The project at a glance
-
Start date:01 Sep 2020
-
Duration in months:72
-
Funding:Others
-
Principal Investigator(s):Philipp SISCHKA
About
Valid measurements are the sine qua non of every empirical science. Researchers from social and psychological sciences often rely on multi-indicator self-report scales in order to study attitudes, emotions, cognitions and behavior – especially in cross-cultural research. However, in order to make substantial conclusions, it is necessary to test whether these scales adequately function within each group that is compared and whether scale scores can be compared in a meaningful way across groups (i.e., whether a scale is measurement invariance across these groups). Thus, the current project aims to evaluate the psychometric properties of frequently used scales in a large-scale cross-cultural context across different datasets. Therefore, an item response theory (IRT) approach is used that offers some advantages compared to classical test theory methods and traditional factor analysis. IRT analysis typically not only incorporate information on how well an indicator represent a latent variable (i.e., discrimination parameters or factor loadings in CFA terms) but also item difficulty (i.e., threshold parameters). Thus, IRT consider the item characteristics when obtaining an estimate of the latent variable. Furthermore, reliability coefficients based on classical test theory (e.g., Cronbach’s α) implicitly assume that the standard error of measurement is uniform across the latent variable continuum. Contrary, by incorporating item characteristics IRT models acknowledge that reliability of person scores across different levels of the latent variable may differ as well and are able to derive the conditional reliability, i.e., the reliability across the latent continuum. Measurement invariance and differential item/test functioning is investigated by applying the alignment method (Muthén & Asparouhov, 2014). Previous studies within this project investigated the WHO-5-Well-Being Index, the HBSC symptom checklist, the ten-item Perceived Stress Scale, and the ten-item Social Provision Scale.
Organisation and Partners
- Department of Behavioural and Cognitive Sciences
- Faculty of Humanities, Education and Social Sciences (FHSE)
- Alina Cosma (Trinity College Dublin)
Project team
-
Philipp SISCHKA
Keywords
- Item response theory
- Psychometric properties
- Reliability
- Cross-cultural measurement invariance
- Differential item functionning