Media Genre More Basic Than Politics Or Personality?

While our personalities are correlated with whether we are liberal or conservative, it seems that neither one of these causes the other. Instead, their correlation is caused by something else that causes both, and that is caused in part by genetics:

The primary assumption within the recent personality and political orientations literature is that personality traits cause people to develop political attitudes. In contrast, … the covariance between personality and political preferences is not causal, but due to a common, latent genetic factor that mutually influences both. … Change in personality over a ten-year period does not predict change in political attitudes. .. Rather, political attitudes are often more stable than the key personality traits assumed to be predicting them. (more)

Amazingly, we know of another variable B that fits this bill, of being correlated with both personality A and political orientation C, and mediating their relation. In technical language, A and C are conditionally independent given B, so that P(C|AB) = P(C|B). This B is preferences for media genres!

Research has consistently demonstrated that political liberalism is predicted by the personality trait Openness to Experience and conservatism by trait Conscientiousness. … Increased preferences for Dark/Alternative and Aesthetic/Musical media genres, as well as decreased preferences for Communal/Popular media genres, mediated the association between Openness to Experience and liberalism. In contrast, greater preferences for Communal/Popular and Thrilling/Action genres, as well as lower preferences for Dark/Alternative and Aesthetic/Musical genres mediated the link between Conscientiousness and conservatism. (more)

So media genre preference is actually a plausible candidate for something closer to whatever causes personality and political orientation, and is caused in part by genes. This makes intuitive sense to me, because I personally feel more aware of, confident in, and comfortable with my media genre preferences than in my personality types or my political orientation.

These authors (Xiaowen Xu & Jordan Peterson) took a survey of 543 US people, and did a factor analysis on their media preferences. They found five factors. The strongest media factor is:

  • Cerebral/Nonfiction: e.g., educational, arts & humanities TV; nonfiction, academic, reference, current events, biography, science, medical books.

This factor doesn’t correlate with political orientation, but it does correlate with the openness personality factor, which has a subfactor of intellect, so this all makes sense.

The other four media factors split nicely into a 2×2 matrix along two dimensions. One dimension is “highbrow” vs. “lowbrow.” (Cerebral/Nonfiction is also “highbrow”.) The other dimension is personality/politics: two factors correlate with both liberals and those with open personalities, and two factors correlate with both conservatives and those with conscientious personalities. Here are these four media factors:


  • Open/Liberal: Aesthetic/Musical: e.g., world, jazz, opera, folk, classical, funk, gospel, blues, new age music; foreign film, poetry book.
  • Consc/Conser: Communal/Popular: e.g., daytime talk, reality, game, soaps, kids TV; made for TV movies; music TV; family films; pop music.


  • Open/Liberal: Dark/Alternative: e.g., horror, science fiction, fantasy, cult, erotic, animation, & independent film; punk, alt, & metal music; sketch comedy.
    Consc/Conser: Thrilling/Action: e.g., action, sports & spy TV; war, western, action film; computer & adventure book.

(My personal tastes favor Cerebral/Nonfiction first, then Dark/Alternative weak second.)

While personality and political variables come from factor analyses of survey answers to relatively abstract attitude and opinion question, media genre variables seem more closely related to concrete real-life choices that people make. And it makes sense to me that our genes (and culture) more directly cause our inclinations to take concrete actions, and that abstract attitudes and opinions result more indirectly, from our trying to rationalize those actions. So it makes sense to me that media genre preferences are closer to more direct genetic (and cultural) causality.

A few more results from the paper: Older people prefer Dark/Alternative less and Cerebral/Nonfiction more, while women prefer Communal/Popular more and Thrilling/Action less. These can explain the personality correlates of age and gender. Agreeableness is higher for those who like highbrow genres (except Aesthetic/Musical has no effect), and less for those who like lowbrow genres. Extraversion is higher for those who like conservative genres, though it doesn’t correlate directly politics directly because personality factors are highly correlated in this dataset. Neuroticism is less for those who like Cerebral/Nonfiction and Thrilling/Action.

GD Star Rating
Tagged as: , ,
Trackback URL:
  • Ilya Shpitser

    “we know of another variable B that fits this bill, of being correlated with both personality A and political orientation C, and mediating their relation.”

    But this is consistent with not only A C, but also A -> B -> C, and A <- B <- C.

    • RobinHanson

      That one fact is so consistent, but the rest of the available data is not so consistent.

      • Ilya Shpitser

        I guess I am not following your argument. You mentioned this independence as evidence, and you also mention something about media preferences (B) being more available to conscious introspection somehow than the other variables (A,C).

        This seems very weak evidence, though. The independence is consistent with lots of causal structures other than the one you named, and conscious introspection as a heuristic for telling causes from effect does not seem plausible to me. What is to prevent people from being more readily aware of consequences of unobserved causes in their mind. This blog talks a lot about this, actually.

      • RobinHanson

        There is also the result I quoted above, that “covariance between personality and political preferences is not causal, but due to a common, latent genetic factor that mutually influences both” That is not consistent with either of the other causal structures you suggested.

      • Ilya Shpitser

        Robin, what are your thoughts on this paper by Clark Glymour:

      • TGGP

        I don’t recall Hanson ever mentioning that book, but Bryan Caplan does and is more interested in philosophical arguments over the methods of social science. You might want to ask him.

      • Ilya Shpitser

        My question is about Glymour’s critique (which I think is relevant to the current discussion as well), not about The Bell Curve itself.

      • RobinHanson

        There are big coordination and network effects that entrench methods in academic fields. These make it hard for a small group to switch to new methods, even better methods. That said, I haven’t studied Glamor’s proposed method enough to have an opinion on if a switch would be better. I’m more interested in the quality of the institutions we use to make such choices, than what particular choice we should make in this case.

      • Ilya Shpitser

        That’s fair. But you were interested in the “causal facts of the matter” enough to make today’s post. Presumably, these facts of the matter are independent of current academic fashions.

        Glymour, being a causal inference person, (and a long list of others, me included) would say that factor analysis will not tell you anything about causality without carefully listed assumptions (this is because factor analysis just thinks about the joint distribution, and the joint distribution alone will not give you causal facts).

        If you ever find the time to read his paper worthwhile, I would be curious about your opinion.

      • Ilya Shpitser

        My question is about Glymour’s critique (which I think is relevant to the current discussion as well) not about The Bell Curve itself.

  • TGGP

    I’ve noticed that comment sections at the AV Club, which is probably less “inherently political” than econ blogs, appear to be more uniform in their politics. Although it could be that those with minority opinions (such as myself) mostly keep them to ourselves in such places.

  • Stephen Diamond

    So media genre preference is actually a plausible candidate for something closer to whatever causes personality and political orientation, and is caused in part by genes.

    Before the advent of media, there was no differentiation of personality and ideology?

    • RobinHanson

      Presumably our innate behavioral tendencies have always expressed themselves in our choices of free time activities from the menu available from at one’s time, place and class.

  • Chris Hallquist

    Weirdly, I also favor Cerebral/Nonfiction first, and Dark/Alternative a weak second.

    (I probably *view myself* as more of a sci-fi fan than I really am, truth be told I don’t read much fiction.)

    Edit: I wonder if this is a common set of preferences among Overcoming Bias readers. If you were too *stereotype* the kind of people who read blogs like Overcoming Bias, you’d probably stereotype them as sci-fi fans, but I wonder how many of your readers similarly don’t read much fiction.

    • Lord

      Birds of a feather, it would help if they laid out frequencies. I am stronger Dark/Alternative, sci-fi, but video, not text.

  • Pingback: Overcoming Bias : Seeking Super Factors

  • Lord

    While I have some stronger preferences, I also have others, so I always wonder how extreme these have to be to be significant.


    You can’t just pretend everyone agrees on a definition of “personality” where you can just strip personality of political preferences and media preferences without changing personality.

    This is all really messy and pretty much pointless since people have such vastly different definitions of the key terms in their heads (and it’s likely the authors also maintain different and conflicting definitions) that these kind of “research” papers just sow confusion rather than increase our knowledge of the world.

  • Stephen Diamond

    Reality and soaps are highbrow?

  • Habakkuk

    I’m politically conservative (in both a social and economic sense), and for me cerebral/nonfiction, aesthetic/musical (eg, I really liked that George Herbert poetry book Tyler Cowen recommended), and dark/alternative are my favorites. I get these are averages and what not, but I have trouble getting on board with things that directly contradict introspection. Is this a bias of mine?

  • steve

    I would be interested in seeing if these relationships have changed through time. I used to read a great deal of Sci-Fi, but not so much anymore. Mostly because it is hard to find new Sci-Fi that doesn’t push watermelon environmentalism.

  • Philip Goetz

    This is interesting, but what do they consider “liberal” and “conservative”? We have the strange situation in the U.S. that taking away the rights to own guns and raising taxes are both considered “liberal”, though both are reversions to medieval norms. I find it hard to entertain such systematic explanations of “liberal” and “conservative” when the positions themselves often seem arbitrary.