## PSY6003 Advanced statistics: Multivariate analysis II: Manifest variables analyses

### Method of teaching:

The course will consist of 5 2-hour sessions, one per week. Notes summarising the material to be covered each week will be made available over the world wide web before the class. . Sample problems will be included in the notes. Students will be expected to have read these notes and worked on the examples before the session, and to come to the class with questions about them. In addition I shall be available for consultation by email. Interesting questions raised over the email will be reflected back to the entire class.

### Method of assessment:

At the end of the course, a test paper will be distributed. It will include short questions testing students' knowledge of various kinds of analytic technique, together with questions about some sample data sets. Students will be expected to complete this test paper in their own time, using books, notes and other resources. Completed test papers should include explanations of why a particular technique is appropriate for that example, computer output, and interpretations of the results.

### Purpose and aims:

The course aims to introduce students to the more common multivariate techniques that use manifest variables, that is, the observed variables themselves, not inferred underlying latent variables as in, for example, factor analysis. The commonest manifest variables analysis is multiple regression but there are numerous others. The aim of the course is not to teach the underlying mathematics but to put students in a position to (a) choose when to use one of these kinds of analysis and (b) use the appropriate computer statistics package to carry it out.

### Outline

Topics to be covered each week
1. Multiple regression: introduction/revision. R2-adjusted, F, regression coefficients, standardized beta-weights, t-values. Using Minitab and SPSS to carry out multiple regression.

2. Multiple regression continued. Dummy variables techniques. Stepwise and Best Regression procedures. Choosing a regression model. Problems: Outliers, Heteroscedasticity, Multicollinearity, Identification problems.

3. Path analysis. Input and output path diagrams. Direct, indirect, and overall impacts of independent variables on dependent variables.

4. Dichotomous dependent variables. Discriminant analysis. Relation between discriminant analysis, multiple regression and manova. Discriminant analysis with 2 and k groups. Logistic regression. The logistic and logit transformations. Goodness of fit versus accuracy of classification. Logits and probits. SPSS DISCRIMINANT, LOGISTIC REGRESSION and PROBIT ANALYSIS commands.

5. Additional techniques. Analysis of covariance. Ordered logit analysis. Use of LIMDEP for ordered logit analysis.

• Bryman, A., & Cramer, D. (1994), Quantitative data analysis for social scientists. Simple and reader-friendly, based around SPSS. Only covers multiple regression and path analysis.
• Bryman, A., & Cramer, D. (1996), Quantitative data analysis with Minitab. The Minitab version of the previous book. Less useful for this course but worth knowing about.
• Tabachnik, B. G., & Fidell, L. S. (1989), Using multivariate statistics (2nd edn.). More advanced but excellent if you can stay with their mathematical treatment. Covers data screening, multiple regression, and discriminant analysis. Discusses use of several statistics packages, including SPSS but not Minitab.
• Norusis, N. J. (various dates), SPSS Advanced statistics. This manual exists in a variety of editions and versions. Do not confuse it with the User's Guide - the Advanced Statistics manual actually gives statistical explanations. The best source for Logistic Regression.
• Greene, W. H. (1992), LIMDEP Version 6.0 user's manual and reference guide. The only source of help on ordered logit and Limdep generally.
• A FAQ file will be used to collect questions from students that seem to be of general interest

### A note to visitors

Stephen Lea

University of Exeter
Department of Psychology
Washington Singer Laboratories
Exeter EX4 4QG
United Kingdom
Tel +44 1392 264626
Fax +44 1392 264623