Use of the Fisher Exact Test for the Design of Observer Tests
Abstract
This paper presents techniques to aid in the design of an observer test. To select an appropriate number of observation opportunities, the test designer can use the Fisher Exact Test to calculate the number of observation opportunities required so that a given experimental difference in probability of detection will be statistically significant. Alternatively, the designer can select the number of observation opportunities to guard against rejecting a real difference in probability of detection. These criteria require calculating the probabilities of so-called Type I and Type II errors in hypothesis testing.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2001
- Accession Number
- ADA458136
Entities
People
- John G. Bennett
Organizations
- Tank-automotive and Armaments Command