Modular Questionnair Designs for Social Surveys: Statistical Modelling of Missingness in Real Social Survey Data 2 (ModuQuD 2)

This project investigates how imputation

procedures can be applied in social surveys collected by using so-called

modular questionnaire designs. Imputing these data can be challenging due to

the typical features of social survey data, such as predominantly low

correlations between variables, a possible large number of variables combined

with relatively small samples, and often a large number of categorical

variables in the dataset. In the first phase of the project, we have identified

promising strategies to deal with these challenges. However, to meet the

requirements for the applications of modular questionnaire designs in social

surveys, the imputation procedures need to be further developed. For example,

this is related to the application of imputation methods for multinominal data,

to the application of the methods with several hundred variables and the

application of complex multivariate models, which are often used in the social

sciences. The examination of these applications is the research aim of the

second project phase.