This module provides the theoretical foundations and applied advice for models for multivariate data - models where there are multiple dependent variables. There are two broad classes of models covered here.

• Multivariate statistical routines - those that model the structure in observed dependent variables with observed independent variables

• Models of the structure in observed dependent variables with unobserved (i.e., latent) independent variables.

The outline for the course is below.

1 Summated Rating Model, Principal Components, Factor Analysis

• Slides pdf
• Code r
• In-class Exercises Code r
• Data zip
• Lab Exercises pdf

2 Spatial Theories, Analyzing Issue Scales

• Slides pdf
• Code r
• In-class Exercises Code r
• Data zip

3 Multidimensional Scaling (similarities data)

• Slides pdf
• Code r
• In-class Exercises Code r
• Data zip
• Comparison of Scaling Outcomes pdf
• BSMDS Source Code tar.gz
• To use the bsmds package, you’ll have to install it from source. This means - download the package, reset R’s working directory to the location of the file you downloaded, then do:

  install.packages("bsmds_0.1-2.tar.gz", type="source", repos=NULL)

4 Unfolding of Ratings Scale Data

• Slides pdf
• Code r
• In-class Exercises Code r
• Denmark Data .rda

5 Parametric Methods for Binary Data

• Slides pdf
• Code r
• In-class Exercises Code r
• EP4 .rda

6 Non-parametric Methods for Binary Data

• Slides pdf
• Code r
• In-class Exercises Code r
• WVS Data .rda
• European Court for Human Rights Data .rda
• 2004 Feeling Thermometers Data .rda

7 Bayesian extensions of scaling models I

• Slides pdf
• Code r

8 Bayesian extensions of scaling models II

• Slides pdf
• Code r

9 Using latent variable estimates.

• Slides pdf
• Code r
• Democracy data .rda
• MCMCpack dynamic IRT results .rda

Code Highlighting by Pretty R at inside-R.org