This paper addresses the relation between poverty and the presence of a child with a disability in the family, using longitudinal data from a secondary source. Two main issues are considered. First, to assess the degree to which explanatory
variables influence poverty, an adequate statistical model must handle the possibility that substantial variation between respondents will be owing to omitted variables, multicollinearity and past behaviour. Second, research often compares ‘‘no poverty’’ with ‘‘presence of poverty’’. We acknowledge that there could be differences between those in moderate poverty and those in severe poverty which might provide additional insights into the relation between ‘‘child disability in the family’’ and ‘‘poverty’’. A multinomial random effects model is fitted to the data. The results suggest that the distinction between moderate and severe poverty
appears to improve the results. Further improvement could be achieved by a clear definition of poverty and disability and of observed variables included in the analysis.