Investigating the Item Nonresponse and Unreporting in a Panel Survey by Adding a Sensitivity Dimension

Abstract

Item nonresponse and unreporting are significant challenges leading to biases, reducing the precision of estimates, and compromising the validity of inferences drawn from the data. Addressing these challenges with the relationship to research design (panel survey), the data collection modes (face-to-face vs. telephone), and sensitivity and emotional burden remains an important research problem to be solved in the field of survey methodology. The main objective of the current thesis is to investigate the effect of data collection mode on the item nonresponse and unreporting at different sensitivity levels and emotion types in a panel survey. The sub-objectives are to examine the impact of the data collection mode on the item nonresponse and unreporting according to other interview characteristics as well as respondents’ some sociodemographic attributes. To reach these objectives, the individual data set of the National Crime Victimization Survey 2022 (the NCVS 2022) and the data obtained from the Expert Opinion Survey were analyzed through descriptive analyses and logistic regression models. The results reveal that the mode of data collection has significant interactions with day, season, and tenure variables as well as age, education level and employment status. However, the associations differ according to the sensitivity level and the dominant emotion of the questions. The findings of this study shed light on the importance of the mode of interview, other interview characteristics, respondent traits, and the complex interplay between these factors as well as the tradeoff between the nonresponse and measurement errors in social survey research, particularly in the context of question sensitivity and emotional burden.