There’s a new article out confirming what we already know: women initiate the vast majority of divorces. It’s being reported all over the place. Here’s the Yahoo! story: http://news.yahoo.com/women-more-likely-initiate-divorce-not-dating-breakups-110114625.html.
Of course, the conclusions reached as to why women initiate divorce are completely wrong but we shouldn’t be surprised. This paper is actually a pretty good example of what’s wrong with academia. Here’s the link if you want to follow along, but I’ll quote what I need: http://web.stanford.edu/mrosenfe/Rosenfeld_gender_of_breakup.pdf.
First, it’s important to note that this paper has not been published in an academic journal but is already being reported on. There has been no peer review of this paper and the available version is only a draft. This is a problem not so much of academia but of the modern press. However, it wouldn’t surprise me in the least if this paper ends up published as is even though it doesn’t stand up to the most basic scrutiny.
So, the basic idea behind the paper is to compare the initiation of divorce in marriages to initiation of separation of non-married cohabitating couples. The author notes that there is more gender parity in initiating separation among cohabitating couples than married couples, where as we all know, women overwhelmingly initiate. Rosenfeld theorizes that this is because cohabitating couples are less traditional than married couples and it is the traditional gender roles that women find offensive. We know the real answers but let’s play along with Mr. Rosenfeld and find out why he’s wrong.
Let’s start with how Mr. Rosenfeld did his study. What is missed by the media reports is that Mr. Rosenfeld didn’t actually do a study. He never conducted a survey himself, rather he used an existing survey and re-interpreted the results to answer the questions he wanted answered. The source for his data is actually called “How Couples Meet and Stay Together” which as you might imagine was focused on different things. There’s nothing inherently wrong in doing this; it’s certainly economically efficient. However, Mr. Rosenfeld’s paper is a good warning of why relying on other people’s research can be a bad idea. First, his sample size ended up being tiny. The original study followed couples for only six years meaning that they missed a good number of the separations that would follow. So among the 2,538 people surveyed there were only 371 break ups or divorces. This number is further reduced because our author rightfully disregarded non-cohabitating couples from his comparison. This leaves us with a paltry 92 divorces and 76 separations of non-married cohabitating couples (separations from now on for simplicity’s sake).
Those are tiny sample sizes and when he does his statistical analysis it shows. He found that women initiate divorces 69 % of the time +/- 8.5%, 95 times out of 100 and initiate separations 56% of the time +/-9%, 95 times out of 100. Rosenfeld claims that this is a statistically significant difference and he’s technically correct. However, when your margins of error are that high, it is a pretty good indication that you need a bigger sample set. Can you imagine if a political poll had a margin of error that big? “Oh, yeah, we had Romney ahead of Obama by 5% in 2012, but the result was within the margin of error!” This man got funding from Stanford, gentleman. Let’s move on though and we’ll assume for the sake of argument that there is more gender equity in initiating separations compared to divorces.
The second major problem with this paper is how it conducts its analysis. He proposes three different hypotheses.
Women are more sensitive to relationship problems.
The second wave feminist idea that marriage is bad for women because patriarchy.
Two opposite theories related to the idea that men have more power in marriages as they age. Either, men should start initiating more divorces/separations later in life or women are initiating because they feel oppressed (which frankly is the same argument as 2 but he has it twice so I feel obliged to treat it separately).
He then uses statistical analysis to determine if these theories hold true. This sounds good. It sounds scientific. But that’s actually my problem with it. Let’s step back. When you’re conducting a scientific experiment the goal is to determine what happens in very specific circumstances or why something happens in very specific circumstances. We use control groups. There are many people in the social sciences who emulate this approach. However, this approach is often foolhearty when it comes to analyzing people. As my first year sociology prof put it, “Amoebas don’t blush.” In other words, when people know they’re being experimented on, they change their behaviour. In this case, when people are answering questions about their break up, they are going to answer it in a way that makes them look good. Rosenfeld, to his credit, admits this. However, he persists in using a scientific method for his analysis when the research wasn’t conducted using control groups or any other scientific precaution.
Furthermore, his hypotheses are way too diverse. Trying to prove five different things with the same tiny sample set is a bad idea. It’s not how you do science. Yes, we learn things unexpectedly in experiments, but a good scientist would conduct an experiment to verify those unexpected findings. Repeatability being an essential part of the scientific method (also impossible with social science: you can’t ask the same person, the same question twice at the same time of their life). In this particular case, he’s scraping every last bit of data of this study to try to test his hypotheses. First, he rejects the women are more sensitive hypothesis because of the statistically significant if less than convincing difference between the 69% initiated by women in divorces and the 56% initiated by women in separations. This isn’t a horrible conclusion, so we’ll move on.
He gives a lot of credence to hypothesis number 2 but we’ll come back to it because his major supporting argument amounts to “it’s the only option left.”
As for number three, he looked at differences in education, age and income and found no strong corelations. Again, given the size of his sample size, I’m not sure how he could tell. He also goes ahead and makes a huge leap of logic and assumes that those who self-described as evangelical must practice traditional gender roles at home because reasons. If you think I’m exaggerating:
“I assume that individuals who identify as born-again or evangelical Christians were more likely to hold traditional views about gender roles, however HCMST lacks the questions on gender role to test this assumption directly.”
Bald assertions? Why not. Again, he’s forced into this bald assertion because he didn’t conduct his own study and therefore has no idea how people view gender roles. So, he just found a category and made an assumption. There’s 21st century academic rigour. If there’s one part that might get taken out by the time this thing gets published it’s this part. It’s pretty egregious and it proves nothing. Evangelical women aren’t more or less likely to initiate divorce. So finding no other reason, he circles back to his second hypothesis based on the second wave feminist idea that there is something inherent in marriage itself that makes women miserable. He boasts very little evidence for this. He says that on average women are marginally unhappier in marriage but the difference is miniscule. One woman anecdotally claimed abuse which he for some reason associates with traditional marriage (I hate traditionalists as much as the next MGTOW but I won’t assume an abusive relationship is traditional).
However, he’s constrained by his own lack of imagination and the ridiculous limits that he constrained himself to. By limiting himself to popular theories of divorce, he missed the point entirely. Even though, he had no evidence to support the feminist narrative he ends up basically endorsing it because he has no real way to test it and disprove it. You can’t falsify the patriarchy after all. This is fundamentally the issue with modern social science. People limit themselves to established theories and cannot fathom a possibility outside of those theories. This is a problem in all disciplines. However, in gender studies this is an even bigger problem. The reason things like gender studies are ridiculous degrees is that they don’t even have competing theories. In economics, you can analyze things using marxist, liberal, conservative or libertarian economic theories. In gender studies, you can analyze things through feminism. If your really lucky, you might learn about different types of feminism but nowadays, there’s a good chance you’ll only learn about gender through the lens of so-called intersectional feminism. That’s the problem. Learning one theory doesn’t teach you how to think no more than you learn to cook by following the same recipe over and over.
Oh, if you’re new here. The reasons for the difference are quite simple. First, the consequence of a separation are relatively gender neutral. Both partners are likely to be restored to their original posiition. Only if there are children involved will the man lose and kids are less likely in non-married couples. In divorce, the consequences are nowhere near gender neutral. Men suffer immensely while women are either unharmed or outright benefit. Of course, women initiate more divorces. Second, women are much more likely to grow bored with their provider husband and want something more exciting. Hypergamy doesn’t stop with the wedding vows. Of course, Mr. Rosenfeld didn’t even consider those factors.