Key points are not available for this paper at this time.
We use panel data on schoolchildren's friendship choices to investigate the effects of sex, race, and academic achievement on the formation of new friendships and on the stability of existing friendships. Cross-sectional analysis of the data reveals commonly expected associations: a child's friendship choices are positively associated with being the same sex and race as the other child and with having a similar achievement level. Panel analysis of the data suggests ways that these cross-sectional associations may arise. Although the stability of existing friendships is significantly affected by the children's characteristics, the formation of new friendships appears to be almost random, except for a positive effect of their being the same sex. Further, the status value of characteristics in the larger society does not seem to influence changes in friendship choices. Characteristics of children appear to affect friendship changes mainly because children tend to favor those like (or similar to) themselves on a characteristic. There is an important exception. Children's being the same race does not significantly influence changes in friendship choices when their achievement is controlled. The effects of race are limited to a tendency of black children to form and maintain friendships more readily than nonblack children.
Tuma et al. (1979) studied this question.