The articles that we read this week all discussed various measures of social capital. The Zwije-Koning and De Jong article discussed the various different methods of network data analysis. They addressed the techniques of sociometric questioning (the most commonly used form), diary research, observation research, analysis of archival records, ECCO analysis and the small-world technique. I found the discussion of sociometric questioning to be particularly interesting. Although the authors tell us that this is the most commonly used method of network analysis, their discussion of the many flaws in this process makes the reader wonder why? I find it so interesting that despite the issues with validity in this process, researchers still continue to use it so heavily. The authors discussed how in any self-reported study, you run the risk of respondents being untruthful. They also talked about how in an office setting, employees might feel threatened and feel obligated to come across a certain way, and therefore to respond to the survey in a certain way. This made me think about the smoking study we read last week, in which the researchers gave students surveys in school. Like handing out surveys in a business environment, handing out surveys in a school environment could drastically influence the results. There was also discussion of the ways the respondents interpret the questions. As we discussed when talking about the number of people with whom we discussed important matters, question interpretation can also lead to much skewed results. Even when the authors compared sociometric questioning to the other methods of network data analysis, it fell far short of the others. In light of these issues it is difficult to understand why this method is still so widely used. While I understand that research of this nature is difficult, and, as the authors explain, this method is very efficient, I find it bothersome that we are still relying so heavily on this method of research.
Question: Since this method is considered to be the most efficient, can you think of any ways that it could be altered slightly so as not to drastically change the efficiency, but to improve the accuracy (particularly in terms of environment like the school and office situations discussed above)?
The Marin and Hampton article discussed issues of reliability with name generators and the differences between single and multiple name generators. One part of this article that I found very interesting was how the authors discussed the use of name generators within the four different approaches, but then how name generators are often misused. They discussed how, particularly with the role-relation approach, name generators can run into issues of reliability. The authors also discuss how name generators can be time consuming and difficult to administer. Similarly to the first article, these authors acknowledge that many researchers search for ways to reduce the time of such methods. It seems from these two articles that researchers often sacrifice reliability for efficiency. Given that shorter surveys may often lead to less accurate results, and longer surveys may lead to antsy respondents, which may also lead to less accurate results…
Question: Do you think that there is any possible way to balance these two? Which would you think is more important to the accuracy of a study: keeping the respondents’ morale up, or asking all of the more detailed questions?
The third article also discusses a generator, but this time they are discussing position generators, rather than name generators. First the authors discuss several more issues with name generators. They then go on to explain position generators and the benefits of using this method rather than name generators. Although the authors seem to find this method to be more accurate than name generators, I found it to be somewhat confusing.
The fourth and final article combines the ideas of the previous two. This article talks about resource generators, which are essentially combinations of name generators and position generators. Van de Gaag and Snijders argued that measuring social capital in a single way may cause us to lose a lot of information, so they aimed to measured social capital in a way that addressed each of its separate parts. The resource generator, they claim, achieves this by combining the good parts of name generators with the good parts of position generators. I found each of this week’s articles a little bit difficult to get through because, when read together, the different research methods that they discussed all sort of blended together. One thing that these articles definitely helped open my eyes to how difficult it really is to find a research method that can accurately measure social networks.
Comments (2)
In terms of your last question, I think it is more important to ask more detailed questions than to be concerned with the respondents' morale. With a dilemma apparent in any field of research (especially marketing and advertising), there is always a constant struggle to appease the respondent and keep them in the best position to remember, analyze, and contribute to the best of their abilities. However, it is so important for a researcher to be able to get to the nitty-gritty details that could be the basis for new findings. As we discovered in these readings, every single methodology has a long list of pros and cons that affect the reliability and accuracy of generalized social networks. However, for those individuals who are able to dedicate their time and effort to responding to longer and more complicated surveys will be able to deliver valuable information, even if it is biased by others.
As discussed in other courses I've taken, the best way to achieve the maximum level of details with minimal respondent antsiness is to offer an incentive, whether as a gift or through letting participants understand how important their participation is to social network research.
Posted by Veronique - y11 | October 25, 2006 9:57 PM
Posted on October 25, 2006 21:57
I find your comment very intrguing because I manage a research lab and I know how it is firsthand to combine both detail and efficiency. I think this is really hard because you have to think about what it's like to take a long survey. Even if you get paid (as subjects in my lab do) you may very likely start to slack off towards the end of the experiment whether out of boredom or shear exhaustion. Thus, it's really hard to know where the line is and it further complicates the matter that each subject is different and will be able to pay attention and focus for different lengths of time. If these claims hold true in reality, I think that efficiency is the most important aspect because if subjects really do get tired there's a higher likelihood that much of this data isn't very well thought through or is done in a sloppy fashion. Therefore you're putting more effort into a survey and not getting that much more in terms of good and accurate data. Considering the economic context of the research and the relative divide of universities it would be very beneficial for different research teams to work toegether as one community and each try out different modes of research. Since many university departments work by themselves it may be most economical to use the most cost effective methods, even though as we see this isn't always the greatest for the research itself. However, by working together you can try out different kinds of methods in shorter, non-stressful surveys/experiments and then combine the results to at least have a better idea of which is best.
Posted by Jason r32 | October 26, 2006 1:04 AM
Posted on October 26, 2006 01:04