Today, I recorded a Substack Live with Mike Pesca of The Gist. Mike’s a great interviewer, and in contrast to the previous Lives I’ve posted here, this was more of him interviewing me than the other way around. Still, I think you’ll enjoy the conversation; the full video is above.
We touched on a bit of poker at the beginning, and New York mayoral stuff at the end. However, most of the conversation centered on the differences in mental attitudes between liberals and conservatives and how this may be affecting political outcomes.
Mike had reached out after I published my SBSQ on why young men have turned against Democrats, which covered some of this territory. Young people in general report fairly miserable mental health, with high rates of anxiety and depression. But this is particularly true for young women and assumptions Democrats make about how to attract younger voters may not appeal to men.
Still, arguably that SBSQ buried the most interesting finding, which is that conservatives have much higher self-reported mental health than liberals. It’s a wide gap: according to the 2022 Cooperative Election Study (CES) — a very large sample survey (60,000 respondents) that provides the opportunity for highly detailed demographic analysis — among people who report “excellent” mental health, conservatives outnumber liberals 51-20. But liberals outnumber conservatives 45-19 among those voters who say they have “poor” mental health:
Could this reflect a spurious correlation? In other words, that voters with characteristics associated with lower happiness tend to be attracted to liberalism, but that political attitudes themselves don’t tell you much on their own?
In short, no. Or at least, probably not. The difference between liberals and conservatives is remarkably persistent even once you control for those factors.
I’m going to show you a very long chart, where I translated the five choices that the CES provided to a 100-point scale: 0 for “poor” mental health, 25 for “fair”, 50 for “good”, 75 for “very good” and 100 for “excellent”. The average American self-reports at a 60 on this scale: in other words, somewhere between good and very good mental health. But liberals average a score of 53 and conservatives a 68. (I’m just going to ignore moderates for the rest of this post because they’re predictably somewhere in the middle, averaging a 58.)
Taking advantage of the large sample size in the CES, let’s see how this difference holds up across a wide array of demographic and political characteristics. This data is just taken directly from the CES; there’s no fancy modeling of any kind.1
As you can see, the liberal-conservative gap is fairly consistent across all of these characteristics. But let’s take a warp-speed tour of them, understanding that we’re reducing a lot of complex factors into a handful of boxes that people checked on a survey:
Gender. Men report slightly higher mental health than women, although it isn’t really a big gap until you get into the younger age cohorts. However, political attitudes are more predictive. Conservative women report considerably higher happiness (66) than liberal men (58).
Race. Not really a big factor, except that liberal Black voters describe higher mental health than other liberals, which cuts somewhat against the thesis that historically oppressed groups are more likely to report mental health problems. Liberals describing themselves as being mixed race or belonging to some “other” race report being particularly unhappy, however. This may be because mixed-race identification is higher among younger voters, and young people who are struggling with their identity are often struggling with other things, too.
Age. This is one of the few characteristics that trumps political attitudes. Old liberals report better mental health (64) than young conservatives (58), for instance. Still, the political gap persists once you control for age. It is especially large among Gen Z voters, as I described in the post about young men.
Education and income. Consistent with a large body of research, having more money does make people happier, at least up to a point. Still, conservatives with household incomes of $30,000 or lower report the same mental health (60) as liberals making $100,000 or more. Educational attainment behaves similarly, although it’s highly correlated with income and therefore can be conflated with it. There is a particularly wide gap between liberals and conservatives among the large number of “some college” voters: those who attended college but didn’t finish it, or who completed a 2-year degree. If I can speculate, perhaps this reflects how liberals generally place a higher value on education, and liberals who don’t complete college may see themselves as having failed to live up to expectations.
Religion. There is no meaningful difference in mental health between Catholics, Protestants and adherents of other organized religions. The differences between liberals and conservatives persist, but religious adherence does soften them somewhat. For instance, there’s only an 8-point gap between liberal Catholics (61) and conservative Catholics (69), about half the size of the overall conservative-liberal difference. Some of this is because of selection effects: atheists, agnostics and people who describe their religious affiliation as “nothing in particular” report considerably lower happiness, and these Americans are much more likely to be liberal. The finding that religious people are happier might not sit well with irreligious people like me2, but it’s well-replicated across the empirical literature.
Marriage and children. Married people and people in domestic partnerships are happier than those who are separated, widowed, divorced or who have never been married, but the differences aren’t as large as you might assume. (Though other studies report a bigger effect.) For child-rearing, the happiest people are actually empty-nesters (people who had kids but no longer have them in their household), perhaps because this is correlated with age (older people are happier). Again, though, the liberal-conservative gap persists across all of these characteristics. It’s particularly wide for Americans who have never had children. Childless conservatives are happier than the average American (63), while liberals without children report a 48.
Sexual orientation. I’m not going to kick a hornet’s nest by delving too deeply into “LGBTQ+” identity, but you do see some interesting differences here. Gay men and lesbians report somewhat lower happiness than heterosexuals overall, but conservative gays and lesbians report higher happiness than heterosexual liberals. Liberals who describe themselves as bisexual or having some “other” sexual orientation report notably low happiness, however. Undoubtedly, this is partly because they often face a society that isn’t very accepting of them. But this is also correlated with age. According to Gallup, 21 percent (!) of Gen Z women and 7 percent of Gen Z men describe themselves as “bisexual”, a much larger number than the more heated question of how many say they’re trans (just 1 to 2 percent). I mentioned before that liberals who report their racial identity as “mixed” or “other” often say they’re unhappy too. Perhaps sexual orientation behaves similarly, where young people who are struggling to define their identity also select into more ambiguous categories.
Political and social media activity. The CES also asks respondents several questions about their overall level of political interest and social media activity. I thought this might explain some of the differences between liberals and conservatives. If Twitter makes people miserable, for instance, perhaps that explains some of the gap because — at least in 2022 when this survey was taken — social media was predominantly liberal. However, people who report higher political interest3 generally report higher happiness. Maybe this is because people have more bandwidth to follow political affairs and other external interests when their life is otherwise going well. Further to my surprise, people who had recently posted to social media at the time of the survey also reported slightly higher happiness than those who hadn’t. And there were no significant differences based on whether or not the voter had recently attended a political protest. I’d be a little careful with these findings because meta-analyses on social media use report mixed results, and it may depend on the particular type of social media use, e.g., Instagram might be particularly bad.4 At the same time, media types like me may overstate the impact of social media on lived experience and political outcomes, perhaps because we often have unhealthy relationships with it ourselves.
Of course, all of this only provides hints as to the question I posed in the headline.
There are ways to frame these findings sympathetically to liberals: for instance, that liberals are more conscientious about the suffering of other people or the conditions of the world, whereas conservatives are happy but selfish.
There’s a relatively neutral framing: that people become liberals because they’re struggling or oppressed themselves and therefore favor change and a larger role for government.
Or you can treat the liberals less sympathetically. As Matt Yglesias suggested in his post “Why are young liberals so depressed?”, there may be a performative aspect to this. Liberals may express negative sentiment as a sign of solidarity with a movement that thinks there is profound injustice in the world. But then this can snowball — misery loves company — while attracting some number of people who face serious mental health challenges.
I’m not going to try to untangle these hypotheses today. And in consideration of how sensitive they are, particularly when taken in combination with other identity characteristics, I’m going to do something rare and leave the comments off for this post.
However, I do think this is a much under-discussed topic — the gap in self-reported happiness between liberals and conservatives is much larger than for other characteristics that often get far more attention. So I hope this can encourage some thoughtful consideration from other writers.
Although it uses the weights recommended by CES for demographic balancing.
For what it’s worth, I’d describe myself as more agnostic than an atheist.
The precise question the CES asks is as follows: “Some people seem to follow what’s going on in government and public affairs most of the time, whether there’s an election going on or not. Others aren’t that interested. Would you say you follow what’s going on in government and public affairs: most of the time, some of the time, only now and then, or hardly at all”. In the chart, I grouped the “only now and then” and “hardly at alls” together into “low political interest”.
Or at least high-variance: great for some users, terrible for others.
