|
Activity Number:
|
437
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #303494 |
|
Title:
|
Sensitivity Analysis for Covariate Misclassification in Logistic Regression via Predictive Value Weights
|
|
Author(s):
|
Robert H. Lyles and Ji Lin*+
|
|
Companies:
|
Emory University and Emory University
|
|
Address:
|
Department of Biostatistics and Bioinformatics, Atlanta, GA, 30322,
|
|
Keywords:
|
Bias ; Errors-in-variables ; Odds ratio ; Regression
|
|
Abstract:
|
Potential bias due to misclassification of covariates in regression is well appreciated by epidemiological researchers, who sometimes seek to gauge the sensitivity of an estimated effect to variations in the assumed values of misclassification probabilities. We present an intuitive and flexible approach to such a sensitivity analysis that is readily implemented using standard software. Observed data on the outcome, error-prone binary covariate(s) of interest, and other covariates measured without error are combined with investigator-supplied values for sensitivity and specificity parameters to produce corresponding positive and negative predictive values. These provide estimated weights in fitting the model of interest to an properly defined expanded data set. Bootstrapping provides a convenient tool for incorporating uncertainty in the estimated weights into standard error calculations.
|
