A two-stage procedure for estimating sensitivity and specificity is de
scribed. The procedure is developed in the context of a validation stu
dy for self-reported atypical nevi, a potentially useful measure in th
e study of risk factors for malignant melanoma. The first stage consis
ts of a sample of N individuals classified only by the test measure. T
he second stage is a subsample of size m, stratified according the inf
ormation collected in the first stage, in which the presence of atypic
al nevi is determined by clinical examination. Using missing data meth
ods for contingency tables, maximum likelihood estimators for the join
t distribution of the test measure and the ''gold standard'' clinical
evaluation are presented, along with efficient estimators for the sens
itivity and specificity. Asymptotic coefficients of variation are comp
uted to compare alternative sampling strategies for the second stage.