Accession Number:

ADP007136

Title:

Model Checking for Logistic Regression: A Conditional Approach,

Descriptive Note:

Corporate Author:

NEW MEXICO UNIV ALBUQUERQUE DEPT OF MATHEMATICS AND STATISTICS

Personal Author(s):

• Bedrick, Edward J.
• Hill, Joe R.

Report Date:

1992-01-01

Pagination or Media Count:

7.0

Abstract:

We review the use of exact methods for checking logistic regression models. We focus on global model checks, outlier detection, and goodness-of-link checks. We discuss approximations to exact conditional methods whenever available. We also contrast exact conditional methods and standard unconditional methods based on asymptotic approximations. The techniques are applied to two examples.

Descriptors:

• *LOGISTICS
• *STATISTICS
• CONTRAST
• DETECTION
• GLOBAL
• MODELS
• STANDARDS

Subject Categories:

• Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE