Saturday, June 30, 2012

Which way to Happy Happy Village?

Oh, so you say all you want is to just be happy? By way of ordinary least squares, here are the multiple regression coefficients for nine variables that are, in the conventional wisdom, arguably putative predictors of happiness, with the obligatory qualifier that some component of sunniness, of yet undefined magnitude, is innate. For contemporary relevance and to avoid racial confounding, responses are from 2000 onward and only non-Hispanic whites are included. Asterisks indicate statistical significance at 95% confidence.

First, for men:

Age -- .27* (youth = happiness)
Marital status -- .24* (marriage = happiness)
Class -- .24* (higher status = happiness)
Church attendance -- .11* (worship = happiness)
Education -- .06 (more education = happiness)
Political orientation -- .04 (conservatism = happiness)
Children -- .04 (procreation = happiness)
Number of female partners -- .00 (no relationship at all)
IQ (wordsum) -- .00 (no relationship at all)

The parenthetical stuff is there to clarify which direction the correlation runs in and shouldn't be interpreted as anything close to an absolute statement on causation.

Being young, married, blue-blooded, and active in church are predictors of higher levels of self-reported happiness (I bet Mitt Romney is and has always been a really happy guy). Lisa didn't quite get it right with her graph showing a vigorous inverse correlation between happiness and intelligence, but what one focuses his cognitive horsepower on probably matters more than the number of stallions he has in the reigns.

This more-or-less meshes with the Game perspective--youthfulness and high socio-economic status are definite pluses, and marriage, while scorned as an emasculating deathtrap, does indicate some level of attraction to other women, which can potentially be used to a philander's advantage. While notches in the belt don't obviously bring happiness, they don't push it away, either.

And for women:

Marital status -- .26* (marriage = happiness)
Class -- .24* (higher status = happiness)
Education -- .23* (more education = happiness)
Age -- .18* (youth = happiness)
Church attendance -- .09* (worship = happiness)
Children -- .05 (procreation = happiness)
Political orientation -- (.03) (liberalism = happiness)
Number of male partners -- .03 (fewer partners = happiness)
IQ (wordsum) -- .00 (no relationship at all)

Not much in the way of substantive differences between the sexes, except that educational attainment is a relatively stronger predictor of happiness among women than it is among men (perhaps because social status is tied relatively more strongly to educational attainment among women while tied relatively more strongly to wealth among men), and youth is a stronger predictor of happiness among men than it is among women (or, more precisely, old age is a better predictor of unhappiness among men than it is among women).



Jokah Macpherson said...

Very interesting. I have lots of questions, though.

Why is the HAPPY variable limited only to options 1 and 3?

What did you do for the MARITAL variable since it is not measured on a continuum (1=Married, 2=Divorced, etc.)?

Also, to the extent I've looked at the question, age is positively correlated with happiness - the graph has a gentle rising slope throughout life and then drops steeply around age 80 but not enough to affect the big picture. Does this multiple regression control for the other variables or something? Otherwise, how did you get this result?

Personally, I'm banking on painting everything I own blue to bring me true happiness.

Anonymous said...

"The parenthetical stuff is there to clarify which direction the correlation runs in and shouldn't be interpreted as anything close to an absolute statement on causation."

Okay, but what happens when you look at it from the other direction? Like, are happy people more likely to get married? Such that happiness=>marriage rather than marriage=>happiness.

For example, I was reading this article:

in which the researcher reports:
" Even among the most highly committed Catholics -- those who go to Mass weekly, who say they would never leave the church, and who say that the church is among the most important parts of their life -- 60 percent agree that one can be a good Catholic without obeying church teaching on birth control."

Okay, but when you look at it from the other direction, it turns out that "Among Catholics who say that you cannot be a good Catholic without obeying the Church hierarchy's teaching on birth control, 49 percent attend Mass weekly or more often." (not in linked article. This info was obtained through email from the article's author)

In this case, there is a significant difference, and a stronger correlation between belief and attendance when you check for causation the other direction.

Back to marriage. Many miserable human beings get married, they probably just can't stay married like happy people can. It is probably harder to check for happy disposition causing successful marriage, and if indeed it runs that way, then what sort of policies could social scientists reasonably advocate?

The default of seeing environment/condition as the first, last and only cause, isn't working, but what is the real alternative? To get people to go that direction, there has to be a way to frame a win-win situation even if it is as inaccurate as the current pronouncements that changing environment/conditions will make bigger improvements in outcomes than they actually do.

Audacious Epigone said...


- To make the happiness measure more stark.

- Treated like it was a continuum, which isn't optimal, admittedly. But when you actually look at it, if you had to order the five in some way, the GSS order looks as good as any, save for swapping divorced and separated.

- Right. If you don't control for anything else, that's what you see. But when all the other factors listed are thrown into the equation, age itself is associated with unhappiness (it just also happens to be positively associated with likelihood of being married, higher class, greater church attendance, greater education).

- That at least one person gets the title's neotenous reference warms my heart!


Great points and questions. The GSS doesn't really allow much of a way to answer them, though.

Clark Goble said...

Sadly WORDSUM is more a measure of stupidity rather than reasonable intelligence. That is it can't measure the degrees of intelligence that I suspect the typical academic would consider the range of acceptable intelligence. One wishes there were a way to have that data (and have enough of a sample size that these folks a standard deviation above the mean could be analyzed)

I think that the Simpsons is more talking about that kind of intelligence and the GSS doesn't really give a way to check on that.