Inference for Predictor Comparisons: Dominance Analysis and the Distribution of R(2) Differences
Azen, Razia
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/82312
Description
Title
Inference for Predictor Comparisons: Dominance Analysis and the Distribution of R(2) Differences
Author(s)
Azen, Razia
Issue Date
2000
Doctoral Committee Chair(s)
David Budescu
Department of Study
Psychology
Discipline
Psychology
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Statistics
Language
eng
Abstract
Dominance analysis is a procedure that determines, for each pair of predictors in the general linear (for example, multiple regression) model, the relative importance based on differences between the R2 values of relevant subset models. Depending on the subset models compared, the dominance measure is defined as either complete, average or global dominance. It is shown that predictor comparisons based on these measures will be unaffected by the use of other model fit criteria that rely on the error sum of squares (for example, adjusted-R2, C p or AIC). Four methods are proposed to construct confidence intervals for the dominance measures. Three of these methods rely on examining the distribution of R2 differences by using (1) the asymptotic estimate of its covariance matrix, (2) the bootstrap estimate of its covariance matrix, or (3) the full empirical distribution (obtained using the bootstrap). The fourth inference method relies on examining the distribution of the probability that one predictor dominates another. These methods are compared to each other in simulations applied to data generated using different multivariate distributions (namely, the normal and lognormal) as well as different sample sizes. The methods are also applied to real data samples. The fourth method is shown to be overly sensitive to the detection of dominance. Extensions of these procedures to the multivariate multiple regression model are proposed.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.