Predictors of support for gay marriage

Audacious Epigone has a great post on attitudes toward gay marriage. It made me wonder what best predicts support (sample size = 3,399, GSS, blacks and whites):

OLS standardized regression coefficients

Age -.11*
Sex .14*
Race -.02
Size of place .02
IQ .08*
Education .18*
Income -.01
Church attendance -.30*

*statistically significant

Irreligiosity is clearly the stongest predictor of support. Education, being young, and being female are in the middle, and a high IQ is the weakest significant predictor. Race, the size of your town, and income don't matter.

UPDATE: If I add political orientation to the model, there are two noteworthy changes: 1) being liberal becomes the strongest predictor of support, and 2) race becomes significant. When you remove the influence of liberalism, blacks become signficantly less supportive of gay marriage than whites.

Gender nonconformity and homosexuality

In the comments of an earlier post, someone wrote that there is very little relationship between masculine or feminine behavior and sexual orientation. From a 2003 review:

There are two main correlates of sexual orientation that point to an early developmental genesis. The first is the relationship between childhood gender nonconformity and adult sexual orientation (Bailey and Zucker 1995), which is generally considered one of the strongest of all developmental associations in humans (Bem, 1996). Homosexual men typically recall having been feminine boys, whilst homosexual females were masculine girls. Prospective work confirms this relationship for men (Green, 1987; Zucker and Bailey, 1995). In their meta-analysis, Bailey and Zucker (1995) report large effect sizes for the degree of this association (d=1.31 for men, d=0.96 for women). The gender nonconforming behaviours appear to be specific to childhood sex-typed activities and interests, rather than playmate preferences. Given evidence for an early formative role for androgens in the development of such behaviours, these data are consistent with neurohormonal differentiation theory (Berenbaum and Hines, 1992; Berenbaum and Snyder, 1995). These large heterosexual–homosexual differences in childhood sex atypicality also extend to adulthood, when assessed by “gender-diagnosticity” measures (the extent of male versus female typicality of interests). On these measures (as well as on traditional masculinity–femininity scales; Haslam 1997) gay men are typically more feminine and lesbian women more masculine in adulthood than their same-sex heterosexual peers (Lippa, 2000; Lippa and Arad, 1997).

IQ and religiosity

In the comments of the last post, it was stated that smart people are less religious. Let's remind everyone what the GSS says (sample size = 4,784, surveys 2000-2010, whites):

Scoring six out of ten on the vocabulary test is the mean. Whites in the highest IQ category are slighly more religious than average. Another way of looking at the question is the Pearson correlation. It is .01--overall the two variables are unrelated. (GSS is not able to isolate really smart people, so we can't talk about those folks.)

On the other hand, I know I've got some readers who think of religiosity in terms of belief rather than attendance. If this is what you have in mind, there is a weak negative relationship between intelligence (as measured by the vocab test) and confidence in the existence of God: it is -.14 (p < .01, sample size = 6,849 whites).

Atheism and crime

In Breaking the Spell, Daniel Dennett claims that atheists are as law-abiding as believers.

The MIDUS study asked people their religious affiliation and if they had ever been in jail. The results (sample size = 2,183):


Percent ever in jail

Atheist 0.0
Agnostic 17.4*
No preference 13.0*
Spiritual 17.6*
Has a religion 4.8

*significantly higher than people with a religion


Zero of 18 atheists report any jail time. By contrast, agnostics, spiritual people, and those with no religion have significantly higher rates of incarceration. On the various measures of functioning that I've studied, the squishy middle categories often have the worst numbers.

UPDATE: Keep in mind that the current trend is toward more people having no religious preference (but still believing in God) or saying they are spiritual rather than having a religion.

While we're at it, MIDUS gives us another estimate of the percent of Americans who are atheist. The survey asked people to describe their religious affiliation, and "atheist" is one option. Only .8 percent of participants gave that answer. (The study was conducted 2004-2006.)

Vitamin B12 and infant brain development

In a new study in the European Journal of Paediatric Neurology, cranial MRI images were taken of 15 infants with vitamin B12 deficiency.

In our study, thinning of the corpus callosum was detected in 6 (40%), cortical atrophy in 5 (33.3%), ventricular dilatation in 3 (20%), large sylvian fissures in 5 (33.3%), hydrocephalus in 3 (20%), asymetric large lateral ventricule in 2 (13.3%) and retardation in myelination in 2 (13.3%) patients and 4 infants had normal MRI findings, respectively.

In conclusion, vitamin B₁₂ is important for development of brain. Severe and various neurological and neuroradiological findings may be seen in infants due to vitamin B₁₂ deficiency. Vitamin B₁₂ deficiency should be considered in infants with hypotonia or neurodevelopmental retardation and with neuroradiological findings such as thinning of the corpus callosum, cortical atrophy and retardation in myelination. The early diagnosis and treatment of vitamin B₁₂ deficiency is crucial for significant neurological impairment and long-term prognosis. Vitamin B₁₂ supplementation of pregnant women may prevent neurological and neuroradiological findings of the infants.

It is infants breastfed by vegetarian mothers who are most at risk of vitamin B12 deficiency. This is true in both undeveloped and developed countries.

Fertility among educated, never-married white women

In the recent post on the fertility of never-married women, Audacious Epigone suggested I look at trends over time. I calculated the mean number of offspring for white women ages 40 to 59 who completed at least 16 years of education:


Mean number of offspring

Fifties .00
Sixties .00
Seventies .00
Eighties .14
Nineties .16
Aughts .14
This decade .18

From the 50s through the 70s, educated, never-married white women had no children. Since that time, perhaps one out of six has a single child. The trend suggests that if all educated white women never got married, the numbers of educated white Americans would dwindle rapidly.

This graph shows trends in marital status among educated white women:

Singlehood among educated white women has been increasing for several decades but is still short of twenty percent.

Infant testosterone predicts toddler behavior

A new study from Hormones and Behavior:

The testes are active during gestation, as well as during early infancy. Testosterone elevation during fetal development has been shown to play a role in human neurobehavioral sexual differentiation. The role of early postnatal gonadal activation in human psychosexual development is largely unknown, however. We measured testosterone in 48 full term infants (22 boys, 26 girls) by monthly urinary sampling from day 7 postnatal to age 6 months, and related the area under the curve (AUC) for testosterone during the first 6 months postnatal to subsequent sex-typed behavior, at the age of 14 months, using the Pre-School Activities Inventory (PSAI), and playroom observation of toy choices. In boys, testosterone AUC correlated significantly with PSAI scores (Spearman's rho = 0.54, p = 0.04). In addition, play with a train and with a baby doll showed the anticipated sex differences, and play with the train correlated significantly and positively with testosterone AUC in girls (Spearman's rho = 0.43, p = 0.05), while play with the doll correlated significantly and negatively with testosterone AUC in boys (Spearman's rho = − 0.48, p < 0.03). These results may support a role for testosterone during early infancy in human neurobehavioral sexual differentiation.

Increases in suicide by sex

These graphs show 1999-2009 suicide trends for U.S. men (top) and women (bottom) separately (CDC) :

Both sexes ages 45-64 show large increases over the past decade.

Head Start does not work

A comprehensive study by DHHS shows that Head Start does not work:

Indeed, Head Start did work well in several pilot programs carefully run by professionals in the 1960s. And so it was "taken to scale," as the wonks say, as part of Lyndon Johnson's War on Poverty.  

It is now 45 years later. We spend more than $7 billion providing Head Start to nearly 1 million children each year. And finally there is indisputable evidence about the program's effectiveness, provided by the Department of Health and Human Services: Head Start simply does not work. 

According to the Head Start Impact Study, which was quite comprehensive, the positive effects of the program were minimal and vanished by the end of first grade. Head Start graduates performed about the same as students of similar income and social status who were not part of the program. These results were so shocking that the HHS team sat on them for several years, according to Russ Whitehurst of the Brookings Institution, who said, "I guess they were trying to rerun the data to see if they could come up with anything positive. They couldn't."

More suicide among people in 40s and 50s

The graph below shows 1999-2009 suicide rates by age group (CDC data):

Rates (not shown) are stable for children, adolescents, and younger adults, but suicides have been increasing for middle-aged people. By contrast, they have been dropping among those 75 and over.

How do we explain these trends?  Have advancements in medical care improved the quality of life for the elderly? Are people in their 40s and 50s increasingly feeling like failures at work?  More relationship, marriage and family disintegration?

Rushton on the skin color/behavior correlation

Phil Rushton and colleagues at the journal Personality and Individual Differences continue to push the the theory that genes explain the worldwide correlation between skin tone and important behaviors:

Pigmentation of the hair, skin, cuticle, feather and eye is one of the most salient and variable attributes of vertebrates. In many species, melanin-based coloration is found to be pleiotropically linked to behavior. We review animal studies that have found darker pigmented individuals average higher amounts of aggression and sexual activity than lighter pigmented individuals. We hypothesize that similar relationships between pigmentation, aggression, and sexuality occur in humans. We first review the literature on non-human animals and then review some of the correlates of melanin in people, including aggression and sexual activity. Both within human populations (e.g., siblings), and between populations (e.g., races, nations, states), studies find that darker pigmented people average higher levels of aggression and sexual activity (and also lower IQ). We conceptualize skin color as a multigenerational adaptation to differences in climate over the last 70,000 years as a result of “cold winters theory” and the “Out-of-Africa” model of human origins. We propose life history theory to explain the covariation found between human (and non-human) pigmentation and variables such as birth rate, infant mortality, longevity, rate of HIV/AIDS, and violent crime.

As Razib Khan has explained on this blog, pleiotropy as the explanation of these correlations simply does not work. So what plausible theories remain? Any ideas?

Marital status and fertility among blacks

As a follow-up to the recent post on marital status and fertility, I used 2000-2010 GSS data to calculate the mean number of offspring for black women ages 40-59 by marital status. I looked at the low- and high-educated separately.


Mean number of offspring

Women with less than four-year degree (sample size = 440)
Married 2.65
Separated/Divorced 2.55
Never-Married 2.34

Women with four-year degree or more (sample size = 111)
Married 2.00
Separated/Divorced 1.62
Never-Married .71*

*significantly lower than corresponding married or separated/divorced groups


In the earlier post, we saw that white never-married women have fewer kids than married or separated/divorced women whether they are more or less educated. For black women, this is true only for educated women. Among less educated blacks, never getting married is not associated with significantly fewer kids.

If marriage disappears completely among blacks, its population will not drop dramatically as we see with whites; by contrast, only fertility among educated blacks--a comparatively small group--will shrink. Let's look on the bright side: the disappearance of your smart people is better than the disappearance of your race. 

Black rate of hate crimes against Hispanics is higher than for whites

Using FBI statistics, Audacious Epigone shows that blacks are much more likely than whites to commit an anti-gay hate crime. From the same table, I can calculate the difference for hate crimes involving Hispanic victims. The population-adjusted rate is 1.6 times higher for blacks than whites. This is another indication that blacks have a more serious problem with hate violence than whites. (The rate difference might be inflated some since the "white" numbers include Hispanics who are less likely to commit hates crimes against their own. On the other hand, keep in mind that "Hispanics" include many national groups, members of whom might not like each other.)

This FBI webpage lists the percentage of offenders of any type of hate crime by race. The population-adjusted rate for blacks is 1.8 times that of whites for 2010--the latest year available.

Marital status and fertility

In a chapter of the Meaning of Marriage, Roger Scruton claims that the move away from marriage reduces fertility below replacement level. Using GSS data since 2000, I calculated the mean number of offspring for white women ages 40-59 by marital status. If Scruton is right, married women will have the most kids, never-married will have the least, and separated/divorced women should be in the middle. Let's look at less and more educated women separately:


Mean number of offspring

Women with less than four-year degree (sample size = 1,759)
Married 2.23
Separated/Divorced 2.14
Never-Married .46*

Women with four-year degree or more (sample size = 866)
Married 1.88
Separated/Divorced 1.63
Never-Married .15*

*significantly lower than corresponding married or separated/divorced groups


Never-married women have fewer kids than married or separated/divorced women whether they are more or less educated. Educated single women have hardly any children at all. A society made up of these type of women would pretty much disappear in a historically short period of time. Next, I'll look at black women.

Growing support for gun rights

This Pew graph shows recent growth in support for private gun ownership. I suspect this is part of larger trend that favors maximum individual lifestyle freedom. Other examples include more people favoring legalization of marijuana and gay marriage. How nice it would be to see the acid of individualism attack minority race-loyalty.

Gender reversal on career goals

These two Pew graphs show how young women are now more ambitious than men. Sixty-six percent of females ages 18 to 34 say career success is "one of the most important things" or very important" in their lives. The corresponding figure for young men (seen below) is only 59 percent.

I expect this new gap will further weaken the institution of marriage. I'm not sure if the Girl Power crowd is aware of this consequence. Radical feminists, of course, are cheerleading the decline. If a man does not earn more than a woman, he is less attractive as a partner. Most women want a husband, but they are less likely to enter marriage and are quicker to exit it if the man offers fewer positives. Marriage is still comparatively strong among high status people, but economic incentives matter.

I watched Senator Lieberman on Fox News Sunday. Am I off-base here, or is he the reincarnation of Droopy the Dog?

Are women catching up to men at being unfaithful?

According to this Huffington Post columnist, women are smarter than men at infidelity, and they are catching up to men in prevelance.

The GSS asked respondents if they ever strayed. I divided answers into two periods: 1991-1998 and 2000-2010. In the first period, 23.0 perent of men and 13.1 percent of women cheated (sample size = 8,103). The difference is statistically significant.

In the second period, the figures are 23.4 and 14.5 percent (sample size = 9,304). Again, men are significantly more likely to be unfaithful than women, but the latter estimates for both sexes are not significantly higher than their respective numbers from period 1. The GSS does not support the view that women are catching up with men. And I don't have any data concerning the cleverness of female cheating.

Are stay-at-home moms unhappy?

This Time columnist thinks that homemakers are so unhappy (and angry), social change is called for. What does the GSS say about their level of unhappiness?

I included mothers since the 2000 survey to maximize sample size. Respondents were asked how happy they are these days. Answers range from "very happy" (1) to "pretty happy" (2) to "not too happy" (3). I calculated means for mothers of all work statuses. Higher means indicate more sadness (sample size = 5,286).


Mean sadness score

Full-time 1.82
Part-time 1.78
Temporarily not working 1.90
Laid off 2.05*
Retired 1.82
School 1.83
Keeping house 1.83

*significantly sadder than full-timers


According to GSS data, there is no difference in mean level of happiness between mothers working full-time and homemakers.

People prefer male politicians with lower voices

Here's a new study from Evolution and Human Behavior:

It may be adaptive for voters to recognize good leadership qualities among politicians. Men with lower-pitched voices are found to be more dominant and attractive than are men with higher-pitched voices. Candidate attractiveness and vocal quality relate to voting behavior, but no study has tested the influence of voice pitch on voting-related perceptions. We tested whether voice pitch influenced perceptions of politicians and how these perceptions related to voting behavior. In Study 1, we manipulated voice pitch of recordings of US presidents and asked participants to attribute personality traits to the voices and to choose the voice they preferred to vote for. We found that lower-pitched voices were associated with favorable personality traits more often than were higher-pitched voices and that people preferred to vote for politicians with lower-pitched rather than higher-pitched voices. Furthermore, lower voice pitch was more strongly associated with physical prowess than with integrity in a wartime voting scenario. Thus, sensitivity to vocal cues to dominance was heightened during wartime. In Study 2, we found that participants preferred to vote for the candidate with the lower-pitched voice when given the choice between two unfamiliar men's voices speaking a neutral sentence. Taken together, our results suggest that candidates' voice pitch has an important influence on voting behavior and that men with lower-pitched voices may have an advantage in political elections.

I remember being annoyed in 2008 with McCain's Bruce Dern-like voice. Romney and Cain had the best voices during the recent primaries. Romney might have a slight advantage over Obama. Cain and Obama are definitely better singers.