The purpose of this study is to examine the friendship composition among high school teenagers and analyze the demographic patterns of the friendship links. Furthermore, the authors estimate if friendship composition have an effect on educational outcomes. The authors intend to close some gaps in the teenage friendship literature by focusing on the effects of the peer composition rather than the cohort quality of the teenager. They use survey data from the National Longitudinal Study of Youth Adolescent Health (Add Health) to conduct the analysis.
With the use survey data they identify teenagers who have mutually identified themselves as friends, and define these friendship links using the demographic attributes of the students who are in that friendship group. Furthermore, the authors exclude friendship links between schools and between grades. Although this may seem too restrictive, the data indicates that 80 percent of all friendship links are within the same grade of the same school. Additionally, only same gender friendship links are being considered to avoid any romantic relationships in the model.
Using this information, a probability model is constructed to describe the probability of a friendship link between any pair of students. This probability is given as a function of demographic composition of each cohort. This model suggest that, the probability of a friendship formation does not exclusively depends on the cohort composition but also on the size of the cohort.
Summary statistics indicate an average of 2.65 one-directional friendships for a male students and 3.1 one-directional friends for a female student. These numbers are much lower when mutual friends are being considered, at 0.69 friends for a male and 1.1 friends for a female student. One limitation of this statistic is that the authors fail to calculate the average number of friends for each age group. This would be...