Studies into Non-Saturated Randomized Response Models
When participants in sample surveys are faced with questions of sensitive matters such as illegal drug usage, drunken driving, sexuality and the violation of a social norm, they often refuse to reply or give evasive answers. The randomized response (RR) design is a promising indirect interview technique that reduces such response bias and maintains the privacy of respondents. This design assumes that respondents adhere to the instructions and answer the questions truthfully, however, it appears that there are respondents who do not follow the design instructions and provide untruthful answers.
Our study aims to evaluate the performance of non-saturated RR models in order to check response validity and thus obtain more valid prevalence estimates of the sensitive attribute of interest. Additionally, facilitating the application of these models to help researchers in analysing the RR data in a user-friendly manner.
This will be achieved through performing 4 subprojects:
Project 1: Assessing the performance of the Extended Crosswise Model with a number sequence randomizer: Three Pilot studies on substance use and compliance with COVID-19 regulations in the UK
Project 2: Multinomial Logistic Randomized Response Regression model
Project 3: The correspondence between the Cheater Detection model and the Self Protective model
Project 4: An R package for analysing the Randomized Response data
Prof. dr. P.G.M. van der Heijden
Dr. M. Cruyff
Souad Al Sabah scholarship, Faculty of Economics and Political Sciences, Cairo University
September 2020 – August 2024