Percentage Points and Power Calculations for Outlier Tests in a Regression Situation.
Abstract
The usefulness of an extra sum of squares statistic Q sub K for detecting K outliers has been discussed in previous papers in te context of two-way tables. That work is extended here to straight line regression situations arising from and motivated by a specific set of research data. Percentage points for the appropriate test statistics are obtained by simulation, approximations for these percentage points are suggested, and power calculations are made for various designs and outlier situations. Correct determination of K and position(s) of the outlier(s) appear to be important in influencing power.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1982
- Accession Number
- ADA124369
Entities
People
- Camil Fuchs
- Norman Richard Draper
Organizations
- University of Wisconsin–Madison