Faces Behind the Numbers

This thesis investigated how interviewer–respondent interaction is related to data quality in the context of psychiatric research. Methodological issues are generally studied within survey research rather than psychiatric research. This thesis illustrates that attention to such methodological questions is also highly important in the context of psychiatric research. Various elements of interaction were examined, including deviant interviewer behavior, interviewer repair strategies, rapport, the influence of sociodemographic characteristics, and their similarities. Also, different aspects of data quality were studied, namely participant attrition, item nonresponse, and socially desirable responses to sensitive questions. The findings showed that, despite measures such as standardized interviewing using CAPI, opportunities remain within the interviewer–respondent interaction to influence data collection. Even though interviewers receive extensive training in standardized interviewing, deviations from standardization and the use of repair techniques still occur frequently. The tendency to be helpful to respondents, as part of rapport building or retention, may explain this behavior. Even in studies that do not specifically focus on methodological issues, it is important to consider how data quality can be ensured. The appendix includes a protocol paper describing a cohort study on mood symptoms and resilience in children of parents with mood disorders. This protocol paper details, among other aspects, how training and supervision are organized to ensure a high level of data quality. Good rapport with participants was also associated with reduced attrition of respondents in follow-up assessments and a lower likelihood of item nonresponse. However, good rapport was also linked to more socially desirable responses. Respondents may be motivated to answer as completely as possible while simultaneously avoiding harming rapport by withholding socially undesirable answers. Another notable result was the association between participant age and rapport: older participants rated rapport with the interviewer less positively. Older participants also exhibited higher item nonresponse and more socially desirable answers to sensitive questions, although the latter may reflect genuinely lower incidence of undesirable situations rather than social desirability bias. The less positive experience of rapport partially explained these findings among older participants. Finally, psychiatric status of the respondent appears to play a role in both rapport building and data quality. Although further improvements in the definition and operationalization of rapport are needed for future studies to provide greater clarity, rapport appears to play an important role in the data collection process. This might especially be the case in psychiatric research, in which a relationship built on mutual trust and understanding is presumably even more important to motivate respondents to participate and discuss sensitive topics. Focus on building and maintaining rapport during interviewer training and supervision is therefore recommended.