Accumetric Test Performance Analysis software, receiver operating characteristic analysis, nonparametric, sensitivity, specificity, likelihood ratio, predictive value, kappa, weighted kappa, intraclass correlation

New Version 1.1 can do ROC analysis on tests with scores directly or inversely related to state being detected.

ROC analysis includes:

    * counts of true and false positives and negatives at each cutpoint

    * sensitivity and specificity, with precise 95% confidence intervals, at each
      cutpoint

    * ROC analysis of a single test, using the method of DeLong, DeLong, and Clarke-
       Pearson (1988), a more completely nonparametric method than that of Hanley
       and McNeil (1982)

    * bootstrap resampling to calculate bootstrap standard errors, bootstrap
       standard normal confidence intervals, bootstrap percentile confidence
       intervals and bootstrap bias-corrected, accelerated confidence intervals, using
       the methods of Efron and Tibshirani (1993)

    * comparison of two independent ROC curves from separate samples using the
       method of DeLong, DeLong, and Clarke-Pearson (1988)

    * comparison of two or three correlated ROC curves from identical samples
      using the method of DeLong, DeLong, and Clarke-Pearson (1988), a more
       completely nonparametric method than the correlated method of Hanley and
       McNeil (1983)

    * graphs of one, two or three ROC curves on the same graph in color or
       monochrome

    * Copying and Pasting of ROC curves to other Windows documents

    * exporting of graph coordinates to graphics packages

    * statistical power calculations to estimate required sample size for ROC studies,
       for both discrete and continuous data