Inductivist

One of the strongest findings in sociology is that people follow the principle of homophily--they naturally associate with people like themselves.  It goes without saying that sociologists never take the next logical step and conclude that this universal tendency is rooted in biology; that it is very resistant to change; that it's counterproductive to swim against a very strong tide. The typical view of a sociologist when it comes to a social universal is, "Yes, we see it everywhere, so naturally, we should get rid of it!"

Nowhere is this truer than in the case of race.  Around the entire globe, people of the same race tend to gravitate to each other, so in America, we're bound and determined to see the day when a random white person has all non-white buddies.

So how are we doing on this?  I looked at a question given in the General Social Survey where participants were asked to list friends (sample size = 1,300).  I assume that the first person mentioned is a close friend, if not the closest friend.  When the respondent is white, how often was the first friend mentioned black?  Two percent of the time.  Black respondents first mentioned a white friend 11.2% of the time.

If making friends were truly random, the distribution of friends would match the racial distribution of the population. (We'll set aside the goal of preferring friends from other races.)  For example, 13% of the first friends mentioned by whites should be black since blacks are 13% of the population.  But when I do the math that takes into account the size of the black and white populations, blacks are 37.1 times more likely to say their first-mentioned friend is black rather than white. A white is 9.3 times more likely to mention a white person.  I doubt the numbers were much different in 1960.

For all America's efforts, biology seems to be stronger than sociology.

A reader in the comments of the last post suggests, reasonably enough, that the link between church attendance and number of friends might be due to extraversion.

The MIDUS Study has measures of Big 5 personality traits, but the closest I can find for friends is the extent to which one gets support from friends. Here are the OLS regression results:

OLS Regression Coefficients (standardized, sample size = 3,885)

Church attendance .09
Extraversion .21
Agreeableness .17
Negative emotionality -.06
Conscientiousness .06
Openness to experience .04

All relationships are statistically significantly, and all except negative emotionality are associated with more support from friends. Even after we adjust for Big 5 traits, attendance predicts more support. It is the strongest predictor after extraversion and agreeableness.

Since beginning to go to church in 2004, I find that I have a lot more friends. By nature, I'm a lone wolf who would spend all my time in a corner with a book if I had my way. Church pushes me to be social which is good for me. Heck, I'm even on the parish council now, so I can pretend to have leadership skills. Plus my friends are always hooking me up with great stuff. Even liberals admit that this form of non-governmental social capital is good for society. 

But is this a general phenomenon?  My situation is peculiar because most of my colleagues (professors) are not my kind of people, so I don't make many friends at work.   

As usual, the GSS enlightens us. Here are the mean number of close friends by church attendance:

Mean number of close friends (sample size = 1,463)

Never attends 6.44
Less than once a year 7.25
Once a year 7.21
Several times a year 5.83
Once a month 6.74
Two to three times a month 7.49
Nearly every week 9.89
Every week 7.90
More than once a week 9.25

Number of close friends tends to rise with attendance. I'm not alone.

GSS respondents were asked: "I'm going to read seven qualities one might look for in a personal friend. All of the qualities may be desirable ones for a personal friend, but I'm interested in those that are most important to you. As I read each one, could you tell me whether it is extremely important (1), very important (2), fairly important (3), not too important (4), or not at all important (5): Intelligent."  For four sex-race groups, I correlated this measure with the WORDSUM measure (a vocabulary test) which serves as a proxy of IQ:

Correlations (sample size = 1,009)

White males -.03
White females -.21
Black males -.03
Black females .24

Weird results. Wanting a smart friend is basically unrelated to IQ for white and black men. I wonder if the non-correlation is due to the fact that many people have inaccurate views of their own intelligence. 

There is a tendency for intelligent white women to want smart friends, but more intelligent black women are more likely than dull ones to say that having a smart friend is not important. I might be tempted to interpret this as anti-intellectualism among blacks, but black men do not show the same pattern. Your thoughts?  

Who has the most good friends? I was interested to know which American ethnic groups say they have the most close friends. According to the General Social Survey, this is the mean number reported:


Mean number of close friends (sample sizes in parentheses):

Norwegians 12.7 (32)
Scandanavians 9.7 (60)
Germans 7.8 (198)
English/Welsh 7.7 (198)
Mexicans 7.4 (38)

USA 7.2

American Indians 7.2 (43)
Irish 6.9 (167)
Poles 6.2 (54)
Blacks 6.1 (91)
Italians 5.0 (75)

This is not the pattern I expected. Comments?