A new sleep study hit the press a few days ago, and my inbox is lighting up. Should I believe it, inquiring minds are asking. The study was presented at a SLEEP conference (SLEEP is what they call the Associated Professional Sleep Societies). Media reports flaunt headlines such as "Marriage linked with better sleep" and "Stable marriage is linked with better sleep in women."
Readers of this blog know how I feel about research studies - I want to read the published versions, not just abstracts, not just press releases, and surely not just what the media claims that the studies say. But this latest study has been a challenge. I can't find a published study corresponding to the press release. I can't find anything longer than an abstract. I emailed the author who is quoted in many of the stories, but she has not responded. My guess is that there is no published article. The results were presented at a conference, and maybe not even a conference paper (rather than just an abstract) is available. So I'll tell you what I think based on the information I can gather, but I still want to see a professional write-up.
Let me start with my hypothetical drug study example, because I think we are better at thinking intelligently, critically, and impartially about drugs than about marriage. A drug company - I'll call it Promarital - has come up with a new drug that is supposed to help you sleep better. I'll call the drug Sleeppiage. The Promarital drug company offered their Sleeppiage drug to 360 White, African-American, and Chinese women (average age of 51). The company measures the women's sleep when they first sign up for the study and again over the course of the next 6 to 8 years.
1. Some women never do take Sleeppiage. The Promarital company calls these women the No Drug group.
2. Some women take Sleeppiage but hate it and refuse to continue taking it. The company calls these women the Losing the Drug group. There are no published reports of the precise size of this group but from other research, we can estimate it as at least 43%.
3. Some women did not take the drug at first but did start taking it later. Promarital describes these women as "Gaining the Drug."
4. Finally, there is the drug company's favorite group. They took the drug from the start and stayed on it. Promarital calls it the Sleeppiage group, or just the Drug group.
Promarital reports, with great fanfare, that Sleeppiage works, because the women in the fourth group, the Drug group, sleep better than all the rest! The Drug group women sleep especially better than the women in the "Losing the Drug" group.
That's right, the Promarital drug company takes out of the Drug group all of the women who hated the drug - at least 43% of them! - and says about the remaining women, "Look at how well they are sleeping - Sleeppiage is just great!"
This, in its broadest strokes, is what the latest sleep study is saying. (I'll go into more detail below.) It is also an example of how decades of research on marital status has been used to suggest, falsely, that if only you get married, you will be happier, healthier, sleep better, live longer, and all the rest.
My Sleeppiage = Marriage equivalencies are probably obvious, but I'll spell them out just to be sure I'm clear:
1. The No Drug group corresponds to women who stayed single the whole time.
2. Losing the Drug: women who got married, then divorced.
3. Gaining the Drug: women who started out single, then married.
4. Drug, or Sleeppiage group: This is the currently married group, or everyone who is left of the people who got married, after setting aside all the people who got married, hated it, and got divorced.
Actually, it is even worse than I just suggested. That's because often, the people who marry, hate it and get divorced, are not just taken out of the marriage group and set aside. Instead, their distaste for the whole thing and their bad outcomes are used as evidence that marriage is good! Seriously. I'm not making this up. Every time you see someone argue (in the media or at professional meetings or in the journals) that divorced people do worse in some way (sleep or happiness or health or whatever) than people who are currently married, that's the case that is being made. It is the same as a drug company proclaiming that their new drug is great because people who take the drug and like it do better than people who take the drug, hate it, and therefore stop taking it.
Would you take a drug based on that evidence?
But can't you conclude that the moral is to get married and stay married? Well, that assumes that if all of those people who got married, hated it, and got divorced had just stayed married instead, they would now be sleeping just fine. Seems unlikely. The only definitive study cannot actually be done - you'd need to randomly assign people to stay married or get divorced and then assess sleep.
Another finding was that women who started out single and later got married and stayed married also slept almost as well as the women who were married the whole time. (Well, sort of. When they first married, their sleep was more restless.) Doesn't that indicate that if you get married, you will eventually sleep better? Again, not really. Because if you get married, you may or may not stay married. If you eventually divorce, you may not sleep any better than if you stayed single the whole time. (Unfortunately, not enough data were reported, so I can't find any precise numbers. In most studies of marital status - I reviewed them in Singled Out - people who were previously married do worse than people who have always been single.)
The Real Summary of the Results of the Sleep Study
From what I can tell from the paltry data that has been made available, here's what the sleep study really did show.
• If you get married, chances are pretty good (probably about 43%) that you will eventually divorce, and divorced people have some sleep problems.
• If you are among the remaining people who got married and did not divorce and did not become widowed over the course of the 8-year study, your sleep will be pretty good. (Again, we don't know how good - data are not reported.)
• If you start out single and then get married, you might eventually get divorced. See above. If you don't get divorced over the course of the study, your sleep will eventually improve, though at first you will sleep more restlessly.
• If you start out single and stay single, we don't know how your sleep will compare to all of the people who got married - that's not reported. Your sleep will not be as good as the people who are left in the married group once all the people who got married, hated it, and got divorced are taken out of that group. If the results of this study are like so many other marital status studies in the literature, then probably the always-single group will sleep better than the previously married group.
• Actually, there are some qualifications of these findings:
o The study only reported on women.
o The study included White, African-American, and Chinese women, but the results were mostly true only for the White participants.
o Oh, and another thing - the currently married sleep better if they are also happily married. If they are having problems, forget about it. So the group that is already given an unfair advantage (because the people who didn't like marriage and got divorced are taken out of the group) still only looks good if you consider the happily married among them.
So start with women only. Set aside the African-Americans and the Chinese. Look at the people who got married. Not when they first got married, because their sleep is restless then. Not the people who got married and then divorced, because their sleep is not so good, either. Also, of the married group that is already rigged (because everyone who got married, hated it, and then got divorced have had their marriages statistically annulled - they are not included in the married group anymore), rig it even more by underscoring just those who are happily married. There you go! That's the evidence that getting married makes you sleep better!
Let's all go out and buy Sleeppiage.
What Was GOOD about the Study
Actually, this study had one very important strength - it was longitudinal. The authors looked at the sleep patterns of the same people at the start of the study and again 6-8 years later. That way, we can see what happened to the same people as they went from being single to getting married, or being married to getting divorced, or any other pattern. That's the next best thing to the methodologically superior study that we cannot ethically do - randomly assign people to get married or stay single or get divorced or become widowed.
So the problem is not with the design of the study. It is with the misleading reports of the results and of the implications of those results.
What Was Good About the Reporting
Most reports that I found in the media used an important word: "linked." They said that marriage was linked with better sleep. That's good because it is not a word that implies causality. It is not the same as saying that getting married results in sleeping better. (My guess, though, is that many readers made that leap anyway. I'd love to see a study of how these headlines are interpreted.)
The best reporting, unsurprisingly, came from the website of the SLEEP group. They reported 3 sleep studies, with this headline: "It's Complicated: Sleep, Marriage, & Relationships."
I also got a kick out of something else I discovered while trying to figure out what the sleep study really did say: One of the best reports came from a website that describes itself as "Celebrity, Sex, Fashion for Women." That site is Jezebel. In the relevant post, all three studies are mentioned, including the one showing that people who earn a lot of money sleep a lot better than people who don't. Here's the closing line: "The study author recommends never going to bed mad, or, you know, just having huge amounts of money."
Once More, Because It Is So Important
If you hear someone say or imply, "get married because married people do better than divorced people," you are at risk for getting scammed. Divorced people DID get married. They didn't like it. They amount to about 43% of the people who got married. Imagine testing a new drug, or a new hypothesis, and letting yourself exclude from the key group the 43% of the people who hated the drug, or who behaved just the opposite to the way you predicted. Then imagine trying to publish that study, with the claim that it shows how great your drug or your hypothesis is!
Same for comparing the group that took the drug (minus the ones who hated it and stopped taking it) to those who never did take it, and saying that the drug wins. Or comparing your experimental group (minus the people in that group who behaved contrary to your hypothesis) to the people in the control group, and declaring your hypothesis has been supported. It wouldn't happen. Unless you were studying marital status.