Wednesday, March 28, 2012

Promiscuity Data: Guest Post.

A while after putting up my first post of subject of promiscuity and divorce risk, I was contacted by an individual who wished to analyse the data further. That individual, whom I shall call INTREPID, was able to analyse several cycles of data from the National Survey of Family Growth with regard to the promiscuity effect on the risk of divorce.

As far as I'm aware the data that has been obtained is not available anywhere in the published literature. I'm not a statistician and cannot vouch for the veracity of the data but I believe the analysis was done truthfully and without bias. 

I want to publicly thank INTREPID for performing the task. I've had access to the findings for a while and have decided not to post them till now so that no trace could be linked to INTREPID. (The other reason was that I lost access to the account which contained the data and only recently was able to gain access to it) 

INTREPID provided a report on his findings which I reproduce in full below.

**********************************************************************************

Following up on your post regarding the Heritage Foundation Study, I thought to explore the same
National Survey of Family Growth (NSFG) data and see if there’s more to the story. I think there is. Like the Heritage Foundation, I find a negative relationship between non‐marital sex and marital stability, but the relationship is more complex than the Heritage Study leads one to think. In particular, the effect of the number of pre‐marital sex partners on marital stability isn’t linear. Rather, it’s a U‐shaped relationship and invites some further exploration.

For starters, the chart below simply reproduces a finding from the Heritage Foundation study. I use the same axis definitions (as best I can tell) that the Heritage Foundation used (for the 1995 NSFG data). But I now add to it two further waves of the NSFG data (2002 and 2006/2008). The pattern is the same across study waves. An increase in the number of non‐marital lifetime sex partners is related to less marital stability. 



I should point out that my data display for 1995 is slightly different from that reported by the Heritage Foundation. I suspect the difference arises in how “sexually active” is defined. I defined it as ever having had sex. It’s not clear how the Heritage Foundation Study defined it. However, I’ve got some gripes with the above axis and definitions used by the Heritage Foundation. For example, the vertical axis showing “stable” marriages includes some marriages that I wouldn’t consider to be “stable.” Also, the horizontal axis includes some sexual relationships that, in my opinion, are unlikely to be clear  causes of marital instability but might instead be the effect of instability. Specifically…,

• Per the Heritage Foundation’s definition of “stable marriages”, second or third marriages that go on to last five or more years (at the time of the NSFG interview) would be considered “stable”. That is,  the Y‐axis could include some women who have divorced at some point and are poor representatives of “marital stability.”

• It’s likely not a big issue (nor is likely to correlate with non‐marital partner count), but marriages that ended with the death of a husband would fail to count towards the Heritage Foundation’s definition of a “stable marriage”.

• The causal direction isn’t clear. Women who have left a marriage (leaving the pool of those in “stable” marriage) may have increased their non‐marital partner count after leaving the marriage, not before the marriage or during it. A knee‐jerk interpretation of the chart, however, suggests that a high number of non‐marital partners leads to marital instability. Yet it may be that a rise in the number non‐marital lifetime partners follows marital instability. Once divorced, women might sleep around more.

• By looking at the count of non‐marital lifetime sexual partners, we count the sexual episodes of women who perhaps got married as virgin brides, then divorced and subsequently slept around before entering into a “stable” marriage (i.e., one lasting five or more years). From the viewpoint of one desiring to minimize his chances of ever experiencing divorce, looking at the number of non‐marital lifetime partners isn’t perfectly helpful. One ought to instead look at the number of pre‐marital sex partners and the rate of first marriages ending in divorce or separation. This is what Jay Teachman looked at in his paper: first marriages that ended in divorce or separation (marital dissolution)

So, for a more careful look at the effect of extra‐marital partner on marriage success, I create Chart 2 that focuses on divorce outcome (vertical axis) rather than the broader concept of a “marital stability” Specifically, Chart 2 below looks at the effect of non‐marital lifetime partner count (same horizontal axis as above) on the rate of first marriages ending in divorce at the time of the NSFG interview. In other words, and compared with Chart 1, the below excludes those women who may have been  divorced and then re‐entered back into a “stable” marriage by the time of their NSFG survey interview. The pattern looks very similar to that in Chart 1 (albeit reflected since the measure is marital divorce rather than marital stability). 
 

For an even further revelation, I create Chart 3 which improves upon the horizontal axis as well as the vertical one. The vertical axis is the same as in Chart 2 and looks at the rate of first marriages ending in divorce (same outcome measure that Teachman looked at). But the horizontal axis now groups women by their count of pre‐marital sexual partners as opposed to lifetime non‐marital sex partners:

 


This tells a different story from that in Chart 1. In my opinion, Chart 3 is more helpful than the  Heritage Foundation chart as it permits one to come closer to drawing a causal connection between partner counts marital outcomes. The most interesting element in Chart 3 is the U‐shaped relationship between pre‐marital partner count and divorce rates. This surprised me so much that I had to chart the 2002 NSFG data alongside to be  certain that the 2006/08 data wasn’t just a fluke, a sampling artifact. This chart most clearly shows that divorce rates are the lowest for those with zero pre‐marital partners. That part is unambiguous. Virgin brides are the least likely to divorce, all else equal. But why do  divorce rates rise with one or two premarital partners only to reverse and drop after two or more sexual partners? And next, why do divorce rates reverse and climb back up after the pre‐marital partner count goes into the double digits?

This is a very interesting finding that neither the Heritage Foundation Study nor Jay Teachman’s hazard model analysis examine. I’m not even sure they’re aware the relationship exists. I’m anxious to hear reader opinions on the drivers here. My own hypothesis is that a higher partner count (up to 5‐9 or so partners) is correlated with age and maturity in dating experience. Older women, and women with more dating experience, are more likely to have learned which personal qualities will work best for them in a marriage partner. As a result, such women choose more wisely and tend to experience lower divorce rates. Now, it also happens to be the case that older women have had more time and occasion for pre‐marital sex! Specifically, I suspect it’s not the 5‐9 pre‐marital sex partner count per se that drives the relative drop in the divorce rates, but rather it’s the maturity and experience that women have acquired while they’ve dated more men.

It’s too bad the NSFG doesn’t have a variable for something along the lines of “pre‐marital dating experience” or something similar to tease this out. I don’t believe that it does. The NSFG does, however, collect age and education data which I’ll look at in a future analysis. Indeed, my own quick look (not charted here) reveals that the age is positively correlated with partner count, offering some support of a dating maturity hypothesis. But it warrants some additional charts and analysis. …But this idea of partner count as a proxy for dating maturity doesn’t hold forever. On the right‐hand side of Chart 3, I suspect that some women with a high double‐digit pre‐marital partner count begin to make poor marriage candidates due to yet other personality traits. I suspect that some women who have had 21+ marital partners may also have a high need for variety‐seeking across mates such that marriage or settling down with one man might be considered boring. A resulting marriage, if it happens, is less likely to last.

To further understand partner count on divorce outcome, I need to show yet one more data result. The question I have is: does the divorce rate for women change if her pre‐marital sex partner becomes, in fact, her later husband? If one sleeps with one’s (otherwise virgin) female fiancĂ© prior to the wedding, what’s the effect?

The table below contains the answer. I build this from the 2006/08 data (the 2002 data show a similar result). One can see that the divorce rate is nearly 50% for women who had only one pre‐marital partner and if that partner did not develop into her husband. The divorce rate falls to half the above rate (25.6%) if a woman later marries her first and only pre‐marital sex partner. However, both these divorce rates are higher than the divorce rate for virgin brides. Pre‐marital sexual experience with one’s future spouse does not beat out having no pre‐marital sex at all.

 

Distilling all the above, the practical advice for a single woman seeking to lower the risk of her first marriage ending in divorce is to remain a virgin bride. If she sleeps with even one person, her risk of divorce increases, at the very least, by roughly 10% (from 14.9% to a value of 25.6%). …And it could easily go higher. Indeed, her overall divorce risk could jump to 50%—which is higher than any value shown in Chart 3—if she doesn’t marry her first partner (and assuming she doesn’t go on to sleep with other pre‐marital partners). And moving on to sleeping with two or more pre‐marital partners doesn’t ameliorate her likely divorce rate over that of her having zero pre‐marital partners. The best option for a women seeking to lower her divorce risk is to remain a virgin bride (all else equal).

The practical advice from the man’s perspective is only a little less clear. A man never truly knows if  his dating partner is a virgin or not unless he sleeps with her—at which point he would only know that she’s not a virgin. (Sleeping with someone to determine their virginity is a destructive test!) But by definition, a man’s own risk of divorce is connected to his wife’s risk of divorce. Therefore, if he  suspects the woman he’s dating is a virgin, his risk of divorce (just like his wife’s) is lower if he  doesn’t have premarital sex with her and she remains a virgin bride. The question is what to do if he strongly suspects (or if she admits) that he wouldn’t be her first pre‐marital sexual partner. If she’s had just one pre‐marital sexual partner, then Chart 3 suggests that her risk of divorcing him is lower if he becomes her second pre‐marital partner (a roughly 40% divorce risk) than if she has sex with just one pre‐marital partner who isn’t her eventual husband (a 50% divorce risk as found from the Table). One could perhaps extend this logic to the case of a man considering a potential wife who’s had two, three, or even more pre‐marital sexual partners. However, the NSFG doesn’t collect the data needed to  extend the Table above and confirm. Besides, this train of thought related to a male’s pre‐marital sex  strategy sounds like an unfolding “tragedy of the commons” type problem. (The strategy that might be  optimum from a personal standpoint is sub‐optimal once everyone tries to pursue it. ) …In addition, we don’t yet know the effect of the husband’s pre‐marital sexual history on divorce outcomes. That  could have a bearing. These male data on divorce and pre‐marital sex are in the 2002 and 2006/08 NSFG and I’ll be looking at those data in the future.

********************************************************************************

Some comments:

The first thing that strikes me about this report is the confirmation that no premarital sexual experience powerfully and significantly lowers the risk of subsequent divorce. Whether or not you agree with the sexual revolution, I don't care, rather, it's that repeated analysis of the data confirms what common wisdom has always asserted; marrying a girl who has not slept around before is the best long term bet from a divorce-avoidance point of view.

Secondly, some of INTREPID's findings seem to contradict. In the last table, INTREPID  indicates that should a woman sleep with one pre-marital partner who later does not become her husband, her risk of divorce increases to near 50%, ye,t the 1 previous partner divorce risk, in chart 3, is listed at somewhere near 35%. If INTREPID can provide an explanation I can be contacted through the usual channels.

Thirdly, the Heritage Study is vindicated. Women who sleep around are far less likely to be in stable relationships.

Fourthly, The U Shaped graph is interesting:
  • I would go with the 2006/8 Figures more than previous ones. I understand that it was only in this cycle that ACASI interviewing was used. Here a computer prompts the questions and the responder enters the data into it. This way there is very limited face to face contact with the questioner and hence, the responder gains a greater degree of anonymity. Previous research has shown that women are prone to give "socially acceptable" responses when interviewed by a real life interviewer as opposed to answering anonymously.
  •  The graph is not controlled for age of marriage.  Girls who marry early will have less sexual experience than girls who marry much later. Age at marriage is a big predictor of divorce and  I imagine that the one to two non-marital partners category is inflated because of this effect.  It would be interesting to see what the age at divorce would be in this  group.
Other reader comments are invited.

Once again, a sincere thanks to INTREPID for all his/her work.