424-434

• Răileanu Szeles, Monica

This paper applies the one and two parameter- item response models (IRT) with a probit link to derive a scale of financial precariousness. A strong focus is on the item testing and selection, which is sometimes ignored in the IRT literature. In order to test whether the items fulfil or not the IRT fundamental assumptions, a number of tests are applied. The Mokken Scale is the most comprehensive one and provides the most complex output. Also, the paper highlights the advantages of using the IRT instead of traditional models when deriving latent measures from a set of observable binary indicators. This is often the case in the economic and social fields where the IRT is rarely used. Using both IRT models and some testing procedures, we derive a scale of four items and estimate item parameters. In the empirical part, the IRT models are applied on data from the Luxembourg socioeconomic panel (PSELL-3, wave 2006) using the modules GLLAMM (for running the IRT models) and msp/loevH (for testing the items) in the statistical package STATA. The software and modules allowing the application of the IRT models and its underlying tests are also examined. At the end, the paper leads to both empirical and methodological conclusions.

• Economics and Business

• Financial management
• Business management

item response theory, scale of deprivation, tests, difficulty, discrimination, parameter, score

Περιέχει γραφήματα, πίνακες και βιβλιογραφία