A multiple-process latent transition model of poverty and health

Summary

Health researchers often use the life-course perspective, exploring how
long-range experiences in one life domain may influence, and be influenced by,
those in another. We develop a multiple-process latent transition model (MPLTM)
to estimate changes in health and poverty dynamics simultaneously, using
repeated measures of self-rated health and income for working-aged adults from
the British Household Panel Survey. We apply the model to quantify concurrent
and longitudinal effects to assess whether changes in these two processes are
related or independent. Model extensions add time-invariant (cohort, gender) and
time-varying (weeks nonemployed in previous year) covariates. We find both
concurrent and bidirectional longitudinal relationships between poverty and
health, with nonemployment appearing to mediate longitudinal health-to-poverty
effects and confound longitudinal poverty-to-health effects. The MPLTM can
provide quantitative estimates of complex interlocking processes that are often
difficult to measure and assess.