I’ve been waiting for reports using the data from 2010 General Social Survey (GSS) for a while now, given that GSS has implemented experiments on the name generator question in their 2010 survey. Sociologists in Univ. of Iowa just released their findings online at here, indicating that the decrease in the network size found in 2004 (as compared to what was found in 1985) may be an product of interviewer effects. Given that the idea of social isolation in America seems to become a common sense in popular press now, this piece is particular crucial as it challenges the validity of the findings from the 2006 AJS article by McPherson and his colleagues.
Interviewer effects or not, eliciting names of contacts in one’s personal networks is a very difficult task. Not only do respondents have to think and decide what constitute important matters, but they also have to recall the people they know and then evaluate whether they had conversations about the important things with the people they think of in the given time frame. Indeed the name generator has its flaws, but it is still a useful technique in large-scale survey settings. The real agenda is how to improve the effectiveness of recall and help respondents to report the information.
I’m excited that my survey experiment works well and my preliminary analysis finds that respondents can provide good amount of personal network data when they are prompted appropriately, without a huge trade-off of engaging in satisficing behaviors during the questions subsequent to the name generator. Stay tuned for a more completed write-up, I’ll report more findings at here as well as in conferences and journals.
I’ve been waiting for reports using the data from 2010 General Social Survey (GSS) for a while now, given that GSS has implemented experiments on the name generator question in their 2010 survey. Sociologists in Univ. of Iowa just released their findings online at here, indicating that the decrease in the network size found in 2004 (as compared to what was found in 1985) may be an product of interviewer effects. Given that the idea of social isolation in America seems to become a common sense in popular press now, this piece is particular crucial as it challenges the validity of the findings from the 2006 AJS article by McPherson and his colleagues.
Interviewer effects or not, eliciting names of contacts in one’s personal networks is a very difficult task. Not only do respondents have to think and decide what constitute important matters, but they also have to recall the people they know and then evaluate whether they had conversations about the important things with the people they think of in the given time frame. Indeed the name generator has its flaws, but it is still a useful technique in large-scale survey settings. The real agenda is how to improve the effectiveness of recall and help respondents to report the information.
I’m excited that my survey experiment works well and my preliminary analysis finds that respondents can provide good amount of personal network data when they are prompted appropriately, without a huge trade-off of engaging in satisficing behaviors during the questions subsequent to the name generator. Stay tuned for a more completed write-up, I’ll report more findings at here as well as in conferences and journals.