The Problem of Collecting Data on Mid-Sized Complete Networks (50 < n < 200) via Questionnaire: A Proposition for Using Subgroup-Based Name Generators

Abstract

Commonly, two approaches for collecting network data by means of a questionnaire are distinguished: rosters (complete name lists) and free recall (name generators) with subsequent merging of the ego-networks. However, both methods are reaching their limits when dealing with larger networks: rosters, on the one hand, increase in length with a larger network size, so that respondents either respond more unreliably due to fatigue, or they even abort answering the questionnaire all together. With free recall, on the other hand, weak ties and unpopular persons tend to be forgotten by the respondents, a problem that is also amplified with an increasing network size. In this paper, I want to propose an alternative method for collecting network data via questionnaire: the subgroup-based recall. With this method, it is possible to reliably collect data on mid-sized networks (50 < n < 200). When employing the subgroup-based recall, the network actors are divided into subgroups and a separate name generator is used for each group; the subgroups serve as cues without letting the questionnaire become too long. The use of the subgroup-based recall, however, leads to new methodological challenges, mainly the appropriate division of actors into subgroups. The discussion of subgroup determination shows that the method is best suited for collecting network data in organizational settings as they already provide detailed formal subgroups like for example departments. The article ends with specific recommendations for when to employ rosters, free recall and the subgroup-based recall.