Exact logistic regression for a matched pairs case-control design with polytomous exposure variables

Research output: Contribution to journalArticle

Abstract

Logistic regression methods are useful in estimating odds ratios under matched pairs case-control designs when the exposure variable of interest is binary or polytomous in nature. Analysis is typically performed using large sample approximation techniques. When conducting the analysis with polytomous exposure variable, situations where the numbers of discordant pairs in the resulting cells are small or the data structure is sparse can be encountered. In such situations, the asymptotic method of analysis is questionable, thus an exact method of analysis may be more suitable. A method is presented that performs exact inference in the case of pair-wise matched case-control data with more than two unordered exposure categories using a distribution of conditional sufficient statistics of logistic model parameters.

Original languageEnglish
Pages (from-to)450-456
Number of pages7
JournalJournal of Modern Applied Statistical Methods
Volume11
Issue number2
Publication statusPublished - 2012

Fingerprint

Matched pairs
Case-control
Logistic Regression
Control Design
Case-control Data
Exact Inference
Sufficient Statistics
Unordered
Odds Ratio
Logistic Model
Exact Method
Asymptotic Methods
Data Structures
Binary
Logistic regression
Cell
Approximation

Keywords

  • Conditional logistic regression
  • Diophontine systems
  • Exact analysis
  • Sufficient statistic

ASJC Scopus subject areas

  • Statistics, Probability and Uncertainty
  • Statistics and Probability

Cite this

@article{6750dc2686c64731afaf802eba38c6a8,
title = "Exact logistic regression for a matched pairs case-control design with polytomous exposure variables",
abstract = "Logistic regression methods are useful in estimating odds ratios under matched pairs case-control designs when the exposure variable of interest is binary or polytomous in nature. Analysis is typically performed using large sample approximation techniques. When conducting the analysis with polytomous exposure variable, situations where the numbers of discordant pairs in the resulting cells are small or the data structure is sparse can be encountered. In such situations, the asymptotic method of analysis is questionable, thus an exact method of analysis may be more suitable. A method is presented that performs exact inference in the case of pair-wise matched case-control data with more than two unordered exposure categories using a distribution of conditional sufficient statistics of logistic model parameters.",
keywords = "Conditional logistic regression, Diophontine systems, Exact analysis, Sufficient statistic",
author = "Ganguly, {Shyam S.}",
year = "2012",
language = "English",
volume = "11",
pages = "450--456",
journal = "Journal of Modern Applied Statistical Methods",
issn = "1538-9472",
publisher = "Wayne State University",
number = "2",

}

TY - JOUR

T1 - Exact logistic regression for a matched pairs case-control design with polytomous exposure variables

AU - Ganguly, Shyam S.

PY - 2012

Y1 - 2012

N2 - Logistic regression methods are useful in estimating odds ratios under matched pairs case-control designs when the exposure variable of interest is binary or polytomous in nature. Analysis is typically performed using large sample approximation techniques. When conducting the analysis with polytomous exposure variable, situations where the numbers of discordant pairs in the resulting cells are small or the data structure is sparse can be encountered. In such situations, the asymptotic method of analysis is questionable, thus an exact method of analysis may be more suitable. A method is presented that performs exact inference in the case of pair-wise matched case-control data with more than two unordered exposure categories using a distribution of conditional sufficient statistics of logistic model parameters.

AB - Logistic regression methods are useful in estimating odds ratios under matched pairs case-control designs when the exposure variable of interest is binary or polytomous in nature. Analysis is typically performed using large sample approximation techniques. When conducting the analysis with polytomous exposure variable, situations where the numbers of discordant pairs in the resulting cells are small or the data structure is sparse can be encountered. In such situations, the asymptotic method of analysis is questionable, thus an exact method of analysis may be more suitable. A method is presented that performs exact inference in the case of pair-wise matched case-control data with more than two unordered exposure categories using a distribution of conditional sufficient statistics of logistic model parameters.

KW - Conditional logistic regression

KW - Diophontine systems

KW - Exact analysis

KW - Sufficient statistic

UR - http://www.scopus.com/inward/record.url?scp=84875257639&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84875257639&partnerID=8YFLogxK

M3 - Article

VL - 11

SP - 450

EP - 456

JO - Journal of Modern Applied Statistical Methods

JF - Journal of Modern Applied Statistical Methods

SN - 1538-9472

IS - 2

ER -