How should we measure that?
This week's readings show just how difficult it is to find a satisfying, standard measuring tool in social research experiments and how different situations call for varying measures. We see that each measure has important strengths and flaws and that in an experiment the smallest changes in wording and context can greatly affect the outcome.
The Konning et. al work was the most comprehensive in terms the different methods of data collection that can be employed. Rather than looking at a specific strategy or generator it gave an explanation of the strengths, weaknesses and methods surrouding sociometric surveys, diaries, observations, etc. This paper served as a sort of guide to the general issues behind social network research (like the Monge reading was to networking terms) and is strong in that it is very organized and extensive. The continuous example of the office setting was very useful in helping to make each of the research methods concrete and it also made it easier to understand how these methods compared to one another in a real world setting. Besdies the attention it gave to each individual type of research tool, it also was very strong in that it reviewed reserach in comparisons between some of the various methods. Like the article I was rather disappointed in the limited between-method comparisons and it seems that there needs to be more of these if we are to determine exactly what methods are best for which contexts. Finally, I think this article made a very important conclusion in that there's a tradeoff between generalizibility and capturing all network characterisitcs and showing which methods correspond with which end of this spectrum. This seems to be a critical component in determining how these methods are assessed in the future and it's very questionable as to which is more important to satisfy, as well as whether or not it's possible to satisfy both with one tool.
The Marin and Hampton paper was much more specific in looking at name generators and how to improve their accuracy. Overall the paper was very comprehensive and prescribed shortcut generators to use as well as the rationales behind them. One issue I have however is the idea that the MGRI provides a "perfect" measure of network size (pg. 2). First off, the readings by Hill & Dunbar and Killworth seemed to show that measures predicting network size are very unreliable and hard to match up at all (yet alone perfectly). Furthermore, I feel that the paper didn't connect this total network size issue back to the research in a clear way and these types of organizational issues make it harder to grip the paper's message as a whole. However, despite this I feel that the paper did serve it's purpose in recommending practical shortcuts and the rationale was very convincing. It also clearly showed that there are advantages and disadvantages to each alternative (e.g. MMG is better at showing number of alters in a particular role, MGRI is better at determining proportion of ties within each role relationship). Finally, I found it interesting that the authors chose to limit the names to 6 for each generator. The rationale and data are clear as to why they chose this number however, I wonder if restricting this compromises a lot of important data. For example I would think according to Gladwell's argument, different types of people (i.e. connectors) would list different amounts of people for each role. Though the six names may be pracitcal it also can greatly restrict the differences we see between people and generalizability of the results and I feel it might be important to reconsider this, considering literature such as Granovetter's and Wellman's (East York Study) on tie strength.
The final two papers looked at generators that differed from the standard use of name generators. Both look more at the social capital contacts can provide in terms of resources and access to information. Specifically the Lin et. al paper looks at position in the occupational structure and shows a clear difference in the way that men and women attain success (men through social capital and women through human capital). This of course brings up the question of whether or not this difference is clearly societally imposed or if there are real differences between men and women (as we saw in Wellman's community piece) that affect their social networks. I think this paper's greatest strength is that it looks at a different culture than most of the studies we've examined so far and adds much validity to the idea that these network trends (like the difference between men and women) are universal rather than just present in a certain type of society. Furthermore, I think this paper is strong in incorporating preceding theoretical and empirical tested ideas into the validation of the model. For example, the connection between Granovetter's weak ties and the fact that name generators tend not to show us this, considers others' research in the field and influences the final model because it makes sure that the model at least has the opportunity to show the effects of prior research. Thus, the model doesn't necessarily blindly agree with prior research but uses it as a guide so that when the model is used, its results can give us further insight into the validity of these theories and past studies. This makes the position generator a very robust and useful measure for social network research. The only issue I had with this paper was that it didn't seem to note clearly the disadvantages of the generator and I think these are important to consider when using any type of measure.
Finally, the Gaag paper is also very comprehensive in it's description of measurement issues and how they were resolved as well as the measures they used to test out the resource generator. This paper claims that it is the happy medium between the name generator and the position generator in that it takes the beneficial parts of each tool and combines them into one. The article is very thorough in its descriptions to the point that it is exhaustive, however these are all important issues to consider when first introducing a measuremement tool. Interestingly, the results from using this generator seem to validate many of the patterns we've seen in this course such as "weak ties are better for finding jobs" and "strong ties are better for social support and discussing important matters." The fact that this type of paper shows this, illustrates the important fact that research on measures not only helps standardize a measure for use in different experiments, but also is useful in adding credence to prior theories and empirical results. Thus, as with the Lin paper it's clear that these types of experiments aren't only necessary but can also help reinforce some of the more interesting topics and ideas we've explored in studying social networks.
Questions:
1. How do you explain the apparent discrepancy between the fact that the Lin paper claims men have more access to social capital and articles like Wellman's claim that women are the main players in maintaining social networks?
2. Do you think it's more important to study specific issues that aren't generalizable or many broad issues that are generalizable? Why? Furthermore, do you think this tradeoff actually exists or is it possible to have a practical study that does both very well?