Parametric models such as the bi-normal have been widely used to analyse data from imperfect continuous diagnostic tests. Such models rely on assumptions that may often be unrealistic and/or unverifiable, and in such cases nonparametric models present an attractive alternative. Further, even when normality holds, researchers tend to underestimate the sample size required to accurately estimate disease prevalence from bi-normal models when densities from diseased and non-diseased subjects overlap. In this thesis we investigate both of these problems. First, we study the use of nonparametric Polya tree models to analyze continuous diagnostic test data. Since we do not assume a gold standard test is available, our model includes a latent class component, the latent data being the unknown true disease status for each subject. Second, we develop methods for the sample size determination when designing studies with continuous diagnostic tests. Finally, we show how Bayes factors can be used to compare the fit of Polya tree models to parametric bi-normal models. Both simulations and a real data illustration are included.