I'm analyzing data from a 2x2 table and need to compute sensitivity, specificity and positive and negative likelihood ratio values. Can SPSS produce these?
Resolving the problem
SPSS currently does not explicitly offer measures for 2x2 tables that include sensitivity, specificity, and likelihood ratios for positive and negative test results. However, the ROC procedure, which produces receiver operating characteristic curves, will provide sensitivity and 1-specificity values, from which the full set of values can easily be computed. Run the ROC procedure: in the menus, Graphs>ROC Curve. Uncheck the box to display the ROC curve and check the box to display the coordinate points of the ROC curve. When this is done with a binary test variable and a binary state or outcome variable, the listing of the coordinate points of the ROC curve will have three lines. The first line will display sensitivity and 1-specificity values of 1, and the last line will have 0. The middle line will give the sensitivity and 1-specificity values of interest. The likelihood ratio of a positive test is defined as sensitivity/(1-specificity) and the likelihood ratio of a negative test value is defined as (1-sensitivity)/specificity