Applying the Item Response Theory to the measurement of financial precariousness

Measuring financial precariousness in Romania using Item Response

1. Objective
   

Lately, different models have been advanced in the analysis and measurement of latent constructs. The Item Response Theory (IRT) is a methodology widely used in the educational field as well as in psychometrics, as to measure abilities or other scores. The IRT is used here in order to derive a scale and a score of financial precariousness in Romania, from a set of observed variables.
The paper advances methodological and empirical issues. First, it is aimed at exploring several directions in applying the Item Response Theory (IRT) to the investigation of latent measures in the financial field.  The IRT allows analyzing the integration of the measurement and analysis of causes into a single model, as a Multiple Cause – Multiple effects model. As the financial deprivation is one of the most important symptoms of poverty in Romania, a detailed investigation using an innovative methodology may bring new insights.
Even though the paper addresses several issues regarding new extensions of IRT in the measurement of deprivation, it mainly derives a deprivation scale, calculates deprivation scores and analyzes the determinants, in order to catch the most relevant aspects of financial precariousness in Romania.

2. Method and data
   

In summarizing precariousness and finding the determinants, some advantages of IRT arise when one compares it with the sum-score approach and multivariate probit regression (Cappellari and Jenkins, 2006), because the IRT belongs to the classical test theory. For instance, the IRT provides more accurate results when applies at the top and bottom of distribution, compared to the sum score approach. As financial precariousness is a matter of concern especially for the individuals at the bottom of distribution, the IRT is appropriate to be used here. In this paper, we particularly focus on the two parameter- IRT and explore its potential to capture different aspects of financial deprivation into a scale of deprivation.

3. Research questions
   

The analysis of deprivation using the IRT framework is rather new. New extensions of IRT in this field are possible, as previous papers analyze deprivation using IRT only at a static level and calculate deprivation scores only for certain dimensions. The extension at a longitudinal level, aggregation of deprivation indicators into a collective index of deprivation, integration of the IRT into the framework of a MIMIC model are also discussed and applied.  

4. Expected results
   

At a methodological level, we consider that the paper may contribute to literature in several regards: (1) We find several differences in the ranking of deprivation items into a scale, according to the one and two- parameter IRT, which is completely different from the findings of Cappellari and Jenkins (2006). (2) The IRT is also applied here as to derive a global scale and a global score of deprivation; (3) Different dataset and variables provide new insights. At an empirical level, the analysis identifies the groups at risk of financial precariousness and analyzes the factors leading to this form of deprivation.