Psychology 6140: Multivariate Data Analysis

Instructor: Michael Friendly (my home page)
- Email: friendly AT yorku DOT ca
- Office: 226 BSB
- Phone: x66249
- Office hour: Tuesday: 11:30-12:30 (other times by appt.)
Class meetings: Thursday, 2:30 am - 5:30 pm, 203 BSB, lab session: 1:30-2:30, Hebb lab, 159 BSB

Course Outline and Class Schedule [PDF] or [HTML]

Psychology 6140 is designed to provide an integrated, in depth, but applied approach to multivariate data analysis and linear statistical models in behavioural science research. There is a strong emphasis on using graphical methods to understand your data. The statistical topics covered will include:

Regression analysis
Univariate and multivariate ANOVA and ANCOVA
Discriminant analysis
Logistic regression
Canonical correlation analysis
Principal components, factor analysis, LISREL models (CFA, SEMs)
Cluster analysis and Multidimensional Scaling (time & interest permitting)

See the more detailed Course Outline for further information, and the class schedule for lecture notes and tutorials. See the Hebb Lab website for information on access to the Hebb Psychology Computer Lab.

Course feature: MVStatsNews Occasional presentations by members of Psychology 6140 on topics of interest, typically from popular news and blogs.

Resources

Statistical software

You can use any statistical software for the course, but in lectures and lab sessions I will be using R software nearly exclusively, together with the R Studio user interface for R. You are well-advised to download and install these to your computer so you can follow along. This web site also contains many legacy examples using SAS.

R Project for Statistical Computing. R is a free software environment for statistical computing and graphics (Windows, MacOS, Linux).
- You can download it from any CRAN mirror site.
- R Studio is a powerful front-end for R, much more convenient than the standard R GUI. Among other features, it offers integrated tools for help, plotting, history and the ability to run an R script to obtain a HTML, PDF or Word document containing your input and output. Using R will be much easier if you install the R Studio user interface for R.
- For this course, use this R script to install useful add-on packages, the same ones as used in the Hebb lab.
- The matlib package provides a number of useful functions for learning matrix algebra for multivariate statistics.
If you insist on SAS or SPSS, see York group licenses for SAS and SPSS for purchasing a copy for your own computer.
- You can also use the Web FAS service to access SAS, SPSS, R, etc. from home or on your laptop. It gives you access to the files on your AcadLabs home drive (F:/).
- See Connecting from Home - WebFAS.

Computing Guides

R software
- An Introduction to R (Official introductory guide: 100 pages)
- R in One Page (well, 2); also: The R Reference Card (4 pages)
- Cheatsheets: R Studio provides a nice collection of two-page RStudio Cheat Sheets
- Quick R for SAS, SPSS, Stata Users Great site for conversions to R!
- R Cookbook, a collection of recipes for analyzing data from psychology, with a detailed section on Graphs with ggplot2. experiments.
- The R matlib package will now be used in the course:
  - The stable version on CRAN has compiled binaries for all systems.
  - The Development version on GitHub is always most current.
  - You can view the vignettes in the package using browseVignettes("matlib").
SAS OnlineDoc, V9 The SAS Online Documentation
- SAS OnlineDoc, V9.3
Local SAS Information Guides including (among others):
- Introduction to SAS for Windows,
- SAS V8 for Windows Tutorial (from University of Texas)
- Introduction to SAS ,
- overview of SAS Procedures and Programming Statements,
- Exploring the Baseball data with SAS/INSIGHT
- Introduction to SAS/IML -- the Interactive Matrix Language.

Methods Notes, and Bibliographies

Matrix algebra
- Matrix Algebra Reference Manual (by Mike Brooks, Imperial College)
- Mathematical preparation (matrix algebra) for the course.
Annotated list of multivariate texts
Regression and linear models
- Regression Analysis bibliography
- Mediator vs. Moderator variables
Factor analysis and Structural equation modeling
- Factor Analysis Outline
- Annotated bibliography on Factor Analysis and Structural Equation Modeling
- notes on Planning a Factor Analytic Study
- Ed Rigdon's Structural Equation Models FAQ
- Jason Newcomb's Structural Equation Modeling Reference List (journal articles and chapters on structural equation models), more up-to-date than my Annotated bib.
- A discussion of Statistical Power in Structural Equation Models by David Kaplan
MDS
- a Multidimensional Scaling Outline
- a Multidimensional Scaling bibliography
Mixed effects models

Resources for Multilevel Modeling

Assignments

The items below give readings and problems for course assignments.
Problems denoted "MAP" come from: Matrix Algebra Problems with Applications in Psychology and Multivariate Analysis [PDF version]

Graded Projects

Matrix algebra (take-home)
Research critique:
- Silver Blaze Problems in Regression
- Bibliography of research studies
Data analysis project
Final data analysis project

SAS Examples

There is a large library of SAS examples, grouped into the following categories. Most now have the printed and graphical output on corresponding web pages. All of these, together with SAS data sets and macros for the course, are available in a single ZIP file, psy6140.zip.

Data Screening