Modelling and generating correlated binary variables

Samuel D. Oman*, David M. Zucker

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

49 Scopus citations

Abstract

Many applications use simple parametric models for the correlation structure of binary responses which are observed in clusters. The usual approach, to use correlation models appropriate for normally distributed responses, suffers from two drawbacks when the marginal probabilities within the clusters differ. First, as it does not explicitly take into account constraints on the second moments which must be satisfied for binary responses, it may fail to model realistically the range of correlations present in the data. Secondly, computer simulation of observations from these models is very difficult. We present an alternative class of correlation models which reflect the binary nature of the responses and allow for simple simulation.

Original languageAmerican English
Pages (from-to)287-290
Number of pages4
JournalBiometrika
Volume88
Issue number1
DOIs
StatePublished - 2001

Keywords

  • Binary variable
  • Computer simulation
  • Correlation structure
  • Generalised estimating equation

Fingerprint

Dive into the research topics of 'Modelling and generating correlated binary variables'. Together they form a unique fingerprint.

Cite this