Department of Public Health Sciences

BMTRY711

BMTRY 711 : Analysis of Categorical Data

DESCRIPTION

This course offers a short review of standard chi-square methods, followed by several special-purpose two-dimensional techniques. Other areas covered include log-linear models, analysis of threedimensional and higher tables, model selection strategies, maximum likelihood and weighted least squares methodologies, logit models, diagnostics, GEE, patterned tables, and treatment of zero cells.

TOPICS:

  • Sampling distributions
  • One-way classification
  • Effects of small expected values
  • Symmetry, Linear trend models
  • Goodness of fit tests
  • 2x2 tables with various sampling schemes
  • Fisher's exact, mid-P tests. Delta method
  • Several 2x2 tables: CMH
  • Measures of association
  • General IxJ contingency tables
  • IxJ tables with various sampling schemes
  • Log-linear models: 2 and 3 way
  • Log-linear models: 3 and multi-way
  • Log-linear models: 3 and multi-way
  • Partial and conditional independence
  • Selection strategies
  • Linear logistic regression models
  • Logistic regression models (contd)
  • Diagnostics
  • Overdispersion
  • Logit models (log-linear equivalence)
  • Logit models and Poisson regression
  • Association models: R,C,R&C, RC,U,O
  • Ordinal and nominal square tables
  • Multinomial logit models (PO,AC,CL,CR)

PREREQUISITES AND OFFERINGS

BMTRY 700, 701, 706. Typically offered every other spring.

 
 
 

© 2012  Medical University of South Carolina | Disclaimer