On Combining Independent Significance Tests
Abstract
How to combine the results of k independent tests of significance has long been an important problem in statistics. The problem can arise, for example, in such diverse situations as when tests on the mean survival time after diagnosis of a terminal disease are made on k groups of patients in different hospitals, or when the sets of observations in k cells of an ANOVA table are separately tested for normality. An important feature of such tests is that often the individual sample sizes will be small, so that asymptotic results will not necessarily be valid. We suggest below that they might even be misleading in some situations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 24, 1991
- Accession Number
- ADA239660
Entities
People
- F. J. O'reilly
- Michael A. Stephens
Organizations
- Stanford University