ESTIMATION AND TESTING PROCEDURES FOR CIRCULAR NORMALLY DISTRIBUTED DATA.
Abstract
The report examines some aspects of circular normal distributions. The analysis of such data requires methods different from those used in studying conventional linear distributions, inasmuch as choice of origin is significant. In the circular normal (Mises) distribution commonly used estimators are the vector mean and the vector strength, more fully defined in the text. Tests of preferred orientation are conducted by the Rayleigh R-test or by the Greenwood-Durand U-test. Methods for placing confidence intervals about the orientation, and testing hypotheses of specific orientations, are based on the distribution of R given U, analogous to student's t-distribution. These methods or various alternatives are applicable to data which are bimodal, concentrated over a small range, or determinable only modulo 2pi. An appendix lists a computer program for conducting various tests and estimating parameters for circular normally distributed data. Rejection criteria for the various tests are also included. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 01, 1967
- Accession Number
- AD0648082
Entities
People
- Thomas A. Jones
Organizations
- Northwestern University