Latent class analysis of diagnostic agreement

摘要: We describe methods based on latent class analysis for and interpretation of agreement dichotomous diagnostic ratings. This approach formulates in terms parameters directly related to accuracy leads many practical applications, such as estimation the individual ratings extent which may improve with multiple opinions. refinements varying panel designs, apply successfully examples medical data that include previously found be poorly fitted by two-class models. Latent techniques provide a powerful flexible set tools analyse one should consider them routinely data.