Comparing Poverty and Deprivation Dynamics: Issues of Reliability and Validity

In this paper we seek to establish if earlier findings relating to the relationship between income poverty persistence and deprivation persistence could be due to a failure to take measurement error into account. In order to address this question, we apply a model of dynamics incorporating structural and error components. Our analysis shows a general similarity between latent poverty and deprivation dynamics. In both cases all unsystematic error is captured as change and we substantially overestimate mobility. Distinguishing between different types of reliability we find that by far the largest component of error is associated with overestimation of the probability of exiting from poverty or deprivation. We observe a striking similarity across dimensions at both observed and latent levels. In both cases levels of poverty and deprivation persistence are higher at the latent level. However, there is no evidence that earlier results relating to the differences in the determinants of poverty and deprivation persistence are a consequence of differential patterns of reliability. Taking measurement error into account seems more likely to accentuate rather than diminish the contrasts highlighted by earlier research. Since longitudinal differences relating to poverty and deprivation cannot be accounted for by measurement error, it seems that we must accept that we are confronted with issues relating to validity rather than reliability. Even where we measure these dimensions over reasonable periods of time and allow for measurement error, they continue to tap relatively distinct phenomenon. Thus, if measures of persistent poverty are to constitute an important component of EU social indicators, a strong case can be made for including parallel measures of deprivation persistence and continuing to explore the relationship between them.