Sensitivity, Specificity, and Misclassification
Here at NeuropsychNow, one of our goals is to provide our readers with a way to simplify complex information. There are certain topics in neuropsychology that, for a variety of reasons, are not well understood by many. One such topic has to do with the statistics used to determine classification accuracy. Oh, and by the way, because we love to show you how to use technology to make your life easier, the figure used in this post was created using an iPad app called Notes Plus, a Bamboo stylus, and some inspiration from Mike Rohde’s The Sketchnote Handbook.
The figure (above) speaks for itself. However, suppose you developed a test to detect the presence of some condition (e.g., malingering, dementia, PTSD, etc.). Of course, you would want to know how well the test can discriminate between those with and without the condition of interest (COI). You would also want to know the rate of false or incorrect classification. The statistical terms used for this type of information include the following:
- Sensitivity - the proportion of individuals with the COI who are correctly detected by the test.
- Specificity - the proportion of individuals who are correctly identified as not having the COI.
- False positive - the proportion of individuals who are misclassified as having the COI when they don’t.
- False negative - the proportion of individuals who are misclassified as NOT having the COI when they really do.
That is all for now. Next time: negative and positive predictive values.