2. Analysis of Categorical Data: The Odds Ratio as a Measure of Association and Beyond

  1. Ming T. Tsuang4,5,
  2. Mauricio Tohen6,7 and
  3. Peter B. Jones8
  1. Garrett M. Fitzmaurice1,2,3 and
  2. Caitlin Ravichandran1,2

Published Online: 14 APR 2011

DOI: 10.1002/9780470976739.ch2

Textbook of Psychiatric Epidemiology, Third Edition

Textbook of Psychiatric Epidemiology, Third Edition

How to Cite

Fitzmaurice, G. M. and Ravichandran, C. (2011) Analysis of Categorical Data: The Odds Ratio as a Measure of Association and Beyond, in Textbook of Psychiatric Epidemiology, Third Edition (eds M. T. Tsuang, M. Tohen and P. B. Jones), John Wiley & Sons, Ltd, Chichester, UK. doi: 10.1002/9780470976739.ch2

Editor Information

  1. 4

    Center for Behavioral Genomics, Department of Psychiatry, University of California, San Diego, 9500 Gilman Drive, La Jolla CA 92039, USA

  2. 5

    Harvard Institute of Psychiatric Epidemiology & Genetics, Harvard School of Public Health, Boston, USA

  3. 6

    Department of Psychiatry, University of Texas Health Science Centre at San Antonio, USA

  4. 7

    Division of Mood and Anxiety Disorders, University of Texas Health Science Center at San Antonio, 7526 Louis Pasteur Drive, San Antonio TX 78229-3900, USA

  5. 8

    Department of Psychiatry, University of Cambridge, Box 189, Addenbrooke's Hospital, Cambridge CB2 2QQ, UK

Author Information

  1. 1

    Laboratory for Psychiatric Biostatistics, McLean Hospital, 115 Mill St, Belmont MA 02478, USA

  2. 2

    Department of Psychiatry, Harvard Medical School, Boston, MA, USA

  3. 3

    Department of Biostatistics, Harvard School of Public Health, Boston, MA, USA

Publication History

  1. Published Online: 14 APR 2011
  2. Published Print: 15 APR 2011

ISBN Information

Print ISBN: 9780470694671

Online ISBN: 9780470976739

SEARCH

Keywords:

  • categorical data analysis - odds ratio as a measure of association and beyond;
  • statistical methods, for analysis of categorical ‘outcome’ data in psychiatric studies;
  • measures of association for 2 × 2 tables - quantifying departures from independence;
  • inference for a single proportion or probability - data from first-episode major affective disorders with psychosis study;
  • treatment or exposure, two levels - data summarised in a 2 × 2 contingency table;
  • substance use disorders (SUDs) - and elderly suicides;
  • first-episode major affective disorders with psychosis study - Axis I comorbidity and 2-year functional recovery;
  • matched pair study design - number of partial tables (J) is large, sample size small;
  • Cochran–Mantel–Haenszel test statistic for data - adjusted OR, patients with Axis I comorbidity lower odds of recovery, than those without Axis I comorbidity;
  • interactions in logistic regression - test for interaction using methods for contingency tables

Summary

This chapter contains an overview of statistical methods for categorical data in general and binary outcomes in particular, with an emphasis on the odds ratio and its applications in psychiatry. Illustrations from a number of study designs common in psychiatry are approached using contingency tables and the methods that arise from their analysis. Logistic regression is considered along with a number of its extensions, including conditional, multinomial and exact logistic regression and methods for clustered categorical data.