The effect of different survey designs on nonresponse in surveys among non-Western minorities in The Netherlands.

Abstract

The present study investigates the impact of survey design choices on the representativity and nonresponse of eight sub-surveys conducted among non-Western minorities in The Netherlands. The surveys were part of a repeated, cross-sectional measurement conducted separately among each of the four largest non-Western minority groups. This study utilizes fieldwork outcomes in conjunction with the R-indicator and maximal absolute standardized bias to show the impact of survey design choices - such as the period and length of fieldwork, the use of bilingual interviewers, the number of face-to-face call attempts and a re-issue of nonresponding sampled persons - on the potential nonresponse bias. Partial R-indicators are used to detect which socio-demographic subgroups contribute the most to a nonrepresentative response, conditional on ethnic group and survey design. The results indicate that long fieldwork periods increase the potential for nonresponse bias among non-Western minorities due to moving and that the timing of fieldwork has an impact on the number of sampled persons who are unavailable during the fieldwork period. Furthermore, the use of bilingual interviewers is necessary to conduct a survey among Turkish and Moroccans due to language problems; otherwise the potential for nonresponse bias can be quite severe. Also, the use of a re-issue phase reduces the potential for nonresponse bias in surveys among non-Western minorities in The Netherlands. Finally, partial R-indicator analyses provide further insight on how future surveys can be improved in order to reduce the potential for nonresponse bias among each of the four non-Western minority groups. Editor Note: The originally published version of this article has been replaced by a corrected version on 13/02/2015. The corrections concerned two numbers in Table 2.