX variable here; the definition is not as clear as to males.)
This having been said, the survey has some very interesting and statistically significant findings (all these are based on the multivariate regressions, which control for age, race, college degree, income, urban/rural residence, religion, weekly religious services attendance, alcohol consumption in the past year, marijuana in the past year, smoking in the past year, history of military service [men only], history of incarceration [men only], health status, body mass index, and sexual orientation):
- There seem to be no statistically significant correlations between virginity and age, so the results aren’t substantially skewed by 25-year-olds who just haven’t had sex yet — and, to the extent they are (and some such skew is inevitable), that would be offset by something else, presumably greater incidence of virginity 20 years ago, when the 45-year-olds were 25.
- Unsurprisingly, people who attend religious services weekly were considerably more likely to be virgins; the ratio was 5 for men, 4 for women.
- The incidence of virginity among men was lower for people with a history of military service (6 times lower) and incarceration (8 times lower). Recall that we’re talking here about absence of heterosexual sex, so either coercive or opportunistic homosexual sex in prison isn’t covered.
- The incidence of virginity was 4 times higher both for men and women who had not had alcohol in the past year, and that’s with a separate control for weekly religious attendance.
- The incidence of virginity was 3.5 times lower for black women than for white women, and 3 times higher for Asian women than for white women. (The latter number is the only one I quote that isn’t statistically significant at the 95% level, but it is very close — it’s statistically significant at the 94% level.)
- The incidence of virginity was over 5 times higher for women who had a college degree than for those who didn’t. And note that this is among single women, so the later marriage age among women with college degrees doesn’t seem to explain all of this.
- Health status, body mass index, income, urban/rural residence, religion (as opposed to frequency of religious attendance), and marijuana use in the past year were not statistically significantly correlated to virginity.
Steve2 says:
Professor Volokh, your link yielded me a “session cookie timed out” error when I tried to follow it. Thus, I have to ask rather than try looking for myself did they include depression/mental illness in the “health status” variable you mentioned in part 7? I’ve hypothesized based on anecdotal evidence that there’s a causal relationship between virginity and depression as well as between virginity and several personality disorders. I’d be interested to see if this study shed any light on that hypothesis.
Quote
November 9, 2009, 6:36 pmHouston Lawyer says:
When a friend of mine died intestate a few years back, they held a hearing to determine who his heirs were. A mutual friend was asked to testify regarding the deceased’s offspring. After testifying that he had no children out of wedlock, the judge asked whether the deceased had ever been enlisted in the military. The judge found it easier to rule as to the lack of his out of wedlock children knowing that he had never enlisted. Until that hearing, I hadn’t heard of that correlation.
I don’t know how you get a statistically valid sample regarding who is a virgin after age 25. I would suspect a large number of people of fudging their answers.
Quote
November 9, 2009, 6:39 pmEugene Volokh says:
Steve2: Health status is general “self-reported health status,” ranging from “fair” to “excellent.” There’s no separate variable for depression.
Quote
November 9, 2009, 6:42 pmCato The Elder says:
It is always good to remember that for these type of discussions that religious behavior, such as church attendance, is not the same as religious belief. Interestingly, the same people who will answer that it is “always wrong” to engage in premarital sex on a survey tend to be the same types who perpetrate the behavior more often.
Quote
November 9, 2009, 6:52 pmalkali says:
After testifying that he had no children out of wedlock, the judge asked whether the deceased had ever been enlisted in the military. The judge found it easier to rule as to the lack of his out of wedlock children knowing that he had never enlisted. Until that hearing, I hadn’t heard of that correlation.
I wonder if the judge assumed “military = morally upright” or “military = you’ve seen the training film about prophylactics.”
Quote
November 9, 2009, 6:53 pmCato The Elder says:
More virginity statistics here if you’re interested, Professor Volokh; the link highlights research showing that self-reported virginity significantly correlates with intended undergraduate major, high school grades, and measured IQ. For example, did you know that 35% of MIT graduate students are virgins?
Quote
November 9, 2009, 6:59 pmNunzio says:
Are the guys just too cheap to get a call girl?
Quote
November 9, 2009, 7:05 pmDjDiverDan says:
So attending religious services reduces one’s likelihood of having sex. Another excellent reason to vigorously resist the urge to attend religious services.
Quote
November 9, 2009, 7:23 pmAnatid says:
Similarly, those who are severely depressed or suffering from other untreated mental illness are often less likely to self-report mental illness than those suffering from milder illness. Self-report is poor measure of this variable.
Quote
November 9, 2009, 7:33 pmLaura(southernxyl) says:
Cato — you have something to base that on, or is it just a slam at people who profess to have morals? I’ve known lots of unmarried people who were active and who never seemed think they were supposed to say premarital sex was “always wrong”.
I read this as the judge, discovering out that the deceased had not been in the military, found it easier to believe that he had not sown wild oats.
BTW, this statistic makes DADT look kind of stupid, as if the military as it is is full of morally upright heterosexuals who don’t need to have their principles corrupted.
Quote
November 9, 2009, 7:36 pmShelbyC says:
That’s cuz it’s so much better when it’s wrong.
Quote
November 9, 2009, 7:39 pmEugene Volokh says:
DjDiverDan: Wait, why are you assuming the causation goes that way? What if virgins go to services to pray for sex?
Quote
November 9, 2009, 7:39 pmAriel says:
The most surprising result to me is that women are less likely to be virgins then men. I’m out of the market (happily married) but my impression was that men tended to have sex younger than women. I wonder if this is true at the younger end of the range.
Professor Volokh — as I’m sure you know, if a measure is not statistically significant at the threshold you’ve chosen, here 95%, it’s not correct to say it’s “almost” statistically significant. My clients used to call this sort of result “directionally accurate,” which drove us all nuts. Query whether a 95% CI is appropriate for this kind of data, as opposed to a 90% or 80%, which would have made it significant, but that’s a different story.
Quote
November 9, 2009, 7:46 pmJacob Berlove says:
What about people single at the time of the study, but previously married? If they are considered “single” for purposes of the study, then the sudy doesn’t mean too much.
Quote
November 9, 2009, 7:54 pme says:
Ariel,
Introverted girls probably get approached more often than introverted guys...
Quote
November 9, 2009, 7:55 pmEugene Volokh says:
Ariel: My assumption was that while women might be less eager than men to lose their virginity, they can more easily do so
I don’t see why it’s “not correct” to say that a figure significant at the 94% level is “almost significant” at the 95% level.
Quote
November 9, 2009, 7:57 pmTNeloms says:
That is really interesting. I wonder if you could disentangle these two factors (that individuals are necessarily less likely to be a virgin over time, and that the overall incidence of virginity is decreasing over time), and figure out if the latter is overtaking the former in any given age group.
Quote
November 9, 2009, 8:04 pmAriel says:
Professor Volokh,
To have something be statistically accurate, there are a few requirements that must take place beforehand: (1) decision on tests to be used; (2) decision on thresholds in deciding significance; other requirements. The idea is that by setting these beforehand, you’ve decided what is significant and what is not. You are not prejudiced by what your results end up being, so your decision has consequences. There’s a risk of error in all statistics, but if you draw the lines first, measure second, you’re not likely to be drawing the lines based on your results, introducing other errors. Therefore, the lines are meant to be statistically significant on this side, not so on that side.
I should note that many statistics are not reported in conformance with the proper way to do so. At least colloquially, it’s easy to say “almost” significant. But things are not “almost,” as a definitional matter. They either are, or are not.
That said, if you’re making business decisions based on this kind of data, I would be prepared to do so. Medical decisions — tougher call — and in my view, you cannot responsibly do so. For this kind of data, unless you’re using it for some unknown medical decision, there’s not much importance in calling it “almost” significant or not. It’s mostly for the times that it matters that I’m saying this.
Quote
November 9, 2009, 8:08 pmRandy says:
“So attending religious services reduces one’s likelihood of having sex.”
What’s interesting is that the ‘no sex before marriage’ has been drummed into people’s heads so much that they actually believe it. And so we see lots of people getting married at age 19 or 20 just so that they can finally have sex.
Which isn’t exactly a sound basis for marriage, and so you have lots of people getting divorces several years later. But of course, if you deny gays the right to marry, that will correct the situation!
Quote
November 9, 2009, 8:12 pmCato The Elder says:
No slam, Laura(southernxyl). My assertion comes from the GSS; you can read the relevant tables in the blogger Half Sigma’s post here. The GSS survey asked its participants the following questions (amongst many others):
The blogger plotted both variables against WORDSUM, a 10 question test which can be capably used as a proxy of IQ score. You can see that the percentage answering PREMARSX 217 as “Always Wrong” monotonically decreases with WORDSUM scores, and that the percentage answering PARTNERS 1291 as “0 Partners within the last year” stays fairly level with increasing WORDSUM score until about 6, at which point it sharply increases. So, the survey shows that the percent giving the first answer is negatively correlated to the percent giving the second — the claim.
Quote
November 9, 2009, 8:33 pmLaura(southernxyl) says:
But PARTNERS 1291 is unmarried people. It is not clear to me that PREMARSX 217 is unmarried people. If not, then these are not the same group. It would be interesting, if PREMARSX 217 is married, or married and unmarried, to ask how many refrained from sex before marriage.
Quote
November 9, 2009, 8:44 pmJay says:
“But things are not “almost,” as a definitional matter. They either are, or are not.”
To paraphrase our President, we need not make a false choice between the two. Obviously something can be both “is not” and “almost is,” even “as a definitional matter.” That is, someone who is 17 years and 11 months old is not 18, but is almost 18.
Quote
November 9, 2009, 9:06 pmAriel says:
Jay — haha. Not a strategy I would try before a judge, but that’s just me. :P
Quote
November 9, 2009, 9:08 pmAnonymous says:
Makes sense to me. At the risk of subjecting myself to internet ridicule, I am a 26 year old female virgin, and no fewer than four of my closest female friends are in the same boat. Four of us are at least somewhat religious, but we come from varied religious backgrounds (two Catholics, one Mormon, and one Jew), and except for one of us, religion has not been the primary rationale for abstaining from sex. What we all have in common is that (1) we are educated enough to be aware of the dangers of casual sex, (2) we are smart enough to resist peer pressure and delay gratification in favor of long-term goals, and (3) we have no interest in having sex with someone to whom we are not romantically attached. We are all highly educated women who now have demanding jobs, and for whatever reason, none of us have met “Mr. Right” yet. Thus, the study’s suggestion that single women in their mid-20s are much more likely to be virgins if they are college-educated is entirely unsurprising to me.
Quote
November 9, 2009, 9:13 pmCornellian says:
I find it surprising that there is no correlation with body mass index. You would think grossly obese people would be less likely to find someone willing to have sex with them, but maybe the people in the study didn’t reach that extreme.
Quote
November 9, 2009, 9:38 pmRicardo says:
Studies based on self-reporting can be problematic. For instance, it has long been observed that when Americans are asked how many sexual partners they have had in their lifetime, the reported number for men is always significantly higher than the number for women. Now, it is possible that lots of American men are having lots of sexual experiences in other countries or with women who are otherwise not represented in the survey, but I doubt that can fully account for the discrepancy. Instead, it seems men are over-reporting or women are under-reporting. Whether virgins lie about whether they have had sex or not isn’t clear but seems likely.
Quote
November 9, 2009, 9:43 pmyankee says:
It doesn’t surprise me. Anyone can get laid unless they’re pathologically shy or insist on sleeping with people who are considerably more attractive than they are.
Quote
November 9, 2009, 10:03 pmTweets that mention The Volokh Conspiracy » Blog Archive » Virginity Among Unmarried Heterosexual 25-to-45-Year-Old Americans -- Topsy.com says:
[...] This post was mentioned on Twitter by Josh Barro, Parley Peck. Parley Peck said: The Volokh Conspiracy » Blog Archive » Virginity Among Unmarried ... http://bit.ly/FunXb [...]
Randy says:
yankee: “Anyone can get laid unless they’re pathologically shy or insist on sleeping with people who are considerably more attractive than they are.”
Uh-oh.
Quote
November 9, 2009, 11:09 pmRicardo says:
Interesting but it’s not clear what this shows. Low IQ is generally correlated with low impulse control — so people may think pre-marital sex is always wrong but just cannot resist the temptation. Just like if you ask a low IQ criminal whether stealing is always wrong, he may well answer yes before knocking off a convenience store.
It’s not clear their views on sex play a causal role in their behavior.
Quote
November 9, 2009, 11:16 pmLeo Marvin says:
FWIW (very little I realize), the only layperson I know over 30 who claims/admits to being a virgin is a regular churchgoer. I have no idea how he’d answer the “always wrong” survey question, but I’ve never heard him express a negative judgment about anyone else’s private behavior.
Quote
November 9, 2009, 11:19 pmBrad says:
Are the people that SAY they are virgins the same people who ACTUALLY ARE virgins? Someone might be a virgin, but say they are not; people who are not virgins may say they are.
Quote
November 9, 2009, 11:29 pmRadford Neal says:
To have something be statistically accurate, there are a few requirements that must take place beforehand: (1) decision on tests to be used; (2) decision on thresholds in deciding significance; other requirements. The idea is that by setting these beforehand, you’ve decided what is significant and what is not. You are not prejudiced by what your results end up being, so your decision has consequences. There’s a risk of error in all statistics, but if you draw the lines first, measure second, you’re not likely to be drawing the lines based on your results, introducing other errors.
I think this is mis-guided. The purpose of writing up the results of a study is to communicate useful information to the reader. Reporting only whether the result is “significant” or not does not communicate as much information as reporting the actual p-value. (By the way, p-values (eg, 0.05), not confidence levels (eg, 95%), are the normal way of reporting the results of signficance tests.) It makes a big difference whether the p-value is 0.012, 0.049, or 0.09. It makes very little difference whether the p-value is 0.049 or 0.051, regardless of whether or not you had decided that 0.05 was your threshold of significance. To think otherwise is to mistake research for some sort of game.
Quote
November 10, 2009, 1:19 amptt says:
What calculation would yield those results?
I’m a gay man, if you asked me if I was a virgin, I’d say no. If a form asked me if I’d ever had “sexual intercourse”, I’d say no.
Quote
November 10, 2009, 1:23 amyankee says:
No, a greater incidence of virginity among the unmarried 20 years ago. There’s a long history of sex causing marriage: e.g. the woman gets pregnant and social pressure (or shotgun) induce her and the man to marry. As nonmarital sex and single motherhood have become more accepted and effective birth control has become more widely available, having premarital sex is less likely to induce a couple to marry, resulting in a lower rate of virginity among the unmarried. Of course, this is not necessarily the entire explanaion; it may not even be a substantial portion of it. However, a lower rate of virginity among the unmarried is not necessarily caused by a lower rate of virginity.
Quote
November 10, 2009, 2:13 amElmer says:
Hyposcoria: the hidden epidemic.
Quote
November 10, 2009, 2:32 amzuch says:
Cato the Elder:
No surprise there. When I was there, it was 90% male to 10% female....
Cheers,
Quote
November 10, 2009, 2:40 amNickM says:
That’s what grossly obese people of the opposite sex are for. OK, that’s what large amounts of alcohol are for too.
Nick
Quote
November 10, 2009, 2:44 amMemo to conservative muslims in America: send your daughters to college « Entitled to an Opinion says:
[...] a Comment Don’t buy the Roissysphere’s B.S, unmarried women aged 25 to 45 are five times more likely to be virgins if they have a college degree than otherwise (yeah, yeah, correlation isn’t [...]
Ricardo says:
Very true. Deirdre McCloskey has some good articles attacking the cult of statistical significance. Even if you take statistical significance very seriously, a p-value of 0.051 often tells you “get a bigger sample size,” not “nothing to see here folks, move along.”
Quote
November 10, 2009, 4:21 amEric S. says:
Re finding No. 3 and the term “coercive homosexual sex”: did some state recently pass a constitutional amendment defining rape as between a man and a woman?
Quote
November 10, 2009, 8:30 amMartinned says:
Indeed. I’ve never understood the preference for reporting significance at some stated level of confidence instead of simply reporting the p value and letting the reader make up his own mind.
Quote
November 10, 2009, 8:34 amMartinned says:
I noticed that one, too. I guess it’s a matter of doublethink: if one were to admit how rampant rape apparently is in American prisons, one would have to do something about it.
Quote
November 10, 2009, 8:35 amPintler says:
Great phrase, thanks!
Ariel is right in one sense — carefully phrased, significance at P=0.XX
means that, given the sample size and variation in the data, the probability the hypothesis is true is 0.XX. That statement is strictly true only if you pick your analysis in advance. In particular, you can’t collect the data and then analyze it 47 different ways until you squeeze a P=0.95 result out somewhere and then go ‘Eureka!’. If you analyze it 20 ways, you’d expect one false conclusion at P=0.95. This is particularly abused with multivariate regression, where people start adding and removing variables until they get a ‘significant’ result from the fourth order interaction of shoe size, blood type, astrological sign and hair color.
The other commenters are correct, though, that 0.94 and 0.96 are effectively the same number, and P=0.75 still means three chances out of four. 0.95 isn’t some magical true/false switch. One of the most brilliant analyses I have seen strung together a sequence of inferences at the P=0.7 to 0.8 level, each of which made good biological sense; that’s a lot more compelling than a P=0.95 result that doesn’t make sense.
Quote
November 10, 2009, 8:43 amSnaphappy says:
So, if you want to have sex:
1. Stay out of church.
2. Drink.
3. If you are a man, avoid college women. If you are a college woman, drop out. (This is nonintuitive)
4. Try not to be asian. If at all possible, be african american.
5. Hitting the gym, moving to or from the city to or from the country, increasing your income, and changing your religion won’t help.
6. Smoking weed won’t help either. (Nearly always true)
7. If all else fails, join the military.
8. If that doesn’t work, get yourself put in jail.
Quote
November 10, 2009, 10:03 amDan Weber says:
I’ve got an ad on the side telling me I can find my Filipina sweetheart today. Is VC trying to tell me something?
Quote
November 10, 2009, 11:19 amMartinned says:
I’ve been getting some weird French dieting site lately, so it looks like there is some system at work here.
Quote
November 10, 2009, 11:46 amCCTrojan says:
Do you have any data to show that divorce rates are higher among people who did not have sex before marriage? If so, please share.
I think everyone here understands how you feel about the application of religious/moral values to the realm of sexuality. There’s no need to harp on the topic on every single thread.
Quote
November 10, 2009, 12:26 pmEugene Volokh says:
Ariel: I guess I don’t find the argument you give about statistical significance to be persuasive. As I understand the argument, if the study says it’s looking for either significance at the 90% level or the 95% level, it presumably then can point to the 90%-significant (or 94%-significant results) as significant at that level — but if the study began with a search for 95%-significant results, it then can’t refer to the 94% results as significant. That strikes me as counterintuitive, and requiring more justification than I’ve seen. It seems to me that 94% significance can be sensibly reported as almost 95% significance, whether or not the study designers had initially said they were looking for 90% significance or only for 95% significance.
Quote
November 10, 2009, 12:44 pmClayton E. Cramer says:
I find it difficult (although not impossible) to imagine that people who waited for marriage to have sex because of religious reasons are going to divorce because the sex wasn’t good. I mean, they let their religious beliefs discourage them from having sex before marriage (which Christianity strongly discourages) but not discourage them from getting divorced because the sex wasn’t good enough? (Which Christianity far more strongly discourages.)
I’ve seen a lot of marriages break up, but usually it is because there are very deep problems based on selfishness. I’m sure that bad sex is a symptom, but since Christianity teaches that selfishness is a sin, I’m guessing that the strongest correlation on divorce isn’t, “We’ve got to wait until marriage because Jesus demands it,” but “I deserve a new BMW every three years! What’s wrong with my spouse not making enough money to pay for it?”
Yes there is, because the alternative is introspection.
Quote
November 10, 2009, 12:47 pmuberVU - social comments says:
Social comments and analytics for this post...
This post was mentioned on Twitter by jbarro: Who are the adult virgins? Eugene Volokh examines the data. http://is.gd/4RizZ...
Martinned says:
How about the hypothesis that a taboo on premarital sex causes people to marry sooner than they otherwise would, and therefore not necessarily to the same person than they otherwise would, leading to poorer average “fit” between spouses and higher divorce. That seems like a more reasonable explanation of what Randy had in mind.
At least the tendency for “blue states”, where premarital sex is presumably more common, to have lower divorce rates is well established.
Quote
November 10, 2009, 1:50 pmLaura(southernxyl) says:
Martinned, the divorce rate on your table is per population, not per marriage. If fewer people in those areas are bothering to get married, then there will be fewer divorces.
Quote
November 10, 2009, 2:42 pmRadford Neal says:
Ariel is right in one sense — carefully phrased, significance at P=0.XX means that, given the sample size and variation in the data, the probability the hypothesis is true is 0.XX.
This is completely incorrect. P-values are not the probability that the null hypothesis is true. The p-value is the probability of getting evidence as strong or stronger than what was actually observed, if the null hypothesis is true. This is not even remotely the same thing.
That statement is strictly true only if you pick your analysis in advance. In particular, you can’t collect the data and then analyze it 47 different ways until you squeeze a P=0.95 result out somewhere and then go ‘Eureka!’.
First, it is small p-values that lead you to rejecting the null hypothesis, not big ones. What is it with this thread? Most of the commenters have it backwards (though some may just be using an opposite convention, rather than being truely confused).
You’re right, however, that you can’t do numerous tests and then declare that finding one that gives a “significant” result justifies announcing a discovery.
The other commenters are correct, though, that 0.94 and 0.96 are effectively the same number, and P=0.75 still means three chances out of four.
A p-value of 0.75 (or of 0.25, which seems to be what you really meant to say) is not considered to be of any significance at all by anyone who understands what p-values are.
One of the most brilliant analyses I have seen strung together a sequence of inferences at the P=0.7 to 0.8 level, each of which made good biological sense; that’s a lot more compelling than a P=0.95 result that doesn’t make sense.
You’re right that a “significant” result that doesn’t make sense is probably wrong. That’s one reason why a p-value can’t be interpreted as the probability that the null hypothesis is true. However, the “brilliant” analysis you refer to was probably total bunk. Even one inference based on a p-value of 0.8 (really 0.2, I assume) is highly dubious. A string of such would have essentially no basis in the data. The conclusion might have been true, but the study provided no evidence for it.
Quote
November 10, 2009, 3:25 pmDjDiverDan says:
Someone needs to tell them that God helps those that help themselves.
Quote
November 10, 2009, 4:31 pmNorthern Dave says:
Number 6 makes sense to me in that women tend to want to couple up the social ladder.
Women also tend to have sex for relational purposes (not all do, but the ones who are just after the physical have already been with the football team and don’t enter this discussion.....).
Several women I have known who have great careers have never even dated to my knowledge because they simply weren’t physically attractive enough to catch the interest of the men in their father’s social circles (who being rich, powerful and tall have a *very* large selection of perky, petite, well-educated, groomed and mannered women competing over them).
Quote
November 10, 2009, 5:47 pmDr. Caligari says:
So a never married 25/45 year old male who attends religious services weekly is 5 times more likely to be a virgin than a never married 25/45 year old male who does not.
Interesting.
Now does anyone know how much more likely is a never married 25/45 year old man who attends at least one Star Trek convention a year to be a virgin than a never married 25/45 year old man who does not? And while body mass index may not affect the likelihood of virginity, what about the size of comic book collection?
Quote
November 10, 2009, 6:09 pmDan Weber says:
If the size of your comic book collection is impeding you getting a mate, you should work hard to make it bigger.
Quote
November 11, 2009, 11:51 amRandy says:
Trojan: “Do you have any data to show that divorce rates are higher among people who did not have sex before marriage? If so, please share.”
Gladly! From Andrew Sullivan’s blog, a university professor who wrote in: “I teach at a large university in a conservative part of the country, and I think a large part of this fear of children learning that — gasp — people can be attracted to the same sex, has to do with the religious right’s emphasis on marriage as a primarily sexual institution. They would not agree with that, of course, but look at how they teach sexuality education to their children: “Abstinence til marriage. Nothing else need be said.” (Thus sending the message that sex and marriage are yoked at the hip.)
The conservative youth group “Young Life” is very active where we live (high schools and college), and I cannot tell you how many young people (18–21) I know who have gotten married because they simply cannot hold out sexually any longer. They get married in order to have sex. They don’t get married because they love the person; they may be deeply in love, but that’s not why they’re getting married at that particular time — they’re getting married before they finish college, before they have decent-paying jobs, before they have health insurance, because they are afraid they won’t be able to control their sexual urges any longer.
Sex is intrinsically linked to marriage. Sex=marriage=sex.
I also have known quite a number of people in their late 20s and early 30s (students) who are now divorced (or unhappily married), who tell me that that is exactly why they got married 10 or whatever years ago — and they are now stuck with two or three kids, trapped in a marriage they recognize was entered based on an immature idea of what “love” is and pressure from their families and conservative churches to, no matter what, NOT have sex before marriage.
This is not to say that many of those young couples can’t make their marriages work, but to point out that the pressure — and the conflation of sex with marriage — is intense.
It is no wonder, then, that for a religious conservative, the thought of their first-grader coming home asking questions about their female teacher’s wife, or reading about Heather and her Two Mommies, scares the bejesus out of them. Marriage is all about sex, and sex is already an extremely scary topic (see again their resistance to any kind of true sex education), and Good God ...! It’s downright terrifying for them.”
If you have a problem with this, you may wish to take it up with Sullivan or the reader who submitted this.
“I think everyone here understands how you feel about the application of religious/moral values to the realm of sexuality. There’s no need to harp on the topic on every single thread.”
Since so many people conflate religious and moral values to sexuality, I think it’s necessary to point to out that sometimes it has an ill effect on sexuality and marriage.
Quote
November 11, 2009, 5:40 pmAriel says:
I agree that it makes very little difference if your result is 0.049 or 0.051. You can report the result of your test at 0.051 and many people will (rightly) conclude that if your sample size was larger, it would probably have been significant. I also agree that the actual p-value is more meaningful than only whether it was significant or not. But that doesn’t mean that reporting the p-value alone makes something “almost” statistically significant.
The whole point of statistics is to say this is statistically significant, that isn’t. It’s pretty elementary probability theory that if you start calling things “almost” statistically significant, you’re going to end up blurring the concept of whether things are significant or not. You will consider many things to be kind of, sort of significant, that you’ve previously determined are outside of your thresholds. That’s why, even in the kind of loosy-goosy statistics that business clients sometimes used (lower CIs), we always decided our CI with the client first — it prevents you from adjusting your CIs afterwards to make things that you think ought to be significant, significant, merely because they were “almost” significant.
When I did biomedical work, you can bet that we decided our CIs before we so much as looked at our experiments.
It may be counterintuitive, but it’s true. The biggest danger you’re trying to prevent, as the one conducting the study, is introducing your own bias in trying to find a result. If you want something to be true, you’ll call it “almost” true. And, frankly, you’re probably right — which is part of why so many people do it.
But it’s not correct to do so. If you set your CI at 95%, what you’re saying is that you’re 95% confident that the result falls within the range: lower bound <= x <= upper bound. If a result is true at a 94% confidence interval, you’ve confirmed the null hypothesis, that you cannot tell if the confidence interval overlaps with whatever you’re comparing it to, at a 95% confidence interval. By definition, you’ve said: I can’t tell if x is different than y or not. That’s why you have to (or should) decide your confidence intervals first — if you don’t, it will be tempting to decide that x is different than y, when in fact, you have not proven that to be true.
In basic addition, we know that 1+1=2. In statistics, we might think that there are about 12 apples in a covered basket, based on knowing it has apples and the shape and size. Let’s say the 95% confidence interval is 9–15 (I haven’t attempted to estimate this), and you find only 8 apples when you do the count. Now, it’s true that you were almost right. But you’ve disproven the null hypothesis, because your actual result was outside of the confidence interval. This is the same kind of reasoning that would lead one to say “almost” significant in the context you’re saying, and it’s the same reason why that’s not true, just simplified.
Quote
November 11, 2009, 8:14 pmCCTrojan says:
Um, none of that is “data.” That is anecdotal evidence (which is probably skewed because of the type of people someone like you would choose to associate with). Besides, what is “quite a number of people,” two or three? I have just as much anecdotal evidence that marriages are stronger when the parties are the type of people with enough self-control to live their beliefs. That would be because I am part of a social group where that is the norm. My experiences, like yours, are not data.
Divorce rates are lower in blue states because people are more likely to live together without getting married. And divorce rates are higher when people are married very, very young, no matter whether they had sex first. But you have ZERO evidence that people who choose to abstain from sexual relations until marriage have a higher divorce rate. That’s because there is none.
Someday you’ll figure out that your opinions are not factual or even based on any relevant data. They’re just the product of your imagination of what it must be like to live with some sort of religious values. It’s not the hell you seem to imagine, lol.
Quote
November 12, 2009, 12:19 pmohwilleke says:
Does the consistent data set say what percentage of the age group are married? The median age of marriage has been creeping up to the mid-20s, and ultimately about 95% get married (probably by age 45 in most cases), but the number matters.
Married rate data for the subgroups matter too. For example, the African American marriage rate is very low comparerd to the population as a whole. The sample would have proportionately many 40 year African Americans who were single, and proportionately few 40 year old Asians who were single. If the number of 40 year old virgins per 1000 people were the same in both samples, the Asian single virginity rate would be much higher than the African American virginity rate. Marriage also probably completely explains the lack of age correlation.
It also isn’t clear from this summary if “single” includes “divorced and not remarried” or “widowed and not remarried” something it wouldn’t in a typical personals ad, but might in this study.
It would also be interesting to see the virgin single, non-virgin single and married groups profiled comprehensively rather than simply predicted with probabilities. Taken together, just how distinctive demographically are single virgins, or single non-virgins.
Quote
November 12, 2009, 4:34 pmRandy says:
Trojan: “But you have ZERO evidence that people who choose to abstain from sexual relations until marriage have a higher divorce rate.”
You are correct. But I never said that, so I don’t know what your point is.
“Someday you’ll figure out that your opinions are not factual or even based on any relevant data. They’re just the product of your imagination of what it must be like to live with some sort of religious values.”
I merely cited someone who reports what they see. Perhaps the person is lying, but I have no reason to believe so. I report, you decide. If you have a beef with anyone’s imagination, it would be with the university prof who wrote the comment.
Additionally, Dan Savage is a nationally syndicated columnist who has made a similar point, that he gets letters quite often from people who ‘saved’ themselves for marriage, and 5 years in, 10 years in, or whatever, they are desperately looking for help. Some want to cheat and still remain true to their vows (which is rather difficult), some want a divorce but have kids, and so on. So no, it’s not based on my imagination, but rather upon at least one person who is in a position to know.
One would think that if we should have a policy that discourages divorce, we should look at the reasons why people divorce, and then try to take a proactive approach. The more information the better, in my view. You can interpret that as being anti-religious, but I don’t see it as such. I just want to be practical and find solutions that keep everyone happy. If that means certain religious values are causing a problem, then that needs to be examined. I hardly see why that would be a problem, but to you apparently it is.
I don’t have any answers, but the fact remains that this is an issue for some people. Ignoring it doesn’t really do anyone a favor, does it?
Quote
November 12, 2009, 5:50 pmPintler says:
If I might be permitted a hypothetical: I believe you said your statistical experience is in the biomedical field. Let’s say your company is searching for a drug by synthesizing zillions of compounds and giving them to rats. The number of rats that survive is compared to the number that survived in the control group. The researchers give you the data, you crunch away, and then you present your results. In many of the groups more rats died than in the control, and in many groups almost as many died as in the control. In some groups, though, very few rats died. In some of those groups, so few rats died that the results are statistically significant at your preselected significance level, and you inform the researchers that those compounds seem efficacious.
A few groups, though, didn’t quite reach significance at your preselected level. Maybe if just one more rat had lived, they would have. What do you report about those? Are they just lumped in as dead ends with the groups where more rats died than in the control group, or do you report them as possibly being worth further study, maybe by varying the compound’s structure slightly. If so, what terminology do you use to refer to those, well, ‘almost significant’ results?
Quote
November 12, 2009, 6:31 pmRadford Neal says:
The whole point of statistics is to say this is statistically significant, that isn’t. It’s pretty elementary probability theory that if you start calling things “almost” statistically significant, you’re going to end up blurring the concept of whether things are significant or not.
Indeed you will. That’s because whether the data demonstrate that the null hypothesis is false, or don’t demonstrate this, is a blurry concept. Pretending instead that results are either “significant” or “not significant” does not change the fact that experiments actually provide varying amounts of evidence, ranging from “no evidence at all” to “evidence that’s as conclusive as anything you’re ever going to see”.
So, you see, the “whole point” of statistics is not at all to say “this is significant” and “that isn’t”. The point is to provide an accurate and understandable indication of what the experimental data has shown.
Quote
November 13, 2009, 1:37 ammarkm says:
“So attending religious services reduces one’s likelihood of having sex. Another excellent reason to vigorously resist the urge to attend religious services.”
That’s because they’re going to the wrong religious services. I lost my virginity to a girl at met at the Unitarian-Universalist Fellowship.
“Do you have any data to show that divorce rates are higher among people who did not have sex before marriage?”
Just anecdotal. The only man I knew who bragged about being a virgin until married eventually became the first divorced Baptist minister I knew...
Q. “why are you assuming the causation goes that way? What if virgins go to services to pray for sex?”
A. “Someone needs to tell them that God helps those that help themselves.”
My understanding is that in most churches, such self-help is frowned upon.
Quote
November 14, 2009, 8:53 pmnaina says:
Quote
November 20, 2009, 7:31 am