Module 5: Further Regression Topics

Prerequisites

This module is concerned with greater knowledge of regression, through extension of the simple linear model; enabling students to assess the models they use, testing for problems such as collinearity, outliers/leverage, and heteroskdasticity.

Mphil Students from participating departments taking the Social Science Research Methods Course as part of their research degree

Students expected to be familiar with basic logic of statistical reasoning and linear regression up to the JSSS standard.

Number of sessions: 4

• Session 1: Recap on simple & multiple regression
• Session 2: The importance of diagnostics I Checking model assumptions
• Session 3: The importance of diagnostics II Outliers, influential cases and residuals
• Session 4: Non-linearity & interaction effects

• The objective is to enable students to access the models they use
• To test for problems ie collinearity; outliers/leverage and heteroskedasticity

• To understand non-linear effects; variable transformations
• to think about how statistical models are built
• to introduce interaction effects

Presentations, demonstrations and practicals

SPSS v. 16 on PWF Windows

Three exercises

• Field, Andy (2009), Discovering Statistics using SPSS. London:Sage
• Jaccard, J.and Turrisi, R. (2003) Interaction Effects in Multiple Regression (2nd ed.) London:Sage
• Student also expected to read Linear Regression chapter on EST: http://www.statsoft.com/textbook/sstathome/html

• To gain the maximum benefits from the course it is important that students do not see this course in isolation from the other MPhil courses or research training they are taking. Responsibility lies with each student to consider the potential for their own research using methods common in fields of the social sciences that may seem remote. Ideally this task will be facilitated by

Four sessions of two hours

Four times in Michaelmas term