Share this:

Jeffrey A Smith 80x108Miller McPherson 80x108 Lynn Smith-Lovin 80x108In recent decades, the U.S. has undergone significant social and demographic change, and has become much more racially and religiously heterogeneous. But have these changes been reflected in who people now associate with for friendship and other relationships? In new research, which measures changes in ‘homophily’, or the tendency for people who are similar to be more likely to know each other, Jeffrey A. Smith, Miller McPherson and Lynn Smith-Lovin find thatmore people now know someone with a different racial/religious background, and that these changes have occurred in line with demographic shifts. They also find that homophily is decreasing along gender lines as men and women occupy increasingly similar roles in society, and that there is less interaction between people in different age groups.

Birds of a feather have always flocked together. Homophily is a basic and pervasive social fact: people who are similar are more likely to know each other than people who are dissimilar. But how does that shift in times of economic and demographic change? And what does that tell us about the salience of the demographic features in society?

The extent to which our social connections reflect our own demographic characteristics reveals a great deal about how well society is integrated across race, gender, age, class and religious lines. Having a homogenous social world may be comforting and secure, but it also limits our social horizon, making it difficult to understand and accept the views of others. In recent research, we ask if the strength of demographic homophily has changed over the past 20 years, as the United States has grown more demographically diverse and economically unequal. Using the 1985 and 2004 General Social Survey, we explore trends for 5 demographic characteristics: race, religion, sex, age and education.

We find that race is the most important dimension overall, while education and sex are the weakest. For example, the probability of two people of a different race claiming each other as a close confidant is lower than a person with high school and a graduate degree being close confidants. Similarly, less than 10 percent of confidant ties are cross-race, while over 40 percent of ties are cross-sex. The overall pattern of salience is unaltered between 1985 and 2004. The changes that we do see are subtle, and not large enough to upset the basic pattern of homophily seen in 1985.

Looking at Figure 1, it is clear that more people report being close to someone of a different racial or religious background in 1985 than 2004. The US is more racially and religiously heterogeneous than it was in 1985, with an increasing Asian and Hispanic population, as well as an increase in individuals claiming a non–Judeo-Christian religion. White Protestants represent a shrinking portion of the population. These changes directly affect the racial and religious composition of individuals’ social networks.  Individuals now have more racially and religiously diverse social contacts.

Figure 1- Proportion of Social Ties that are between Demographically Similar People

Smith Fig 1

The results suggest, however, that the number of cross-race and cross-religion ties would not have increased without large changes in the population’s composition. Thus, once we adjust for the demographic changes, the shifts in racial and religious homophily are quite minor. The increasing number of out-group racial and religious ties are based on changes in opportunity, where one has more chances of meeting someone of a different race/religion than in the past—as opposed to more fundamental changes in the relationship between different racial or religious groups.

Still, we should not downplay the importance of increasing cross-race and cross-religious social contact. Nothing had to change, of course. Even with more opportunities for out-group interaction, we may not have seen more interracial ties (for example, if the economic or cultural differences were too large between demographic groups). The fact that we do see more cross-race and cross-religious ties is important: more people now know someone with a different racial/religious background. This may, ultimately, lead to a more integrated country.

The trends we see for gender invert the racial/religious story: while the gender composition of the country changed very little, the proportion of cross-gender ties increased (see Figure 1). Men and women are now more similar along economic and occupational outcomes; women earn higher salaries and are more likely to work full time, while men are (somewhat) more likely to participate in childrearing. Given this economic convergence, men and women, particularly spouses, are more likely to see each other as confidants. Thus, we see decreasing homophily along gender lines as men and women occupy increasingly similar roles in society, and not because of increases in opportunities for interaction.

The changes for age are more subtle. There are few changes to the overall salience of age. The changes that we do see are concentrated primarily among the young, who are increasingly isolated from older cohorts. Younger individuals are increasing delaying “older” responsibilities of marriage and family formation. Younger individuals may look increasingly different from older cohorts, especially in terms of residential and associational choices. Couples with kids live in different neighborhoods and join different clubs than singles without kids. We thus see a larger block of “young” people, incorporating those in their 20s and early 30s, who are increasingly more likely to interact with each other, but less likely to interact with those who are older.

The changes in education are similarly circumscribed. Over the last 20 years, we have witnessed an unprecedented increase in income and wealth inequality (see the work of Neckerman). Wage inequality along educational lines was much higher in 2004 than 1985, as were differences in wealth and consumption disparities. The social consequences of not having education are higher now, and this should, theoretically, lead to fewer ties between individuals with low and high education. In actuality, educational salience is largely the same in 1985 and 2004. The changes that we do see are concentrated on the lower end of the educational spectrum: those with less than high school and high school education are socially closer than before, and less likely to have confidants with higher levels of education. Individuals without the requisite educational credentials are thus increasingly socially isolated.

Overall, the implications are that: the composition of one’s social network is highly contingent on the composition of the population; homophily, net of the opportunity structure, is quite stable over time; the changes that we do see are relatively small, or isolated, and do not upset the general trend of similarity between social confidants; and changes are small because homophily depends on a number of intertwining factors, all of which must move in the same direction to see a large change in the salience of a demographic dimension. Such systematic upheavals are rare, and the integration (or disintegration) of society will tend to move at a comparatively slow pace.

This article is based on the paper ‘Social Distance in the United States: Sex, Race, Religion, Age and Education Homophily among Confidants, 1985-2004’, in the American Sociological Review.  

Featured image credit: Amanda Venner, Flickr, CC-BY-NC-SA-2.0

Please read our comments policy before commenting. 

Note:  This article gives the views of the author, and not the position of USApp– American Politics and Policy, nor of the London School of Economics.  

Shortened URL for this post:


About the authors 

Jeffrey A Smith 80x108Jeffrey A. SmithUniversity of Nebraska-Lincoln
Jeffrey A. Smith is an assistant professor in the Department of Sociology at the University of Nebraska-Lincoln. His research interests include social networks, quantitative methodology, and stratification. He has done methodological work on network sampling and simulation, and more substantive work on social distance, status, and race/ethnicity.


Miller McPherson 80x108Miller McPhersonDuke University
Miller McPherson is Professor Emeritus at the Department of Sociology at Duke University. His research focuses on voluntary groups and the social networks that draw people into and out of them.




Lynn Smith-Lovin 80x108Lynn Smith-LovinDuke University
Lynn Smith-Lovin is a Professor of Sociology in the Women’s Studies Program at Duke University. Her research interests include identity, action and emotional response.