Evaluation of Interviewer Characteristics and Analysis on Interviewer Effect in TDHS-2013
xmlui.mirage2.itemSummaryView.MetaDataShow full item record
Recently, while being few in numbers, there have been studies focusing on survey quality and discussing sources of survey errors. Studies in the literature have emphasized on relatively controllable sampling errors, instead of non-sampling errors, which are costly to determine and difficult to control. This thesis aims to reveal profiles of data collection staff candidates involved in 2013 Turkey Demographic and Health Survey (TDHS-2013) recruitment process. This process consisted of different stages, in which the candidates filled in various forms. In order to reach the aim of this study, a special data set named “TDHS-2013 Data Collection Staff data” (hereafter, “staff data set”) was constructed using these forms. Descriptive analyses are carried out using nationally representative household and women data sets from the TDHS-2013 along with the staff data set. Moreover, pre and post processes of the fieldwork are compared by means of certain information obtained from recruitment forms. Furthermore, effects of interviewer characteristics on both number of completed interviews and interviewer performance are considered. Additionally, results of interviews, response rates and completion rates, along with sample provinces and assignments of the staff, are evaluated on the basis of data collection staff for household interviews and women interviews. Results indicate that information on the candidates should be gathered fully and kept updated to carry out further analyses in detail. The constructed staff data set let us examine the profiles of all applicants for the TDHS-2013 fieldwork, and evaluate the relation between interviewer characteristics and their performances. However, the constructed staff data set with the available information of interviewers was not suitable to make assessments regarding the attitudes, behaviors, and expectations of interviewers. Additionally, calculated calendar ratios on displacement of child birth dates imply that there is a potential interview bias on child birth date information. We expect this study will be a starting point for further studies analyzing interviewer profile, and investigating for a relation between data collection staff and survey estimates.