Fieldwork Decisions and Their Implications for Data Quality

Abstract

This cumulative dissertation examines how the implementation and adaptation of survey design during fieldwork affect overall data quality. It is based on the idea that data quality is not just determined by the design of the questionnaire or the analytical techniques used, but is also influenced by how the survey is carried out in the field. Fieldwork decisions influence who participates in a survey, the conditions under which information is provided, and whether respondents remain part of a study over time. These processes operate simultaneously and interact with one another, particularly in longitudinal survey contexts. Drawing on the total survey error framework, data quality is conceptualized as a multidimensional outcome resulting from a series of design and implementation decisions involving necessary trade-offs between competing objectives. <br/> The empirical analyses draw on three papers based on the Cologne Dwelling Panel, each focusing on a distinct yet interrelated aspect of fieldwork decision-making. The first paper examines adaptations to interview conditions introduced in response to the constraints of the COVID-19 pandemic, focusing on whether modifying the implementation of face-to-face interviews affects measurement quality while maintaining longitudinal comparability. The second paper investigates recruitment strategies aimed at addressing nonresponse among out-movers, using an experimental design to assess the influence of incentives and mode combinations on participation, sample composition, and survey costs. The third paper focuses on the monitoring of panel attrition in a longitudinal design with non-standard primary sampling units. It examines how attrition unfolds as a cumulative and design-dependent process and how it can be monitored when standard attrition frameworks are inadequate. <br/> Overall, the papers demonstrate that fieldwork and design decisions influence various aspects of data quality, and that the impact of these decisions cannot be assessed in isolation. Adaptations made to address one challenge often introduce new trade-offs in other dimensions of survey quality. The overall discussion therefore argues for a more reflexive and transparent approach to the implementation of fieldwork, in which design assumptions are continuously questioned, and adaptations are evaluated not only in terms of immediate gains, but also with regard to their long-term implications for data quality. By highlighting fieldwork as a central site of data production, this dissertation contributes to a more nuanced understanding of how survey data quality is shaped in practice.