Approximations of the Normal-Logistic Convolution Integral,
Abstract
The normal-logistic convolution arises in several statistical applications, including logistic regression models and multinomial logit models. We begin by characterizing the logistic distribution as a scale mixture of normals. We then construct least maximum approximations of the logistic distribution function using finite discrete mixtures of normal df's using the Remes algorithm. The convolution integral follows by convolving this approximation with the normal.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1992
- Accession Number
- ADP007186
Entities
People
- John F. Monahan
- Leonard A. Stefanski
Organizations
- North Carolina State University