I go to update an old trend with the 2012 General Social Survey data, and a sleepy little slope suddenly pops. First it was a rise in people identifying as “lower class.” Now it’s marital happiness.
You might have predicted that marriages would be growing happier after the 1970s, since divorce allows the unhappily-married to drop out of the sample. But the trend until recently had been boring — a slight drop in happiness in the 1970s and 1980s, and flat since. But now we’ve got a bonafide trend back toward happiness from 2006 to 2012 (individual data points with five-year moving averages):
To test the curves, I did a simple set of logistic regressions with year and year-squared, controlling for sex (women are less happy with their marriages; sex interactions were not significant). The curves for “very happy” and “pretty happy” are highly significant — meaning the U-shapes you see are highly unlikely to be due to chance; “not too happy” tests flat:
|Logistic regression coefficients for marital happiness states, 1973-2012|
|Very happy||Pretty happy||Not too happy|
|Years since 1973||-.024||.026||-.013|
|All coefficients significant at p<.001 except time trends for “Not too happy” (p>.20). N = 27,207.|
I have no explanation for this, and no position to argue. I just thought you should know. To keep it in perspective, it’s a modest change, and the majority were already “very happy.”