Analyzing n Samples of 2 Observations Each
Abstract
When data are scarce, it is common to combine small samples from a number of sources considered to be reasonably similar. When sample sizes are extremely small, testing the assumption of similarity of sources is often only attempted by subjective means. This paper provides a method to add quantitative risk assessments to the study of this assumption, using two observations per sample. In addition to general compatibility testing of the sources using modified versions of Westenberg's Interquartile Range Test, and the Westenberg- Mood Median Test, a new hypothesis test has been developed to aid in identifying whether one (or more) of the sources of data provides a substantially larger or smaller set of values due to its underlying population. Because the probabilistic risk assessments provided here address a situation so commonly found in analyzing military operations as well as in test and evaluation, details are provided to simplify the implementation of this methodology. A major goal here has been that power analyses be described in terms meaningful to the user and the decision maker. The new hypothesis test makes use of simulation- aided power analyses.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1983
- Accession Number
- ADP001574
Entities
People
- G. Petet
- Jim R. Knaub Jr.
- L. M. Grile