For instance, Moore (2022), writing in “Teen Vogue” claimed that “it comes as a surprise to many to discover that white women have benefited more from affirmative action programs and policies than any other demographic.”. Her evidence for this is the fact that Crenshaw (2007) said “the primary beneficiaries of affirmative action have been Euro-American women.”. Crenshaw provided no empirical evidence for this statement. 

Similarly, Massie (2016), writing for Vox, stated that: “”The primary beneficiaries of affirmative action have been Euro-American women,” wrote Columbia University law professor Kimberlé Crenshaw for the University of Michigan Law Review in 2006. A 1995 report by the California Senate Government Organization Committee found that white women held a majority of managerial jobs (57,250) compared with African Americans (10,500), Latinos (19,000), and Asian Americans (24,600) after the first two decades of affirmative action in the private sector.”

This is the same Crenshaw citation as above. As for the California Senate report, it of course does not follow that just because white women have had more managerial jobs since affirmative action began that affirmative action, as opposed to more white women trying to enter the work force, or more white women going to college, etc., caused all those occupational gains. 

Kohn (2013), writing in Time, argued that “While people of color, individually and as groups, have been helped by affirmative action in the subsequent years, data and studies suggest women — white women in particular — have benefited disproportionately. According to one study, in 1995, 6 million women, the majority of whom were white, had jobs they wouldn’t have otherwise held but for affirmative action.”

Of course, since white women are the majority of women this does not show that minority women did not benefit more than white women from affirmative action. Further, the article they cite as the source for this figure is another article by Tim Wise which states that “According to a 1995 study, there are at least six million women — the overwhelming majority of them white — who simply wouldn’t have the jobs they have today, but for the inroads made by affirmative action (Cose 1997, 171).”

The book by Cose in turn states, on page 171, that “The Working Woman special report on affirmative action also cited a N95 study by Alfred Blumrosen, a professor at Rutgers University Law School and consultant to the Labor Department, suggesting that an “estimated six million women wouldn’t have the jobs they have today were it not for the inroads made by affirmative action.”

Notice that this says nothing about the racial composition of the women who got jobs thanks to affirmative action. Wise was thus lying about his source, which itself is only citing another report which supposedly cites the primary source.

Turning to the primary source, on page 118, Blumrosen (1995) states that “It is modest to point out that 5 and one half million minority employees are in higher level jobs than they would have been under the occupational distribution of 1960, and that 6 million women have moved into executive, managerial, professional and sales jobs since 1972”. This paper makes no note of proportion of jobs that went to white women nor does it show that this change in occupational distribution was entirely due to affirmative action.

Moreover, this comparison fails to account for the population sizes of these groups.  Using census demographic data from 1980, roughly the mean year of the time period under consideration, giving 6 million jobs to white women would mean that one in every 17 white women got a job via affirmative action and giving 5.5 million jobs to non-whites would mean that one in every five non-whites were given a job via affirmative action. Thus, even if we assume that these numbers are a valid measure of jobs gained due to affirmative action, which has not been shown to be true, and that literally all the women given such jobs were white, which has also not been demonstrated, it is still misleading to claim that white women benefited the most from affirmative action. 

Turning away from the reasons given in popular discourse to serious attempts to measure the impact of affirmative action, the best methodology available is to compare the rate of at which various groups employment grew among firms doing federal contract work, and who are mandated to follow an affirmative action policy, to the rate at which their employment grew among non-contracted firms. Using this approach on data covering 1970 – 1980, Smith et al. (1984) found that affirmative action had a small positive effect on employment for white women but a much larger positive effect for blacks of both sexes. The most dramatic benefit was for black men who in 1980 accounted for 11.6% of managers and officials in non-contractor firms but 44% of them in firms with federal contracts. 

Covering a larger time period, Kurtulus (2016) analyzed data from 1973 to 2013 and found that white women were the primary victims of affirmative action in federal contract work writing that “affirmative action increased the employment of black and Native American women and men at the expense of white women”. The paper also notes a trend for white men to become the managers of federal contract work, but obviously the explanation for this is some factor other than attempts to fill federal affirmative action requirements. Thus, this approach to measuring the impact of affirmative action produces findings which totally contradict the narrative that white women were the primary beneficiaries of affirmative action. 

A somewhat less rigorous approach is to survey private firms, regardless of contractor status, and see whether firms claiming to support affirmative action predicts who they employ. This method warrants less confidence than the previous method because we may doubt the validity of firm’s self reporting who they employ and whether they utilize affirmative action.

The results of these studies vary. Holzer et al. (1999) found that “Affirmative Action is associated with increases of about 15% in the probabilities of hiring white women and black men. On the other hand, the last row indicates that the probability that a white male is hired is lower by about 20% under Affirmative Action.”. So white women in this paper tied with black men for the group that benefited the most and the negative effect on white men was large enough such that the net effect on whites was still negative. 

Button (2006) took this same approach and found that support for affirmative action was unrelated to the hiring of white women. 

By contrast, the same authors found that support for affirmative action did positively predict the hiring of black employees (Button et al., 2003). 

Kalev et al. (2006) produce still different results. This paper begins by documenting the rise in various forms of affirmative action in private firms between 1970 and 2002.

They then go on to show that which group benefits most from an affirmative action depends on which policy you look at. The most consistent pattern across policies is the negative impact on white men.

That said, in this data set white women are evidenced to have significantly benefited from affirmative action. However, this data is restricted to the impact affirmative action has on who is hired as a manager and is based on relatively older data. This makes it not comparable to the studies referenced above on the net impact of affirmative action on who is hired across all employee positions. And as already noted the federal contract data also showed a lesser anti-white impact in the case of managers. So this study does not contradict the other reviewed research to the degree that it might initially seem to.

In total, this line of research also fails to support the contention that white women have benefited most from affirmative action. And as we’ve already seen, the most solid data available actually suggests white women have on net been hurt by affirmative action.

Moving away from the economy, in studies on admissions bias they generally don’t break the data down by sex and race. However, the net effect is massively negative for whites so if there is any benefit to be a white women, which there is no reason to think is true to begin with, then it is more than made up for by the negative impact on white men. 

(The race columns show the odds of admission compared to those of white applicants when qualifications are held constant.)

For these reasons, the relationship between white women and affirmative action does not contradict the general anti-white bias that is present in our society. 

To begin with, it’s worth noting that most acts which constitute discrimination, ostracizing someone, insulting someone, emotionally or physically hurting someone, etc., are essentially forms of bullying. This fact matters because even though there is a substantial correlation between bullying and suicide rates the causal relationship between bullying and suicide is much weaker than this correlation suggests. To be specific, a one standard deviation increase in bullying predicts a 68% increase in the odds of someone attempting suicide. However, if you use a twin study design to control for genes and the family environment, and also control for psychopathology that pre-existed the bullying, we find that a one standard deviation increase in bullying only predicts a 25% increase in the odds of attempting suicide (Oreilly et al., 2021). Thus, only roughly a third (37%) of the commonly noted association between bullying and suicide is plausibly causal.

As we will see below, sexual minorities attempt suicide at a rate that is at least three times the rate at which heterosexuals attempt suicide. Given the weak causal impact that bullying has on suicide, for bullying to explain the elevated LGB suicide rate the average LGB person would need to be bullied at a level that fewer than one in one hundred thousand heterosexuals are. This is obviously completely implausible.

In fact, a meta-analysis of 15 studies found the vast majority (74%) of LGBT youth (age: 12 – 25) who have attempted suicide answer “no” when asked if they’ve personally been the victim of various sorts of anti-LGBT discrimination and bullying (Williams et al., 2021). And of course even for the 26% that did report having been victimized we can’t assume that this victimization caused their suicide attempt since, as already seen, the relationship between bullying and suicide is mostly non-causal.

That discrimination does not completely explain gay suicide rates is also evidenced by several studies finding that sexual orientation continues to be a risk factor for suicide after self reported victimization is controlled for.

For instance,  Garofalo et al. (1999) found that homosexuality continues to be a risk factor for attempting suicide after controlling for drug use and a scale measuring the degree to which one has been the victim of violence.

Similarly, Shields et al. (2012) analyzed data on 2,154 students attending high-school in San Francisco and found that LGB sexual orientation was associated with a 2.6 times elevated risk for suicide even after controlling for self reported victimization, race, gender, and drug use. Finally, Ballard et al. (2016) found that LGB status significantly predicted drug use and suicide risk even after controlling for self reported victimization in a large sample of American gay youth living in Appalachia.

It is also worth noting that the suicide rate of homosexuals is far greater than average even when only looking at those homosexuals who claim to have not significantly been victimized by discrimination. For instance, Ferlatte et al. (2015) analyzed data on 8,382 gay Canadian men and asked their participants about experiences of bullying, sexual violence, verbal abuse, physical violence, and discrimination at work. While experiences of marginalization correlated with suicidal behavior, the rate of suicidal ideation among those who reported having experienced no discrimination was roughly twice than of heterosexual Canadian men.

Similarly, Russell et al. (2010) found a very high rate of suicide attempt even among gay adolescents who self reported having experienced low levels of LGBT related school victimization.

Once again, I should add that the suicide rate of non-bullied LGB individuals should not be taken to be representative of what the LGB suicide rate would be if we got rid of discrimination because individuals who experience little bullying generally have better mental health than those who are bullied for reasons independent of bullying itself.

Another relevant observation is that the self reported happiness of homosexuals in the United States has slightly declined since the 1980s. Over the same time period we’ve seen a huge decrease in anti-gay discrimination and a large increase in gay legal rights, but all of this progress has seemingly done nothing for the psychological well being of homosexuals.

We reach a similar conclusion looking at suicide rates over time. Meyer et al. (2021) used data from a sample of 1,518 LBG American adults surveyed by Gallup. They divided their data into three age cohorts and found that the rate of life time suicide attempt had increased over time from 21% among those born in 1956-1963, to 24% among those born in 1974 – 1981, to 30% among those born in 1990-1997. The increase between the oldest and youngest cohort was statistically significant. This change is especially striking given that people born in the 50s have had far more time to accumulate a lifetime suicide attempt than have people born in the 90s.

In an analysis of data on 96,893 LGBT veterans, Lynch et al. (2020) found that the rate of suicide mortality had been fairly constant between 2006 and 2017 after going up and down in the early 2000’s.

Turning to data on adolescents, looking at data from Massachusetts, Liu et al. (2020) found that between 1995 and 2017 the trend in how the rate of suicide attempts was changing did not differ between LBG and heterosexual adolescents.

By contrast, using data from the states of Delaware, Illinois, Massachusetts, Maine, North Dakota, and Rhode Island, Raifman et al. (2020) found that the suicide attempt gap between LBG and hetero kids had decreased from a factor of 5.2 in 2009 to a factor of 3.8 in 2017. However, this change was not statistically significant. Moreover, the same paper finds that a different picture emerges if we define adolescents sexuality in terms of sexual contact rather than mere identity. Over this period, the gap in rates of past year suicide attempt between those who have and have not had sex with members of their own sex increased. In 2009, non-hetero behaving kids were 2.6 times as likely as hetero behaving kids to have attempted suicide. By 2017 this figure had increased to 3.3. However, this change was not statistically significant.

On the whole then, the evidence suggests that the gay-straight suicide gap has probably not narrowed over time while the absolute rate of suicide among gay people has either remained the same or increased.

The suicide rates of homosexuals in gay friendly areas also gives us reason to doubt that their mental problems are entirely a result of discrimination (Caputi et al., 2017; Shields et al., 2012; Bjokenstam et al., 2016; Ferlatte et al., 2015; Meyer et al., 2008; OMH., 2016; Bergen et al., 2013).

All this data suggests that, if anything, homosexuals in leftist nations and cities are at a higher relative risk for suicide than is the average homosexual in the United States. This may be surprising since bullying does have an effect, even if weak, on suicide. These findings can be reconciled by positing that the subcultures that gay people tend to establish in left wing communities are even more damaging for mental health than is being bullied, though this is just speculation.

Regardless, on the basis of the lines of evidence I’ve just reviewed it seems likely that discrimination either causally explains a modest fraction of the straight-LGB suicide gap or close to none of it.

As is well known, non-hetero sexualities has increased in popularity in recent years. The fact that suicide rates for non-hetero people are remaining so high even as we’ve created a world in which increasingly normal people identify as LGB and a culture which is generally accepting of homosexuality is cause for concern. It may be that abnormal sexualities, at least as they are currently being manifested in culture, cause defects in mental health. To the degree that this is plausible we should view this change in sexuality, at least in so far as it is real, as a problem. In the past we had a set of norms such that homosexuals were evidently no less happy than they are now but their numbers were not rapidly gaining. I’m naturally inclined towards a fairly liberal attitude about homosexuality, but the concerns I’ve just explained make me more sympathetic towards a more right wing approach.

But according to a recent WJS poll, Hispanics report planning to vote for Biden over Trump by a statistically insignificant one point margin (44% for Biden, 43% for Trump). The same poll finds that 37% of Hispanics would support a generic republican or democrat for congress. This tie would represent a radical departure from 2018 when Hispanics broke for the democrats by 40 points (Krogstad et al., 2018).

The extremity of these results is caused mostly not by Hispanics coming over to republicans but by them abstaining from favoring democrats. Similarly, according to Gallup polling, the Party ID of Hispanics was inline with expectations in early 2021 but over the course of last year a modest move away from democrats, but not to republicans, was evident (Newport, 2022).

This is consistent with favorability data showing that over the course of 2021 Biden’s favorability fell among Hispanics from nearly 70% to just below 50% (Skelley, 2021).

So multiple sources seem to indicate that Hispanics are moving away from the Democrat party possibly by so large of a margin that they on net wouldn’t benefit either major party. These shifts are so large they more than make up for the usual degree of error in polling for elections a year away (Erikson, 2016).

We might also note that according to an exit poll by Fox, Hispanics voted republican in the Virginia Governor race by a 12 points margin. This too would be a large change relative to historical norms, but these exit polls will be somewhat biased in favor of republicans because a large number of people voted by mail in that election.

Such a change requires an explanation. For years, Republicans have wanted Hispanics to vote republican on the basis of right wing social views, but such social views have been unrelated to how Hispanics have voted (Last, 2018)

(Columns 5 and 6 are Hispanics party ID is being predicted with a positive coefficient indicating the variable is related to increased republicanism, columns 1 and 2 are Asians, 3 and 4 are Whites, 7 and 8 are Blacks).

More generally, even among self identified conservatives Hispanics on net have identified as democrat (Last, 2018).

This has even been true of Hispanics who oppose immigration (Last, 2018).

And historically analyses of support for congressional candidates based on their NumbersUSA grade finds that Hispanics are no more likely to vote for Republicans when Republican politicians are pro-immigration (Hawley, 2013). (A is very anti-immigration, F is very pro-immigration).

Against this background, a shift in Hispanic voting seems surprising. A point to raise in favor of skepticism is that the GOP currently has a rather ambiguous existence and there’s not a strong sense of what they are about or who they are best represented by. The departure of Trump has also made it harder for the media to constantly engage in plausible intense negative coverage of the GOP. These numbers might move back towards the norm once the GOP takes a more concrete form.

It’s also worth noting that there hasn’t been any historical analysis of how Hispanic voting relates to COVID since it is a new issue. According to some polling, COVID is now the top issue for minorities. But polling shows only 24% of Hispanic Americans disapprove of how Biden has dealt with COVID policy.

Critical Race Theory is also a new issue, kinda. At least, it being talked about a lot in those terms is new. YouGov polling finds that, among those who know what Critical Race Theory is, 53% of Hispanics have an unfavorable view of it compared to 22% of Blacks. Past research on how Hispanic Americans feel about affirmative action and related policies has produced mixed results, so it is not that shockingly that they might oppose CRT. Moreover, while Hispanics who are right wing on issues like abortion, immigration, and gay marriage still vote democrat Hispanics who are right wing on race do not (Last, 2018).

Evidence on how economically right wing Hispanics vote has been mixed, it depends on the specific question being asked. But satisfaction with the economy is usually important in how people vote. Gallup finds economic satisfaction fell a good deal in the last year. And YouGov polling indicates only 29% of Hispanics think the economy is getting better. Dissatisfaction with the economy might explain why Hispanics seem to be moving away from democrats more than moving to republicans.

If I had to guess, I’d say that the polling data probably is overstating the change in Hispanic voting but that a more moderate change has occurred and that the most probable explanation is some combination of racial politics and the economy. However, figuring out how groups will vote is hard and finding out why they vote the way they do is even harder, so I wouldn’t put much confidence in these conclusions even if I think they are the best guesses based on the data available at this time.

A few things are clear. First, assuming someone got the right answer, the probability that any given person got the right answer is only 10%. Second, each person could not point to the fact that their reasoning seemed clearly correct to them and other’s reasoning seemed clearly in error to justify thinking they got the right answer since this perception will be had by all the people including those who got it wrong. They would need some further independent justification and without that they should simply note that there is an answer which seems most reasonable to them but there is only a 10% chance they are right and a 90% chance they are wrong.

There are important areas of human thought that seem somewhat analogous to this. The odds of someone being correct are not always as low as 10%, but they seem clearly too low to warrant confidence and make clear that the feeling that your line of thought is obviously right and other’s obviously in error don’t track truth very well in these domains. Some obvious examples are religious faith and moral intuitions. If we think of these as “methods” it is obvious that they are unreliable methods in the sense that different individuals and populations using the same methods produce very different results.

Methods of knowledge where this sort of analogy clearly does not apply include simple perception and basic mathematical intuition. These methods produce basically the same results for everyone and so can be considered reliable in the relevant sense.

None of this is very concerning to me, but what does bother me is the potential for this sort of analysis to be applied to a method we can broadly call rationality, trying to look at the relevant evidence and infer the most probable conclusion, in the domains of social science and philosophy. Comparing individuals within cultures and people between cultures at different places and times it is clear that many people who thought they were rationally forming beliefs about many philosophic and social scientific questions were wrong, so unless we have some additional reason for certainty we have to conclude that if we are such a person then there is a high chance, for many questions greater than half, that we are wrong. This will be true no matter how clearly we seem right and other seem wrong so long as, so far as we can tell, people who think we are wrong have similar feelings about their own line of reasoning.

One thing we’d hope to be able to appeal to in order to get out of this is knowing lots of facts or trying to follow some more specific methodology like “the scientific method”. But a quick look at highly informed laymen, intellectuals, and people who claim to be engaging in a scientific approach to social science or philosophy even within current western culture, let alone across cultures from elsewhere, makes it clear that a great diversity of opinion exists even among this subset of people. This problem is even more severe if, like me, you think most intellectuals and social scientists have failed in their use of reason and consequently believe lots of false things.

We might be tempted to say that our preferred belief was produced by the more correct understanding and application of the scientific method and by a superior selection of relevant facts. We might even feel that people who disagree with us would agree if they could see our facts and understand our method. We all know this doesn’t happen often in debate when we attempt to do exactly that, but we can blame that on the irrational mental state of the people we disagree with. But of course, lots of people think all these things while holding a diverse range of views on many topics and so us feeling this way doesn’t help us justify trusting our own reason more than theirs when applied to these sorts of issues.

If this argument is valid, it may not shake our feeling that our reasoning on these sorts of topics is trustworthy. But it would imply that this is implicitly a belief that we have a special ability to tell apart good and bad arguments and form beliefs which most people do not have and this belief is not only unjustified but, so far as the evidence we have indicates, is probably false.

This argument is fairly simple and it seems like the sort of idea that certain people would have thought worth talking about so I assume it has a name. I don’t know its name though and my hope is that someone who reads this will be able to inform me of its name so that I can see what sort of responses people have thought up to it. Beyond that the point of this article was just explaining the argument in a hopefully clear way without trying myself to rebut it.

First, let’s look at the graph he uses to visualize this trend.

Murray (2021)

Something which immediately pops out is that the proportion of data points coming from “cognitive batteries” significantly decreased around 1990. The black dots mostly represent NAEP data and are not based on formal IQ tests. How problematic this is will depend upon how much formal IQ tests share with the NAEP.

But the truth is that IQ tests don’t even share enough with each other for this sort of aggregation to make sense. On average, roughly half of the variance in a given IQ test overlaps with the variance in ability measured in other IQ tests. The other half of variance in ability an IQ test measures usually reflects specific abilities which that test measures but which are not common across most IQ tests.

(Herrnstein and Murray, 1996)

The variance which IQ tests do share is due to the g factor. After extracting a g factor, the residual correlation between mental abilities is, on average, roughly zero (Gignac, 2016). While full scale IQ scores taken from different tests only share roughly half of their variance, g factor scores taken from different tests share virtually all of their variance in common. This fact will be important later.

Returning to Murray’s analysis, he admits that the non IQ test data he is using “tend to understate differences in g that would be revealed by full-scale cognitive test batteries”. Given this, such tests probably don’t even share half of their variance in common with IQ tests. Consequently, even though they are plotted on the same axis the variable being measured is meaningfully changing with time. For this reason, comparisons of earlier and later periods will obviously be invalid.

For Murray’s estimate of the current racial IQ gap, he uses ten tests, only one of which is an actual IQ test. This is because he restricts himself to only using data from 2010 or later and it just happens than not many IQ tests have been normed since then. This decision to restrict the analysis to very recent data might make since if we thought that the gap before 2010 was significantly different than the one after it. But Murray’s own story is that the gap has been pretty stable since the mid 1980s. Given this, such a restriction makes no sense.

Murray calls his choice to only use very recent data points which are almost entirely not derived from IQ tests “conservative”. Normally, a conservative decision would be one which minimizes the risk for error, but in this case it seemingly involved introducing unnecessary and obvious error into the analysis in order to alter the result.

Turning now just to IQ tests, before I said that they themselves are too heterogenous to be simply aggregated and this fact is easily demonstrated.

Consider the following:

• According to the WISC, the B/W gap stayed at 1.2SD from 1972 to 1989 and then fell to 0.89 in 2002.
• According to WJ data, the gap fell from 1.33d to 0.75 between 1976 and 1987 but then increased back to 1.02d by 1998 before falling back to 0.76 in 2012.
• The WAIS data says the gap was only 1.00d in 1978, fell to 0.92d by 1995, but then rose to 1.06d, its all time high, in 2006.
• The SB shows a decline from 1.11d to 0.98 from 1986 to 2002
• The AFQT shows a decline from 1.35d (1972) to 1.24d(1980) and then 0.98d(1997).

Notice that these tests generally don’t show the same trends over time and, perhaps most importantly, literally none of them show the B/W IQ gap falling significantly below 1SD in the 1980s and staying put since. Murray’s narrative is an artifact of combining data that should not be combined to produce results that no coherent dataset actually shows.

I think we can produce a better estimate of the current racial gap, but before turning to that I want to note some oddities in Murray’s data. First, it is interesting that in 2007 Murray reported that the WJ1, 2, and 3, samples showed B/W gaps of 1.23d, 0.90d, and 1.05d.

Murray (2007)

The reason this is interesting is that in his new dataset Murray reports the gaps from WJ1, 2, and 3, as 1.33d, 0.75d, and 1.02d. I don’t know what explains these differences but the effect of them is to increase the degree to which the WJ data shows a decline in the B/W IQ gap over time. The WJ4 data Murray is now reporting is based on unpublished data, so not much can be said about that.

I also want to note that the inclusion criteria Murray used when selecting tests is non-obvious. For instance, he included a sample of the KAIT, even though previous research on trends in the B/W IQ gap have excluded this data due to the test being constructed so as to minimize racial gaps (Dickens and Flynn, 2006). There are also various tests (Wonderlic, DAT, etc.) which could have been included but for whatever reason weren’t. Because Murray’s work is aimed at a popular audience, him lacking an explanation of this is not surprising, but I thought I should note that the inclusion criteria being used here is not self evident.

Anyway, as we’ve seen unlike is the case with comparing full scale IQ scores generated from differing tests, or IQ scores to scores on non-IQ tests, g factor scores should be virtually the same regardless of the test they happen to be extracted from. For this reason, g factor scores are far better suited for this kind of analysis. After briefly looking for estimates of the g factor gap between Blacks and Whites, I was able to find a handful of studies, each using a large and quality sample, and they all showed the B/W gap still being moderately above 1.0d even after the early 80s.

Because it is g that drives racial disparities in IQ to begin with, as well as the role IQ plays in explaining racial inequality, I feel that this gives us a better picture of what is happening with the racial IQ gap over the last few decades. Racial differences in non-general abilities are normally small and sometimes favor blacks, and the gap in those abilities may have changed with time, or with time we may be including tests that more heavily measure such abilities, but any investigation of this would require using the same specific measures repeatedly and by its nature is not susceptible to aggregation across tests. Moreover, it would likely not be of much more than academic interest, since it is differences in general ability that produced the gaps we all know and care about.

Incomes Have Not Stagnated 

In the early 2000’s research was published which claimed that the median American’s income hadn’t changed much since the 1970s. This research was heavily flawed, it failed to account for changes in household size, taxes, in-kind transfers from employers, and government transfers, and made no effort to measure income often not reported to the government. Today, we have far better research which accounts for each of these flaws and finds that the median American’s income increased by quite a lot over the last few decades.

Rose (2018)

Looking at a longer time span, the differences that these methodological adjustments make is made especially clear in this chart taken from Elwell et al. (2019), which shows that adjusting for things like household size and transfer payments makes the difference between thinking the median income has only increased by around 5% since 1959 and seeing that it has actually more than doubled in that time. 

Contemporary estimates of how the incomes of people at the bottom of the income distribution have changed since 1970 also display a growth of 44%. 

(Horowitz et al., 2020)

Even this research is misleading though, because we’ve introduced new people into the American population over that time period. This fact can lead to counter-intuitive scenarios. For instance, if immigrants come here and increase their income relative to what it was prior to immigrating, but still have incomes below the median American, then, all else being equal, this will lead to a “decline” in the median American’s income even though no one experienced a decrease in income. 

We can account for this flaw by looking at research which follows the same people over time. Such data shows that the median income of Americans in every class increased greatly between 1980 and 2012. Moreover, the greatest proportional gains went to those in the bottom quintile of the income distribution. Their incomes more than doubled over this period. 

Roberts (2018)

Other research has found that nearly all households in the bottom 20% of income in 1975 were no longer in the bottom 20% by 1991. 

Horwitz (2015)

Income mobility was lesser between 1996 an 2005 (a shorter time span), but most households in the bottom 20% still managed to escape that quintile over that just 9 year period. 

Horwitz (2015)

It’s also worth noting that poverty statistics often give a misleading impression about how common poverty is in America.  The problem is this: often times people will report incomes below the poverty rate because they had a sudden, temporary, drop in income (for instance, due to regular, non-long term, unemployment) and as a result their income won’t reflect their life style. If we look at how many people were continually poor between the years 2009 and 2012, we find that only 2.7% of households meet this criteria of chronic poverty (Henderson, 2016). 

We should also note that poor people’s lives have been materially improved with time. For instance,, unlike in 1984, by 2005 most poor households had a clothes dryer, a microwave, and air conditioning and by 2013 most had a home computer (Horwitz, 2015; Wagner, 2015). So even if the number of poor people remained the same, which it has not, and their incomes had stayed the same, which they have not, their material conditions would have still improved. 

On the whole then, the material standards of living have risen over the last few decades for the median American and for poor Americans as well. 

Unemployment is Slightly More Common and Lasts Longer

Turning to unemployment, we can see that, prior to COVID, the unemployment rate was similar to what it was 40 years ago. But over the long run there is clearly a trend towards fewer people working. 

Hipple (2016)

This is significantly due to college and retirement. If we look at the labor participation rates of people in prime working age, 25 to 54, we see that there’s been a large increase in the proportion of women working and a slight fall in the proportion of men working from roughly 98% down to roughly 88%. This does represent an increase in unemployment for men, but it also means that nearly all men still manage to find work. 

Hipple (2016)

It is also worth mentioning that the average duration of unemployment is now around 20 weeks while it used to be below 15. 

BLS (2021)

Importantly, neither the change in unemployment nor the change in unemployment duration of the last twenty years can be blamed on a lack of job openings. As can be seen, prior to COVID there were actually more job openings than unemployed people and this was the best ratio of openings to unemployment seen since the statistic began to be recorded around 2000. 

BLS (2021)

It thus seems that, to a significant degree, the slight increase in unemployment that we’ve observed is due to some combination of people not wanting the jobs that are available or not having the skills needed to perform said jobs. Given this, the data doesn’t allow us to tell to what degree unemployment among working age men has moderately increased due to a change in employment preferences as opposed to employment opportunity. Regardless, this moderate change falls well short of the picture often painted by populists. 

People Spend Less on Necessities 

Some people will say that even though certain luxury technologies have become cheaper, thus raising our total “material standard of living”, living essentials have greatly increased in price leaving normal people with nearly impossible bills to pay. Before looking at some of these living essentials in detail, I want to note that the proportion of income spent on housing, clothing, food, and utilities, has been declining for decades.

Pinker (2019)

Of course, the two major items not included in this list are college and health insurance. But notice that these things were not thought of as necessities until very recently, and so it wouldn’t make since to put them on a chart going back to a time when almost no one had either of these things. I will look at education and health insurance individually, but in both cases it is important to remember that we aren’t just talking about a change in our material condition, we’re also talking about a change in our standards for what is considered necessary. But before turning to these items let’s look at housing since it is the other major cost that people commonly talk about increasing with time. 

Homes Cost More Because They Are Larger

Adjusting for inflation, living spaces are more expensive than they used to be. For rent, prices increased by more than 60% between 1960 and 2014. 

Woo (2016)

However, our homes are also much larger than they used to be.

Perry (2015)

If you adjust for dwelling size and look at price per square foot, we find that the median home price was basically the same in 2014 as it was in 1973. 

Perry (2015)

Granted, it has increased slightly in the last five years, time will tell if this is another housing bubble, but even if it isn’t this represents a very small increase over a long period of time which saw many improvements to our housing standards (e.g. regulations concerning building materials like lead, clay, and asbestos) which could easily justify a small increase in price. 

Latham (2020)

It’s also worth noting that despite our homes being larger and safer than in the past, the rate of homeownership has pretty much stayed constant, though it has shifted up in the age distribution. To some degree, this simply reflects the fact that more people go to college and marry late than used to be the case. 

Drum (2019)

On the whole, housing prices leave little to complain about outside the fact that we as a society have shifted our standard for how large homes should be and so now it would be more difficult for an individual to purchase a home at the size that was typical 50 years ago. 

Education Costs More But Is Manageable 

Turning to education, the cost of college has more than doubled since the 1980s (NCES, 2018). The economic value of going to college, the degree to which each dollar spent on college predicts an increase in your future income, is higher than it used to be too and makes college a better investment than most plausible alternatives. But of course this is largely because so many employers now require a degree for jobs that used to lack such a requirement. So now college both costs more and is more necessary than in the past. 

Abel et al. (2019)

How much college costs varies greatly by college type. For public universities, the typical tuition and associated fees comes out to roughly 11k per year while that figure is roughly 38K for private universities. On top of tuition students must find a way to pay for normal living expenses not directly caused by school while attending classes. 

College Board (2019)

In part, the solution society has set up for students is grant money. For the typical student at a public two year school, the grant money they receive will be greater than is the cost of their tuition so that it will cover all those expenses and a little of the cost of living. 

College Board (2019)

For the typical student of a four year public school, grant money will account for around 65% of their tuition cost. Due to the remaining tuition cost and living expenses, they’ll still spend around 15K per year  of their own money to get by. 

College Board (2019)

For private schools, grant money typically covers only around a third of their tuition and these students spend another 27K of their own money on tuition and living expenses. 

College Board (2019)

Students also often cover living expenses by living with their parents. This is true of roughly one in four students in general, and almost half of students from low income backgrounds. 

Roughly 43% of students also work while in college (NCES, 2018). The median income of student works is $3,900 among full time students living with their parents and $13,880 among students not living with their parents (Urban Institute, 2019).

These figures suggest that, between grants and work, students can generally make enough to fund their life style and college classes before taking out any debt. It is therefore unclear exactly why the median college graduate takes out roughly $25,000 in debt. 

(Cilluffo, 2019)

Possibly, this is because many students prefer to not work while in college or to go to very expensive universities.

Regardless, the average starting salary of college graduates is $50,944 (NACE, 2019). The median may be a bit lower but reaches this level fairly quickly and rises far above it in people’s 30s and 40s. 

Day (2019)

So most people could pay off their college debt in a handful of years if they really wanted to. For the median college graduate, this would simply require living for a few years at the material standards of people who haven’t gone to college. Despite this,  Hess (2019) says that “The Department of Education reports that the typical repayment period for borrowers with between $20,000 and $40,000 in federal student loans is 20 years, and a 2013 study of 61,000 respondents conducted by One Wisconsin Institute found that the average length of repayment for student debt borrowers is 21.1 years.” This obviously reflects a financial preference among college graduates and is not because the debt is so high that it necessitates decades to pay off. 

In sum, despite rising prices there’s nothing in this data that makes education seem wildly or unmanageably expensive. There are irresponsible decisions someone could make about what schools to go to, whether to work while in school, and what kind of life style to have during and after school, that could land someone in a large amount of debt but this is by no means a necessity nor is the possibility of irresponsible decisions landing you in a bunch of debt particular to school. School is expensive, but it also has a large pay off for most people and the costs are typically very manageable. 

Healthcare

Next let’s turn to healthcare. As is well known, spending per person on healthcare has increased dramatically over the last half century or so. 

Cox et al. (2020)

Of course, people also live longer than they did half a century ago, despite the increased prevalence of obesity, and that is in large part because health care has improved and more people have gotten access to it. So it’s not as if we are paying more now for the same set of services we got in 1970.

In fact, Dunn et al. (2020) found that health care costs actually declined between 2000 and 2017 if you adjust the cost of health care by its quality. In other words, its become cheaper in terms of the actual expected medial benefits but the range of quality you can pay for has improved so that the raw cost is still higher. 

In large part, our response to the rising cost of healthcare has been to give everyone insurance. By the 1990’s we’d accomplished this for old people. 

Moon (1996)

Among the total population, the rate of being uninsured fell by more than half between 1963 and 2014, leaving around 10% of people uninsured. 

CDC (2015)

And this 10% is disproportionately made up of young adults. 

McCarthy (2019)

So we need to look at how much people are paying for healthcare both directly and through insurance. Among non-elderly people who have health insurance through their employer, the median cost is pretty low and even the 90th percentile isn’t as high as you might think. 

Hayes et al. (2019)

The cost for old people on Medicare is higher, but still not incredibly high even at the 90th percentile: “In 2017, the average beneficiary in traditional Medicare spent $5,801 on insurance premiums and medical services. One in ten people with Medicare spent at least $10,268.” – Miller (2020)

So, in the US, you can expect to pay a something like 3 – 6 grand a year on health care during your working years and a bit more when you’re old. Probably, you won’t need to spend much money at all when you are young. And if you’re poor the government will almost entirely cover your health care costs through Medicaid. But you should probably save some money for years down the line when you’ll have to spend more than average. All this is a significant expense, but, as with the other expenses I’ve looked at, it also seems obviously manageable for the vast majority of people. 

Speculations and Conclusions

When it comes to income, education, housing, and healthcare, our system works reasonably well for the vast majority of people. So why has populist rhetoric to the contrary become so popular? Here are some speculative explanations:

First, in each case there are exceptions to “the system” generally working well. Some people are greatly hurt by these systems, sometimes through no fault of their own, and they are talked about an awful lot. By contrast, no one talks about how these things worked out fine for them. There are probably a few reasons for this. First, things being fine isn’t an interesting story. Second, if you’ve heard lots of people talk about how these systems have screwed them over then you might feel like it would be rude to talk about them treating you well as opposed to condemning them. Third, outing yourself as a non-victim might be socially unprofitable because victimhood is seen as a virtue in our society. And forth, everyone can convince themselves that these systems haven’t been good enough to them because no matter how good of a life you’ve had there will always be ways it could have gone better. So the stories people hear, both on the news and in person, tend to support populist rhetoric.

A second contributing factor is the fact that social scientists and politicians frequently tell people that the data shows the economy has screwed over all the regular people. For politicians, saying this is profitable because then you can say you will fight back against the elites for the people and thus try to capture votes. Social scientists are mostly left wingers who fail tests of basic statistical literacy, so their involvement in this is entirely unsurprising (Last, 2020).

These are the sorts of forces I’d guess are at play which explain how it is that economic populism has become so popular in a society where regular people’s economic position has been improving for decades and the economic system continues to work reasonably well for most people. Of course, we might still be concerned about the rare exceptions to the system working well, or we might argue that there are signs that the system will cease working well for most people in the future. But neither of these concerns could fuel an economically populist movement the way saying that most people are getting screwed over right now does.

To conclude, I’d like to reiterate that I’m not saying the modern system of politics and economics doesn’t have serious problems. I’m just saying that not producing enough material stuff for a sufficient number of people is not one of those problems.

The basic idea behind the tests is this: due to culture, races have been exposed do differing sorts of information and so to get an unbiased test we should base the tests on information neither race has been previously exposed to and give them a brief exposure period to it in a controlled environment so can ensure that each race is given an equal opportunity to learn said information. For instance, we could allow people of each race to study the definitions of words they’ve never heard before and them give them a vocabulary test on those words after their study period. When we do this, advocates of the theory say, we find that races do not differ in their ability to acquire new information and so are equally intelligent.

There are many problems with this approach. Firstly, unless participants are made to acquire the novel information in a complex way or to perform a complex operation on the information in order to score well on a test, this will be a test of rote memory which is a pretty simple cognitive ability. Standard IQ tests include a test called a “forward digit span” which consist of reading participants a list of numbers and then seeing how many they can repeat back. Forward digit span tests seemingly measure the same cognitive ability that these “culture free” tests do, and we already know that black and white Americans either do not differ in forward digit span score or differ in score only slightly. By contrast, if you have people repeat back the list of numbers in the opposite of the order which they heard the numbers in, the so called backwards digit span test, you see significant racial gaps. This has been shown numerous times.

Citation	N	FDS	BDS
Dalliard (2013A)	864 / 407 (b)	0.05 (ns)	0.41
Dalliard (2013B)	1609 / 265 (b)	0.01 (ns)	0.43
Dalliard (2013B)	2904 / 481 (b)	0.04 (ns)	0.41
Dalliard (2013B)	3085 / 582 (b)	0.16	0.36
Jensen (1975)	669 / 621 (b)	0.25	0.57

Of course, both digit span tests concern the same information, so the difference in how each race performs on the two tests cannot be due to a difference in the information required. Rather, the difference reflects the fact that backward digit span tests requires test takers to perform a slightly complex operation on the information they’ve been given and the races primarily differ in their ability to perform complex operations on information rather than their ability to learn simple new facts. Thus, these culture fair tests may cause racial gaps to disappear because they remove complexity from IQ tests.

Another problem with this research is that the samples used all range from so small as to be useless to moderately inadequate. No study, including meta-analyses done with this research, contains even 200 black participants. Most research in this area has utilized samples of fewer than 50 blacks.

Troublingly, when samples are large enough to merely be moderately inadequate, Fagan and Holland consistently fail to actually report what the mean racial differences in their culture fair test scores are. Instead, once the tests were slightly improved to add a small degree of complexity to them and larger samples began to be utilized, Fagan and Holland only ever reported racial differences in their test after controlling for SAT scores.

The single exception to this involves a sample which is provable unrepresentative with respect to racial gaps in intelligence. And this brings up another problem, the representativeness of the samples utilized are usually suspect and in some cases demonstrably invalid. There is an instance where Fagan and Holland try to show that their samples are representative, but it relies on faulty statistical reasoning.

Another problem with this research is that it cannot test whether racial differences in information exposure are due to racial differences in behavior which itself may be caused by genes (e.g. the propensity to read) or some ultimately environmental factor. For this reason, it is obviously not a valid test of the hereditarian hypothesis.

With those general criticisms laid out, let’s turn to analyzing this literature one paper at a time (there are, to my knowledge, only four papers in this literature and they’re all done by the same two people and this actually constitutes another notable problem with this line of evidence).

Fagan and Holland (2002)

This line of work began with Fagan and Holland (2002). In this paper, participants in the trained group were given time to study the definitions either of rare words they were unlikely to have heard before or words taken from the PPVT, a common test of verbal intelligence. They were then given a test to how many words they remembered the definition of. Participants in the “untrained” group were either tested using PPVT words with no training or were tested on normal vocabulary words, again with no special studying of the definitions before hand.

As can be seen, in the trained condition significant racial gaps occurred (usually this means white vs blacks but sometimes the authors combine other groups like Hispanics into a “minority” category and minorities like Asians into a “majority” category, gaps in SD units). By contrast, in the untrained condition racial gaps were consistently statistically insignificant.

Obviously, this lack of significance shouldn’t be taken to mean much of anything given the sample sizes we are dealing with. In any case, these results are consistent with the trained condition being a test of simple memory ability.

To determine whether their samples were representative, Fagan et all administered the PPVT-R to a new sample of 97 students (67 whites and 26 blacks). They found that “In the present study, the IQs of the 67 Whites averaged 98.1 (S.D. 12.4, range 67 to 131) and the IQs of the 26 Blacks averaged 82.6 (S.D. 13.7, range 62 to 112), a difference of 15.5 IQ points (t = 5.3, df 91, P < .0001).” These groups means are a bit low, but the racial gap is indeed representative of the general population. This finding cannot, however, be taken to demonstrate that the previous four experiments utilized a representative sample because you cannot determine the representativeness of these populations using sample sizes of 26 and 67. Even if this specific sample was representative, the sample is so small that the difference between this sample and one which was significantly unrepresentative would obviously not itself be statistically significant. Consequently, we lack justification for thinking that the first four experiments used a representative sample and, given that the participants all attended the same colleges, we have some reason to doubt their representativeness until better evidence is provided.

Clearly, all of my general criticisms concerning sample size, sample representativeness, the complexity of the tests, and the inability to actually test hereditarianism, apply to this paper.

In their paper, Fagan and Holland also tro to evidence that IQ tests are racially biased. By biased, I mean that the same IQ score does not indicate the same underlying level of intelligence for blacks and whites. Fagan et al. report that they compared 57 blacks and 87 whites who had roughly the same scores on the general vocabulary test and found that in this same group of individuals blacks scored higher than whites on the new word vocabulary test. This is not a compelling demonstration that IQ tests are racially biased for at least three reasons:

1. The sample size is to small to justify any conclusion.
2. Neither test is an adequate measure of IQ or general intelligence.
3. Ignoring 1 and 2, this would evidence that lack of exposure to information is a cause of the black-white IQ gap rather than evidencing that vocabulary tests don’t measure the same thing in both groups.

Fagan and Holland (2007)

Let’s turn now to the various experiments reported on in Fagan and Holland (2007)

In experiment one (n=77 college students), participants were either tested on their ability to recognize faces to which they were recently exposed to for the first time or they were tested on their knowledge of vocabulary words taken from the PPVT. The black-white gap in PPVT score was 1.1 SD. Surprisingly, we are not told what the racial gap in simple memory ability is. Instead we are just told that a regression model using race and native language status to predict memory ability is statistically insignificant while a model using PPVT scores achieve significant predictive validity. No regression output is displayed so that we can see what the beta coefficients are. Instead, the authors give the following verbal report of the analysis: “Specifically, a stepwise multiple regression in which the independent variables of race, native language, and PPVT-R scores were used to predict recognition memory ability revealed that knowledge (PPVT-R scores) was the only significant predictor of processing ability (recognition memory) with a Multiple R of .36 (F(1/71)=10.4, Pb.002) and a beta value of .36 (t=3.2, Pb.002). If PPVT-R scores are omitted from the analysis, and race and native language are the only independent variables employed to predict information processing, no significant variance is obtained (F(2/70)= 1.9, PN.15).” Given the sample size of 77, this is obviously totally inappropriate. In sum, no attempt was made to show that this sample was representative, the sample is too small to test anything useful, and this is very obviously just a simple memory test and not a test of general intelligence.

In experiment two, Fagan et al gave 65 community college students one of two tests, one of which measured whether they understood various sayings requiring specific knowledge based on past exposure to specific information and one of which measured whether they understood sayings which required only general knowledge. The give knowing the meaning of “an apple a day keeps the doctor away” as an example of general knowledge and knowing that “home of the bean and the cod” refers to Boston as an example of specific knowledge.

Unlike in experiment one, this time we are given the results of each test by race: “comprehension based on specific knowledge on the part of the Whites at 65% correct (S.D. 20.7) was superior, t (35)=2.7, P<.01, to the comprehension of the African Americans at 48% (S.D. 10.0). Thus, when opportunity for exposure to information is allowed to vary, Whites are more apt to know the meanings of sayings than are African-Americans. But the same was not true when opportunity for information about the meanings of the sayings was generally available. Specifically, The performance of the Whites at 72% correct (S.D. 16.5) was, if anything, somewhat inferior to that of the African Americans at 80% correct (S.D. 13.1), although not significantly so, t(26)=1.4.”

Speaking of their second experiment from their 2007 paper, they write that “Whites and African-Americans, also matched for PPVT-R scores, differed significantly in general comprehension, t(16)=2.0, P<.03, one-tailed test. The African-Americans at 85.4% (S.D. 8.2, N=9) were superior to the Whites at 75.7% (S.D. 12.4, N=9).” of course, a sample size of 9 per group makes this worthless, a one sided significance test was presumably used because the difference wouldn’t have been significant other wise, and if we ignored all that we’d just conclude that African Americans score better on their test of “general comprehension” than they do on IQ tests which is something we already knew.

This experiment was largely repeated in experiment three where 68 community college students took a test on both sorts of sayings. They found that “comprehension based on specific knowledge on the part of Whites (a mean of 16.2 items correct, S.D. 3.4) was superior, t(84)=2.6, Pb.01) to that of Blacks (13.9 items, S.D. 3.3) while general comprehension was equal for Whites and Blacks at 17.8, S.D. 3.1 and 17.1, S.D. 3.3, respectively, t(84)=0.8.”

Again, the sample size make this work largely worthless. And understanding a saying is generally no more complicated than understanding a word meaning this measure still lacks any cognitive complexity. If we set the sample size issue aside, we might say that this is evidence that blacks and whites are equally likely to learn the meanings of sayings that are common to their environments, but that is not the same thing as saying that they are equally intelligence when given equal opportunities to become smart because knowing sayings is not the same as intelligence.

Fagan and Holland actually (seemingly unknowingly) prove this point. They claim that both of their sayings metrics correlate at 0.53 with IQ. This means they share about 25% of their variance with IQ and are primarily a measure of something other than the variety of cognitive abilities, and general intelligence, measured by IQ tests.

From their third experiment “A sample of 16 African-Americans and 17 Whites were closely matched on their comprehension of sayings requiring specific knowledge with mean scores of 60% (13.9 items correct out of 22, S.D. 3.3) and 62% (13.6 items, S.D. 2.9), respectively. Again, as in the second experiment, on general comprehension, the African-Americans were superior to the Whites, t(31)=2.1, Pb.05, with 77.3% correct (17.1 items correct, S.D. 3.3) for the African-Americans and 66% correct (14.6 items, S.D. 3.4) for the Whites. ” Again, the sample size is a joke and it is obvious that if you have two tests that are correlated and whites do better on one while both races do the same for the other, that if you compare people from each race who have the same score on the former then there will be a gap favoring blacks on the latter. This will be true, all else being equal, for any such two tests and is not evidence of test bias. (Even more obviously, it is not evidence that IQ tests are biased since IQ tests are not being utilized in this experiment!)

Moving on, in experiment four, 223 college students (130 from a private university and 88 from a community college) were given tests of their understanding of sayings, similarities, and analogies. To test knowledge of sayings, participants were given multiple choice questions directly about what sayings meant. In the similarities test, participants were given two words and then asked to select what they shared in common from several possible answers. For analogies, participants were given questions of the form X is to Y as Z is to what? And they were given several possibilities to choose from when selecting their answer.

Of the 223 participants, 179 were given these tests in the normal way and 44 were given the test based on nonsense words which they learned the meaning of just prior to the testing. In the group tested in the normal way, a 0.75 SD gap emerged favoring whites. In the group working with novel words, there was a statistically insignificant 0.20 SD gap favoring whites.

Much can be said of this experiment. First, the move to a sample largely from a private university makes it virtually certain that this sample is not representative with respect to IQ. Secondly, because only 44 participants were in the novel word condition this finding is basically worthless. Thirdly, the gap that was found in the novel word condition, 0.20 SD, is practically significant. Because of their small sample size, it was did not differ in a statistically significant way from zero, but it also doesn’t differ in a statistically significant way from a much larger effect size than 0.20. To say this is evidence of racial equality on this measure is to simply misunderstand the relevant statistics.

This experiment does improve on past designs in that the analogies test introduced a small degree of cognitive complexity into the test. Of course, this test also included a good deal of simple memory testing. It may be that the gap was significantly larger on, for instance, the analogies test than the sayings test. Unfortunately, the results are not given to us broken down by ability so we can’t tell.

Participants in experiments 3 and 4 were also given various other cognitive tests from which a G factor score was extracted. To determine their relation to general intelligence, Fagan and Holland’s tests should have been included in the set of tests from which G was extracted. We could then see the G loading of each of their tests. For unknown reasons, this is not what was done.

Instead, for each of the two experiments, they built a model predicting G scores using race, age, sex, educational attainment, family size, birth order, general knowledge, and specific knowledge. The total model explained 38% of the variance in G scores in both experiments. General knowledge had (standardized) beta values of 0.38 and 0.30 in these models while specific knowledge had beta values of .24 and .37. So in this model both general and specific knowledge didn’t share the vast majority of their variance with G.

In neither model was race a significant predictor of G, but that doesn’t tell anything useful given all the other variables included in the model and the small sample sizes involved.

The fourth experiment also involved another attempt to evidence test bias. They write “a sample of 45 African-Americans and 49 Whites were closely matched for their knowledge of sayings, similarities, and analogies based on specific information with mean z scores of −.08, S.D. 0.71 and mean z=− .06, S.D. .82, respectively, t(92)=−0.1. Given such equality, the African-Americans were now greater than the Whites, t(94)=5.0, Pb.0001, in knowledge based on general information with a mean z for the African-Americans of .29 (S.D. .81) and a mean z for the Whites of −.62 (S.D. .92). This z difference of .91 would translate into an IQ advantage of about 13.7 points for these particular African-Americans over these particular Whites.”

This comparison lacked an adequate sample, in terms of both size and representativeness, and this just tells us that if you control for a large group gap, a small group gap in a correlated variable flips direction, which is obvious and since neither correlated especially strongly with G as measured by normal intelligence tests this of course tells us nothing about whether IQ tests are biased.

Fagan (2008)

Let’s now turn to Fagan (2008), a paper which reported on two relevant experiments.

In experiment 1, the sample consisted of 484 students most of which attended a private university. These participants were given three tests measuring their ability to learn the meanings of new words, sayings, and analogies.

With respect to new word meanings, participants were first allowed to infer the meaning of these words through seeing them used in sentences. Importantly, these sentences made the meanings of the terms very clear and there is no indication that this inference was ever meant to be difficult. For this reason, we don’t have good grounds for supposing that there is much difference between this and simply giving them the definitions of the words. An example can be seen below.

After studying the meaning of the new words, participants were given a simple test measuring how many definitions they had been able to memorize in the study time given. On its face, this looks like a test of rote memorization.

To learn the meanings of sayings participants were simply given explicit definitions of sayings to study.

They were then tested, via multiple choice questions, to see how many sayings they could remember the meaning of. This is obviously mostly a test of rote memorization.

In the case of analogies, participants were first given the definitions of sets of new words to study.

Then they were tested to see whether they could discern the analogy between these two words and two familiar words. This test is partly a function of simple memory ability, but also probably measures the ability to think in terms of analogy as well.

In the sample, the race gap Fagan’s “new knowledge” tests was insignificant. Moreover, this new knowledge test correlated at .58 with SAT-V scores, and 0.72 after correcting for measurement error. Granted, verbal intelligence is only one aspect of full scale IQ, so this test probably does not correlate highly enough with FSIQ to share the majority of its variance with it (0.70). But this test is clearly a better measure of intelligence than were some of the simpler tests Fagan was using earlier, probably because it is partly a measure of thinking in analogies rather than just simple memory ability.

However, in this same sample the racial SAT score gap was only around 0.33 SD, meaning this sample was not representative of the normal cognitive gap seen between races and this finding can be explained entirely by the combination of the test still partly being a function of simple memory as well as the elite nature of the sample in which normal racial gaps in IQ are not present to begin with.

In experiment two, the sample consisted of 696 students of which 153 were minorities. These students mostly went to community college meaning they are a less elite sample than what was used in experiment one. The tests given were the same as in experiment one. Once again, this test correlated quite well with SAT-V scores (0.63 / 0.79 after correction).

As for the results, now that we are using a less elite (but still not necessarily representative) sample with a test that highly correlates with IQ… we are not given the racial difference in any test given. Instead, we are told that race was not predictive of new learning ability once SAT scores were controlled for: “The regression analysis yielded a multiple R of .63, F (2,690) = 231.3, p < .001, with Beta values of .03 (t = 0.9, n.s.) and .63 (t = 2 1.0, p < .001) for minority status and brief SAT scores, respectively, for the prediction of new learning.”

Of course, this is hardly surprising since the SAT-V test and the new knowledge test share most of their variance in common! No theory about racial IQ gaps predicts that there will be large racial gaps in one cognitive test after controlling for another cognitive test that highly correlates with the former cognitive test. In fact, there is evidence suggesting that if you control for general intelligence minorities have roughly the same working memory scores as whites and score more higher than whites on tests of long term memory.

For all these reasons, this paper gives us no good reasons to doubt a genetic model of group differences in IQ.

Fagan and Holland (2009)

The final paper to consider in this literature is Fagan and Holland (2009).

The sample consisted of 633 college students who were enlisted in undergraduate psychology courses. This included only 121 minorities. Also, it is worth noting that Asians were considered to be part of the “majority” even though they are literal minorities.

As in previous experiments, the participants were tested on their understanding of a novel set of words, sayings, and similarities. For the tests on sayings and similarities, participants were given definitions of new words or sayings to study and then later asked multiple choice questions about what the sayings meant or what two words had in common. In the case of novel words, participants studied sentences that clearly demonstrated the word’s definitions without being given explicit definitions and were then tested on their understanding of the word’s meanings. Scores on all three measures were then combined into a single measure of “new learning ability”.

Participants were also again given a subset of the verbal section of the SAT. Their new learning score correlated at 0.66 with this SAT-V score prior to correcting for measurement error and 0.83 after said correction.

Regarding the relationship between race and new learning ability, we once again are not told anything useful or interesting. As Fagan did in his 2008 paper, they just say that race doesn’t predict new learning ability once SAT scores are controlled for : “Past knowledge (i.e. brief SAT scores), but not race, was expected to be related to the ability to process new information. The results were as predicted. A regression analysis yielded a multiple R of .65, F (2/625) = 230.9 P<.0001, with Beta values of .04 (t= 1.3, P>.18) and .65 (t=21.0, P<.0001) for majority–minority status and brief SAT scores, respectively, for the prediction of new learning.” As I’ve already explained, results like this give us no reason to doubt any theory of racial IQ differences.

Final Thoughts

From the outset, this line of research has obvious theoretical problems stemming from the fact that it can’t test what the causes of racial differences in information exposure are. As I’ve tried to show, a close look at the research also reveals many problems in terns of sample size, sample representativeness, and the strategy of data analysis undertaken. As a whole, this line of research cannot justify thinking that IQ tests are racially biased or that information exposure is an important cause of racial IQ differences.

By contrast, meta-analyses consistently show that brain size is positively correlated with IQ.

Moreover, a correlation of 0.19 between brain volume and fluid intelligence was found in a pre-registered analysis of 13,608 Brits (Nave et al, 2018). The fact that this study was pre-registered implies that publication bias cannot explain the finding.

Research also shows that this correlation persists within families meaning it is not the result of brain size being a proxy for socio-economic status or any other variable shared by members of the same family.

The psychologist and noted opponent of hereditarianism Richard Nisbett has denied this. He wrote on page 219 of his popular 2009 book “Intelligence and How to Get It“”: “The correlation found within the white population probably does not indicate that greater brain size causes higher IQ. Within a given family, the sibling with the larger brain has no higher IQ on average than the sibling with the smaller brain”. To substantiate this claim Nisbett cited Schoenemann’s study of 72 subjects, a sample size too small to be worth much, and simply ignored the rest of the research literature.

The fact that brain size also predicts intelligence in non-human species should make us even more confident that this correlation reflects something causal.

Still further evidence comes from the fact that research consistently finds brain size to be genetically, rather than just phenotypically, correlated with intelligence.

There also may be a correlation between an IQ subtest’s g loading and its correlation with IQ. I say “may” because we don’t have much evidence on this question and the conclusion that said limited evidence will lead us to depends on whether we include in our analysis a study which is a statistical outliers from the rest.

Regardless, the evidence suggests that there is a pretty robust relationship between IQ and brain size. To counter this claim, various arguments have been offered.

For instance, a common argument goes like this: men have bigger brains than women, but men are not smarter than women. Therefore, larger brains must not cause people to be smarter.

Of course, brain size only explains a fraction of the total variance in IQ so two groups can be of equal intelligence with unequal brain size so long as the smaller brained group is sufficiently advantaged by some other factor, For instance, some autopsy data has suggested that women’s neurons are more tightly packed than men’s are. Thus, men and women may have the same neuron count even though they have different brain sizes (Lipton 2002 page 193). Or, there may be some other difference that we, or at least I, don’t know about. Either way, this is not a very compelling argument.

It’s also worth noting that the assumption behind that argument, that men and women are equally intelligent, may not be true. Depending on which meta-analysis or large study you look at, you can find that men have higher IQs than women or that the sexes have equal IQ (e.g. Jackson and Rushton 2006, Flynn and Case 2011, and Irwing and Lynn 2005). If there is an IQ advantage for men, then this argument against brain size and IQ becomes even weaker.

Another argument sometimes made is that people with megalencephaly, a neurological disease, have huge brains but do not have high IQs. Therefore, large brains must not cause high IQ. This argument has been made by some well known academics (Pietschnig et al 2015), and I think it’s about as compelling as saying that bigger hearts don’t pump more blood because people with enlarged hearts pump less blood than average. Both lines of reasoning are flawed for the same reason: people with rare conditions like these have abnormally functioning biology’s and so should not be used as a model for how human biology normally functions. In the case of people with abnormally enlarged brains, any tension between this and brain size research can be eliminated by noting that they we have no reason to think that the brain areas which are involved in intelligence are especially large and normally functioning in such people.

With the link between brain size and intelligence established, let’s turn to race. In 1994 Harvey et al preformed the first study comparing the brains of different racial groups using MRI technology to measure brain size. They confirmed previous findings: Blacks have smaller brains than Whites. The same finding was reproduced by Jones et al (1994), though the difference wasn’t statistically significant. Chee et al (2010) found that Whites have larger brains than East Asians.  Finally, Tang et al (2010) found that Whites have longer brains than East Asians while East Asians have wider and  taller brains than Whites. Unfortunately, all of these studies have very limited sample sizes and, with the exception of Tang et al, failed to control for differences in the sexual composition of racial samples. Thus, if, for instance, the Black samples had a higher female to male ratio than the White samples then, because men have larger brains than women, Black’s brain size will be depressed relative to Whites.

But these findings correspond well with what older studies measuring skull volume found. Beals (1984) aggregated data on roughly 20,000 subjects from past studies which showed East Asians having the largest skulls followed by Whites followed by Blacks.

Beals (1984)

A famous critique of this line of work was launched by Gould (1981) who argued that researchers involved in this work, most famously the 19th century anthropologist Samuel Morton, unconsciously allowed their racist views to cause them to back material more tightly into White skulls than into Black skulls and thus inflate the racial difference. Gould also accused Morton of excluding data from his tables that increased the racial disparity in brain size in favor of Whites. Both of these accusations have been shown to be false. Morton’s skulls were re-examined by modern researchers who reproduced Morton’s results, and the data that Gould accused Morton of omitting was actually included by Morton in the same book that Gould cited, just on different pages (Lewis et al. 2011).

Of course, a third way to measure brain size is to rip a brain out of a skull during an autopsy and measure its volume. Autopsy work done in the late 2000’s found that Whites averaged the largest brains, followed by, East Asians, followed by Blacks (Rushton and Ankney, 2009).  This same method has been used to confirm the Back/White/Asian brain size disparity for well over a century:

Tobias (1970) offered a highly influential critique of this literature which in turn was cited and popularized by Gould. In it, Tobias argued that comparisons of the brain size of racial groups based on autopsies were invalid because they failed to control for a wide variety of variables that could impact brain size. Said variables included, but were not limited to, age of death, nutritional intake early in life, occupational status, cause of death, time of death, temperature the brain was kept in after death, and the exact place the brain was cut from the spinal cord.

Some of these controls would obviously be misleading because of the correlation between things Tobias wanted held constant and intelligence (occupational status for instance) and so holding them constant would involve holding a part of the racial IQ gap constant. Some of the proposed controls also had nothing to do with measuring brain size but rather had a plausible connection with impacting brain size (e.g. early life nutrition). Such controls would not be appropriate when asking whether, as opposed to why, racial groups differ in mean brain size.

That being said, some of these criticisms are more plausible. However, there is no reason to think that any of these problems would bias the results in favor of one race or another. And random error introduced can be overcome via aggregation. Thus, Rushton (1995) combined all the data from the studies Tobias critiqued and found the now familiar pattern of Whites having the largest brains, followed by Asians, followed by Blacks. Tobias can be said to have pointed out that racial bias in measurement is logically possible, but he did nothing to justify the belief that this was actually taking place to a significant degree.

A fourth way to measure brain size is to estimate it based on external head size. The major advantage to this method is that researchers can cheaply measure the brain sizes of large and representative samples of living people. One of the largest of these studies was Rushton (1992) which had a sample of 6,325 and found that Whites had the largest brains, followed by East Asians, followed by Blacks.

Rushton (1997) analyzed data on 17,000 white Americans, 19,000 black Americans, and 100 Asian Americans taken from The Collaborative Perinatal Project, a longitudinal study which followed subjects from birth to seven years of age. This data set also showed whites having the largest brains, east Asians having the largest relative to body size, and blacks having the smallest brains in both absolute and relative terms.

Rushton combined data from these sources to produce the following graph showing consistent racial differences in relative brain size from birth to adulthood.

These differences are present among new borns and even unborn fetuses. Such has been demonstrated numerous times. (Several of these paper were initially compiled in Kirkegaard (2018)).

The fact that racial differences in brain size are realized at such early stages of development constitutes the first line of evidence favoring the view that such brain size differences are partly caused by genes rather than just environments.

Several studies have shown that mulattoes have an average brain size in between that of Blacks and Whites (Pearl, 1934; Bean, 1906). This finding has been established on multiple occasions and is what a hereditarian hypothesis would predict since mulattoes are half White and half Black genetically speaking.

Furthermore, many traits which tend to co-evolve with larger brains also differ racially in a way that mirrors the body size adjusted brain size pattern. Rushton and Rushton (2003) looked at 41 anatomical features which 3 textbooks on human evolution identified as tending to co-evolve in the hominid line with larger brains.

For instance, larger pelvic size tends to co-evolve with brain size so that mothers can give birth to larger brained infants. The pattern of the distribution of such traits by race fit what we would expect given evolved racial differences in brain size at a rate far greater than what random chance would likely produce.

Finally, there is some evidence that the races evolved different brain sizes in response to climate. Specifically, various studies have found that a population’s brain size correlates with climate related variables. For instance, Pearce and Dunbar (2011)’s data set produces a correlation of .74 between a population’s brain size and its latitude. Similarly, Ash and Gallup (2007) found a correlation of .48 between the size of 109 fossilized human skulls and the latitude at which they were found. Further still, Bailey and Geary (2009) analyzed 175 skulls ranging in age from 10,000 years old to 1.9 million years old and found a correlation of -.41 between brain size and winter temperature and -.61 between size and latitude (larger brains were found in areas more distant from the equator).

Lynn (2015) used Smith and Beals data set of 20,000 skulls from 122 populations to estimate that roughly 30% of the African-European IQ gap can be statistically accounted for by brain size differences. By contrast, brain size differences would actually predict an Asian-European IQ gap 35% larger than the one that actually exists. Thus, brain size is probably one of many factors which account for racial intelligence differences.

The best known remaining counter to the claim that racial groups differ in brain size due to genes comes from Boas (1912). Boas, who was a founder a modern cultural anthropology, looked at the American children of European immigrants and found that their mean brain size was very different from that of their parents and more similar to the brain size of non-immigrant American children. This study has gone on to be widely cited and pointed to as the first crack in the armor of genetic determinism.

However, in his analysis Boas neglected to account for age differences between groups and this distorted his conclusions. Sparks and Jantz (2002) re-analyzed Boas’s data after accounting for the age difference between parents and children and found that the brain size of immigrants and there children did not differ in any statistically significant way and the brain size of immigrant children did significantly differ from those of Americans. In fact, based on the correlations between parent and children brain size, the heritability coefficient derived from Baos’s own data is greater than 50%. This is consistent with the conclusions of modern research on the heritability of brain size (DeStefano et al., 2010).

Thus, when considering the evidence, objections, and counters to those objections, together we are justified in thinking that data on brain size constitutes a significant line of evidence favoring a genetic model of racial difference sin IQ.

The last decade has seen a radical increase in the cultural visibility of transgenderism. At the same time, the rate at which people identify as transgender or non-gender conforming has risen to something like 2% of adolescents (Struass, 2019).

For the most part, Americans have responded to this shift by adopting more liberal attitudes towards trans-gendered individuals. Today, most Americans say they would be comfortable having a transgender friend or learning that a local teacher is transgender. Americans also favor letting trans people into the military and nearly half of the population says they would be comfortable with having a transgender child.

PPRI (2019)

The public is pretty divided concerning bathrooms with a slim majority saying that people should use bathrooms according to their biological sex rather than their gender identity.

Gallup (2019)

Among adults under the age of 35, the vast majority say they would be comfortable having a gay or trans child, though there has been a moderate increase in the last year or so in those expressing discomfort with this.

Glaad (2019)

The public hasn’t shifted such that the super majority of people take a perfectly left wing view on transgenderism, but the state of public opinion is obviously a great deal more left leaning than it was ten years ago when trans-acceptance wasn’t even a mainstream political issue.

The last decade has also seen the popularization of certain theories, explanations, and narratives surrounding transgenderism. While not monolithic, many on the left have often said something like the following:

Transgenderism occurs when people want to be a gender that is discordant with their prescribed sex. This is caused by a mismatch between people’s neurological and psychological sex on the one hand, and their genetic sex on the other. Many people are not accepting of trans-people, and their discriminatory behaviors and ideas cause trans people to exhibit a multitude of behavioral and psychological problems. These problems should be solved by lessening anti-trans discrimination, and maximizing the ease with which trans people can transition into their preferred gender. Cis people have no good reason to resist changing norms in this way, because accepting the validity of trans-genderism implies no significant imposition on them and is really just a matter of letting other people live as they wish to.

The rest of this post will be aimed at showing that each part of this story is false.

Discrimination and Trans Suicide Rates

Looking at data from the 2015 US Transgender survey (n=27,715) on suicide rates, we find that among the general population only 0.6% of people have attempted suicide within the last year, but among trans people this rate is 7.3% or 12 times the rate of the general population. In this section, I am going to argue that most of this gap cannot be explained by discrimination. 

A first reason to doubt that discrimination is a complete explanation for the mental health problems of trans people stems from the fact that trans people who report not having experienced discrimination still exhibit elevated suicide risk. Across seven studies, the average lifetime suicide attempt rate among such trans people was 24% or 5.2 times that of the general population (4.6%, from Herman et al). 

Similarly, in Herman et al. (2019) the past year suicide attempt rate among trans people reporting having been the victim of zero cases of significant discrimination within the last year was 5.1% or 8.5 times the rate of the general population. 

Importantly, most acts which constitute discrimination, ostracizing someone, insulting someone, emotionally or physically hurting someone, etc., are essentially forms of bullying. This fact matters because even though there is a substantial correlation between bullying and suicide rates the causal relationship between bullying and suicide is much weaker than this correlation suggests. To be specific, a one standard deviation increase in bullying predicts a 68% increase in the odds of someone attempting suicide. However, if you use a twin study design to control for genes and the family environment, and also control for psychopathology that pre-existed the bullying, we find that a one standard deviation increase in bullying only predicts a 25% increase in the odds of attempting suicide (Oreilly et al., 2021). Thus, only roughly a third (37%) of the commonly noted association between bullying and suicide is plausibly causal.

Given this, it is likely that trans people who experience no discrimination are advantaged above the average trans person in terms of their pre-existing psychopathologies or genetic predisposition towards suicide or both. So the above figures represent suicide rates which, though much higher than average, are lower than what we would rationally expect if the average trans person experienced no discrimination. 

A second reason to doubt a mostly-discrimination explanation for trans suicide rates is that suicidal LGBT people usually don’t cite discrimination as a cause of their behavior. In fact, a 2021 meta-analysis found that the vast majority (74%) of LGBT youth (age: 12 – 25) who have attempted suicide answer “no” when asked if they’ve personally been the victim of various sorts of anti-LGBT discrimination and bullying (Williams et al., 2021). Of course, a short coming of this line of research is that it includes non-trans LGBT respondents, but it is still relevant enough to be worth mentioning. 

A third line of evidence against the discrimination hypothesis consists in the fact that the relationship between discrimination and suicide among trans people is statistically quite weak. If discrimination did play a big role in trans suicide rates, we might expect for trans people who have experienced a great deal of discrimination to report suicide attempt rates far higher than those who have not. But when reviewing all the literature I could find in 2019 on the association between suicide and non-violent discrimination among trans people (22 effects), I found, in the first place, that the effect sizes reported were generally small, especially in larger studies. Among studies with at least 1,000 participants, experiencing non-violent discrimination predicted a roughly 10% increase in suicidal behavior. Moreover, a majority of these effects (64%) were not statistically significant, so most of the research did not find that non-violent discrimination significantly elevated suicide risk among trans people.

I found something similar when looking at 22 reported associations between a lack of social support and suicide, with the effect sizes being very weak, meaning that can’t possibly explain more than a tiny fraction of the association between transgenderism and suicide, and once against the effects were statistically insignificant most (78%) of the time.

The data concerning physical violence is less clear. Across ten effects, the link between physical violence and suicide was more consistently statistically significant. However, the size of the effect reported varied a great deal between studies. The degree to which experiencing violent discrimination elevated suicide risk ranged from 1.43 to 4.72. Moreover, there was an tendency for larger studies to find smaller effects. The largest study I found reported odds ratios of less than 2.

Consistent with my review, a 2022 meta-analysis found that experiences of discrimination correlate at only 0.24 with a trans person’s risk of suicide attempt. Expectations of rejection was an even weaker correlate at 0.15 and internalized transphobias exhibited a still weaker and marginally significant correlation of 0.09 (Pelicane et al., 2022). These effect sizes entail that no measure of discrimination explains even 5% of the variance in the rate at which different trans people attempt suicide.

Given the extremely weak effects of other sorts of discrimination on suicide, it is clear that the combination of physical discrimination, non-physical discrimination, and lack of social support, cannot account for anywhere near half of the trans-cis suicide gap. And this is prior to adjusting our estimates for the fact that, as mentioned previously, most of the association between discrimination and suicide is likely non-causal. 

Finally, international data seems to imply that discrimination has very little at all to do with trans suicide rates. I say this because there is no obvious association between the suicide rate of trans people and the trans-acceptance of the place they live in. Sweden has a lower rate of trans suicide than the US, but the US has a lower rate than Canada, and no region I found data on has as low of a rate as china. Thus, lower national rates of trans-acceptance don’t seem to lead to lower levels of trans suicide, a finding hard to square with the notion that the link between discrimination and suicide is causal.

Obviously, we can’t run randomized experiments to see how much discrimination influences suicide. Instead, we have to rely on observational data. That data, such as it is, makes it fairly implausible that discrimination explains the majority of the trans-suicide gap. Moreover, the data is surprisingly consistent with the notion that discrimination plays only a very small role in trans suicide rates. Perhaps most importantly, the data makes very clear that it would be irrational to expect the trans suicide rate to mirror anything like that of the general population’s in a world with no anti-trans discrimination. 

Gender Expression and Trans Suicidality 

Of course, the other major tenant of the mainstream narrative on transgenderism is that trans people’s mental health is also brought down by the fact that they often cannot engage in a mode of gender expression which is consistent with their internal gender identity. In this section I’ll argue that it is unclear whether aligning trans people’s gender identity and gender expression improves their mental wellbeing and that it is clear that doing so does not improve trans mental health nearly enough to bring them inline with the general population. 

The first thing to say is that there are some studies which find that transitioned trans people have lower levels of suicidality than do non-transitioned trans people. For instance, in a sample of 314 transwomen from San Francisco, Wilson et al. (2014) found lifetime suicidal ideation rates of 75% for trans people who hadn’t received any medical intervention, 74% for those who had genital surgery, and 51% / 52% for those who had taken hormones or had breast surgery. However, though this provides some support for transition it is also important to note that around the time at which these trans people were interviewed, the lifetime suicidal ideation rate for the US population was 15.6% (Nock et al., 2008). Thus, even when trans people both had transitioned and were living in possibly the most pro-LGBT city on earth, their rate of suicidal ideation was still 3.3 times that of the general population. 

Other research has found transitioning and open gender expression to be unrelated to trans suicidality though such studies have often suffered from small sample sizes and low statistical power (e.g. Bauer et al., 2015; Zeluf et al., 2018; Lehavot et al., 2016; Amand et al, 2011). 

That said, the highest quality research has found that trans people’s transitioning is associated with elevated rates of suicidality. For instance, in Haas et al. (2014)‘s analysis of data on 5,885 trans Americans it was shown that life time suicide attempt rates were higher among trans people who have transitioned, or who want to someday, compared to trans people who had no plans to actually transition despite the tension between their assigned and internally felt gender identity. 

Relatedly, in a sample of 1,309 trans Chinese, Chen et al (2019) found that trans people seeking sex reassignment surgery had a rate of suicidal ideation 76% greater than did trans people not seeking such a surgery. And in Rood et al. (2015), planning to transition and having already transitioned were both associated with a nearly 3x increase in the odds of lifetime suicidal thoughts (n=350).

The repeated finding that even wanting a transition related surgery is a risk factor for suicide is important to note as it implies that the high suicidality among post-operation trans people is caused by something which pre-dates the operation itself. 

Most significantly, Herman et al. (2019) looked at data on 27,715 trans Americans and found a past year rate of suicide attempt of 4.1% among trans people who say they don’t want to actually live in their gender identity someday compared to 5.2% among those who were unsure and 7.4% among those who did. The study also recorded a past year suicide attempt rate of 5.1% among trans people who had already had a transition related surgery and 8.5% among trans people who wanted such a surgery but who had not yet been able to have one. 

This research brings forth two questions: First, why do trans people who are currently living in alignment with their own gender identity, and trans people who plan to at a later date, have elevated suicide rates compared to trans people who decide they don’t want to do this? Second, why do trans people who want to get a transition related surgery but can’t have the most elevated suicide rates of any group I’ve mentioned? 

One possible explanation is that those trans people who want to actually transition have greater levels of gender dysphoria than do trans people who do not making comparisons between their suicide rates misleading. We might add that the high suicide rate of trans people who want to transition but cannot makes a fairer comparison because these are people who have as much gender dysphoria as do those who have transitioned. 

There are two major problems with this line of reasoning. First, the placebo effect, and the distress someone would likely feel if they thought there was medicine which could help them but which they could not access, and the fact that whatever is stopping someone from getting a surgery they want is probably itself negatively related to well being, make trans people who want but cannot get transition related surgeries obviously invalid as a comparison group. Second, the fact that suicide rates don’t significantly differ between trans people who have had such surgeries and who are merely planning to is hard to explain on this sort of hypothesis.

Actually, there are two reason’s we’d expect post surgery suicide rates to be lower than pre surgery ones even if the surgery had no effect. First, there is the placebo effect which is virtually universal across medical research. Second, transitioned people are presumably older than trans people who merely plan to transition in the future and in general older trans people have lower suicide rates than do younger ones (Herman et al., 2019). Given this, the equality of the rates of suicide across these groups is more plausibly interpreted as the surgery having a modest detrimental impact on suicide risk than it is interpreted as having a protective one. 

My preferred explanation for this pattern of findings is this: in general it is unhealthy to attempt the change things which you cannot change and which are outside of your control. Moreover, it is unhealthy to allow your happiness to be contingent on things like what clothes you wear, how you speak, the details of how others speak to you (pronouns), your body image beyond the concern of physical health, and the other concrete things which define gender roles. This is what a man is doing when they place great significance on attempting to be a woman or vice versa. It is possible, and optimal, to recognize the desire to be a different gender without letting this desire control you and have you attempt the impossible. The more you lean into this desire, the more the impossibility of its satisfaction will harm you, though perhaps no condition will harm you as much as thinking that the desire could be satisfied if only you could have an operation which is available to others but not yourself. 

Of course this is just speculation, but it is consistent with the full pattern of data and a more general view of how people should relate to their desires. Regardless, it is now hopefully clear that the data does not justify the view that transitioning lowers suicide rates and refutes the view that transitioning leads trans people to have a level of wellbeing comparable to cisgendered people. 

Proponents of the standard treatment for transgenderism may respond by noting that people say they feel better after undergoing medical transitions (Murad et al. 2010; Millet et al., 2017). Throughout this post so far, I’ve followed a methodological choice to use suicide as the outcome of interest rather than self reported scores on scales for outcomes like depression and anxiety. This is because people don’t like to admit it when they make mistakes. Everyone is naturally biased in favor of decisions that they have already made. Thus, looking at subjective ratings by people who have undergone various procedures is not very informative. Data on suicide  rates are preferable because suicidality is a more concrete and objective criterion.

I’ve argued that the data on suicide suggests that transitioning does not help trans people, but the data from self-reported emotion and satisfaction suggests the opposite. I feel the best explanation for this divergence of results is exactly the sort of bias that caused me to only look at suicide data in the first place. Someone offering an alternative to this analysis will need to come up with a better explanation than this for why the outcomes diverge, as well as an explanation for how it can be that these treatments improve trans people’s mental health but not their suicide rates.

In combination, the last two sections imply that the left’s explanation for trans suicide rates, discrimination and a lack of free gender expression, clearly fails to account for the vast majority of why it is that trans people have such high suicide rates. 

The Mismatched Biology Hypothesis

Let’s next examine the notion that trans people, due to genes and the prenatal environment, have brains which match the typical member of the gender they identify with. 

Guillamon et al. (2016) provides a useful review of the research on transgenderism as of 2016. Research then indicated that while trans people were similar to the opposite sex in a few brain regions most of their brains more closely resembled the typical member of their own sex. 

The same result was found in Frigerio et al. (2021)‘s review of brain imaging work done up to 2018. 

Importantly, both reviews note that this research is massively confounded by the prevalence of homosexuality among trans people. So, it could be, for example, that MtF people have somewhat feminized brains simply because they tend to be gay, while their transgenderism may be causally unrelated to these neurological trends.

I’m aware of only two studies which have directly accounted for this. First, Savic et al. (2011) from the brains of heterosexual male-to-female trans people were not abnormally feminized. Second, Burke et al. (2017) found that after controlling for sexual orientation trans people’s brains were mostly typical of their sex. The one exception to this was the right inferior fronto-occipital tract. However, even here the brains of transgender people were not typical of those of any sex. That said, all this research is based on small samples and so not much can be said with confidence about the neurological correlates of transgenderism. 

The brain story is also complicated by the fact that sex differences in the brain, while real, are not that large. For most brain differences, there is a good deal of overlap between men and women, so that there are presumably lots of people with sex atypical brains and the vast majority of them are not trans.

Moving on to hormones, the ratio of the length of people’s 2nd and 4th finger is a correlate of pre-natal testosterone, and so trans people having digit ratios typical of the other sex would be evidence for them having an atypical pre-natal environment for their sex.

Voracek et al. (2018) meta-analyzed the research on this topic and found the following:

“MtF cases have feminized right-hand (R2D:4D) digit ratio, g= 0.190 (based on 9 samples, totaling 690 cases and 699 controls; P= .001, 95% confidence interval [CI]: 0.076 to 0.304), whereas the directionally identical effect for left-hand (L2D:4D) digit ratio was not significant, g= 0.132 (6 samples, 308 cases, 544 controls; P= .07, CI: –0.012 to 0.277). FtM cases have neither masculinized R2D:4D, g= –0.088 (9 samples, 449 cases, 648 controls; P= .22, CI: –0.227 to 0.051) nor masculinized L2D:4D, g= –0.059 (6 samples, 203 cases, 505 controls; P= .51, CI: –0.235 to 0.117).”

So, this story doesn’t work for FtM trans people and only works for MtF trans people when we are talking about the right hand. Moreover, the effect size here is 0.19. This is a quite small effect size. Assuming digit ratios are normally distributed, this would imply that the average MtF trans person has a digit ratio more masculine than 42% of males. This variable may have some role in a full explanation of transgenderism, but the vast majority of males with hands that feminized are not trans, and so this is at best a weak explanatory factor.

Twin studies should be informative both with respect to genes and the prenatal environment, but researchers have only been able to find a handful of twin pairs in which at least one twin is transgender. Heylens et al. (2012) aggregated data from previous studies and found a concordance rate for transgenderism of 0% among 21 non-identical twin pairs and 39% among identical twin pairs. A later study from Japan produced similar results in terms of low concordance rates among twins for Gender Identity Disorder (Sasaki et al., 2016). The fact that concordance rates are higher among identical twins than among non-identical twins implies that genetics does play a role in transgenderism, as it does in all human behavior, but the fact that concordance rates even among MZ twins are well below half suggests that genetics and the pre-natal environment are far from a sufficient explanation of transgenderism. Moreover, to the degree that pre-natal environments and genetics do play a role, that role does not primarily seem to be one of creating people whose biology matches the other sex, as evidenced by neurological data and data on digit ratios.

Aside from being empirically unsupported, the idea that transgenderism is caused by trans people having brains typical of their preferred sexual identity implies some very strange things about the relationship between brains and sexual identity.

If my brain were to become feminized, I would probably acquire a more female-typical personality, for instance I might become more agreeable and less emotionally stable, and perhaps I would develop different pre-dispositions about who to have sex with, how many people to have sex with, and the role I’d want to play in raising children. This all seems plausible.

However, there is no obvious connection between having a female typical brain and wanting to wear female typical clothing, or wanting to posses a female body, or wanting to be called a woman, etc. Plausibly, people identify with their own body because the brain is wired to identify with whatever body it finds itself in. This would explain why I feel a sense of identity not only with my sex, but also specifically with the body that is mine. It would be very non-parsimonious, and entirely speculative, to suggest that brains are built to identify with certain sorts of bodies and that if a part of my brain where changed in shape or size to be more typical of a woman then I would desire to have a female body.

It is equally speculative to suggest that women would want to wear feminine clothing even if they didn’t have the feminine bodies that such clothing is made for.

Of course, it would be entirely unreasonable to suggest that women want to be referred to using feminine pronouns because they have female typical brains. Generally speaking, women want to be called women because they are women, in the most essentialist since of the term, and this is true even of women who are psychologically abnormal for their sex.

Thus, the very notion that a mismatched brain causes transgenderism implies speculative seeming assumptions about the nature of gender identity in general. Of course, sometimes surprising things turn out to be true, but we should only accept such claims as true in response to rigorous evidence and never in response to political bullying. 

Transgenderism, Tolerance, and the Meaning of Gender

Finally, let’s consider the idea that acceptance of trans genderism amounts to little more than letting other people live and identify as they wish to. Framed this way, acceptance of transgenderism seems like little more than an act of basic tolerance.

I regard this framing as misleading because transgenderism is often as much about how trans people want other people to treat them as it is about anything internal to themselves. Specifically, trans people often want others to treat them as if they are typical members of their preferred sex.

Consider their insistence that people use their “preferred pronouns”. If I call a trans-man a woman, the two of us both know that I mean they have no Y chromosome, and we both agree on this fact. We both also obviously know that the person in question calls themselves a man. So what is the disagreement that causes trans people to be upset in such instances? Presumably, it is because using a non-preferred pronoun amounts to a more general refusal to treat trans people as if they are a typical member of the sex they identify with.

Pronoun use, on its own, isn’t that big of a deal. It is therefore unfortunate that so much time as been spent focusing on pro-noun use, one of the most shallow aspects of what it means to be treated like a man or a woman. This focus has led to the impression that trans people may be asking something of cis people, but that it doesn’t amount to much more than a simple change in language.

But treating someone like a typical member of their self identified sex is about more than mere words.

This is most obvious with respect to sex. For a heterosexual male to treat a trans woman as if they were a normal female would mean that they would consider them to be people they’d potentially like to have sex with. While I’ve not seen a survey on the matter, I feel fairly confident that most heterosexual men feel as I do and find the notion of having sex with a trans-female to be rather unpleasant. This would be true even if the trans-female looked indistinguishable from a cis-female.

Something similar occurs with respect to instincts surrounding protection. Males instinctively feel protective of females in a way that we do not for fellow males. For instance, if a female is physically assaulted by a male most males feel that they should intervene and this does not occur when males are seen fighting each other even if there is a significant physical disparity between the males in question. Relatedly, we place a higher value on sustaining the innocence of young girls than we do young boys. The idea of treating a biological male as a female in these ways seems obviously absurd.

This takes us naturally into the debate surrounding trans people in sports. Trans people argue that they should be able to participate in sports leagues intended for the gender they identify with. To make things fair, trans-females have proposed various medical routes by which they might eliminate the physical advantages that males naturally have over females. The efficacy of such attempts is controversial, but I don’t think it really matters because this whole argument rests on a misunderstanding of why sports are segregated to begin with.

If the argument of trans-advocates made sense, we should generally allow males into female sports so long as their ability was within the plausible range of ability for female athletes. This could occur for any number of reasons. Some males are naturally weaker and slower than average. Some males simply put very little effort into athletic ability. Or it could be because a male injected themselves with hormones or a drug that lessened their physical capabilities.

If the point of sex-segregated sports was just to stop people of unequal ability from competing then we wouldn’t even have sex segregated sports. We would have ability-segregated sports and women would just disproportionately end up in the low-ability leagues for many sports. But we dont, we have sex segregated sports and this reflects the fact that people generally think the sexes should be treated differently independent of any feature other than sex itself. 

At its most fundamental level sex is about mating and reproduction. In large part, the reason for which men treat women like we do is because we see them as potential sexual partners, mothers, daughters, and grand mothers. I assume a similar thing is true of why women treat men the way they do. Truly treating trans people as a member of their preferred sex would therefore clearly involve more than a mere shift in language. It would require us to do things which naturally cause reactions of disgust and discomfort for many cis gendered people. In some cases, such as the domain of sexual attraction, treating trans people as their preferred sex would even violate many cis people’s sense of their own identity. 

This gives us good reason to refuse to treat trans people as they want to be treated. Of course, you might say that this is just a mental hang up of cis people and that they should learn to get over it rather than imposing things on others. But then cis people might say the exact same thing to trans people. This may make it seem as if both groups want something which may cause the other group suffering, and so they have fundamentally divergent interests such that any compromise is impossible. By necessity, one group will win and the other will lose. Unfortunately, this seems to me like it is probably true so long as trans gender people are convinced that pursuing their desire to be a member of the opposite sex is in their interest. 

Conclusion

In this article I’ve tried to show that trans-genderism is not primarily the result of “mis matched” biology, that the desires of trans people require a good deal more than mere tolerance from cis people, and that the mental suffering of trans people is not primarily the result of discrimination nor can it be cured by trans people attempting to look and act like their preferred sex. I have not tried to explain why it is that trans people want to be a member of the opposite sex, or why they experience so much mental suffering. I don’t think we have good answers to either of these questions, and in any case removing existing explanations about trans genderism is useful to do prior to offering any alternative explanation. 

Before turning specifically to the 2020 election, it is important to note that election fraud is not unheard of in the United States. This is worth establishing because if we think election fraud almost never happens then we will have a higher bar of evidence for the justification of any particular claim of fraud.

In recent decades, there have been over 1,200 known instances of voter fraud due to which 20 US elections had to be overturned to declare a new winner (Heritage, 2020). Of these, 15 were overturned due to fraud involving mail based ballots (Lucas, 2020). This isn’t surprising since mail in ballots have long been noted to be particularly vulnerable to fraud. This is why most European nations have either banned absentee voting or require an ID for someone to obtain such a ballot (Lott, 2020).

This fact has been attested to by American media on both the right and left. For instance, in 2012 the New York Times wrote the following: “While fraud in voting by mail is far less common than innocent errors, it is vastly more prevalent than the in-person voting fraud that has attracted far more attention, election administrators say. In Florida, absentee-ballot scandals seem to arrive like clockwork around election time.” – Liptak (2012)

The piece even went as far as to cast doubt on the 2000 US presidential election results on the basis of such concerns:

“Voting by mail is now common enough and problematic enough that election experts say there have been multiple elections in which no one can say with confidence which candidate was the deserved winner. The list includes the 2000 presidential election, in which problems with absentee ballots in Florida were a little-noticed footnote to other issues…

Voting by mail also played a crucial role in the 2000 presidential election in Florida, when the margin between George W. Bush and Al Gore was razor thin and hundreds of absentee ballots were counted in apparent violation of state law. The flawed ballots, from Americans living abroad, included some without postmarks, some postmarked after the election, some without witness signatures, some mailed from within the United States and some sent by people who voted twice. All would have been disqualified had the state’s election laws been strictly enforced.” – Liptak (2012)

Similarly, in December of 2016 a majority (52%) of democrats agreed with the statement “Russia tampered with vote tallies in order to get Donald Trump elected President” (YouGov, 2016).

So being worried about voter fraud, and even suggesting it could alter the results of a US presidential election, is not outside the norm of US politics.

Before, I said that there are roughly 1,200 known cases of recent American voter fraud. Of course, this doesn’t tell us how common voter fraud really is because it could be that the vast majority of instances are never detected. County level analysis shows that there are at least 1.8 million excess registered voters relative to the eligible population size and this at least opens the door for the possibility of voter fraud on a truly massive scale (Judicial Watch, 2020).

More importantly, non-citizens in America report voting at rates that imply at least nearly a million cases of voter fraud per year. This suggests that only a tiny proportion of the voter fraud that takes place is normally caught and that fraud at a scale that can change elections is actually the norm. Of course, these figures leave out voter fraud not involving immigrants, and so is surely a significant under estimation of the total amount of fraud that takes place.

The typical response to such figures is to say that survey respondents could be lying or could have read the question in correctly. That they are not lying is attested to by the fact that researchers have verified the voting of about 2% of the non-citizens in these surveys (Richman et al., 2016; Richman et al, 2014). That they are not confused is attested to by multiple lines of evidence, such as the fact that respondents give the same answers multiple times when they are followed for years and asked the same questions repeatedly (Richman et al., 2016). Thus, until better evidence to the contrary is presented we are justified in thinking that voter fraud is actually very common in the United States.

Turning to the 2020 election, there are multiple known instances where election workers were caught entering wrong numbers into the system that inflated Biden’s vote county relative to trump. Of course, in each case this is claimed to have been an accident, and, of course, there is absolutely no way of showing this is true.

Election officials have said that these errors being caught is evidence that they are really good at catching errors. But there are at least two good reasons to doubt this. First, these errors were all huge and it seems unlikely that people only make huge errors and never small ones, so the lack of stories about small errors suggests that they are actually pretty bad at detecting “errors”. Second, the sheer number of voter fraud convictions that have taken place, the rate at which non-citizens seem to vote, and the number of excess voters illegitimately kept on voter rolls, all seem to imply that, at the very least, this is a highly error prone process.

So we have good reasons to think they aren’t great at catching errors. By contrast, we don’t have any good reasons for thinking these were actually errors. The fact that all three stories about such errors each favored Biden is reasonably improbable to happen by chance (12.5%). But it’s highly probable given mass election fraud, and so these stories should increase the probability we assign to such fraud having occurred.

We also have multiple eye witness accounts, mostly from election or postal workers, who claim that fraud was being engaged in and ballots were being wrongly discarded.

There are also multiple aspects of how the election was openly carried out which are seemingly illegal and which made it easier for voter fraud to take place.

There are also five significant lines of statistical evidence suggesting fraud.

First, in PA, where by exit polls Biden won those over 65 by only 3 points (CNN, 2020), 69 people recorded as being born before 1900 requested (and most returned) ballots of which 80% were registered democrats and 16% of which were registered republicans, meaning this isn’t well explained by random birth day entry errors. This can all be seen using data published on the PA government’s web domain (PA.gov, 2020). Here’s a screen shot of what that data looks like:

Of course, these are instances of plausible fraud that are extremely easy to catch simply because no on alive was actually born in 1800. Presumably, if there are dozens of instances like this there are far more instances not so easily detected.

Speaking of which, this data set also tells the dates on which ballots were requested and returned. As pointed out by some guy on twitter, there are 109,421 instances in which the return date is earlier than the request date.

So that’s one line of statistical evidence. A second line of evidence concerns mail-in ballots not only in PA but also in MI. Specifically, Biden’s advantage in Mail-in/Absentee votes in PA (+60.5) and MI (+37.9) was several times what it was in any other swing state (+6.3). This on its own is hard to explain without positing fraud.

As pointed out at The Red Elephants, this result is made even more unlikely by the fact that republicans requested more ballots than democrats did in MI.

Given these numbers, for Biden to have a 38 points lead in MI he would have had to have got every democrat vote, every independent vote, and roughly one in four republican votes. But common sense and exit polls make it obvious that that did not happen.

To some degree these numbers will be thrown off by the fact that not everyone who requested a ballot returned a ballot, and by the fact that there are still many mail in ballots to be counted, but a huge partisan skew in this would be required to make these numbers seem remotely plausible and so this data should increase our skepticism about mail in ballots in PA and especially in MI.

A third line of statistical evidence involves Wisconsin. After the nighttime pause in reporting, vote shares shifted in favor of democrats *within* city wards, meaning this was not due to more democratic areas reporting later, and this was more true the more behind democrats were in a given ward, contrary to what an explanation based on late mail in voters would predict (Holmes, 2020).

The fourth line of statistical evidence also involved Wisconsin, and specifically Milwaukee. As noted by the BBC, extremely high turnout in certain areas is a classic sign of voter fraud. In Australia, where voting is legally mandatory, voter turnout rates reach 90-95%. Non compulsory elections should not approach these sorts of numbers (BBC, 2020).

Yet, turn out figures in many city wards in Milwaukee were initially reported to be over 100%.

This was later said to be a mistake and revised, but the revised numbers still include various wards with turnout levels that would normally be considered evidence of fraud.

A fifth and final line of evidence concerns so called bell weather counties. These are counties where the person who wins the county almost always wing the presidential race. As a set, they’ve correctly predicted every US election of the last 50 years, until this one.

There are other lines of evidence out there, but this article has presented the one’s I find most compelling. (See here and here for problems with the Benford analyses going around). And these lines of evidence cast hundreds of thousands of votes in doubt, more than enough to alter the result of the presidential election.

As I’ve argued, voter fraud is reasonably common in the US and so we would be unjustified in dismissing the case for fraud in this election out of hand.

We should also note that, if fraud were to occur, the sort of evidence we’d expect to have is exactly the sort of evidence we do have. Even if fraud were to occur at a mass level, it would be surprising if we had evidence even better than this, if we had video of fraud being committed clearly in view of the camera, or we had a mass wave of admission to thousands of instances of fraud. Given this, we should not require this sort of utopian evidence to be convinced that fraud likely occurred. Instead, we should require the exact sort of evidence I’ve presented and so we are justified in thinking significant fraud occurred until we are presented with equally powerful arguments to the contrary.

Importantly, these arguments have not shown that Trump would have certainly won the election absent any voter fraud. But they do show that the official election results do not justify us in thinking that Biden would have won a fair election either. Given this, to ensure fairness we need to either find a way to remove ballots that are fraudulent, or throw out the entire election and have a re-do. Of course, something being in the interest of fairness is no guarantee that it will occur. I am not predicting that these things actually will happen, but hopefully I’ve been successful in showing that they should happen.