|
Activity Number:
|
315
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #305591 |
|
Title:
|
A Fresh Look at the Discriminant Function Approach for Estimating Crude or Adjusted Odds Ratios
|
|
Author(s):
|
Robert H. Lyles*+ and Ying Guo and Andrew N. Hill
|
|
Companies:
|
Emory University and Emory University and CDC
|
|
Address:
|
Dept. of Biostatistics and Bioinformatics, Atlanta, GA, 30322,
|
|
Keywords:
|
Bias ; Efficiency ; Logistic regression ; Minimum variance
|
|
Abstract:
|
Assuming a binary outcome, logistic regression is the most common approach to estimating a crude or adjusted odds ratio corresponding to a continuous predictor. We revisit a method termed the discriminant function approach (DFA), which yields closed-form estimators and standard errors. In practice, we show that this method suggests a multiple linear regression of the continuous predictor of interest on the outcome and other covariates, in place of standard logistic regression. If typical diagnostics support the assumptions behind this linear regression model, one can tap into demonstrable advantages over the usual logistic regression MLE. These include improvements in terms of bias and efficiency based on a novel minimum variance unbiased adjusted log odds ratio estimator. This version of the DFA requires less stringent assumptions than those for which it was historically criticized.
|
