A European measurement scale for material deprivation

Abstract

Abstract The predominant practice of establishing a material deprivation scale as the raw sum of a relatively small set of binary items has obvious returns in terms of communication and simplicity of use. However, it can be misleading when basic measurement properties are not met. Only within the Rasch measurement approach, these requirements can be properly addressed. In Rasch models, material deprivation is treated as an unobservable numerical variable (latent trait) that can be inferred from a set of observable items, each representing a different aspect of the same latent trait. Based on Rasch model analytical procedure and using the Survey on Income and Living Conditions (EU-SILC) pre-Covid data, this paper develops a European measurement reference scale for material deprivation which ensures validity and comparability across different countries. Specifically, from a pool of potential items, we introduce a stochastic method to select the largest number of items that guarantee the property of uni-dimensionality of the latent trait, not necessarily ensured by the Rasch model. We also adjust for possible differential item functioning (DIF) to better ensure comparability of the scale across European countries. We finally estimate prevalence rates of material deprivation in 28 European countries. Prevalence rates estimated using the traditional counting approach substantially differ from the figures estimated with the new measurement scale when DIF occurs.