Social Network Analysis as a Tool for Assessing Respondent Burden and Measurement Error in Establishment Surveys*Alfred D. Tuttle, US Census Bureau
Diane K. Willimack, U.S. Census Bureau
Keywords: Establishment surveys, respondent burden, measurement error, organizational survey response process
Establishment survey respondents often must obtain assistance from other persons in their organizations in order to complete a survey. This is especially true in larger organizations where the requested data may be distributed among multiple organizational units, data systems and/or personnel with specialized knowledge or system access. Qualitative research suggests that obtaining the assistance of other (secondary) respondents contributes significantly to the perceived burden of a survey request. In this paper we propose a unique approach using social network analysis (SNA) methods to obtain quantitative and qualitative data on the primary and secondary respondents who prepare an organization’s survey response. We will use case studies from qualitative research projects to illustrate how SNA could be usefully applied and the types of data that would be collected. By gathering quantitative data on types and numbers of respondents contributing to responses, we may produce an objective measure of the burden associated with a survey request, and an indicator of possible measurement error. We will discuss prospects for improving data quality and the effectiveness of survey collection.