MEASURES OF DISCREPANCY BETWEEN TREATMENT AND CONTROL

Professor Ingram Olkin
Stanford University

When combining the results of independent studies, the starting point is the choice of an effect size by which to measure the effectiveness of a treatment over a control.  The recent discussion on the effectiveness of a mammogram for women between the ages of 40 and 50 is an example of this problem.  In the case of continuous data there is consensus on the definition of effect size. However, for proportions (binomial data) there are a number of metrics in use. This raises the question: does the choice of metric matter? What are the properties of each metric? The focus of this talk is to discuss the various properties by which we might evaluate each metric.
 

  2:45p.m. to  3:50 p.m.
  THURSDAY, OCTOBER 3, 2002
  BRIGHTON HALL - 202
  (MATH BUILDING)

For directions and parking information go to: http://www.csus.edu/pa/directions.html