What kind of birds should we all be?
In “Birds of a Feather”, McPherson et al introduce the discussion and presence of homophily in social networks. They mention the different levels in which homophily is present, and also go over possible causes that could be driving the creation of homophily in people’s relationships. The authors say that homophily in different types of relationships, including both strong and weak ties (418). If this is true, I wonder what Granovetter and Burt would say since they say that the main strength of weak ties lies in the fact that these can provide us with new information and networks that could benefit us. The point that struck me the most was the fact that race and ethnicity were mentioned as the biggest dividers of networks, followed by “sex, age, religion and education also strongly structur[ing] our relations with others” (429). I find it a little controversial that both race and ethnicity AND education can be high dividers at the same time. It would seem that high education would actually act as a buffer to social division due to race and other socially ascribed status.
It would have been interested if McPherson et al had applied the notion of homophily to the increasing trend of Internet social networks and the different groups that emerge because of similar interests. How would these new-media networks be similar and/or differ to the way in which homophily is found to the levels studied in the article?
The Pearson et al paper examines the roles that both assimilation and the structure of an adolescent’s network, homophily, play in determining that adolescent’s use of substances. The results of the study showed that both homophily and assimilation are important in the friendships and substance use in adolescents. I was struck by the fact that selection was the main explanation behind drinking in the sample. Could there be an age cohort effect going on behind here that the authors failed to mention? The study was carried out in Scotland, with students beginning the study at age 13, and ending 3 years later, which means that they were never legal. Thus, it would seem natural that kids who tended to drink would seek out others who also drunk for several reasons: maybe those who didn’t drink could potentially make the drinkers feel uncomfortable or deviant, and also, the drinkers may believe that the non-drinkers would tell on them if they found out. Therefore, I feel that this piece of data should be analyzed a bit further, and see whether selection still is more important among drinkers once they reach legal age. (This could be taken a step further and compared to/studied a sample that starts to drink once they are of legal age to drink, since it could be said that those who have been friends since adolescents and started to drink together then could have stayed together even when they are older).
The authors say that “the apparent increased smoking levels among girls could be partly accounted for by their reduced sporting activity” (56). Could this be the other way round: because girls smoke more, they play less sports? What other variables do you think could be playing in girls increased smoking levels?
I thought that the study done by Hill and Dunbar was very original in that they decided to measure social network size by the exchange of Christmas cards. However, there are many points that could be made against such a choice of measure. To begin with, by tracing who Christmas cards are sent to, people from other religions who do not follow the tradition are automatically excluded from the sample. Also, 43 returned questionnaire is not a big sample from which one can draw generalizable conclusions from. (On a side not, I was very struck by the fact that the number of means cards sent was almost 70!) Furthermore, the fact that Christmas cards were traced also leaves out other groups of people, such as those who can’t afford to send Christmas cards, who don’t follow the tradition for personal and cultural reasons, etc.
Killworth et al’s paper studied and applied statistical methods to phone books of Jacksonville Florida and Mexico to estimate sizes of personal networks. Phone books can be tricky things: as the authors mention it on their article, phone books not only contain people, but also companies and services which could potentially be a problem for this kind of studies in which the researchers are trying to find information about personal networks. Even though the authors did manage to go around this problem by randomly selecting subjects, the question of whether the majority of the population, and what part of the population is not listed on phone books has to be considered. Especially nowadays, in which cellphones are predominantly used by younger generations, and in which the Internet even has come to replace some of the traditional paper phone books through online phone directories, this method of estimating personal network sizes through phone books could lead to a sample bias and maybe be based on a somewhat out of date information.
Q: What other methods can u think of that would be useful to study the sizes of personal networks?