The contribution of rapport between interviewer and respondent on interview quality from non-sampling error perspective: Evidence from 2014 research on domestic violence against women in Turkey

The dissertation aims to introduce two latent constructs in surveys which are rapport between interviewer and respondent, and interview quality. The third main objective is to investigate rapport impact on interview quality. Furthermore, secondary objectives are to determine levels of rapport and interview quality, to reveal differences among women whose interviews were completed with high rapport according to selected characteristics, and to investigate effect of interviewer characteristics on interview quality. From quantitative perspective, sophisticated measurement models are utilized to achieve thesis objectives. Data set of Research on Domestic Violence against Women in Turkey, which is a household survey conducted in 2014, is used for statistical analyses within the dissertation. The nationwide, cross-sectional, face-to-face survey was carried out by Hacettepe University Institute of Population Studies (HUIPS) in collaboration with Turkish Republic Ministry of Family and Social Policies, the General Directorate on the Status of Women (GDSW). The Field Staff data set was constructed separately and merged with women data set. In the first stage of the thesis, rapport and interview quality operationalization were given with several models. Afterwards, all phases to reach final measurement models for latent constructs were discussed and, finally, rapport impact on interview quality was investigated on the basis of structural equation model. In the second stage of the results section, differences among women interviews in terms of high rapport according to selected characteristics were revealed based on significance levels. Lastly, interviewer characteristics' impact on interview quality was investigated under the control of covariates, and influential interviewer characteristics on the likelihood of high interview quality were given. Operationalization of rapport and interview quality are found useful to make interpretations on survey related constructs which we cannot measure directly. Findings showed that timing and frequency of visits, socio-demographic and socio-economic similarity between interviewer and respondent, and dynamic interview factors contribute to measurement of rapport. Interviewer workload, respondent burden and response quality are detrimental factors on interview quality in a negative way. Not only factors behind rapport and interview quality, but also the impact of rapport on interview quality showed importance of interviewer recruitment, workload allocation, interviewer training, field management, and motivation.