On the identifiability of structural parameters in mixtures: Applications to psychological tests (Q1329704)

The author deals with the problem of identifiability of the parameter \(\theta\) from the mixtures \(P_{\theta, M}= \int p_{\theta,\eta} dM(\eta)\). He specifies conditions for families with finite support under which this phenomenon occurs if the mixing distribution \(M\) is from a large (nonparametric) family and points out on several examples that the results for the families with finite support cannot be easily extended to the families with countable support. Finally, various results on identifiability and nonidentifiability problems arising in models for psychological tests with dichotomous response are given.