On the Imputation of Missing Data in Surveys with Likert-Type Scales
Abstract
En
The aim of this paper is two-fold: to propose the imputation procedure named ABBN for replacing missing data in likert-type scales and to compare its performance with some well-known imputation methods. ABBN is a hot-deck imputation procedure which modifies the Approximate Bayesian Bootstrap method by sampling the donor in the neighbourhood of the nonrespondent. The comparison among the imputation procedures is based on a simulation study with data on job satisfaction and procedural fairness scales coming from the recent survey of workers employed in the Italian social cooperatives (ICSI2007). The effects of the imputation procedures on the respondents’ score and on the quality of the scales are investigated.
The aim of this paper is two-fold: to propose the imputation procedure named ABBN for replacing missing data in likert-type scales and to compare its performance with some well-known imputation methods. ABBN is a hot-deck imputation procedure which modifies the Approximate Bayesian Bootstrap method by sampling the donor in the neighbourhood of the nonrespondent. The comparison among the imputation procedures is based on a simulation study with data on job satisfaction and procedural fairness scales coming from the recent survey of workers employed in the Italian social cooperatives (ICSI2007). The effects of the imputation procedures on the respondents’ score and on the quality of the scales are investigated.
DOI Code:
§
Keywords:
missing data; single imputation; Likert-type scales; latent traits
Full Text: PDF