Bootstrap procedures for linear, robust, and nonlinear regression: application to pharmacokinetics

Thesis/Dissertation · Wed Jan 01 04:00:00 EST 1986

OSTI ID:7106648

Alkuzweny, B M

The bootstrap is a computer-based resampling procedure for estimating the correct variance of an estimator directly from the data obtained rather than from assumptions on the underlying error distribution. The primary objective of this research is to study the small sample behavior of the bootstrap procedure for estimating the variance-covariance matrices of estimators for the biexponential (nonlinear) regression model, and for modified M-estimator using Huber's function. A second objective is to study the bias associated with the bootstrap and to consider several alternative procedures for correcting this bias. This is accomplished via an extensive Monte Carlo simulation study in the linear-regression context. This simulation involves a range of underlying error distributions, a variety of structures for the design matrix,and a range of sample sizes. Three new correlations for the bias in estimation of the variance are considered, and a significant contribution of this research is that one of these is demonstrated to be an improvement over the usual Bickel and Freedman's correction. The remaining two are demonstrated to be less desirable and are based on an inner/outer loop bootstrap procedure. The simulations conducted for the nonlinear model and the M-estimator include the improved bias correction as well as Bickel and Freedman's correction.

• Research Organization: Wyoming Univ., Laramie (USA)
• OSTI ID: 7106648
• Country of Publication: United States
• Language: English

Related Subjects

550600 -- Medicine
62 RADIOLOGY AND NUCLEAR MEDICINE
99 GENERAL AND MISCELLANEOUS
990230* -- Mathematics & Mathematical Models-- (1987-1989)
BOOTSTRAP MODEL
COMPOSITE MODELS
DATA COVARIANCES
MATHEMATICAL MODELS
MATHEMATICS
MATRICES
MONTE CARLO METHOD
PARTICLE MODELS
PHARMACOLOGY
REGRESSION ANALYSIS
SIMULATION
STATISTICS
VARIATIONS