From Interviews to Survey Experiments: Innovating Flexible Coding Approaches

• In Event: Mixing Experiments and Qualitative Methods

Abstract

Semistructured interviews provide powerful evidence of how subjects express political concepts in their own words. The narratives that emerge can reveal how identities and experience—beyond simple demographic characteristics—influence variation in attitudes. As such, semistructured interviewing provides a valuable opportunity to ensure that scholars are asking the right questions and operationalizing concepts in meaningful ways. Furthermore, analysis of interviews can inductively identify central hypotheses worthy of further, deductive testing.

While the benefits of multi-method research that moves from qualitative theory-building to quantitative theory-testing are increasingly clear, several obstacles often prevent researchers from adopting this approach. Multi-method projects require diverse skill sets and ample time, often necessitating large teams and complex sampling. Traditional, grounded theory methods for analyzing semistructured interviews can be difficult to implement in team-based projects with large samples. This paper shares insights and recommendations from a team-based, multi-method research study that employs, but also expands upon, the flexible coding approach introduced by sociologists Nicole Deterding and Mary Waters (2018).

Specifically, the paper draws on lessons from the AmeRicans’ Conceptions of Health Equity Study (ARCHES), an interdisciplinary, two-phase investigation of how Americans form and change their views about who deserves what in the health realm. The study includes three components in two Phases. Phase I involves (a) ethnographic observation of a multisectoral health equity initiative, and (b) 170 semistructured interviews with a diverse range of community leaders, health professionals, and residents in Greater Cleveland, Ohio, including those exposed and unexposed to the health equity initiative. Participant observation and interviews aim to generate a deeper understanding of underlying sentiments that shape public opinion on health policy, beyond what polls currently measure. In Phase II, this grounded understanding of what shapes Americans’ conceptions of health-related deservingness is tested on a nationally representative sample using online survey experiments. Building on ARCHES’ experience, this paper demonstrates (1) how flexible coding can be coupled with systematic, team-based analytic memo-writing to generate hypotheses from semistructured interviews for further quantitative testing, and (2) how this approach can simultaneously improve operationalization of concepts in subsequent survey-based research. As textual data is increasingly archived and digitized, understanding how researchers can capitalize on the richness of qualitative data to inform quantitative analysis will improve measurement and generate deeper insights.

Authors

• Abigail Fisher Williamson, Trinity College

• Sarah S. Willen, University of Connecticut

• Colleen C. Walsh, Cleveland State University