Item analysis of the KIDSCREEN-10 using Rasch modelling

Abstract

Aim: To test the psychometric properties of the KIDSCREEN-10. Background: It is important to assess mental health and well-being in children for an early detection of psychological problems or hidden morbidities. There is limited knowledge about the psychometric quality of the reduced version of the KIDSCREEN questionnaire with only 10 items. Methods: Analysis of psychometric properties was done by fitting Rasch models and graphical loglinear Rasch models to data collected in a study on acculturation of primary school children and their teachers in 2017. Results: The data did not fit a Rasch model but did fit a graphical loglinear Rasch model. There was local dependence for four item pairs and differential item functioning for gender and citizenship. Conclusions: The KIDSCREEN-10 provides essentially valid measurements of health-related quality of life in children if local dependency and dif ferential item functioning are taken into account. Reliability and targeting were less than satisfactory, especially for certain subgroups but reliability was adequate for most groups.