Network Analysis For Discovering Political Conflict In Standardized Interviews

This paper proposes a new usage of standardized interview data: instead of studying contingency tables of individual attributes, the author suggests to analyze the hidden, implicit network structures among the respondents of these interviews. The proposed analysis is based on the idea of virtual encounters between survey-respondents, which allows to identify relations of conflict or consensus between dyads of persons. The relational information gathered this way can easily be complemented with additional survey-data about the social attributes and the group-membership of the persons representing the nodes of the analyzed network. Hence it becomes possible to calculate inter-group conflict of different status-groups and to compare it with their intra-group conflict. Similarly, by looking for isomorphic coincidences of different types of network relations, this kind of analysis may also be used to identify reinforcing social cleavages.

A standard survey has generally about 1 to 2 thousand interviews, which corresponds to a universe of 0.5 to 2 million possible dyads of persons. Hence, the analysis of virtual network structures requires to focus on samples drawn from this universe. Consequently, this paper also describes the construction of such samples by presenting an exemplary analysis of the European Values Study (EVS) 2008. Among others, this survey contains a political self-evaluation on a left-right scale, which is used in the example for the construction of relational arcs of conflict and consensus: dyads of persons with both a leftist or both a rightist political orientation are linked by a relation of political consensus. Other dyads with very different positions on the left-right continuum are similarly linked by relations representing political conflict. Hence, by this exemplary analysis it will be possible to assess, how much different income groups (classes) are politically polarized and how much consensus there is within each of these groups.