THE JOYFUL REDUCTION OF UNCERTAINTY

Highlights From The Comments On Survey Harassment Rates

[Content warning: harassment. This discusses the comments to SSC Survey Results: Sexual Harassment Levels By Field]

brmic writes:

Thank you for posting this and the data file. FWIW, I tried to reproduce the results and couldn’t reproduce the correlations between female victimization, male victimization and male perpetration. fem vic vs. male vic is 0.65, same as yours. fem vic vs. male perp is 0.01 for me, and male vic vs. male perp is 0.21 for me. Everything else more or less checks out.

As a reviewer, I’d say the combination score is not convincing, especially since it ignores all considerations of different male to female ratios in the various industries.
Also, if you have two measures with r = 0.8, Fig 6 is not a good idea IMHO. It’s probably just noise. (Also, it should be a dotplot centered around 1, because the relevant info is distance from 1:1 ratio.)
Instead, I’d focus on the correlation between female victimization at work and female victimization outside work of 0.65 (for me) and the same for males at 0.59, which also leads to the conclusion that there’s a strong ‘people in fields’ effect, without having to go through the combination score. If you’re so inclined, you might then do the at-work by outside-work ratios and end up a kind of cross-validation set, where you can see whether the bad fields for women are bad for men as well. Of course, once you then consider sex ratios per field. it’s story time all over again. Still, e.g. men report similar levels of out of work victimization in computers (20%) and Health Care (24%), but at work victimization of 4% and 12% respectively, which strongly suggests that Health Care is worse.

Their code is available here. Thanks for doing the work to try to replicate my results. I’ve removed the non-confirmed correlations from my post until I can figure out what’s going on with them. I agree that Figure 6 was barely worth it, which is why I tried to make Figure 4 (the unadjusted version) the center of my thesis.

Chris quotes a TIME article that argues that predominantly-male communities generally have lower harassment rates than predominantly-female communities:

Given the epidemic of campus rape, teenage girls and their parents are justifiably concerned about safety, just as teenage boys and their parents are worried about false accusations. What does any of that have to do with gender ratios? Well, there have been multiple studies showing a correlation between gender ratios and rates of sexual assault. As counterintuitive as it may sound, elevated rates of sexual assault are a predictable feature of communities with oversupplies of women, according to studies by sociologists Nigel Barber and Robert O’Brien.

The opposite is true of communities with oversupplies of men. Columbia University economics professor Lena Edlund investigated the impact of lopsided sex ratios in China, where young men now outnumber women by 20% due to sex selection, abortion, female infanticide, and other outgrowths of China’s old “One Child” policy. Edlund and her co-authors discovered that although overall crime rates went up in China as the gender ratio skewed more male—not surprising given that men are more prone to criminality—there was a precipitous decline in rape. It seems that men treat women better, and protect them more, when women are in shorter supply.

Can I prove beyond all doubt that Edlund’s and Barber’s findings also apply to college campuses—i.e. that rape is less common at schools that are at least half male? No, because the available data on campus rape tends to reveal as much about how forthright colleges are in handling sexual assaults—and how comfortable women feel reporting them—as it does about the actual frequency of assaults on a particular campus.

That said, I was intrigued by a recent Washington Post story on the topic. The article ranked 27 top colleges by their sexual assault rates, and I couldn’t help but notice which college had the lowest rate.

It was CalTech, a school that is 59% male.

But some commenters in the subreddit bring up contrary evidence. DinoInNameOnly links to a Pew poll finding that women in predominantly-male workplaces report more harassment. And Hepatitis Andronicus brings up Alaska, a heavily male-skewed state considered to be in “a sexual assault state of disaster”.

Rolaran argues:

I feel like a person who has had bad experiences in the tech industry is less likely to have continued reading a blog that argues (among other things) that problems in that field are overemphasized, and a person who has had nothing but good experiences in tech is more likely to have begun hanging around a blog where they felt like the author was sticking up for them. This is useful data for determining the experiences, preferences and demographics of SSC readers, but I would be considerably surprised if that could be generalized without some serious legwork.

I did consider this, but first of all, only the tiniest percent of my posts are on tech, and I think I might have only had one previous post that could by any stretch of the imagination have been considered about harassment in tech. Second, it’s not clear which direction this should skew things – some would argue that guilty people are more likely to push the “we have nothing to apologize for” line, and innocent people are more willing to critically self-examine their (or their tribe’s) problems. Third, I’m not sure this would really skew self-reported female victimization, self-reported male victimization, and self-reported male perpetration all in the same way.

Rachael writes:

I agree with most of your conclusions, but there’s still a major flaw in the survey methodology. It asked what industry people currently work in, and whether they’ve *ever* been harassed at work. Plenty of people have switched industries over the course of their career, and nearly everyone worked in retail or food service or similar as a teenager or young adult. So we have no idea how much of the reported harassment took place in industries other than the ones people currently work in, and probably skewed towards the industries that hire teenagers.

This is not only a good point, it’s a good point that would have been easy for me to fix if I had thought about it. Sorry.

Leah Velleman writes:

I wonder whether something gets lost here when we conflate the startup world (stereotypically an unsupervised bunch of pushy, risk-taking young single people) with — and I use this term lovingly — dinosaur tech companies (stereotypically a bunch of mild-mannered old married people, with a strong HR department looking over things).

The cultural gap between startups and dinosaur tech seems much bigger than, say, the gap between independent and chain restaurants, or small and large universities. If that’s true, then that’s an argument for treating startups and dinosaur tech as separate sectors. (And while some companies might be hard to sort into one category or the other, that’s true for any set of categories. “Is this job in econ or finance?” and “Is this job in business or law?” will turn up ambiguous cases too, but that’s not a reason to ditch those distinctions.)

I’m pushing this question because I have a hypothesis, which is that startups really do have a harassment problem — but that the people speaking up about it, and the news outlets quoting them, have misrepresented this as a problem across all of tech, ignoring the fact that most tech workers are at boring old companies with low harassment rates. I suspect this misrepresentation has happened partly because startup folks like to represent themselves as The Real Tech Industry, The One That Counts, and partly because companies tend to transition from “startup” to “dinosaur” over time and news outlets can’t be bothered to keep track.

Riceowlguy on the subreddit writes

I think it would be worthwhile to try and distinguish between “people whose job is actually writing code or keeping IT systems running smoothly” versus “people who have jobs at tech companies but whose work is not technical”. My recollection is that a lot of the anecdotal horror stories about harassment in Silicon Valley involve VCs, sales/marketing people, upper management, etc.

One question is whether people in that latter category (if any of them read SSC at all!) answered “computers” or “business” for their profession.

Doug writes:

Maybe men in these groups are just less sexually aggressive and more introverted. It’d be interesting to aggregate # of lifetime sexual partners by industry and correlate with levels of sexual harassment. Not that I’ve ever been involved in sexual harassment on either side of the table… But my guess is that most cases don’t look like Alice walking over to Bob’s desk and grabbing his crotch out of the blue. Most situations probably escalate from some fun, flirty, or social interaction that one party takes to far. Programmers don’t know how to flirt and avoid socializing, so they never get into these situations. Existing literature on sexual assault shows that it disproportionately occurs in environments of “revelry” and “carousing”. I don’t see any reason workplace harassment would deviate from this pattern. Pharma reps get drunk with their coworkers a hell of a lot more than mathematicians.

I’d been thinking of “introversion” as the default hypothesis, not requiring explanation in the same way as the others, but I realize I might have forgotten to mention it at all. Sorry.

But Terran has analyzed (original comment, small correction) whether these results are explained by the Big 5 personality traits of people in each field. He finds they generally are not – profession remains important even after looking at traits (including introversion). Stezinec on the subreddit tries the same thing – their analysis finds that, after adjusting for traits, the only significant result is that computer-related occupations have lower harassment – but guessing this is just a sample size issue and the computer industry was the only one with a good sample size.

In case you are interested, here are some other factors that affected sexual harassment in the survey. I’m not putting error/significance bars on because that would be too much work – but if I have to have a nice-sounding explanation, it’s because I want to discourage people from using these to say “Your ingroup is more harassy than my ingroup, proving you’re bad”. The best one can do without error bars is disconfirm statements like that – which I think is generally a more prosocial activity. This is also part of why I am making this section less prominent than it might otherwise be. I think this especially important to remember given that groups will be penalized for honesty – ie people who are more likely to confess to sexual harassment, or people who are more willing to admit that certain behavior crosses the line.

All analyses were limited to cisgender subjects except where stated otherwise. All analyses of women have a sample size around 600, all of men around 7000. Each subgroup has a sample size of at least 100 except where stated otherwise. Note that each graphic combines information on victimization and offending by one gender; offending results are often very low and slight variations should be taken with a grain of salt.

Rates by relationship status (male, female)
Rates by sexual orientation (male, female)
Rates by transgender status (male, female)
Rates by asexual vs. sexual (male, female)
Rates by political affilitation (male) [not enough data for full female analysis, but binary left/right graph here]
Rates by feminist self-identification (male, female)
Rates by social class of family of origin (male, female)

If you want to analyze this further or double-check any of my results, you can download the raw data here.

This entry was posted in Uncategorized and tagged , , . Bookmark the permalink.

73 Responses to Highlights From The Comments On Survey Harassment Rates

  1. tailcalled says:

    Did you make a mistake in your “rates by transgender status” graphs? The trans groups seem to have the exact some values.

  2. fion says:

    In your “rates by relationship style” male graph I think your key is wrong. The options are “married”, “relationship”, “single” instead of “monogamous”, “polyamorous”, “no preference”.

    • The Element of Surprise says:

      Not only the key, the data is identical to the “relationship status” data, so probably the graph uses the wrong data column.

  3. Murphy says:

    Re:

    >whether they’ve *ever* been harassed at work

    On the survey I work in tech/research and ticked yes because as a teen I worked in a restaurant and hen parties would occasionally get a bit grabby.

    Never experienced any grabby people in tech or research in all the years since.

    • wintermute92 says:

      This seems like a pretty serious issue for the whole dataset. An awful lot of people have worked retail or service jobs, and service work is absolutely infamous for systemic harrassment. And in at least some branches of STEM (e.g. environmental engineering), it’s pretty common to have worked some sort of manual labor earlier in life.

      Thinking about it, I have several friends who would have been ‘yes’ because of their time on construction sites or in restaurants, but have mentioned that tech is a big improvement on that front.

      • AnarchyDice says:

        I wonder if this also works in healthcare too, were doctors all had to start as residents and interns?

        • wintermute92 says:

          That’s a really interesting point. If we just want to handle service workers turned engineers, it’s easy enough to ask about “in your current career”. But staged careers like this are another matter – if 0% of doctors are harassed, but 100% of doctors start as residents and 100% of residents are harassed, we’d ideally want a question which can distinguish that fact.

          I’m not sure what you’d have to ask – thoughts?

          • AnarchyDice says:

            Lots of industries have different structures so to get some apples to apples, it can’t be title specific but maybe it could be seniority specific by years of experience?
            Maybe ask what year or range of years did the harassment happen with a set of toggleable ranges like 0-5, 5-10, 10-15, etc. That would be combined with a question about how long they have worked in the industry so there is some sort of denominator for each of those ranges. Then you could pre-register predictions about various industries (so you can’t be accused of fitting your data post hoc to say “of course the mining industry has this disparity, all those new miners getting their certifications are so vulnerable”) like “the medical industry and media are places that I think have significant power imbalances for interns/residents so they would have a more disproportionate time disparity in harassment rates”.

  4. wintermute92 says:

    which college had the lowest rate. It was CalTech, a school that is 59% male.

    This seems pretty far from compelling. Caltech only has 1,000 undergrads, so you can pretty easily know all of your classmates, at least in passing. It’s far enough from a “party school” that I suspect students simply spend less non-academic time together than most colleges, and it has more alcohol-abstinent students than the national average. It doesn’t have fraternities or sororities at all.

    The obvious comparison would be RPI, a 68% male tech-focused college with far more typical demographics. It’s sadly not on that list, but I’d take a modest bet that it has more assault than Caltech.

    • herculesorion says:

      The issue is that people seem to have this feeling that once you hit 50.05% male you will immediately devolve into a hell of ass-slapping wolf-whistling transphobia; and that tech is stuck in a vicious cycle of “harassment drives away nonwhite nonstraight nonmen, which means there’s more white straight men, which makes the harassment worse, which drives away nonwhite nonstraight nonmen even more, and so on”.

      The fact that a school specifically focused on tech and very definitely more-than-half male is the least harassing school in the survey is a useful reply to that feeling.

      “[RPI is] sadly not on that list, but I’d take a modest bet that it has more assault than Caltech.”

      which, there it is. No actual data, but a feeling, a really strong feeling.

      • wintermute92 says:

        which, there it is. No actual data, but a feeling, a really strong feeling.

        I recognize the idea you mean, but I think you’re misunderstanding me. I don’t mean “I think that because it is heavily male, RPI would have more harassment than Caltech”. I mean “I think that Caltech is exceptionally low on harassment because it is unusual in many ways, and gender makeup is not the most distinctive of those factors”.

        The fact that a school specifically focused on tech and very definitely more-than-half male is the least harassing school in the survey is a useful reply to that feeling.

        This is a solid point I agree with. I think Caltech is nonstandard in basically all relevant ways, but it’s still solid as a proof of existence – it’s definitely possible to create a heavily-male tech space with very low harassment. That alone is worth pointing out.

    • Clegg says:

      I was a grad student at Caltech when they did the survey, and read the report carefully at the time. The tables are here: https://s3-us-west-1.amazonaws.com/www-prod-storage.cloud.caltech.edu/AAU_Campus_Climate_Survey_on_Sexual_Assault_and_Sexual_Misconduct-Caltech_Final_Tables.pdf

      On thing I noticed was that while the rates of sexual assault are about half the average of other colleges (e.g. 5.7% female undergraduates ever experienced attempted or completed penetration by force or incapacitation, Table 3.1a on p. 16, vs. around 12% nationally), rates of thinking sexual assault is a very/extremely serious problem are 1/10th the national average (under 2%, Table 1.3 on p. 11, vs. around 20% nationally). This leads to the surprising implication that there are more female undergrads at Caltech who have been raped there than who think sexual assault is a serious problem there.

      One possible explanation is that people have a common threshold for “sexual assault is a very serious problem here”, like a double-digit rape victimization rate, and since Caltech undergrads are all very good at Bayesian reasoning the victims correctly reason that their own experience is atypical and not a sign of a larger problem.

      Another, objectively similar but normatively different explanation is that there’s a widespread belief that “it never happens here”, no readily accessible process for describing it when it does happen, and a resulting feeling of victims that they are alone, without recourse, and perhaps somehow at fault.

      My impression is that the second is the more likely, but this is mostly subjective.

      This suggests a possible rationale for the media focus on sexual harassment in tech. If tech, now or in the recent past, had relatively high rates of sexual harassment compared to public perceptions, then news stories focusing on tech might have offered the best bang for the buck in terms of improved public understanding. This is the exact rationale that Scott gives in the closing sentence here: http://slatestarcodex.com/2013/06/09/all-debates-are-bravery-debates/

      • Aapje says:

        I think that you are giving the media way, way too much credit. I don’t believe they are capable of that level of reasoning.

        • Clegg says:

          The required level of reasoning is only, “Here’s something that happens that nobody seems to talk about, while people do discuss similar things (i.e. harassment in other industries); let’s talk about it,” which doesn’t seem like too much to attribute to “the media.”

          This is the pure motive which undergirds the analysis, even of other motives (e.g. as Scott describes in Untitled) carry the story.

          • pansnarrans says:

            I took it that Aapje was referring to the closing paragraph, where you suggest that the media focus on tech because they think that’s the best way to raise consciousness of harassment. Newspapers and so on don’t really seem that conscientious, so it seems more likely that tech stood out as being male-dominated and the media focused on the field in an attempt to establish a narrative to explain that.

            Also, you make a really good point about how different groups might assess a statement like “sexual assault is a very serious problem here”, and this should be taken into account. It reminds me of the idea that people in low-corruption countries tend to report high levels of corruption when asked, because any amount of observed corruption shocks them.

          • Aapje says:

            Pansnarrans is right. I don’t believe that the media has the capability to actually figure out how the rates of sexual harassment of an industry compare to public perception. They would be dependent on academic or think tank researchers doing that comparison and I’m not aware of this having been done.

            From my perspective, there is no evidence that the media have been ambitious enough to tackle this nor do I find it a remotely reasonable explanation for them focusing on Tech/IT.

            Here we have WaPo comparing the level of sexual harassment charges between industries. We see that the number of charges in ‘Information’ is 1/5th of the number in ‘Accomodation and food services’, 1/4th of the number in manufacturing, 1/4th of the number in healthcare, 1/2 of the number in public administration, etc.

            Here we already see a low level of sophistication on the part of WaPo, as they seem unaware that the percentage of female workers differs per industry and/or that this matters. They also don’t consider cultural factors that may cause the ratio of actual harassment to charges to differ between industries. Miners are probably not people who are the least likely to sexually harass (female) co-workers.

            This issue is probably causing them to make IT look less prone to sexual harassment than it is, so if the goal was to increase the perception of the level of sexual harassment in IT, then it seems unlikely that they base it on their own assessment of the level of sexual harassment between industries.

            If we take their shitty reporting as representative of what they believe the actual level of sexual harassment for the various industries is, then to make the argument that sexual harassment in ‘Information’ needs most attention because the perception is wrong, people would already have to believe that the level of harassment in ‘Accomodation and food services’ is more than 5 times the level in ‘Information,’ that the level in manufacturing is more than 4 times higher, that the level in public administration is more than twice as high, etc, etc for all 11 industries with a higher rate of recorded harassment than ‘Information.’

            Basically, they would have to believe that the populace has an immensely positive view of IT as a safe haven for women, where nerds are nicer to women than all those other groups, including public administrators. I have seen no indication that anyone in the regular media has this opinion, to put it mildly.

          • albatross11 says:

            It’s worth noting that the Washington Post is on the high end of newspapers in terms of quality and resources. If they don’t do a good job of analyzing this, it seems unlikely that many other media outlets will.

  5. Assistant Village Idiot says:

    The Alaska numbers are likely to be skewed by high incest and assault rates in the native population, especially the villages.

    I think there are too many variables to yield much at present. Each employment category has a wide range of types of environments within it, for example. I think it’s all worth looking at, but only large differences, which can in turne be examined more closely, are going to tell us much.

    • Alexander Turok says:

      Yep, from the source he links:

      “Native Alaskans make up 61 percent of rape victims in the state, making Alaska Native women 9.7 times more likely than other Alaskans to be victims. “

  6. bean says:

    But some commenters in the subreddit bring up contrary evidence. DinoInNameOnly links to a Pew poll finding that women in predominantly-male workplaces report more harassment. And Hepatitis Andronicus brings up Alaska, a heavily male-skewed state considered to be in “a sexual assault state of disaster”.

    I was going to suggest that this was due to the difference between rape and sexual harassment, but I’m not sure that’s the case. I went to a school that was even more male than CalTech, and I’d give good odds that our harassment/assault numbers were pretty good. But, as you’ve probably guessed, I went to an engineering school, and on second thought, that’s the really obvious explanation for everything we’re seeing here. STEM people are generally low-harassment, but I could see other male-dominant workplaces (military, trades) being high-harassment. Ditto with the population of Alaska.

    On the other hand, I could also see the rape/harassment dichotomy being important. STEM generally has an excess of men with poor social skills. Their attempts to interact with women may be unintentional harassment, but they seem less likely than the general population to cross the line into assault/rape.

    • dndnrsn says:

      I don’t know about the trades, or about Alaska, but isn’t sexual assault/harassment in the military frequently explained with reference to the explicit hierarchy? At least anecdotally, a lot of the assault/harassment cases in the military involve people coercing, propositioning, etc subordinates.

      • bean says:

        Even if that’s the case, and the hierarchy raises rates somewhat, I’d bet you’d find a big gap in terms of harassment rate between, say, infantry and intelligence. The geeks and jocks are the same, in or out of uniform.

  7. userfriendlyyy says:

    Bisexual men most likely to be victim and victimizer. Homosexual men 2nd most likely to be victim and least likely to be victimizer. Now I’m wondering how much people are classifying mocking of sexual orientation as sexual harassment. Suggestions for next time:
    1.a How many people have you been harassed by?
    (for each person)
    1. Was the person who harassed you of the same gender or a different gender?
    2. Check all that apply; (You could add ‘was this experience in your current field? if not what field:______’ here too.)
    a. This person made unwanted advances privately.
    b. ” ” publicly.
    c. This person made inappropriate comments about my personal life to shame me.
    d. This person went beyond words and did something physical.

    Because many people might not consider 2a to be sexual harassment in all situations I’d recommend not showing part 2 until someone enters a number for 1a. Then block changes to 1a so people don’t inflate, or at the very least monitor if they do go back to change it.

  8. bbeck310 says:

    The cultural gap between startups and dinosaur tech seems much bigger than, say, the gap between independent and chain restaurants, or small and large universities. If that’s true, then that’s an argument for treating startups and dinosaur tech as separate sectors.

    Interestingly, this applies to law too, but probably in reverse. There’s a huge cultural gap between BigLaw, small law firms/solo practitioners, and the public sector, but in my experience–4 years in a BigLaw patent boutique (i.e., 150+ lawyers in NYC, all in patent law), 4 years in a BigLaw general practice firm (1,000+ lawyers nationally), and 2 years in a true patent boutique (fewer than 10 lawyers, only 1 equity partner)–the middle environments were more likely to have sex scandals. I’m pretty sure it’s a matter of power. BigLaw equity partners command millions of dollars of business, and associates make enough money that they put a lot at risk by complaining and risking firing.

    The national general practice firm was huge enough that they had a real HR department and well-established policies for handling problems, though there were still powerful enough attorneys to cause problems.

    The small firm hasn’t had that kind of problem; the office is too small, the associates don’t have big enough golden handcuffs, and apart from the managing partner, none of the non-equity partners have the clout to protect themselves if they tried anything.

    The BigLaw patent firm had the most issues–fewer established policies, only 1 real full-time HR employee, and more married partner couples than was probably healthy.

    Government law offices have some bad reputations, though I don’t know of them firsthand–judges have a ton of power with few checks, and we’ve heard horror stories about some judges’ treatment of law clerks (the Judge Kozinski matter was huge last year).

    • Clegg says:

      Maybe sector doesn’t matter much at all, and it’s firm structure that’s the more important determinant of harassment rates. This seems obvious as I write it.

      It’s not obvious what about firm structure should matter, though. From bargaining theory, it should perhaps involve men in positions where the firm needs him more than he needs the firm, and women in opposite positions where the firm can find another woman easier than she can find another job.

  9. userfriendlyyy says:

    I’m not saying this has anything to do with sexual harassment here, but this is truly heartbreaking and tangentially related to situations with more men than women. And it makes it that much more frustrating that they look down on homosexuality, which IIRC is more common in men or at least more likely to generate M-M relationships than F-F ones. Like if 1 is straight and 10 is gay; men who are 6 or 7-10 end up in M-M relationships but for women its 8 or 9-10 who end up in F-F relationships. Lesbians must have it REAL bad there.

    China, India grapple with the consequences of too many men

  10. Squirrel of Doom says:

    > It seems that men treat women better, and protect them more, when women are in shorter supply.

    This is not my first thought.

    Rather it’s that when women are in short supply, most of them have boyfriends.

    • Doctor Mist says:

      But isn’t that just part of it, one of the mechanisms in play?

      A boyfriend, especially one who feels lucky to have that status, is more likely than most to treat the woman well and protect her as needed. And a man who might aspire to be a boyfriend to an (uncommon) unattached woman is similarly motivated to treat her well and protect her as needed, in the hopes of being chosen as a boyfriend.

  11. Stezinec says:

    I wasn’t aware of Terran’s analysis. A couple of differences with mine: they controlled for personality traits of the reporter, whereas I used the average personality (Extraversion) of everyone in the profession. Also, I was looking at female reporters only.

    Scott’s point about sample size is accurate. Particularly when you limit it to the ~600 cisgender females, there aren’t enough participants in any profession except Computers for good statistical power. A tried a different type of logistic regression and I could get the effect of Computers to survive control for Extraversion there (p = .006).

    • terran says:

      Nor was I aware of yours until Scott mentioned both of us, so it was interesting to see we reached somewhat different conclusions. Another important procedural difference may be that I looked at the explanatory power of Profession in its entirety, using an Chi^2 test and without merging any categories, whereas you looked at each profession separately; it’s a known phenomenon that sometimes the Chisq or F test can be significant while the individual P tests are not or almost all are not (and vice versa) once there are more than a handful of categories. The other difference could be in your choice of baseline level – p-values for a categorical variable are highly dependent on the choice of the baseline, although in this case it looks like you picked Other, which has both a high number of respondents and a high value for harassment, so it seems like a good choice.

      I think one of the most interesting findings was that Profession is powerful to explain harassment outside of work, even after controlling for all the personality and demographic traits of the reporter that I could think of which seemed in any way relevant. I believe that until we can crack that, we can’t interpret any of the other relationships. I can think of no plausible mechanism by which the Profession of the reporter can explain their harassment outside of work unless it is a proxy for some kind of personality or demographic trait, either of the reporter or of their surroundings. Unless we can establish that relationship and find a way to control for it, we cannot interpret the reports of harassment at work in a way that says anything about the conduct of the people in the profession.

      I liked your idea of looking at the average personality values for the profession, so I gave that a try as well. I could not introduce them directly as explanatory variables in the previous framework because they were redundant with the direct coding of the professions, so I compared two models – one of them used 6 degrees of freedom for the professions (mean values of the big 5 components plus AutismSpectrumTest), and the other used the full 22 degrees of freedom. The additional degrees of freedom remain signficant for predicting a Yes on Harassment2 (victim of harassment, outside of work), with a p-value of 0.005 – still significant but much lower than the 10^-8 from before.

      However, when I filter the data to only “F (cisgender)” respondents, to match your analysis, the results change dramatically – now only the BigFiveE and Social Justice identification of the respondent are significant for explaining harassment outside of work, and Profession is not. Below is for everyone and then just for cisgender females:

      > glm(Harassment2=='Yes' ~ SexualOrientation + GenderConformity + Status + FinancialSituation + SocialSkills + AutismSpectrumTest + BigFiveE + BigFiveN + BigFiveA + BigFiveC + BigFiveO + Trustworthy + SJID + Profession ,data=ssc,family='binomial') %>% Anova
      Analysis of Deviance Table (Type II tests)

      Response: Harassment2 == "Yes"
                         LR Chisq Df Pr(>Chisq)    
      SexualOrientation     21.35  4   0.000270 ***
      GenderConformity      23.39  1   1.33e-06 ***
      Status                 0.01  1   0.933634    
      FinancialSituation     2.98  1   0.084079 .  
      SocialSkills           8.91  1   0.002830 ** 
      AutismSpectrumTest    12.02  1   0.000525 ***
      BigFiveE              15.57  1   7.96e-05 ***
      BigFiveN               3.16  1   0.075616 .  
      BigFiveA               5.57  1   0.018301 *  
      BigFiveC               0.64  1   0.425449    
      BigFiveO               1.64  1   0.200932    
      Trustworthy            8.58  1   0.003405 ** 
      SJID                  22.51  3   5.12e-05 ***
      Profession            78.23 22   3.16e-08 ***
      ---
      Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1


      > glm(Harassment2=='Yes' ~ SexualOrientation + GenderConformity + Status + FinancialSituation + SocialSkills + AutismSpectrumTest + BigFiveE + BigFiveN + BigFiveA + BigFiveC + BigFiveO + Trustworthy + SJID + Profession ,data=filter(ssc,Gender=='F (cisgender)'),family='binomial') %>% Anova
      Analysis of Deviance Table (Type II tests)

      Response: Harassment2 == "Yes"
                         LR Chisq Df Pr(>Chisq)    
      SexualOrientation     2.797  4   0.592291    
      GenderConformity      1.875  1   0.170876    
      Status                0.623  1   0.429949    
      FinancialSituation    0.858  1   0.354394    
      SocialSkills          0.005  1   0.943427    
      AutismSpectrumTest    1.212  1   0.270881    
      BigFiveE              8.658  1   0.003256 ** 
      BigFiveN              0.121  1   0.727961    
      BigFiveA              0.327  1   0.567163    
      BigFiveC              0.545  1   0.460284    
      BigFiveO              0.005  1   0.944428    
      Trustworthy           0.716  1   0.397363    
      SJID                 18.534  3   0.000341 ***
      Profession           27.387 22   0.196913    
      ---
      Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

      I think this is largely due to the fact that there are roughly 10x as many male survey respondents, so it’s going to be much easier for anything to reach statistical signfiance when we include the data from males. There are slightly over 500 female respondents who filled out the Big 5 personality information. However, that does not fully explain what’s going on, because we see that for the all-genders vs female result, the relative significance of profession changes.

      If we wanted to pursue this further, we could try to convince ourselves about whether or not it is OK to lump male and female respondents together to assess the impact of profession on harassment.

      I still think that the Really Weird Thing here is the explanatory power of Profession on being a victim of harassment outside of work, even if only for males, which persists after I throw in everything I can think of to control for (I also tried several things not shown above, to save space in this post – if there is interest I can put up all the code on Github or equivalent). I remain suspicious there has got to be some relevant personality and/or demographic/location/subculture trait that I’m failing to control for. Any suggestions/ideas?

      P.S. You are welcome to refer to me with male pronouns.

      • terran says:

        I had another idea – maybe respondents thought “Well, my coworker harassed me, but it was at the bar down the street, so I’m going to classify that as ‘not at work’ “. The questions simply said “at work” and “not at work” and said nothing about “by coworker” so I think this interpretation is plausible. Scott, perhaps you could put that on the list of things to make more specific next year, along with whether one was harassed in one’s current profession.

        • Stezinec says:

          Great work terran. That’s a good point about the wording of the survey question.

          I still think that the Really Weird Thing here is the explanatory power of Profession on being a victim of harassment outside of work, even if only for males, which persists after I throw in everything I can think of to control for (I also tried several things not shown above, to save space in this post – if there is interest I can put up all the code on Github or equivalent). I remain suspicious there has got to be some relevant personality and/or demographic/location/subculture trait that I’m failing to control for. Any suggestions/ideas?

          My only thought here was to look at the residualized harassment levels in men by profession to see if the effect is concentrated in certain areas or not, and what this might suggest.

          Also, there were some distributional issues with the personality test results, which could have hurt their reliability: https://www.reddit.com/r/slatestarcodex/comments/7pbs62/collected_findings_from_ssc_survey_data/dsgdxas/

          • terran says:

            I had a go but nothing obvious appeared from it. I found the deviance residuals very hard to interpret, so I instead calculated the modeled fraction of Harassment2==Yes (Victim Outside of Work) without using Profession in the model, and compared it to the actual values. Here’s a graph organized by actual-minus-model for each profession:http://role.consistent.org/images/2018/201804_ssc_harassment_outofwork_fig1.png

            and here’s a scatterplot with model on the abscissa and actual on the ordinate:http://role.consistent.org/images/2018/201804_ssc_harassment_outofwork_fig2.png

            . Art, Health Care(mental health) and Media are the three were the actual harassment rate most exceeds the modeled value from other attributes, but it feels contrived to come with any reason that these three would be fundamentally different from “Health Care (other)” or “Other Social Science”.

  12. herculesorion says:

    Given this story about Theranos I’m entirely willing to believe that Startup Culture is a problem, and that people are incorrectly extending their impression of Startup Culture to tech as a whole.

    • Caleb says:

      Thirding this.

      As a genderqueer woman, I’ve encountered a lot of sexism in STEM, but very little has been explicitly sexual in nature. It’s condescension (and turtles??) all the way down.

      Caveat for academia: I do think the tenure system protects serial harassers in a truly toxic way that can poison whole departments, but that’s not specifically a STEM problem.

      • cuke says:

        I second this caveat, as someone who experienced harassment by a tenured serial harasser in academia and watched it continue long past my time and that when consequences finally happened, they were really anemic up against how egregious the behavior was.

    • Chris says:

      I’m confused. That story doesn’t have anything to do with sexual harassment…

      • Doctor Mist says:

        It’s at best unseemly, and suggests that the people at Theranos are fecal.

        Granted, seeing evidence of people who I wish were not part of my species doesn’t mean they are prone to every single bad behavior I can imagine. Still, though.

        • The Nybbler says:

          They made a Space Invader style game with someone who cost them a lot of money as a target. This is just the high-tech equivalent of putting someone’s face on a dartboard. It hardly seems to rise to the level of making people “fecal”

      • herculesorion says:

        Given that story about Theranos I’m entirely willing to believe that Startup Culture is a problem, and that people are incorrectly extending their impression of Startup Culture to tech as a whole.

  13. Caleb says:

    With respect, Scott, I think you may be too quick to dismiss the claim that people who find the tech world sexist probably don’t read SCC. Anecdotally, I run in a pretty social-justice minded, STEM-focused crowd, and opinions on SSC run the gamut from “irredeemably sexist” to “worthy opponent/guilty pleasure”. I would bet dollars to donuts that “Scott (falsely) believes that sexism in tech is not a serious problem” is a sentiment that everyone I know who is familiar with this blog would have agreed with, even before this series of posts. To be clear, I realize that “sexism in tech is not a serious problem” isn’t something you explicitly stated in this post. Maybe it’s not something you’ve ever said! It’s just…you know…tribal allegiance, pattern matching, reading between the lines, and so on.

    [That said, I don’t actually disagree a priori with the conclusion of this series, I just think it has too many limitations to be the final word on the subject.]

  14. Icedcoffee says:

    This may be getting too in the weeds, but I’d be interested to see “average number of hours a day worked” as a variable. (Possibly also controlling for “working from home/remotely” vs. “in office.”) Law and healthcare both have notoriously long hours (in office), whereas every tech job I’ve held has allowed for relatively lower total hours (and was much more likely to allow working remotely).

    This would certainly skew the ratio of “at work” vs. “not at work” numbers, since these professions literally spend a disproportionate amount of time at work. But I could also imagine harassment being more likely at the end of a long shift.

    • Aapje says:

      That makes me wonder: does domestic harassment count as ‘at work’ when you are working at home? 🙂

  15. The Element of Surprise says:

    It would probably be interesting to also have some data on the (perceived) severity of sexual harassment. If the rate of harassment in some fields is relatively low, but the distress it causes is very large, then the media could have its priorities right after all. I could imagine, for example, that there might be greater social network support for female a victim in a place where gender balance is more even, or that a typical person in a people-oriented field copes better with transgressions of their personal boundaries than one in a thing-oriented field.

  16. Douglas Knight says:

    When the Washington Post published an AAU survey of 27 colleges, it is silly to pull out one school, even the school with the lowest rate of assault. Instead, you should make a scatterplot.

    data
    everything is for undergrads
    survey from spring 2015
    sex ratios from usnews, today, 2018
    correlations:
    .37 % f with assault of f
    .34 % f with (attempted) penetration of f
    .26 % f with assault of m
    .74 assault of f with penetration of f
    .53 assault of f with assault of m
    .38 penetration of f with assault of m

    • baconbits9 says:

      Seeing my school (cwru) on the scatter plot makes me think of so many different possible con founders.

      1. CWRU shares a campus with the Cleveland Institute of Art and Cleveland Institute of music with different M/F ratios.
      2. CWRU has a nursing school that is 90-95% female (or at least was), on the same campus and sharing some classes but there is a distinction there.
      3. There was at least one instance of CWRU ‘hiding’ assaults, specifically those they believed were committed by non students on students.

      • Douglas Knight says:

        3. As the Time article said, there is reason to be skeptical of numbers reported by the administration, but this was a survey of students.

        • baconbits9 says:

          If you assume that student culture and faculty culture are totally separate, and the administration aren’t selecting for types of students, and that students who feel marginalized by the administrations response aren’t more likely to have dropped out, etc, etc.

  17. ChelOfTheSea says:

    Selection bias seems like a huge problem for this.

    I’m a pretty big fan of the more…I guess “social dynamics” style of post around here (e.g. Meditations on Moloch), so I still read the blog – but I stopped posting on the subreddit, and don’t usually post here, in part because of a culture I perceive as rather sexist. I know I’m not alone in this, because I got quite a few PMs to that effect on Reddit from other frustrated women. A group that is literally 90% male is probably working under some strong selection effects – that’s as male-dominated as midwives or elementary school teaching is female-dominated (http://www.oecd.org/sdd/37964549.pdf) – and it seems plausible that this may be a major basis for such selection.

    It seems plausible that women who have had problems with e.g. harassment are more likely to have entrenched views on the subject, and outspoken feminist-ish trigger a hostile response from, if not all, a substantial portion of this community and the rationalist-o-sphere in general. And that’s if they even get here – an entrenched feminist is unlikely to be much of an SSC fan in the first place, considering Scott’s ambivalent-at-best feelings on the movement. Assuming that people generally don’t want to hang around folks who they could quite reasonably perceive as blowing off their problems (cf Radicalizing the Romanceless), or that they tend towards politics that conflict with Scott’s centrism-as-dogma, that would mean this community selects against victims, particularly women victims.

    There’s another problem, and that’s that “sexual assault” is ill-defined (in the survey question) and heavily political, in a way that further enhances selective effects. There are surely some behaviors considered sexual harassment by people at some points on the political spectrum and not others (huh, maybe I should poll on this), so selection for politics (definitely in play) would suggest selection in response rates too.

    • Aapje says:

      Your comment is rather frustrating to me. Radicalizing the Romanceless is a complaint that feminists have a tendency to blow off the problems of men/male nerds, so it suggests that you yourself may have a tendency to not consider victimized men/male nerds to be victims (or to a much lesser extent than is warranted).

      Your accusation that this community selects against victims may reflect that you have a different opinion on who are the victims in our society and to what extent, compared with (some of) the people here. I’ve certainly seen multiple comments that this community is a safe space for (abused) aspies/nerds. Perhaps this community selects against some kinds of victims, but select for other kinds of victims.

      People who particularly care about the kinds of victims who are selected against here may then draw the conclusion you draw: that this space is hostile to victims in general. However, that would be just as uncharitable as me arguing that feminist spaces are hostile to all victims because victimized men often feel hostility in feminist spaces.

      As for the effect of a strong over-representation of men: I have noticed that men and women tend to deal differently with victimization (and many studies back me up). I would argue that much of this is due to gender roles (that are encultured at both the conscious and unconscious level). People tend to respond differently to victimized men than women, also on an emotional level. People also tend to victim blame men more than women, demanding agency from men more than from women.

      The consequence of this is that it is fairly rational for victimized men to behave differently from victimized women, especially for certain kinds of victimization, which in turn may feed the perception that they are not victims, because they don’t match the stereotype of the victim. Even if this behavior is not rational in a specific situation, many years of enculturation may make men disgusted by themselves if they act like the stereotypical victim.

      When men diverge from & complain about the negative aspects their gender role, this is very commonly seen as a threat to women, because it typically also means that they challenge benevolent sexism. I think that any space where men partially shed this enculturation and openly speak out about their concerns is thus going to be seen as misogynist by many in society and especially feminists, just like feminists spaces are often considered misandrist by many in society and especially MRAs.

      I personally prefer to have feminists rationally challenge (and not shame) opinions that they consider misogynist, just as I prefer to have anti-feminists rationally challenge (and not shame) opinions that they consider misandrist, so people keep each other honest.

      • ChelOfTheSea says:

        > Radicalizing the Romanceless is a complaint that feminists have a tendency to blow off the problems of men/male nerds, so it suggests that you yourself may have a tendency to not consider victimized men/male nerds to be victims (or to a much lesser extent than is warranted).

        This accusation strikes me as a bit odd. The fact that I went out of my way to mention it was an intentional signal on my part that I am aware that that kind of thing happens and consider it a problem. The point is that your average feminist feels in this space about how those people did, and that’s naturally going to induce bias.

        > I’ve certainly seen multiple comments that this community is a safe space for (abused) aspies/nerds. Perhaps this community selects against some kinds of victims, but select for other kinds of victims.

        So…selection bias, then?

        I’ll agree this is a safe space for a certain sort of guy. I’m not altogether sure that’s a good thing, considering what I’ve seen grow out of it. But that’s not the point – the point is that this data is wildly nonrandom and trying to reason about the general population from it is pointless.

        > People who particularly care about the kinds of victims who are selected against here may then draw the conclusion you draw: that this space is hostile to victims in general. However, that would be just as uncharitable as me arguing that feminist spaces are hostile to all victims because victimized men often feel hostility in feminist spaces.

        Setting aside for a moment that I have an object-level disagreement with you here – every sexual-assault-supportive space I’ve ever been in proximity to has gone way out of its way to mention that men can be victims too – it’s not really relevant whether or not the perception is *justified*. All it need be is *present* to produce sampling bias.

        > I personally prefer to have feminists rationally challenge (and not shame) opinions that they consider misogynist, just as I prefer to have anti-feminists rationally challenge (and not shame) opinions that they consider misandrist, so people keep each other honest.

        I am willing to do so, to an extent. I have a personal Overton Window, both in terms of object-level views and in terms of meta-level epistemic hygiene, within which I am broadly willing to meet people on their terms. But a large chunk of this community is well outside that window, and so I’m usually no longer a participant in it.

        The best analogy I can make for my take-no-prisoners approach to a certain sociopolitical bloc is that of old-school ‘outlaws’. We have rules of debate, culture, and epistemic hygiene, that we agree to for the purposes of having a debate at all. People who stay within those rules should be heard and treated seriously as we ask them to hear and treat us seriously. People who don’t do so should be ostracized, deplatformed, and silenced, because that is how we preserve the rules, in the same way that Old West law retracted its protection from those who grossly failed to honor it.

        I get that that’s not a very popular opinion around here, and that many of you – even non-members of that bloc – will not agree with me (and to them, I say “good luck carrying that scorpion across the river”). But that isn’t the point – the point is that people like me usually do not stick around, and that makes these survey results dangerous to generalize to the public at large.

        • Aapje says:

          @ChelOfTheSea

          The fact that I went out of my way to mention it was an intentional signal on my part that I am aware that that kind of thing happens and consider it a problem.

          I did not read it that way, because you explicitly stated that this community might select against victims in general, which seems inconsistent with a reading where you referred to Radicalizing the Romanceless to argue that victimized men might find solace in a community like SSC. Perhaps your comment seems more ‘outgroup-friendly’ to you than it seems to me.

          Setting aside for a moment that I have an object-level disagreement with you here – every sexual-assault-supportive space I’ve ever been in proximity to has gone way out of its way to mention that men can be victims too – it’s not really relevant whether or not the perception is *justified*. All it need be is *present* to produce sampling bias.

          My experience is that feminists are generally quite willing to accept that men can be victimized by men, but that a substantial group is not willing to entertain the possibility that men can be victimized by (somewhat typical) women, that this happens in significant numbers and/or that this may require structural solutions. Furthermore, my experience is that even the feminists who don’t believe this, generally don’t want to exclude these beliefs from their Overton Window, while they tend to aggressively exclude beliefs that they perceive as hostile to women (or to put it bluntly: I think that these spaces tend to be very oversensitive to misogyny and very undersensitive to misandry).

          I doubt that you would consider a space supportive to victimized women, if they had a significant contignent who acted that way to women who were victimized by men or to people who spoke out for that group.

          Anyway, I agree with you that the actual truth of the matter is indeed less relevant for population effects than the gap between how people think they ought to be treated and their perception of how they will be or are treated. Some people may actually only consider a space supportive if it is aggressively sexist against the gender of their perpetrator, for example. So when (certain) victims seek out a space, that in itself is not evidence that the space is healthy for them. However, this can both be true or not true of SSC and of feminist spaces.

          I am willing to do so, to an extent. I have a personal Overton Window, both in terms of object-level views and in terms of meta-level epistemic hygiene, within which I am broadly willing to meet people on their terms. But a large chunk of this community is well outside that window, and so I’m usually no longer a participant in it.

          My perception/experience is that feminists tend to have fundamentally different debating norms, which are strongly focused on preserving a far smaller Overton Window. You seem to claim that having a strict Overton Window creates epistemic hygiene and I think that is true only to a certain point, where people feel encouraged and safe to speak out and believe that they will encounter pleasant debate. However, an overly strict Overton Window results in the protection of false beliefs against challenge. Furthermore, aggressive policing of the Overton Window can itself discourage people to speak out, make them feel unsafe and make them self-censor, rather than challenge questionable claims.

          I believe that the proper way to protect an Overton Window is to very strongly focus on the quality of the argument and the tone, but only fairly weakly to adherence to orthodoxy. I would argue that SSC strikes a better balance than most feminist spaces, although my view may be colored by my personality traits.

          In any case, I think that most of feminism and SJ in general, has adopted an Overton Window that disallows serious consideration of crucial criticisms and that this has fatally wounded these movements in various ways (including making them incapable of achieving their goals).

          But that isn’t the point – the point is that people like me usually do not stick around, and that makes these survey results dangerous to generalize to the public at large.

          I agree with you that feminist spaces probably are more pleasant to many (but certainly not all) female victims and that it’s plausible that SSC is relatively unattractive to this group. So it is then questionable to use this survey to make statements about the absolute level of sexual harassment in the general population.

          However, Scott compared different groups of SSCers to make (weak) claims about relative differences. One might reasonably assume that even if (female) SSCers are relative unlikely to be have been harassed, because of selection effects, that this effect will be the same for all SSCers, mostly independent of industry. So the comparison of one industry to another could then reasonably be made.

          • ChelOfTheSea says:

            I’m going to avoid the culture-war portions of this post, because honestly I’ve trodden this ground eighteen zillion times and I’d rather not re-litigate the usual “well my experience with X group is Y” argument in favor of saying that because people *think* their experience with X group is Y their behavior is non-random.

            Sometimes I wonder if the Bay Area – which I’ve never been to and have no particular desire to visit – is just totally insane on these things. Problems that seem like common complaints from people there are totally non-existent to me even within a bubble of people I filter pretty strongly for liberal/SJ-ish politics, so either Bay Area SJ people are very different from my own or someone here is wrong, misinterpreting, or lying.

            > So when (certain) victims seek out a space, that in itself is not evidence that the space is healthy for them. However, this can both be true or not true of SSC and of feminist spaces.

            Yes, I agree. But the fact that other data can be skewed does not mean this survey’s can’t. This question is very, very hard to get objective data on – you need a highly general survey with rigorously worded questions and high response rate, at a minimum. Something akin to the census.

            > However, an overly strict Overton Window results in the protection of false beliefs against challenge.

            No, not necessarily.

            I started out as a hardcore fundamentalist conservative, in line with my upbringing. I found that some things did not make sense, but my personal window did not then include liberal politics, so I shifted *within* that window along a gradient of things-making-more-sense to a semi-libertarian moderate conservatism. This, in turn, shifted the window, so that when semi-libertarian moderate conservatism’s problems cropped up, I adjusted further to the now-within-the-window “socially-liberal/fiscally-conservative” sort of position. This process repeated itself, with incremental change over many years, and got me to my current politics.

            My beliefs were open to challenge – to a point. Once updated, the new beliefs were open to *different* challenges. And so on.

            > I believe that the proper way to protect an Overton Window is to very strongly focus on the quality of the argument and the tone, but only fairly weakly to adherence to orthodoxy.

            Overton Windows within societies or communities – as opposed to the ‘personal overton window’ discussed above – work differently. Because you’re involving multiple people and multiple subcultures, you’re opening the door to game-theoretic strategy, and one of my current primary concerns is that some people who are looking for truth are being, in my view, very successfully out-gamed by people spamming radical claims in the hopes that they’ll hit enough cognitive biases to occasionally stick.

            As an extreme example: suppose a community holds two beliefs: (A) you should absolutely never lie and (B) you should always trust what someone else says is an accurate representation of their internal state (if not necessarily of the world – people can be wrong but not lying, of course). This community actually works great on its own! Its beliefs are justified and its results wonderful, even utopian. But the problem is that such a community is so vulnerable to abuse of its priors that it cannot stand.

            Overton Windows on a *social* level are not just truth-seeking in a vacuum, they are game-theoretic defenses against attack.

          • Aapje says:

            @ChelOfTheSea

            Sorry for doing the culture war thing again, which you may want to not reply to, but…

            Sometimes I wonder if the Bay Area – which I’ve never been to and have no particular desire to visit – is just totally insane on these things.

            I’m not in the Bay Area or even the US & in my country they are (in part successfully) pushing for job discrimination against men. So…

            Let me be clear: I am primarily judging activists with power. Those in academia and organisations that do research, philosophy and such, those that run programs to help certain people, those in the media that spread memes and narratives, politicians that win (mainly women’s) votes with lies and support for discriminatory policies, etc. I am very unhappy with what most of these people do.

            My perception is that these activists tend to often be in very radical bubbles, that make them not so reasonable, but that outside of those bubbles, there exist mostly fairly powerless feminists that on average are more reasonable.

            My primary beef is with the former, not so much the latter, although the latter tend to give a lot of passive support, because they are often unaware of the extent of the negative behavior that is going on. So I think that the former need to be delegitimatized, by truth-telling* and such.

            * Truth to power 😛

            As an extreme example: suppose a community holds two beliefs: (A) you should absolutely never lie and (B) you should always trust what someone else says is an accurate representation of their internal state (if not necessarily of the world – people can be wrong but not lying, of course). This community actually works great on its own! Its beliefs are justified and its results wonderful, even utopian. But the problem is that such a community is so vulnerable to abuse of its priors that it cannot stand.

            I would argue that this vulnerability already exists in a subset of SJ, because there actually is a seemingly somewhat common belief that the experiences of minorities should be treated not merely as an honest description of their own perceptions, but that these descriptions are factually correct (including when they consist of a judgement call) and can be generalized to all people with the same trait.

            Given that people are actually individuals and not people with identical experiences and traits, as well as being prone to bias & error, this then often creates a conundrum, when different people have a different description. Who is then right?

            This conundrum seems to then be often solved with ‘oppression olympics,’ where there is a power struggle over who is more oppressed and thus more correct by virtue of not being blinded by their privilege.

            I would argue that this is not merely an absolutely horrible way to discover the truth, but also tends to create a very strong, fixed hierarchy with immense risks when challenging those on top. To me, such communities appear immensely authoritarian and anti-progressive.

            You say that such communities are so vulnerable to abuse that they cannot stand, but here I must disagree. They can indeed not function as they should, but they can function as dysfunctional entities. One can compare it to Putin’s government. The dysfunctional Russian government can indeed not govern well, but they can hang onto power.

            one of my current primary concerns is that some people who are looking for truth are being, in my view, very successfully out-gamed by people spamming radical claims in the hopes that they’ll hit enough cognitive biases to occasionally stick.

            I have the exact same concern, but am probably more pessimistic, since I think that Enlightenment values are under siege. For example, an increasing number of academics seem not content with being right and with leaving politics (making choices based on facts) to others, but instead want to ensure that the ‘right’ choices are made and are willing to compromise their research to make the ‘right’ choice seem inevitable.

            Of course, you also have people who for example, create dumb climate change skepticism by triggering cognitive biases (‘It’s a cold day, we can’t have global warming’). However, in cases like that, one can actually look to the scientific consensus to figure out the truth, if one wants. If science is corrupted, even quite smart and not too gullible people may become incapable of figuring out the truth.

          • Zorgon says:

            @Aapje – your patience and charity are impressive.

        • albatross11 says:

          ChelOfTheSea:

          I’m quoting one pretty noncentral part of your comment above, because I find it interesting on its own:

          The best analogy I can make for my take-no-prisoners approach to a certain sociopolitical bloc is that of old-school ‘outlaws’. We have rules of debate, culture, and epistemic hygiene, that we agree to for the purposes of having a debate at all. People who stay within those rules should be heard and treated seriously as we ask them to hear and treat us seriously. People who don’t do so should be ostracized, deplatformed, and silenced, because that is how we preserve the rules, in the same way that Old West law retracted its protection from those who grossly failed to honor it.

          It seems like there’s a pretty obvious failure mode in this style of debate: What happens if the truth is outside the window of your culture/epistemic hygiene? It seems dangerously easy for true but unpalatable facts to end up outside your window, and then stay there because people following your strategy are successful at ostracizing, deplatforming, and silencing them.

          This seems a bit like if you have a probability model to describe some data, and anytime you get a datapoint that your model says is impossible (like a +10 sigma sample from a normal distribution), you classify it as an outlier and throw it away. This prevents incorporating a bunch of measurement error/mistakes into your picture of reality, but may also mean that you don’t notice when your model is a bad description of the thing you’re modeling.

          • ChelOfTheSea says:

            > It seems like there’s a pretty obvious failure mode in this style of debate: What happens if the truth is outside the window of your culture/epistemic hygiene?

            First: this is irrelevant to the sampling bias point at the end of the day.

            I agree that this is a potential failure mode. But I think it’s a less likely and less dangerous mode than the risks of disregarding social notions that we have adopted with very, very good reason (e.g. don’t discriminate based on race/sex/etc, don’t give all your power to one guy).

            But perhaps more rigorously, the problem is that a 10-sigma result – and its associated p-value under current priors of 2.088*10^-45 – does not and should not modify my priors by 10^45 or anything close to it. My prior on being outright psychotic or otherwise entirely unable to evaluate evidence correctly is much higher, at about 1% give-or-take (either I’m able to correctly read up on rates of psychosis at any given time, which seem around there, or I can’t, in which case I must indeed be in that state). Moreover, an outside-view approach tells me that nearly every time I’ve gotten a result anywhere near that surprising, I’ve screwed something up somewhere.

            What this means is that a 10^-45 ostensible result does not shift the likelihood ratios among my priors by 10^-45. In fact, I tend to see anything over ~5-6 sigma as more or less irrelevant, because minuscule p-values link to a commensurate rise in the chance of experimental error. I haven’t sat down to do a fully rigorous look at this, but I kind of wonder if a 20-sigma result is actually *less* Bayesian evidence than a 10-sigma one, because there’s nothing preventing the rise in experimental-error probability from actually outpacing the (supra-)exponential decay of p-values.

            In short: if you think you have a 10-sigma result, you probably don’t. This sucks, but is the appropriate outside-view approach to learning, and it means you should have a relatively narrow window within which to accept (not necessarily consider, but accept) radical ideas. And yes, it is possible that there exist global truths not locally approachable with such a meta-prior, but that’s the hand we’re dealt as fallible creatures.

        • Mark V Anderson says:

          The best analogy I can make for my take-no-prisoners approach to a certain sociopolitical bloc is that of old-school ‘outlaws’. We have rules of debate, culture, and epistemic hygiene, that we agree to for the purposes of having a debate at all. People who stay within those rules should be heard and treated seriously as we ask them to hear and treat us seriously. People who don’t do so should be ostracized, deplatformed, and silenced, because that is how we preserve the rules, in the same way that Old West law retracted its protection from those who grossly failed to honor it.

          Well yes, all groups have Overton Windows concerning their rules of debate, including SSC. I think the Window for SSC is a lot more about process than content though. You don’t describe what you mean when you talk about the rules, so I don’t know if your rules are more content or process. I am sometimes pretty shocked at extreme comments that posters make, but it is mostly accepted by the group as long as they follow the accepted process. For example, one thing I almost never see here are ad hominem attacks, because I think anyone using those attacks would be shunned by the other posters (and maybe banned too). I very much like the SSC Overton Window accepting almost any content, but pretty strict about how civil the debate. Your mileage may vary.

          I do agree that an SSC survey can’t be considered a proxy for the world, because we are a lot different on average than the rest of the world (otherwise I wouldn’t be here).

          • ChelOfTheSea says:

            Ugh, alright, fine, let’s CW.

            > I very much like the SSC Overton Window accepting almost any content, but pretty strict about how civil the debate. Your mileage may vary.

            I want strong tone policing and moderate topic policing, basically. But the subreddit, at least, fails to even provide the former.

            No ad hominems is great, but it baffles me to say “no personal attacks” and simultaneously allow claims that the target’s class is inherently inferior. Or to ban attacks on “conservatives” based on things demonstrably (by polling) favored by that group, but allow attacks on “SJWs” (a small minority even of the social left) despite the very obvious Superweaponing of that term.

            Just from me personally, I’ve been told in this community that I’m too emotional and had it suggested that being on estrogen must be interfering with my Pure Logic, that the only reason I’m in STEM is that I really have a man-brain and therefore am not a Real Trans Person, told that people telling me point-blank they’d like to rape me can’t be Real Sexism because I’m trans (a fact said people did not know), and in probably my favorite example, a PM about how genetic racial differences are obvious that went out of its way to call me “boy” (I am, for the record, as lily-white as can be).

            The one that actually got me to leave was an extended debate over whether or not the statement “black people are inherently genetically of lower intelligence” is racist. Quite a number of people seemed to think it was not. Their excuse was “oh, we’re not saying they’re less valuable”, a claim laughably absurd in the face of them using that belief as an input for their beliefs about e.g. immigration (“well, they’re stupid, so we shouldn’t take more of them, but bring on the Asians!”).

            Maybe the comments here are better. But I think that bloc is abusing the openness of the rest of this community to debate to inject ideas/memes into the opposition while conversely allowing none to enter their own community.

          • The Nybbler says:

            I want strong tone policing and moderate topic policing, basically. But the subreddit, at least, fails to even provide the former.

            They certainly let you get away with a lot.

            No ad hominems is great, but it baffles me to say “no personal attacks” and simultaneously allow claims that the target’s class is inherently inferior.

            Here’s the contrast: “personal” attacks and target’s “class”. Calling someone “shorty” is a personal attack in some circumstances. Noting that Italians tend to be shorter than Dutch people, given adequate nutrition for each, is not.

          • albatross11 says:

            What about factual claims with actual evidence backing them up? For example, racial differences in average IQ and crime rate are really well-established. There’s no way to make young-Earth creationism (believed by a nonnegligible set of people) consistent with observable reality. Using those statements as a club to attack someone else on their identity is shitty and shouldn’t be tolerated, but rejecting actual correct statements because they’re hurtful to some people or groups is blinding yourself.

            What about factual claims that are plausible–consistent with the available evidence, not obviously way less likely that competing theories, etc., but that many people find hurtful? Like arguments that all religious are delusional, or that the rise of western industrial democracies was only possible because of the exploitation of slave labor and colonialism? Those upset a lot of people. Should we bar them from polite discussion?

            I am not especially interested in CW type arguments, but I’m pretty interested in having discussions where the rules aren’t engineered to make us dumber or to put a thumb on the scales in favor of one side of a question.

          • Mark V Anderson says:

            Just from me personally, I’ve been told in this community that I’m too emotional and had it suggested that being on estrogen must be interfering with my Pure Logic, that the only reason I’m in STEM is that I really have a man-brain and therefore am not a Real Trans Person, told that people telling me point-blank they’d like to rape me can’t be Real Sexism because I’m trans (a fact said people did not know), and in probably my favorite example, a PM about how genetic racial differences are obvious that went out of its way to call me “boy” (I am, for the record, as lily-white as can be).

            Well yeah most of this stuff is ad hominem. Did you say you have been on the SSC reddit, not SSC itself? I don’t read the SSC reddit myself, but I’ve heard that posters there are less civil than they are here. I would think that most of the stuff above would get disapproval comments from third parties on SSC itself, because we try to police this stuff. Not always — we’re far from perfect.

            I’m not saying that SSC is necessarily the place for you, even if you like the process. Even though we don’t police on content, there are some belief systems that seem pretty prevalent here. The most obvious to me is the dislike of many social justice advocates and identity politics. Even the leftists here seem to hold these SSC beliefs, or at least don’t disagree with them enough to argue much against them. I for one want to have as wide a variety of ideologies here as possible, as long as all advocates are civil and make rational arguments. But no one should be part of our merry band if they don’t find value in reading Scott’s postings and/or commenters’ postings.

      • Simon_Jester says:

        Your comment is rather frustrating to me. Radicalizing the Romanceless is a complaint that feminists have a tendency to blow off the problems of men/male nerds, so it suggests that you yourself may have a tendency to not consider victimized men/male nerds to be victims (or to a much lesser extent than is warranted).

        Surely it is a well understood phenomenon that two groups may symmetrically see either group complaining about their own problems as an instance of them not caring hard enough about the other group’s problems? It’s a phenomenon that pops up just about every time two interest groups come into conflict over an issue.

        We’ve all seen it, it’s a fairly obvious human cognitive bias to assume that people who refuse to talk about our problems because they’re too busy talking about their problems secretly hate us. So we should try to compensate for the tendency of this bias to color our thinking.

        As for the effect of a strong over-representation of men: I have noticed that men and women tend to deal differently with victimization (and many studies back me up). I would argue that much of this is due to gender roles (that are encultured at both the conscious and unconscious level). People tend to respond differently to victimized men than women, also on an emotional level. People also tend to victim blame men more than women, demanding agency from men more than from women.

        I think it may depend on the nature of the victimization. Women are expected to put up with different things than men, and are expected to take in stride inconveniences and forms of harm that a man would be fully in his rights to protest. The reverse is also true. There is ample room for both men and women to feel victim-blamed when others question whether they have handled a situation appropriately.

        And again, this is a fairly obvious cognitive bias we should be able to watch out for: one man’s “I, the victim, am being unfairly blamed!” is another man’s “Um, you know, you really shouldn’t be walking into dark alleyways with $100 bills taped to the outside of your clothing ”

        Because of the well understood phenomenon where we adopt an internal locus of control as our model when things are going well, but an external locus of control if things go badly.

        The consequence of this is that it is fairly rational for victimized men to behave differently from victimized women, especially for certain kinds of victimization, which in turn may feed the perception that they are not victims, because they don’t match the stereotype of the victim. Even if this behavior is not rational in a specific situation, many years of enculturation may make men disgusted by themselves if they act like the stereotypical victim.

        When men diverge from & complain about the negative aspects their gender role, this is very commonly seen as a threat to women, because it typically also means that they challenge benevolent sexism. I think that any space where men partially shed this enculturation and openly speak out about their concerns is thus going to be seen as misogynist by many in society and especially feminists, just like feminists spaces are often considered misandrist by many in society and especially MRAs.

        I personally prefer to have feminists rationally challenge (and not shame) opinions that they consider misogynist, just as I prefer to have anti-feminists rationally challenge (and not shame) opinions that they consider misandrist, so people keep each other honest.

    • albatross11 says:

      I think the selection bias criticism is almost certainly right. SSC readers are weird in about a hundred ways that make results from a poll here inapplicable to the wider world.

  18. Alkatyn says:

    RE startups vs. big tech firms, seems like it would be the same across different industries. Larger companies tend to have HR departments and more of an emphasis on process and professionalism. Smaller companies in any sector are going to be less formal.

    There may also be something to be said for industries with big power differntials. E.g. servers in restaurants are in a significantly less powerful position in relation to their managers than programmers at a generic tech company.

  19. Matt M says:

    Chris quotes a TIME article that argues that predominantly-male communities generally have lower harassment rates than predominantly-female communities:

    I was thinking of something like this, but with a bit of a twist, that communities in which women are rare self-select for the type of woman who is more predisposed to “tolerate” (for lack of a better word) harassment.

    Women in general are unlikely to become miners, sure. But when we start to imagine the type of woman who becomes a miner, what comes to mind? The stereotypical angry feminist SJW who views every sideways glance as “eye rape?” Hardly. Given what we know about mining (both factual, and imagined), the type of women who might consider becoming miners are probably the tomboyish, “tough” women who are unlikely to be living in a perpetual state of victimhood.

    This difference in attitude is unlikely to effect the actual incidence of harassment behaviors, but it is very likely to influence ones perception of said behaviors. And given the obvious and admitted difficulty that all measurement in this space is reliant on self-reporting surveys and that respondents have wildly different views on what counts as harassment or not, this seems relevant.

  20. freedomfan says:

    Reading through all of this. So many trees, so hard to see the forrest.

    Harassment will only occur when the harraser is too powerful to be fired or reprimanded sufficiently to make the harrassment not worth it to him/her. Here are the common cases where that happens a lot:

    – Tenure (saw that mentioned in the comments)
    – Rainmaker (this I’ve witnessed, can be lawyers or sales of any kind)
    – Celebs (Weinstein, et al)
    – Media – (powerful editors, star writers, common #metoo abusers)
    – Politicians
    – Top doctors

    The last may be surprising but is it? Is the highly coveted surgeon with a long waiting list of pro athletes going to be fired for disrespecting the staff? Please.

    The only correletion with ‘industry’ is how much the industry is structured such that some people are very hard to discipline or fire. If the harrassed staff are easily interchangeable, all the moreso.

    • Aapje says:

      Trainers in pro sports too. It’s a pretty perfect setup for the crossing of boundaries:
      – The athlete tends to be socially isolated
      – The athlete tends to be very dependent on the trainer
      – There tends to be a lot of physical contact as part of the job

      • albatross11 says:

        All these seem to be driven by the ability of these high-value people to get away with bad behavior generally. The same guy who gets away with screaming curses at his secretary when he’s in a bad mood is more likely to be able to get away with groping her in the elevator.

    • Simon_Jester says:

      There may be a multiple-lobe distribution here.

      For example, you might have another category of industries with high harassment rates because they habitually find themselves sorting through the dregs of the employment market. Which means they “get to” (have to) put up with employees that don’t respond as reliably to discipline. So they’re locked in an equilibrium where they have more employees getting written up for all kinds of abuses and violations of procedure- petty theft, random squabbles, sexual harassment.

      In that scenario, no individual employee is getting away with sexual harassment for being too big to fire. But collectively there’s an ongoing pattern of harassment because trying to stamp out all the harassment in such a workplace is like trying to stamp out all of any other deficiency in an inherently deficient slice of the labor force: it’s like trying to sink a floating cork with a sledgehammer. Fire one bozo for harassment and another one pops up, and meanwhile the bozo you just fired probably slips past the guard of someone else’s HR department sooner or later and gets hired at another place much like yours.

  21. Apollyon says:

    I think one of the biggest problems in measuring sexual harassment is the difference in opinion of what exactly constitutes harassment. More than that, some do the disparities may be linked to differing attitudes towards sex and sexual harassment in different fields. Case in point: I’d be willing to bet women in the adult film industry have a vastly different attitude towards sex and harassment than women who work in a Title IX Office. This could both disguise and exacerbate the true amount of harassment in the workplace.

  22. mack says:

    One possibility that I’m surprised no one mentioned: say I’m a heterosexual male interested in sexually harassing female coworkers. Why would I try to get a tech job, where there are few female coworkers available for me to sexually harass?

    I remember a conversation with another student my during my freshman year as an undergrad. A guy told me that his choice in college came down to Fordham and [unnamed university we attended that is universally regarded as better than Fordham]. Why did he consider strongly consider Fordham? Their undergraduate student body is/was so predominantly female that apparently any man with a pulse can get laid if they want.