SOME TEST FOR COMPARING PERCENTAGE POINTS OF TWO ARBITRARY CONTINUOUS POPULATIONS
Abstract
Samples are considered from two continuous populations, the first with unique 100 alpha % point theta sub alpha, the second with unique 100 beta % point phi sub beta. The two populations are not necessarily the same or even related. This paper presents some easily applied significance tests for theta sub alpha - phi sub beta which are approximately valid for moderate and large sized samples. The exact significance level of a test is not known but its value is determined within reasonably close limits. Efficiency properties of these tests are investigated for the special case of normal populations with known ratio of variances. The tests are found to be reasonably efficient if alpha and beta are not too large or too small. Since these tests are often valid for moderate as well as large sized samples, they may be of practical value.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 02, 1950
- Accession Number
- AD0603875
Entities
People
- A. W. Marshall
- John E. Walsh
Organizations
- RAND Corporation