http://www.mysocialnetwork.net/blog/mt-tb.cgi/233
http://www.mysocialnetwork.net/blog/481/y35/2006/09/small_university.html
A similar theme in the results of the Stevenson et al. article and the class Small World Experiment results is the pattern of homophily. I predicted that, despite the diversity of the Penn population and my social network, my chain would still consist mainly of females that are similar to myself. Stanley Milgram also notes the prevalence of homophily in social circles in “Small World Problem.” “Small World of the University: A Classroom Exercise in the Study of Networks” exhibits the same tendency toward females passing their folders to other females: “this homophilous tendency is especially pronounced among undergraduate women: 6 out of the 8 paths that originated from undergraduate female students went to other females.” Our results with the S.Y. female folder reflect an even stronger pattern of homophily. Of the ten folders targeted at S.Y., there were only two exchanges from female to male. Our results could possibly be skewed because we have a higher proportion of female participants with a female target in our experiment than in the Stevenson et al. project. Folders targeted at A.P., the male target, also exhibit the tendency to pass to folders to members of the same sex. For this target there were only two completed chains; neither of them show the same degree of homophily. However, the incomplete chains do still echo the prediction that the folders will pass between members of the same sex.
Another reflection of the tendency toward homophily lies in the decision to pass the folders to members of the same class and rarely to members of a lower grade. Results of the Stevenson et al. experiment show that 7 out of the 16 chains had all intermediaries sharing the same class. The results of the A.P. folder somewhat reflect the same tendency to pass the folder to members of the same class or a higher class. In the A.P. chain, 43% of transfers were student to student in the same year, and 28% passed student to student in lower class, as well as 28% of transfers that were student to student in a higher class. Though these results are not dramatic, we have a smaller sample in our experiment that could explain why the trend is not more pronounced in our data.
Differences in design between the two experiments can account for much of the difference in results. While the Stevenson et al. experiment shows the differences in the transfers between class and includes freshmen, our class has Juniors and Seniors, making it difficult to draw conclusions about classes. It would be interesting to see if our results would have replicated the Stevenson et al. data if our experiment involved more folders and lower classes. “Small World of the University: A Classroom Exercise in the Study of Networks” concludes that freshmen are more socially isolated than other Undergraduates because only one out of the 37 starters or immediate intermediary links in the completed chains was a freshman, and no folders in any of the chains were passed to freshman or members of a lower class. When I was planning who I would pass my folder to first, I certainly considered class. I wanted to pass is to someone who is my year or older because I felt that it is more likely that they will know someone in a Graduate program who could access my target. It makes sense that members of the University community would have fewer ties and may be socially isolated. I suspect that my folder did not reach my target because of my last intermediary’s first-year status. While I hoped and predicted that my trustworthy close tie who is also a female would pass the folder to another Senior female, the folder ended up in the hands of a first year student which probably impacted the path of the folder.
A major limitation in the Stevenson et al. research was that they were unable to expand their information about the incomplete chains and about the nature of the ties because they did not use postcards and they did not collect information about tie strength. While our experiment suffered because we did not have as many participants, we are able to draw some conclusions about the incomplete chains. In our experiment, there was no major difference between success rates in chains that began with a strong tie transfer versus chains that began with a weak tie transfer. This is a surprise in some ways because I predicted that closer ties would me more successful in passing the folder to an appropriate intermediary than weaker ties.
Eighty percent of the folders with S.Y. as the target reached her successfully, while only two folders reached A.P. With such a great disparity between the two success rates, one has to wonder what aspect of the chain failed consistently with the A.P. folder. Looking at the two sets of aggregate data, there is no major difference in the nature of the paths that could account for the lack of success in the A.P. folders. The difference could be attributed to the differences in the characteristics of the individual targets. Wasserman and Faust in “Centrality and Prestige” outline the different factors that contribute to degree centrality and degree prestige that impact the position of various members of the Penn community, including both of our targets. I would predict that personal characteristics related to access and prominence of our targets accounts for the different success rates. I would argue that S.Y. is more easily accessible than A.P. because she is a Graduate at the Education school which is geographically closer to Annenberg and is closer in subject matter to communications. A.P., on the other hand, is a staff member who works in a lab that is less related to communications by subject and also more distant geographically.