|
Activity Number:
|
41
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, August 6, 2006 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #305138 |
|
Title:
|
Double-Smoothing Local Linear Estimation in Partial Linear Models with Application to Environmental Health Data
|
|
Author(s):
|
Li-Shan Huang*+ and Christopher Cox
|
|
Companies:
|
University of Rochester and Johns Hopkins University
|
|
Address:
|
601 Elmwood Ave., Box 630, Rochester, NY, 14642,
|
|
Keywords:
|
nonparametric regression ; partial linear models ; local linear regression ; hypothesis testing ; environmental health
|
|
Abstract:
|
New estimators and F-type hypothesis testing procedures are proposed for partial linear models. The new estimator for the nonparametric component is based on the local linear estimator with "double smoothing." The proposed F-tests formally evaluate significance of the nonparametric component by testing a no-effect null hypothesis and whether the nonparametric function can be simplified to a linear relationship. The Seychelles Child Development Study is an ongoing longitudinal study of child development after prenatal exposure to methylmercury through maternal fish consumption. A possible nonlinear effect of methylmercury exposure was seen in Huang et al. (2005), but its significance was not confirmed. We apply partial linear models to the Seychelles data with the proposed new estimators and formally examine the significance of the nonlinear exposure effect.
|
