Social Network Blog - y7

The first article, “Social Isolation and the Underclass” looks at the relationship between poverty and social isolation, the role that neighborhoods play in social isolation and also the gender differences in the structure of social isolation. Fernandez & Harris assess social isolation and also Wilson’s theory of the underclass. They have defined “social isolation” as a lack of personal contact between members of the underclass and mainstream society, isolation from institutions/lack of participation in local organizations and also marginalization from the social networks that extend beyond the local community.

They obtained some pretty interesting results: Firstly, Fernandez & Harris found that not only are women more likely than men to be poor, but women have more constricted networks than non-poor women. In addition, “there are no class differences among males, but poverty serves to isolate females” (273). Could some of these effects be attributed to the disproportionate number of single-mother families in impoverished neighborhoods.

According to the article, the survey respondents were asked to name up to 3 non-kin, non-spouse friends and up to 6 people they could “depend…for everyday favors” and up to 6 people they could turn to for help in a major crisis. I think that one issue with their measurement is that they measured “percentage of kin in support network” solely based on the relationships found in “daily” and “crisis” support networks. What are some of the issues with this measurement?

Finally, Fernandez and Harris also conclude that “ if the experience of poverty is more isolating in the context of very poor neighborhoods, then an argument can be made for policies that redistribute the poor among nonpoor neighborhoods.” This made me think back to Gladwell’s article on social networks and his statement that “poverty is not deprivation. It is isolation”. (6) In the article, he quoted Lois Weinberg saying: "I don't believe poor kids can advance in any way by being lumped together with other poor kids (10). She started a program where poor kids would be able to mix with middle class kids in their afterschool extracurricular activities and it was a great success.

This links to the next article about social resources and mobility outcomes. Marsden and Hurlbert discuss the effects of social network resources & social capital on job changes. They looked at several measures: occupational prestige, wages, industrial sector, firm size, possession of authority and closeness of supervision.

It is interesting that they found that tie strength had a significant negative effect on status and an insignificant positive effective on prestige. But perhaps one of the most significant findings was that the industrial sector of the contact has a substantial effect on the respondent’s current job. This supports the homophily principle and the idea that employers have core contacts.

The debate on weak vs. strong tie continues with this article. Marsden & Hurlbert found that the use of weak ties is associated with experience and firm size and that “the finding for experience is consistent with a network-building argument that suggests that useful contacts are accumulated over time” (1051). But at the same time, contacts reached by weak ties tend to be of higher prestige than those reached via strong ones (1052). It is interesting to note that in this situation, the strength of the tie is not a very strong predictor. The authors conclude that the strongest effects on access to social resources reflects homophily patterns: higher prestige persons reach higher prestige contacts and people tend to find contacts in the industrial sector within which they are employed. Is this advice useful if we are looking for a job as a college grad? If we are looking to work in another industry?

It all ties together...
An analysis of social network structures

In this assignment I have divided the analysis into 4 main topics and for the majority of the topics, analyzed the data in the 18-22 group and 33+ group as well as analyzing the data by sex (male/female).

(1) Social Support & Network Density

:: Social Support ::
Social support in this study is measured by the size of the core discussion networks - the number of people listed that the respondent discusses “important matters” with. Network density, which is defined by Monge and Contractor (2003) as the “ratio of the number of actual links to the number of possible links in the network” (33), is measured by how well the respondent’s ties know one another (strangers vs. know each other vs. especially close).
I found that with the 18-22 group, there were an average of 5.4 ties for males and 5.86 for females. This is a surprisingly high number considering that McPherson et. al (2006) found in 2004 that the typical American had 2.08 discussion partners. This may be related to several factors: (1) the fact that all the respondents I surveyed currently attend college and would skew the results in comparison to the GSS survey participant pool (2) probing them to reconsider if they had less than 6 ties may have had an effect and (3) social pressure & design of the survey (see limitations section).
In the 33+ group, I found results much closer to those found by McPherson et. al. The women averaged around 3 ties and the males around 2.75 ties. This may be due to the fact that with age, comes a “shrinkage of the friendship network…the older people are, the fewer friends and fewer friendship contacts they report” (Kalmijn, 247). Limitations of my sample may have also skewed these results, and I will discuss this in further detail in the limitations section.

:: Network Density ::
In the 18-22 age group it seems that the women have more densely connected networks. The majority of the ties were listed to “know each other”, followed by “especially close” and then “strangers”. There was a surprising lack of connectivity in the male group and 4/5 respondents had listed the majority of their close ties as “strangers”, which goes against Granovetter’s theory of the forbidden triad. Granovetter believes that if A and B are strongly linked and A has a strong tie to some friend C, the tie between C-B should not be absent (1363). This could be caused by the discrepancy in the percentage of kin found in both groups- men had a significantly higher proportion of kin than females did in my sample as will be explained in the section below. In a college setting, parents rarely have the opportunity to meet our friends, which would have made many of these ties “strangers”.
In the 33+ group I found that the networks were much more densely connected. Majority of the ties either “know each other” or are “especially close”. One interesting difference is that the people listed in the men’s networks were more likely to be “especially close” than with female respondents. This is partially explained by Kalmijn’s study (2003) in which he found that “women have more separate contacts than men and that they less often share friends with their spouse than men” (247).

:: Kin in our social networks ::
McPherson et. al also found that with this remarkable drop in size of core discussion networks, also came “ a shift away from ties formed in neighborhood and community contexts and towards conversations with close kin” (353). This can be seen in both age groups but there was a significant difference in gender. In the 18-22 age group, all of the males listed at least 1 kin member in their core discussion network and had an average of 40% of kin in their overall core discussion networks. For females, only 3 listed at least 1 kin member in their core discussion network and females had an average of 12% of kin in their overall core discussion networks. This is interesting because according to McPherson et. al’s study (2006), they saw that “women still have significantly more kin in their networks than men do, but they no longer have fewer non-kin confidants than men”(362). While the latter part of the findings was true in my case, it is interesting that there would be such a huge discrepancy in the data. In fact, McPherson noted that “the kin-dominated nature of women’s networks is one of the staples of the social capital literature” (362). I think that this can be partially attributed a cultural/ethnic factor because the respondants I surveyed were unfortunately all of Asian ethnicity. In Asian culture, it is not common for the male children to be closer to their parents, which may have skewed the data in that way. In the 33+ group, there is a significantly higher proportion of kin in their social networks (males 79%, females 89%).This is again supported by Kalmijn’s findings that with age, social networks shrink and become more kin-based.

(2) Network Size

:: Network Size ::

The position generator is used to measure network size and also access to social capital. Position generators, first proposed by Lin and associated used a sample of ordered structural positions salient in a society and asked respondents to indicate contacts that they knew on a first-name basis. Based on the Lin et. al (2001) article, I ranked the jobs from 2 (lowest, laborer) to 30 (highest, judge) and from these responses it was possible to construct measures of (1) Extensity of heterogeneity of accessibility to different positions (total number of positions accessed; (2) Upper reachability of accessed social capital (status of the highest position accessed) and (3) Range of accessibility of different hierarchical positions (distance between the highest and lowest accessed positions.

Firstly, the 33+ age group had significantly more access to social capital/resources than the younger group As we can see from the tables, the older respondents on average have higher levels of extensity, upper reachability and also range. Lin et. al found that “white males tended to access more position, there was no significant difference between males and females in terms of upper reachability (the highest prestige score) or the range of scores” (67). Although I did not survey white males, the gender discrepancy was also quite significant in my study. Males overall ranked higher in terms of all three categories with the exception of upper reachability in the 33+ age group. Two of the women I surveyed- my mother and my aunt are very much “connectors” like Lois Weisberg (Gladwell) and may have contributed to this upper reachability difference.

(3) Community

Scholars like Wellman argue that “community is much less likely now to be locally based and locally observed” (15). Interestingly, I found that that may not necessarily be as clear-cut with the data I collected. In the 18-22 group I found the following results:

The majority of the people that the respondents discuss important matters with actually either live in their neighborhood (33.8%) or house (16.2%) as well as live in the same country (29.4%). Also, all of the kin that were listed did not live in the same neighborhood, city or state, but were in the same country.

There were also significant gender differences in the distance of their core discussion networks. For females, their discussion partners tend to live in the same building/dorm (66.7%) neighborhood (65.2%) or city (62.5%). For males, they tended to live farther away with 65% only living in the same country. I’m not sure why this is the case, but it is probably attributed to the fact that in my sample, many of the male respondents listed more kin, and therefore would result in a higher percentage of data in the “same country” category.

(4) Homophily

:: Age ::

Age and education levels are pretty homophilious across the board. What I did find interesting was the gender homophily.

:: Gender ::

Firstly, in the 18-22 age group, it seems that there were slightly more female to male interactions (35.3%) and in the 33+ age group, there are significantly more males seeking female discussion partners (40%). In fact, for males aged 33+, 72.7% are more likely to turn to a female than a male. That seems to be reflected in the general trend as well. This could be interpreted to reflect Kalmijn’s findings that show “women are socially less dependent on the marriage than men, a pattern which is directly the opposite of dependencies in the economic domain” (247) and thus men are more likely to turn to their spouses (hence the higher M-F %).

Limitations & Conclusions

:: Measurement ::

Sociometric questioning does have some limitations in its validity. Firstly, Zwijze-Koning & Jong (2005) noted that “an important issue is the truthfulness of respondents’ self-reports…for reasons of social desirability they may report different network relationships than they actually have” (434). I think that in our Social Network Survey this was definitely the case and also because of the structure of the survey itself. Having a table with 6 rows made a lot of people write in more discussion partners than I think they would’ve had it not been presented that way. There also may have been some social pressure to “give the impression that they have a broader network or play a more central role than they do in reality” (Zwijze-Koning & Jong, 435). Also, the probe question may have pressured them into writing more names than they would’ve had I not asked them the “anyone else” question.

In addition, Zwijze-Koning & Jong also noted that forgetfulness may be a concern and that they “may also forget to mention certain exchange relations they have in an organization” (435). It may also be because of status/proximity of these contacts- it is highly possible that many in the 18-22 group listed their college friends because of the proximity/geography factor.

:: Analysis ::

I think that some of the measurements that were left out of our particular survey that could’ve been interesting are (1) race and (2) where the respondent lives. Firstly, I think that racial/cultural factors did play a part in the results I got and also my comparison with many of the studies, which usually surveyed upper-upper/middle class white people. It would have been interesting to see if race does account for certain trends or if there is as much racial homophily we think there is. Secondly, Fischer had some interesting points in his articles from week 5 that address the issue of urbanity/urbanization and our social relations. Since some of the people I surveyed do not live in the city, it would have been interesting to see if that did play a role. For example, does living in a suburb account for the fact that you may know your local dry cleaner, truck driver, etc? It would be an interesting point of comparison.

:: Data Collection & Interpretation ::

I found that I had a lot of trouble with the data collection process. Firstly, many people I feel interpreted the question “discussed matters important to you” differently. I know for a fact that one respondent thought that meant work-related issues and not personal issues, and some felt it was financial issues. As an objective data collector, I could not tell them necessarily what “important matters” constituted, but their interpretation may have skewed the results.

There was also a significant amount of error in filling out the survey. Many people left out boxes or filled it out wrong and I had to get them to read it over again, or in some cases chase them down to correct their mistakes. One unforeseen mistake was misinterpreting the distance question- 2 respondents checked “same house” for their kin members when I know for a fact they don’t live with them now.

Also, I know that my analysis is limited because I only managed to collect 3 females and 4 males over 33. I tried to compensate by surveying more in the 18-22 range, but that will skew my data significantly.

In addition, because of my workload this week I was really unable to survey strangers. I completed the majority of my 33+ age group surveys with respondents on the phone. I had emailed them a copy of the survey beforehand and asked them to have a copy in front of them while I guided them through it and recorded their answers. I understand that this wasn’t really the point, but it was the only way I realistically could get the data.

Lastly, I feel that my data was skewed because I didn’t get a very wide range of people to fill it out. In the 18-22 age range I got many of my friends to fill it out and many of their networks are overlapping and very similar. In the 33+ range I surveyed 1 friend and the rest were family/extended family. These factors may have led to a more homophilious and less representative data set.

:: Conclusions & Final Thoughts ::

It was really interesting to see everything tie together with this final project. I regret not being able to put more time into it and the same goes for the work over the course of the semester. Having to work part-time and juggle a full class load and extracurriculars really proved to be too much for me to handle and I would’ve liked to have been able to devote more time to the material learned in this class. I hope you don’t take that as a sign of disinterest/boredom, because I truly feel that there was a lot of value in the readings, our discussion and also the learning style (blogs and such).

Conducting our own research certainly proved to be an effective way of having to synthesize the course materials. I think that with the Social Network Survey I was able to see the trends we found in the reading apply to “real life”- the social networks around me. At the same time I realized that are so many limitations to data collection (type of data, how you collect it, etc.) that every study is only able to show a slice of the bigger picture.