Networks in Time: Testing Contextual Effects in the 2004 General Social Survey Network Items
Download data and study materials from OSF
Principal investigators:
Byungkyu Lee
Columbia University
Email: bl2474@columbia.edu
Homepage:
http://www.byungkyulee.com/
Peter Bearman
Columbia University
Email: psb17@columbia.edu
Homepage:
https://sociology.columbia.edu/content/peter-bearman
Sample size: 1055
Field period: 02/06/2016-10/13/2016
Abstract
How do political contexts in the real world shape individuals’ interpretation of network name generators? Although the finding on increasing social isolation in American based on the 2004 General Social Survey (GSS) has garnered scholarly and public attention, prior scholarship has largely focused on potential methodological artifacts as driving the reported decline in network size. However, what is missing in these debates is consideration of political context, which would shape what are “important matters” to discuss. We propose a prospective experimental design by combining the survey timing and political priming instruments with two different network name generators. The result will help to confirm or disconfirm the contextual sensitivity of “important matters” name generator as well as provide insights into the role of survey timing in network studies.
Hypotheses
Hypothesis 1a. Political priming that would frame important matters as political matters will decrease the size of core discussion networks.
Hypothesis 1b. Political priming that would frame important matters as political matters will increase the size of political discussion networks.
Hypothesis 2a. The first presidential debate that would frame important matters as political matters will decrease the size of core discussion networks.
Hypothesis 2b. The first presidential debate that would frame important matters as political matters will increase the size of political discussion networks.
Experimental Manipulations
We propose a prospective 2 x 2 x 2 between-subject experimental design that incorporates the combination of survey timing as essential part of treatments along with two types of name generators (political/important matters name generator) and political priming. First, we will consider the First presidential debate in 2016 (September 26th Monday), which is the first and foremost salient political event at the beginning of presidential election, as a major instrument for contextual priming. We will randomly assign subjects into two periods – before and around the first presidential debate (September 26th). Second, to assess the effect of psychological political priming, we will vary the order of presenting political questions. In the priming condition, a bunch of political questions -- political interest, political discussion, voting behavior, political information channel, political activity -- will be administered before asking the name generator. In the control condition, political priming questions will be administered after the name generator. Finally, we randomly assign half of the subjects to the “important matters” name generator and the other half to the “political matters” name generator.
Outcomes
Network size (the number of names provided by respondents in network name generators) and other network characteristics including discussion topic and network homophily.
Summary of Results
Both important matters and political matters network name generators provide almost identical estimates for network size (=1.38), which suggests that people thought about “important matters” as political matters throughout the highly polarized 2016 election season. Average network size was smaller in the psychological priming condition (1.44 versus 1.31), though this effect is not statistically significant. In contrast, network size becomes significantly larger in the contextual priming condition (1.23 versus 1.52). Our experimental design enables us to identify the precise role of political priming in shaping both political and social networks. We find that there is a marginally significant effect of the psychological priming that arises from asking political questions before asking name generators when both discussion networks are assessed simultaneously, though it loses its significance when they are assessed separately. With regard to the framing effects that arise from switching conversational topic from “important matters” to “political matters,” we find that they are not significant across both types of political priming because there has been no gap between the size of the important matters and political matters networks.
References
Lee, Byungkyu, and Peter Bearman. 2017. “Important Matters in Political Context.” Sociological Science 4:1–30.
Lee, Byungkyu, and Peter Bearman. 2020. “Political Isolation in America.” Network Science 8(3):333–55. doi: 10.1017/nws.2020.9.
Lee, Byungkyu. 2021. “Close Relationships in Close Elections.” Social Forces. doi: 10.1093/sf/soaa101.