Did you know that physically strong men are more likely to be socialist? Shocking, I know. But there’s no other conclusion that can be reached after reading all the articles on my social feeds that claim the opposite.
The Blaze: Study – Physically weak men more likely to be socialist, strong men more likely to be capitalist, MrCTV: New Study Finds Wimpy Guys Are More Likely To Be Socialists, same from The Daily Wire, and the hilarious counter-narrative framing by Vice: Gym Bros More Likely to be Right-Wing Assholes, Science Confirms.
The Daily Wire article concludes with a challenge:
Anyone with two eyes would be hard-pressed to refute such findings.
Well, I have two eyes and I don’t mind being hard-pressed — just today I did bench press at the gym while remaining liberaltarian. So, challenge accepted.
Putanumonit has a recurring Defense Against the Dark Arts section, helping my readers escape unscathed from an onslaught of gruesome statistics and preposterous science reporting. I most come across left-leaning BS (Bad Statistics), because the left creates a higher demand for “science proves we’re right” smugpieces. Or perhaps because science is a liberal conspiracy. Either way, I was excited to see those sneaky scientists giving conservatives a break for once.
We’ll take a closer look at The Blaze article, which is pretty typical of the genre, and the study itself by Michael Price et al of Brunel University in London. “Brunel” sounds like the love child of American Ivies Brown and Cornell, although it’s hard to imagine an Ivy League university publishing something so, unprogressive.
Anyway, here’s The Blaze:
Brunel University academics studied 171 men aged 18 – 40…
That sounds like a diverse selection of men. But here’s another way to describe it: 171 men, mean age 21, 98% of who are students and one of whom is a 40-year-old dude who needed the £5. Doesn’t sound quite like a “representative sample” now, does it?
I know the composition of the subject group because Dr. Price and his associates put their entire dataset, for free, online.
There’s going to be a lot of snark about this research in the paragraphs ahead, and I want to make one thing clear. While I think that the researchers may be guilty of overhype and statistical carelessness, I cast no aspersions on their integrity. Publishing your data is the gold standard in honest research, even as it leaves you vulnerable to criticism by dickish bloggers with no skin in the game.
One last note on the authors: I love the nominative determinism of two gentlemen named “Price” and “Pound” studying economic attitudes, and co-author Jennifer Sheehy-Skeffington having an awesome name all around.
Here’s how Price, Pound, Sheehy-Skeffington and final co-author James Sidnaius explain their hypothesis:
“We believe that this link between perceived formidability and egalitarianism could be explained in a number of ways,” Price said in the report on Brunel University’s website. “It could be the result of men calibrating their egalitarianism to their own formidability. It could be the case that less egalitarian men strive harder to become muscular. Or there could be a third variable at play affecting both egalitarianism and muscularity. […]
This is about our Stone Age brains. Our minds evolved in environments where strength was a big determinant of success. If you find yourself in a body not threatened by other males, if you feel you can win competitions for status, then maybe you start thinking inequality is pretty good.”
The problem with evolutionary psychology is that how plausible a story sounds says very little about whether the story is true because these stories are very easy to come up with. In fact, I can come up with a better explanation that doesn’t rely on any extrapolation of the ancestral environment at all.
I don’t know if 21-year-old guys spend a lot of time thinking about the capital gains tax, but they certainly spend a lot of time thinking about hooking up with 21-year-old girls. Male political psychology aside, the females of the “early twenties university student in a cosmopolitan city” species show a strong empirical tendency to lean socialist.
Trust me, I spent half a decade dating educated young women in New York.
If you’re a young guy on campus, there are two strategies that increase your chances of getting laid: a fit body, or a Bernie Sanders/Jeremy Corbyn sticker on your MacBook. If you don’t possess the former, you may be inclined to profess your support for the latter. This plausible-sounding story refutes any direct link between strength and socialism on an evolutionary basis.
The questionable just-so hypotheses, the irrelevant sample of British 21-year-olds, and the general brainless way in which this study is reported by partisan media all raise more red flags than the communist party on Mao’s birthday. But we can’t dismiss the study just because it’s dressed in a politicized clown suit, we need to look at the numbers.
Here are the numbers:
Our study tested both models with 171 men, using two sociopolitical egalitarianism measures: social dominance orientation and support for redistribution. Predictors included bodily formidability and attractiveness and four facial measures (attractiveness, dominance, masculinity, and width-to-height ratio). We also controlled for time spent lifting weights, and experimentally manipulated self-perceived formidability…
Two models, two outcome variables, six predictors plus a manipulation, several controls, and interactions! This is not a garden of forking paths, this is the fricking Cretan Labyrinth with the Minotaur in the middle eating your p-values. There must be at least twenty predictor-outcome pairs that are implicitly and explicitly tested, any of which if confirmed at p<0.05 would be reported credulously as “weak men are socialist”.
In fact, if you test 20 hypothesis you don’t need to even bother collecting data to get a p-value of 0.05, the number of hypotheses virtually guarantees it.
Of course, it doesn’t cross the authors’ mind to correct their p-values for the multiplicity of hypotheses. Multiplicity correction isn’t some esoteric question of research methodology. If you test multiple hypotheses but report p-values as if you tested a single one, your math is wrong and your conclusions are wrong.
The main finding of the study is that “bodily formidability” affects “social dominance orientation” (p=.04) but the very same table shows a non-significant effect of formidability on “support for redistribution” (p=0.4). No matter how you spin it, if you test 2 hypotheses at the same damn time you have to multiply the p-values by 2. This minimal correction would make the main finding non-significant at p=0.08. If you multiply all the p-values by 20, as you really should given the multiplicity of the study, they all end up equal to get the f%#@ outta here.
Now this all just shows that there’s no reason whatsoever to conclude that strong men are less likely to be socialist. But I promised that I will not just refute the article’s claim, but demonstrate its opposite: that stronger men are in fact more prone to socialism. I’ll accomplish this using nothing but the original study’s dataset.
Here we go.
Are stronger men more likely to be socialist?
J. Falkovich, M. Price et al, May 2017
Our arbitrary extrapolation of political concepts across thousands of years predicts that political attitudes today are determined by strategies that maximized resource acquisition in the ancestral environment. For a large chunk of human developmental history, from about 2-1 million years ago, homo erectus used persistence hunting in tracking prey for long periods of time on the savannah (Frey, 2002, Liebenberg, 2008). Persistence hunting requires a physique optimized for endurance running, known in the technical literature as small, skinny dudes (Aloia et al, 1978).
In this environment, men with large muscular physiques could not compete for resources by themselves. They relied on tribal norms of economic redistribution, and subsequently evolved to support those norms.
Previous research on body types and politics (Petersen et al, 2013) relied on bicep circumference as a measure of strength. That methodology proved flawed because arm circumference is strongly dependent on body fat percentage (Caines, 2013), and isn’t a good measure of strength. For our study, we used grip power as the main measure of upper body strength in men while controlling for bicep circumference, using a methodology outlined by Tsatsouline, 2015.
Here are the results of regressing support for redistribution (as measured on a 10-item scale) on the normalized z-scores of grip strength and bicep circumference:
And here’s the same result in chart form, for our readers who are more visual learners:
Perhaps we tried a bunch of other regressions, perhaps we didn’t, you’ll never know. What’s important is that grip strength in men leads to increased support for economic redistribution with a p-value of mother-effing 0.0086.
Eat your heart out, Bernie haters.
I didn’t edit any of the data. I didn’t throw out any data points as “outliers”. I used the same technique (linear regression) with the same data set used by the researchers, and I reached the opposite conclusion with a p-value that’s 5 times more significant. It took me less than two hours.
That’s the power of multiplicity — it can “prove” any hypothesis you throw at it, and the opposite of that hypothesis as well.
My goal isn’t to shame the researchers, who ran a solid experiment based on a hypothesis they believed in and published the raw data. I don’t even think the partisan magazines breathlessly reprinting this stuff should be shamed, because shaming them obviously doesn’t help.
My goal is to protect you from BS.
If an online article makes bombastic, politically-charged claims that are “based on a study”, it’s BS. When you hear a just-so story of evolutionary psychology that pretends a prehistoric tribe is just like a bunch of 21-year-old students, it’s BS. When a research paper measures twenty things and doesn’t correct for multiplicity, it’s BS.
And it’s not hard to smell this BS from miles.
My current position on male strength and socialism is the same as my position on ovulation and republicanism, or power posing and anything else: zero effect until proven otherwise.
But still, if push comes to shove, I’d rather shove a socialist.
14 thoughts on “Strong men are socialist, reports study that previously reported the opposite”
“In fact, if you test 20 hypothes[es] you don’t need to even bother collecting data to get a p-value of 0.05, the number of hypotheses virtually guarantees it.”
That’s a bit of an exaggeration, isn’t it? 1 – 0.95^20 is quite a bit less than 1. See Gelman, Hill and Yajima (2012) on this (and the downsides of Bonferroni corrections, plus alternatives):
“The classical perspective would have us worry if we performed 20 tests at a .05 significance level that there is a 64% chance that at least one of these will yield a statistically significant result inappropriately.”
Click to access 3fd2f43f1c3210fdce08968fd18f76478710.pdf
Nope! Sheehy is an Irish name (anglicised from Mac Síthigh), and the original Sheehy-Skeffingtons were (the man) on the Irish side in 1916 (though it’s complicated) and (the woman) Ireland’s best-known suffragette. Recently a descendent of theirs won a case against her employer, an Irish university, she’s well-known as well. The author on this paper will be another descendent.
LikeLiked by 1 person
But Ireland was still British in 1916.
What would happen if I try 20 combinations and they all give me a p value less than 0.05? Do I still need to apply a correction?
I’m guessing it depends if the conclusion is “at least N of these combinations are true”. If you want N=20, you’ll still need a much lower individual probability to be confident that none of them are by chance.
This depends on the relationship between the hypotheses, because they’re not just by themselves in a vacuum. If the hypotheses are pointing in the same general direction (i.e. you’d expect them to correlate), then getting a p-value for each is very strong evidence that most of them are true. It would be even better to look at it as getting very strong confirmation for the “general direction”.
For example, in this study, all the hypotheses (facial masculinity, strength, male figure…) pointed in the general direction of “formidable masculinity” and that combined hypothesis was not really supported by squeezing a p=0.04 for one of the variables.
If your hypotheses were negatively correlated (or independent), you probably messed something up. In any case, you would need a lot more than just the p-values to get a good idea of what’s going on. No good scientist will just look at 20 p-values, they would be doing multilevel and multiparameter modeling (ideally of the Bayesian kind) to get a comprehensive picture.
Subtle tense error: “So far I covered” (should be “I’ve”)