When the results of a reference (or gold standard) test are missing or
not error-free, the accuracy of diagnostic tests is often assessed th
rough latent class models with two latent classes, representing diseas
ed or nondiseased status. Such models, however, require that condition
al on the true disease status, the tests are statistically independent
, an assumption often violated in practice. Consequently, the model ge
nerally fits the data poorly. In this paper, we develop a general late
nt class model with random effects to model the conditional dependence
among multiple diagnostic tests (or readers). We also develop a graph
ical method for checking whether or not the conditional dependence is
of concern and for identifying the pattern of the correlation. Using t
he random-effects model and the graphical method, a simple adequate mo
del that is easy to interpret can be obtained. The methods are illustr
ated with three examples from the biometric literature. The proposed m
ethodology is also applicable when the true disease status is indeed k
nown and conditional dependence could well be present.