08/19/2011 12:45 pm ET | Updated Oct 19, 2011

Reports of Singles' Early Demise are Greatly Exaggerated (or Just Plain Wrong)

A just-published article claims that people who stay single are headed straight to the grave -- and fast. Faster than people who are currently married. The media is already on it, as with today's MSNBC headline, "Single people may die younger, new study finds""and the many breathless sequels.

If you are a regular reader of my blog posts here at The Huffington Post, of my "Living Single" column at Psychology Today, or if you read "Singled Out" or "Single with Attitude," you will recognize something in that very first paragraph that tells you all you need to know about why this latest scare story is bogus. See what it is?

The suggestion is that if you get married, you will live longer. To make that claim with any methodological rigor, at a minimum you would need to compare everyone who ever got married to those who did not marry. Instead, the research in the journal article in question is based on the typical cheater technique, in which a group comprised of everyone who stays single -- whether they want to be single or not -- is compared to the group of people currently married. So the comparison group is not all of the people who ever got married. That's the appropriate group to use if you want somewhat respectable scientific grounds for saying to single people that if they get married, they will live longer. No, that's not the comparison group. Instead, all of the people who got married, hated it, and then divorced (and all of the people who got married and became widowed) are removed from the comparison group. That's not good science -- it's a set-up.

If it is still yet clear what's wrong with that sort of comparison (all people who stay single vs. only those people who got married and are still married), I'll give you the short version here. The more detailed explanations are in Chapter 2 of "Singled Out" and the section of "Single with Attitude" called, "If Marriage Were a Drug, the FDA Would Not Approve It." There is also a less-detailed version in this post.

I typically use the example of the hypothetical new drug, Shamster, but it's summer, so I'll try a cruise line variation. Festival Cruise Lines is in competition with Royal Treatment Cruises. Festival has access to satisfaction ratings from all of the people who ever cruised on either line. Some people tried Festival, hated it and never cruised with them again. Maybe they, along with most of the other people onboard, got sick. Maybe the service was lousy or the food was disappointing or the rooms were tiny. Whatever.

So, Festival decides to include in their analyses only those people who rated their Festival experiences positively and continue to book cruises with Festival. They compare those satisfaction ratings with the ratings of everyone who ever cruised on Royal Treatment, whether they liked their experiences or not. What's more, the number of people who hated their Festival experiences was close to 50 percent. Knowing all that, what would you make of the claim that people who currently cruise with Festival are happier cruisers than Royal Treatment customers? Would you assume that you should cruise with Festival, too? I didn't think so.

Don't expect social scientists to get all apologetic about the bogus comparison. It's their standard practice and has been for as long as this sort of research has been conducted. Their defense might be that they are clear about who is in each group -- all people who stayed single vs. those who are currently married. Their claim is that people who are currently married are happier (or live longer, or whatever today's bogus story may be) than those who are single and always have been. Technically true, perhaps. But again, what would you make of the comparable claim that the current customers of Festival are happier than all of the customers of Royal Treatment? (I know, my drug example works better, but I just wanted to do something different for once.)

In scientific papers, there is almost always a section in which the authors are made to fess up about the limitations of their study. If the authors recognized the cheater technique I have described so often, or if they were willing to admit it, you would find that concession in the limitations section. It is not in the review paper on mortality that I'm discussing, and it is hardly ever in any other paper. That concession would not sit well with the ruling narrative that "married people win."

It is interesting in a way, because there are two popular "explanations" for the bogus claim that "married people win" that should nudge at least some social scientists into realizing what's wrong with their comparison. Those explanations are "selection" and "protection." The selection argument says that you can't compare currently married people to single people at one point in time and say that marriage made people happier, because perhaps the married people were happier (or healthier, or whatever) than the single people even before they married. The protection explanation says that married people win because their spouse protects them from unhappiness, ill health, an early demise or any other bad outcome. The fact that those same social scientists don't realize that there is also selection out of marriage (divorce, widowhood), and that selection out could potentially be even more important than selection in, continues to astound me.

In a longer critique I wrote in my Living Single column, I described more of the details of how the research was done and the problems with that work. I also took the authors to task for the ways they discussed their research -- lots of issues there, too. Finally, based on my reading of the original research report, I describe some of the findings from the report that have not made it into any of the stories in the popular press. For example:

For studies that started in 1950 or earlier, the relative mortality risk was not significantly different for the currently married compared to the always-single. In fact, for studies that started between 1940 and 1949, the always-singles lived non-significantly longer than those who were currently married when the study began.

Take a look at the rest of that post to see why those particular results are especially telling.