Fieldwork in social surveys can be adapted in ways that reduce occurrences of nonresponse bias, where systematic differences between respondents and nonrespondents can result in invalid inferences being drawn from the data. In this thesis, we examine changes to fieldwork procedures which aim at reducing the risk of nonresponse bias by raising response rates, particularly by gathering more from units associated with lower propensities to respond to surveys, or by collecting more balanced response sets through prioritising cases in fieldwork. The thesis consists of four research-based chapters, dealing with two major themes: on the one hand, the design and wording of survey communications and on the other hand, changes to fieldwork procedures in repeated cross-national surveys. First, we investigate how mailed invitation letters can be affected by respondent burden. We show that short letters and letters with links in the middle can raise response rates. Second, we examine the effects of including targeted appeals based on demographic background and of framing the survey request in terms of losses and benefits (informed by prospect theory) in survey communications. Our treatments mostly failed to affect response propensities. Third, we study the utility of interviewer observations in monitoring fieldwork centrally in repeated cross-national studies. Our results indicate that these variables are consistently associated with the propensity to respond, but that the strength is likely insufficient to inform post-survey weighting adjustments. Finally, in a simulation study, we explore how different case prioritisation strategies and response propensity models affect survey outcomes under an adaptive survey design protocol in repeated cross-national studies. We show that sample balance and survey costs can be affected, even when few harmonised covariates are available to inform fieldwork efforts, and marginal improvements in response rates can be achieved. However, the simulations did not affect the distribution of target variables.
Date of Award | 1 Aug 2023 |
---|
Original language | English |
---|
Awarding Institution | - The University of Manchester
|
---|
Supervisor | Natalie Shlomo (Supervisor) & Alexandru Cernat (Supervisor) |
---|
- Respondent burden
- Interviewer observations
- Nonresponse error
- Paradata
- Survey data collection
- Adaptive survey design
COLLECTING REPRESENTATIVE SURVEY DATA: FIELDWORK ADAPTIONS FOR CO-OPERATION WITH LOW-PROPENSITY RESPONDENTS
Einarsson, H. (Author). 1 Aug 2023
Student thesis: Phd