Sex, models and housework

Sorry, not sex with models (or models doing housework) but mathematical models about sex and housework.  The first study I wanted to look at in my discussion of the use of models in data reporting actually got sent to me a while ago.  The headlines around this were things like "Want to have more sex?  Men, stop helping with the chores!", and a concerned (male) reader sent me the link with a "what's up with this???".

Essentially, the study took answers from a large survey that asked people about their household division of labor, their marriages in general, and their sexual frequency.  The authors were attempting to prove or disprove several notions about how housework and sex relate in marriage. They came to the conclusion that the more housework conformed to traditional gender roles, the more sex was had by all.  A few notes about the study up front:

1. The data was collected in 1992, with a mean age of 43 for women/46 for men.  This is notable because people's expectations for marriage have change dramatically over the past few decades (divorce rates peaked in the 80s), so the generalizability may be limited.  However, this data set was used because it's the largest in existence that has all this information.  The authors acknowledge this limitation.
2. The authors divided chores in to traditionally female (core) and traditionally male (non-core) tasks.  Core tasks include meal prep, cleaning, grocery shopping, etc and non-core tasks include lawn maintenance, bill paying and driving.  The finding was that the more men did the core work, the less sex the couples had, but the more non-core work they did, the more sex they had:So the headlines that more chores = no sex are wrong...it was the "wrong" kind of chores that influenced things.
3. The authors never studied (nor claimed to study) the effects of changing chore arrangements on sexual frequency.  In fact all of their conclusions are based on the entire marital arrangement, so do NOT take the headline writers advice and start shaking things up assuming that this will have a particular result.  
4. I found it fascinating that the authors specifically ruled out coercion as a factor here.  Satisfaction was fairly high across the board. 
5. As the data is presented here, I do not argue with their conclusion.  While I think we could all quibble about the mechanism that causes this to be true, the data as presented in the paper supports what they say it does.

Getting back to the modelling stuff....the graph above shows the model they came up with, after controlling for all other factors.  Kind of nifty, right?  But what concerns me about this is that it's so nice and linear.  When I look at graphs that are supposed to model certain phenomena, I take a look at the extremes.  Now, I know quite a few super-egalitarian couples, but I actually don't know any couple in which the male does 100% of the cooking, cleaning, laundry, etc.  Even with over 4000 couples in this survey, how many did they really have at that end?  What would cause that arrangement to evolve?  Unemployment? Disability? I would be very suspicious that any couple would actually settle in to that arrangement long term...so I'd wonder how things would really behave at that end of the chart.  

Another note on this model: the key phrase is "controlling for all factors".  From a research perspective, the researchers appear to have done this quite diligently.  From a real life perspective though, people attempting to extrapolate this data for their own lives would do well to remember we don't live in vacuums (no pun intended).  Spending time with your spouse, having a higher income, not having small children, and being religious all are positively correlated with higher sexual frequency.  When I was researching this article, I was interested to find that the WSJ had taken a different tactic with their article, and mentioned that those who do more total chores also have more total sex.  Work hard, play hard they called it.

Finally, we have to consider why the researchers likely went with a mathematical model at all over just reporting the data directly.  My guess is outliers.  When I followed up with the guy who sent me the study, I mentioned that it was key to remember that this was not a straight up reporting of data, but rather an extrapolated model.  He asked why they would do that.  The only response I could think of is that it's likely the data simply wasn't this clean when they put it together the other way.   That doesn't mean the conclusions are wrong, it just mean the reporting isn't quite as direct as we might presume from the headlines.  

Blame the journalist.

Now if you'll excuse me, I have to go do some dishes.