SIMULTANEOUS CONFIDENCE INTERVALS FOR DIFFERENCES OF CLASSIFICATION PROBABILITIES
Abstract
Given a group of individuals with a common characteristic, such as having (or having had) a given disease, there can be an interest in the probabilities a sociated with a specified set of classifications for these individuals. For example, the common characteristic could be lung cancer and the classifications could be cigarette smoker, cigar or pipe smoker, cigarette and cigar or pipe smoker, and nonsmoker. Often, the indivi uals can be considered to represent independent sample values from a multinomial probability distribution that has these classifications as categories. Given two separate groups having a different characteristic ut t e same classifications, there can be an interest in comparing their multinomial probabilities for each of the classifications. One way to make this comparison is to consider the differences of the probabilities for the classifications. Separate confidence intervals for the differences of the two probabilitie are developed for each classification, with these intervals being such that the large sample probability is at least a specified amount that all the relationships are simultaneously satis ied. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 05, 1962
- Accession Number
- AD0288585
Entities
People
- John E. Walsh
Organizations
- System Development Corporation