In 2012, Scott Alexander defended social sciences against the claim that they can’t figure anything out. He gave a long list of well-established findings across a variety of social science disciplines.

12 years later, how well did that list hold up?

I evaluated the list off the top of my head without doing any research,¹ so please don’t take this too seriously.

The text before the colon on each numbered item is Scott’s words; everything else is my words.

Anthropology

1. Humankind evolved in Africa, gradually settled the Old World, and crossed the Bering land bridge to America around 20,000 years ago: still good
2. Languages form large families like Indo-European that can be used to trace the history and development of different peoples: still good
3. People have an almost-miraculous language instinct that can for example turns a pidgin into a creole in the second generation: still good
4. There are various human universals, but people tend to overestimate how universal their own culture’s norms are: I don’t know
   • I don’t know how to interpret this claim. It’s well-established that there are various human universals and also that cultural norms vary a lot, but I have no idea whether it’s true that people tend to overestimate how universal their own culture’s norms are.
5. Any biological mental differences between groups are less important than previous believed and quickly overwhelmed by within-group differences: still good

Verdict: 4/5 or 5/5. Anthropology is doing well.

Economics

1. Prices in the marketplace are determined by supply and demand: still good
2. Capitalism leads to faster economic growth than the alternatives: still good
3. Unless you have very strange priorities, free trade is a good idea: still good
4. The gold standard is a bad idea: still good
5. Rent control decreases the quality and quantity of available housing: still good
6. Minimum wages increase unemployment: still good
   • There are some observational studies contradicting this and there’s debate as to whether minimum wage is currently high enough that raising it would cause noticeable unemployment but the basic principle of this claim is still true. (I found Bryan Caplan’s The Myopic Empiricism of the Minimum Wage persuasive.²) And anyway, the counter-arguments people would raise today are the same ones they would’ve raised in 2012.
7. Cutting taxes will not increase government revenue in normal conditions: still good

Verdict: 7/7. Economics is doing well.

Psychology

1. Personality is about 50% biologically determined: still good
2. People’s self-image is constructed on the spot and varies widely depending on the situation: still good
3. People exhibit various cognitive biases that deviate systematically from rational thought: partially replicated
   • Some cognitive biases have replicated (base rate neglect, status quo bias, loss aversion, hindsight bias), others haven’t (priming, nudge theory, implicit bias). From scanning thru Wikipedia’s list of cognitive biases, looks like 70–90% have replicated. I will give this one half credit for a partial replication.
4. IQ correlates to all kinds of important life outcomes: still good
5. Many mental disorders correspond to disordered brain chemistry and can be partly treated chemically: still good
6. People lack privileged access to their own mental processes: still good
7. Cognitive behavioral therapy does better than placebo in treating mental disorders; Freudian therapy does not. still good (?)
   • IIRC some RCTs have found good benefits to the modern version of Freudian therapy but I don’t think it’s well-established.
8. Babies are not a blank slate but have various built-in behavioral patterns; they develop new mental abilities in an orderly fashion: still good
9. Animals react to reward and punishment in extremely predictable, almost mathematical ways: still good
10. Strong relationships and driving purpose are very important to happiness; material goods less so after a certain point: unclear
    • As I understand it, the happiness research does consistently support this claim, but it’s not clear that we are doing a good job of measuring happiness.

Verdict: 8.5/10 or 9.5/10. Psychology is doing fairly well. I found this one surprising considering how much stuff in psychology has failed to replicate, but Scott did a good job of identifying the claims that would hold up (a much better job than my college psychology textbook did).

Sociology

1. People are racist as heck: failed to replicate
   • I believe Scott is talking about Implicit Association Tests, which do consistently show implicit associations but don’t reliably predict behavior. Some other things like reume name bias failed to replicate. There’s some version of this hypothesis (the “all bad racial outcomes are caused by racism” hypothesis) that’s basically unfalsifiable and therefore hasn’t been disproven but that’s not a point in its favor.
2. No, really, they’re really racist: failed to replicate
3. Even the ones who don’t think they are: failed to replicate
4. Even the ones who swear up and down that they’re not racist and donate to the NAACP: failed to replicate
5. There are major disparities in the income levels of social status of various ethnic groups: still good
6. Discrimination explains a lot of this, sometimes in surprising ways: was never established in the first place
   • I don’t entirely understand what Scott meant by this but as far as I know, it was never established and the evidence almost entirely contradicts it.
7. Most social problems are closely correlated with one another. Scandinavia has the fewest social problems of developed countries, and the US has the most: still good
8. Poor people and uneducated people tend to suffer more social problems and commit more crimes: still good
9. Religious people tend to be happier and better-adjusted than others: still good
10. Social class is a big deal, even in societies that make a big pretense of being classless: still good

Verdict: 5/10. Sociology are you ok? (To be fair, this list is a bit skewed because Scott spent four bullet points reiterating the same claim that failed to replicate.)

Epidemiology

1. Smoking causes cancer and many other problems: still good
2. Alcohol causes liver disease and many other problems: still good
3. Bad diets (in some sense of the word) cause heart disease, Type II diabetes, and many other problems: still good
4. Two zillion other correlations between risk factors and diseases: still good
5. Exercise is really good for you: still good
6. Vaccines are extremely effective at controlling infectious diseases: still good
7. And they don’t cause autism: still good
8. Stress causes or exacerbates many diseases: still good
9. Poor people suffer from more diseases, even in ways that are not directly linked to them not being able to afford medical care: still good
10. Many diseases seem to be part genetic and part “other factors”, including mental diseases: still good
11. Lifestyle changes can decrease your chance of getting mental diseases: still good(ish)
    • As I understand, this is broadly supported but there’s some contradictory evidence on e.g. whether exercise helps with depression.
12. Multivitamins don’t work: mixed evidence
    • There are some RCTs on both sides (for a quick summary, see here under the heading “Most unusual claim”). Multivitamins are cheap so IMO they pass a cost-benefit analysis. I am giving zero credit on this one because, while it might be true, it’s not definitely true, and this was presented as a list of definitely-true findings.
13. Low-dose aspirin (probably) helps prevent cancer: still good
    • The evidence on this is weak but Scott did say “probably” so I’ll give this one full credit.

Verdict: 11/13 or 12/13. Epidemiology is doing pretty well.

Notes