Bryan Caplan  

Is a Bad Marriage Better than None at All?

Iraq and the Corruption Trap... Models of the Model Minority...

Will Willkinson reports that it is. But this conclusion doesn't check out in the General Social Survey.

In this data set, the average married person is indeed happier than the average never married person. But people who are only "pretty happy" with their marriage are actually a smidge less happy overall than the never married, and people who are "not too happy" with their marriage are even more miserable than people experiencing a separation. The numbers, from most happy to least happy ("very happy"= 1, "pretty happy=2," and "not too happy"=3 - i.e., lower numbers mean more happiness):

Marital Situation Average Overall Happiness
very happily married 1.466
married 1.735
never married 1.922
widowed 1.946
pretty happily married 1.982
divorced 1.985
separated 2.118
not too happily married 2.498

Bottom line: Unless you are "very happily married," you are happier if you don't get married at all. Fortunately, almost two-thirds of married people affirm that their marriage is very happy.

Needless to say, there are dozens of obvious concerns with these simple comparisons. But the exercise is still worth doing - and the simple results are consistent with common sense.

Comments and Sharing

COMMENTS (9 to date)
Will Wilkinson writes:

I haven't checked out the study cited in my post, but it's a 2005-er, so the data might be newer/better than the GSS, for all I know.

You can check out the actual paper, here, to find out.

PJens writes:

Shakesphere said something like: "It is better to have loved and lost, than to never to have loved at all."

Marriage is an adventure. Some work, some are doomed to fail, even predictably.

Good Grief Caplan, you have been blogging so expertly on economics. Please do not think it transfers into sociology:)

Peter Leppik writes:

I'm not sure what the point of this exercise is, exactly.

If you wanted to make a true comparison, wouldn't you also have to break out single people into buckets, too? For example: "Happily single," "Single and miserable." Fairness requires that if you're going to compare "unhappily married" to any group of single people, it has to be "unhappily single." You've excluded the happy people from the unhappily married subgroup, so you have to exclude them from the comparison groups, too.

Then you just wind up proving that people who say they're happy also say they're happy.

Finally, the simple fact that the researchers (apparently) asked married people if they were happily or unhappily married raises big methodological issues, since the answer to one question (how happy is your marriage) will very strongly color the answer to the other (how happy is your life) (caveat: I haven't read the study, just this blog entry, so I'm merely raising the issue and not saying that this is clearly a problem). Depending on how the questions were posed, in fact, this could render the entire data set worthless.

(Why? Because if they first ask "are you happily married" and most people say "yes," then they've primed those people to rate their life overall as happier than they might have if they asked the life-happiness question in a different context. Singles get no such priming.)

Paul N writes:

Bryan may be right about the conclusion (I am skeptical, as I am about the paper Will cites), but I think it was a poor decision to post this in any case because there are so many potential points of criticism for his line of logic. I'm sure there are more, but most obvious include (1) you have to control for age (2) you have to analyze what portion of "never married" people are happy and unhappy, like Peter notes (3) marriage doesn't happen by accident, you have to WANT to get married, so there are inherent differences in populations of married and non-married people. e.g., happy people may prefer to stay single, or unhappy people may want to get married.

By posting this, Bryan implies that he thinks there is some value to his analysis, and that makes him look like a worse scientist (in my opinion). You can get some reasonable conclusions from GSS data (none are perfect) but this is not even close to one.

Will Wilkinson writes:

By the way, the paper about marriage to look at is Lucas, et al "Reexamining Adaptation and the Set Point Model of Happiness: Reactions to Changes in Marital Status," Journal of Personality and Social Psychology, vol. 84, no. 3, 2003.

They look at a longitudinal survey of Germans and find, on average, marriage has no effect. But, the important thing is that there was very high variance. The effect of marriage for some people was very positive and lasting, and for others very negative and lasting. So sets like the GSS are likely averaging out a lot of individual differences, and are not a good sort of advice. Marriage doesn't appear to be good or bad for you as such, but it depends on the kind of person you are.

Matt writes:

Funny that it's "better" on average to be widowed than pretty happily married. The data hold up for divorced, separated and not too happily married--surely many of these folks would rather be widowed.

Chris Bolts writes:

When I look at how sad some of my single friends are that they're single and can't hold on to relationships and how happy I am now that I'm married, I question the validity of the data. One point of criticism, though what the hell does it mean to be "not too happily married" but be the most happiest of the group? Maybe they're getting a little kaching on the side, if you know what I mean. :)

margaret schaut writes:

If the numbers above are accurate, the happiest of all are the unhappily married. Maybe our joy is having someone ready to hand to fight with and blame everything on...

Adriane writes:

Methodology would be a key here, I think. If I am married, for example, to a spouse with a terminal illness, my mariage might be unhappy, but not as a consequence of the marriage.

Comments for this entry have been closed
Return to top