As McPherson et al explains the important matters question as a name generator elicits “strong personal ties” (355) and mostly does not look at weak ties, which are more likely to show up when a roster is used. The mean number of strong personal ties turned out to be 5 for my sample and the modal number 6, which is the highest possible number. These findings about network size do not coincide with McPherson et al’s findings at all. In their sample, the mean number is 2.08, while the modal number is 0 for the year 2004.
I find it unlikely that these numbers have gone up so significantly over the past couple years. I think the reason for this discrepancy is mainly due to the non-random nature of my sample, since I administered the survey based on convenience sampling. All the respondents of my survey are members of Penn Latin & Ballroom Dance (PLBD), which is a student organization on campus. Although the members of this club are very diverse, this might make them more likely to have larger networks and a higher number of close ties than the average person since they seem to be socially active people. As Fischer suggests “formal organizations are more often supplements to already active lives” (113).
Another explanation could be that for reasons of social desirability the respondents may have reported a higher number of close confidants and give the impression of a broader network. As Zwijze-Koning & Jong suggest “truthfulness of respondents’ self-reports” (434) could be problematic in sociometric questioning.
Although the name generator yields some interesting results, it has its. As Lin et al suggest there are at least three areas where the name generator falls short. To begin with it is bound by a specific content area and it does not measure what people talk about in their relationships. Secondly, it elicits strong ties rather than weak ones, which could be integral to someone’s network (especially in terms access to information) as Granovetter would argue. Finally, it only provides access to individuals as an ego-centric network measure and does not look at people’s social positions.
Some of these weaknesses can be overcome with the position generator. With this method, in all three categories of range, extensity, and prestige, females overall had higher scores than males, which suggests that women have higher social capital in general. However, this finding does not coincide with Lin et al’s findings. The reason for this could be that the Taiwanese society is different than the American society in terms of the “advantage of being in the labor force for males” (75). In my sample every person I surveyed is either a student or in the labor force which eliminates this difference between genders.
One of the weakness of the position generator is that it doesn’t indicate the strength of the ties, which is problematic since it would reveal a lot about the nature of these relationships and types of social support that are exchanged. Also, the list of occupations in descending social prestige could have biased people to pick the higher occupations from the list. Another weakness of this method is that it does not provide detailed information about the “social resources and the diversity of this collection” (Van Der Gaag & Snijders, 4).
Based on the results of the position generator, the value of overall extensity (heterogeneity) is 6.4 out of 15, which indicates low network diversity for this sample. The rest of my findings reaffirmed that networks are homophilous, indicating a low degree of network diversity, as I found 53% gender homophily, 61% age homophily (+/ – 5 years) and 63% education homophily. Also considering that McPherson et al suggest “having kin in one’s network tends to increase contacts across age categories, education strata and sex” (361), these values would have been much higher if kin were excluded. So it seems like, when strong ties are concerned, network diversity tends to be low, as homophily plays an important role in these types of relationships.
To analyze the density matrix, I have used McPherson et al’s approach of looking at the average level of interconnectedness among named confidants by assigning values from 0-1. Based on the survey results, the mean density for all the respondents turned out to be 0.374, which is much lower than McPherson et al’s finding of 0.66.
Moreover, while females and younger people seem to have more densely-knit networks, males and older people seem to have more sparsely knit networks. Since densely-knit networks are more likely to consist of strong ties and sparsely-knit networks are more likely to consist of weak ties, this is an interesting finding. When we look at the duration of relationships as an indicator for tie strength, as Granovetter suggests, we see that the longest relationships of the younger group (excluding kin) are much shorter than that of the older group, even when we look at the duration of their relationships as a percentage of their age. This is not surprising since a lot of people meet their close friends in school and age is bound to become an important factor, while gender does not lead to such differences. However, there seems to be some discrepancy in terms whether these ties are indeed strong ones when we compare the density vs. duration measures, which made me question the strength of ties obtained my the name generator.
Also, there seems to be many “forbidden triads” present in these networks based on the density matrix. In this respect, Granovetter would probably conclude that these ties are not strong ones since “forbidden triads” should not occur between these kinds of ties. However, Burt might say that this is a result of structural holes, which put certain nodes at an advantage in terms of controlling the information flow. It’s also likely that as Marin & Hampton claims the single name generator approach has some validity and reliability issues, and therefore, is not a good measure of network size or density. However, it was simply not feasible to administer a multiple name generator since it is a very time-consuming process.
In terms of McPherson et al’s findings about community ties, there were mixed results in my survey. They found that co-members of a group and neighborhood ties had very low percentages which led to weak community ties. In my sample, no one named a neighbor as some someone they discuss important matters with, which is parallel to the findings of McPherson et al. However, 40% of the respondents named at least one co-member as someone they discuss important matters with, which contradicts their findings. Since the sample for my survey is non-random and since all of the surveyed people were part of PLBD this result is not surprising though.
These types of community relationship are important in terms of looking at privatization as well. An abundance of relationships with spouse and kin as close confidants indicate a high degree of privatization, and the nonexistence of neighborhood relationships supports this. Even though co-membership is very high in my findings, since the sample is biased in this respect, it’s hard to claim that this could indicate a lower degree of privatization.
Another indicator of privatization seems to be the distance and method of communication between the respondent and his/her close ties. 57% of the total close confidants named are in the same state or further as the respondent, which makes regular face-to-face contact very hard and in these cases people use new media such as email, IM and cell phones to keep in touch. It seems like these methods of communication are replacing traditional communication methods.
Only 2% of the total relationships mentioned by the responded are solely based on online communication, which confirms that online and offline communication take place simultaneously, supporting each other. Accordingly, Baym, Zhang & Lin found that in their social interactions college students are supplementing face-to-face interactions and phone calls with Internet interactions. With the advent of new media technologies geographic/physical location has lost its importance in determining network ties, which is reflected in the findings of the survey with the abundance of long-distance relationships that do not allow too much face-to-face interaction. Hampton would argue that all these aspects of online communication could be a positive factor in maintaining community ties that could be at risk by the increased privatization. Based on these analyses, Putnam’s view that people are losing their social support networks seems to be a little too pessimistic. It’s also possible that since the sample consists of respondents with high public participation, these effects are not as pronounced to begin with.
Furthermore, the abundant usage of new media supports Wellman’s suggestion that communities are turning into larger, global, “sparsely-knit and loosely-bounded” social networks from the traditionally conceived notion of fairly small, local, and highly connected groups. Plus, as I have indicated before, while network size and distance between close confidants are pretty large when compared to previous studies, network density is much lower in this respect, all of which suggest that communities as Wellman’s defines them could exist in my sample.
My findings also suggest that people are still getting various types of social support both online and offline. According to Wellman & Wortley, people “get most of their social support –of all kinds- through their small number of strong ties,” (566) and all of the respondents to my survey named at least 2 close confidants. Furthermore, 55% of the respondents named at least one parent as a close confidant and as Wellman & Wortley claims parents are “broadly supportive, usually providing all dimensions of support except companionship” (573). 95% of the respondents named at least one friend as a close confidant and friends do provide companionship, which indicates that most of the respondents should be able to get support in all 5 categories defined by Wellman & Wortley. Moreover, as Wellman & Gulia suggest online relations, which are pretty common among both age groups, provide all kinds of support.
Besides the sampling issues mentioned before, this study also has some weaknesses in terms of administration and response of survey takers. I let the respondent see the survey as I went over it with them and since the survey looks long and complicated, I did get some cringes, which might have led people to put down fewer names or at least go through the answers quickly without thinking too much. This might have affected the multiplexity aspect of their relationships, as people were likely to check only the most relevant box for type of relationship. Also, their communications in the last month were all estimates, as it is impossible to remember every single interaction over the course of a month without a diary. Also, names could have been left out as there was only room for six. Another problematic area was the position generator. I could tell that people who didn’t know many people in different occupations felt bothered by it and this might have led some of the respondents to check more boxes than they should have. Finally, the interpretation of the questions (such as a the definition of important matters) by the respondents might have led to discrepancies in the results.