Abstract

The detection and estimation of outliers in a saturated factorial experimental setting is researched and discussed. Methods for detection and estimation are developed and tested in both simulated and "real" data sets. The techniques developed proved themselves useful in both detection and estimation of outlying values. The simulated data sets also showed how an outlier can affect the interpretation of an experiment depending on the number and sizes of active effects and on the amount the outlying value is actually shifted.

Library of Congress Subject Headings

Outliers (Statistics); Factorial experiment designs

Publication Date

1-1992

Document Type

Thesis

Student Type

Graduate

Advisor

Joseph Voelkel

Advisor/Committee Member

Daniel Lawrence

Advisor/Committee Member

Hugh Wood

Comments

Physical copy available from RIT's Wallace Library at QA276 .C73 1992

Campus

RIT – Main Campus

Download

Share

COinS