Overview

Description

CALENDAR DESCRIPTION:

Focuses on multivariate regression and correlation models. Deals with ways of answering questions when direct experimental manipulation is not feasible, and demonstrates the utility of the principles involved for solving problems other than those for which they were first proposed.. Quantitative.

COURSE DETAILS:

Course Outline Psyc411/911

Instructor: Professor Michael Maraun RCB 4202 maraun@sfu.ca 778-782-5685

Sources:
The readings will be articles and sections of books assigned on an as needed basis (and, when assigned, available for duplication in the copier room)

Tentative  Structure:

 Psyc411/911 will be comprised of two parts:

i. Fundamentals/background/review

 
a. Matrix algebra
b. Multivariate geometry
c. Random vectors/multivariate distributions
d. Expectation and covariance operators
e. Linear function of random vector

ii. Survey of multivariate techniques (provisional)

  a. Relationships within a single set of p variables   
    - Principal Component Analysis (PCA)         
    - Linear Factor Analysis (LFA) ?  

b. Two set association (symmetric)/dependency (asymmetric)            

   bi. Set 1 (1 DV): Set 2 (q IVs)                      
        - [DV quasi-continuous] Multiple linear regression     
        - [DV dichotomous[0,1]] Logistic regression

  bii. Set 1 (p quasi-continuous variables): Set 2 (q quasi-continuous variables).  

   
        - symmetric: Canonical Correlation Analysis (CCA)                        
        - asymmetric: Set 1, DVs; Set 2, IVs: Redundancy Analysis (RA) ?
             

biii. Set 1 (p quasi-continuous DV): Set 2 (1 Nominal IV)       
 - 1-way Multivariate Analysis of Variance (MANOVA)                                    
  w/canonical variate analysis (discriminant analysis)  

Grading:  
4 assignments 60%
   short paper 40%  

A+ 95 and higher A    90-95 A-   85-90 B+   80-85 B     75-80 B-    70-75 C+   65-70 C     60-65 C-    55-60 D     50-55 F      less than 50  

Assignments:
Each of the four assignments will be constituted of one or more sets of questions  handed out during the semester.  Each question set you receive will make reference to an assignment number, and you will hand in, on the relevant due date (see schedule below), all questions making reference to the assignment that is due.   

Short Paper:  
The short paper, no greater than ten pages in length, will be due at the end of the semester (final day of class).  In it, you will provide a technical description of a multivariate problem indigenous to your research (the description should be at the population level, of course), and its inferential solution (this, I anticipate, will be via employment of some particular multivariate procedure or procedures, the application of which to the multivariate problem in question, you will describe in a careful technical fashion).  

Microlab:  For the purposes of demonstrating analyses, and practicing some applied aspects of multivariate analysis, we will be spending some time in the microlab.  There are 14 machines in the microlab, and something around 20 students in 411/911.  That is to say, there will have to be some sharing, during our sessions, and it will be, perhaps, a touch cramped. 

Tentative Schedule  

Jan. 
1 3                                                                            
2 10                                                                  
3 17                                                                              
4 24     Assignment 1 due                                                               
5 31                 

Feb.     6 7                                                        
14        No class(reading break) Assignment 2 due                                                                               
7 21              
8 28     Assignment 3 due     

Mar.    9 7                    
10 14                            
11 21                                                                
12 28   Assignment 4 due                 
13 4     Short Paper Due

Grading