Cochran-Weiss-Shanteau (CWS) Home Page

What is CWS?

CWS is the Cochran-Weiss-Shanteau index of performance. Our project team, headed by Jim Shanteau and David Weiss, has successfully developed and tested this new approach to the evaluation of expert performance where there is no independent criterion.

The CWS project is administered by James Shanteau and David J. Weiss, who conceived the index as a way to quantify expert performance. The late William Cochran (pictured left) was never involved in the project. His name is commemorated in the acronym for his (1943) suggestion that a ratio of variances be used to assess the quality of a response instrument. CWS similarly employs a ratio, in this case the ratio of discrimination to inconsistency within a set of judgments, to assess the performance of experts.

CWS provides a high degree of predictive accuracy in the identification of expert performance. To date, CWS has been successful in a variety of contexts, including research of experts in auditing, livestock judging, personnel selection, and high and low-fidelity simulations of air traffic control. In addition, CWS has been successful at identifying expert performance in our own Microworld Research Laboratory (MRL). Indeed, we have yet to find a study of expertise where CWS has not done well in identifying who is, and who is not, an expert performer. (Of course, we have only examined data from a small fraction of the total number of studies done on experts. As data from more studies become available to us for reanalysis, we will continue to apply our approach retroactively.) For more information on how to use CWS, you can download the (PDF) CWS User's Guide.

More questions? See our FAQ.

If you would like to know the history behind the CWS index, please see the following paper.

Collaboration

The CWS team is interested in collaborative studies with researchers who gather data from experts. We have collaborated with researchers in academia, the Federal Aviation Administration (FAA), and the Office of Naval Research (ONR).

If you are interested in collaborating with our team, please contact a member of our team. Suitable data sets include responses from two or more candidate experts who have each evaluated a set of stimulus objects more than once. The stimulus objects can be "anything" - air traffic samples, brain scans, diagnostic medical cases, wines, Olympic performances - so long as the expert is asked to make judgments that are applicable to all of the objects.

Site designer: J. Shawn Farris

Webmaster: C. Vowels cvowels@ksu.edu