Module Database

Information for module PSYC3301

This module is available for 2017/18

If you're a member of UCL you can add this module to your personalised course list

This information is for guidance only. If you are an UCL undergraduate interested in studying one of these courses, you must seek permission from both the providing department and your 'home' department. Appearance in this database is not a guarantee that a course is running in any particular academic year.

Module code:PSYC3301 (Add to my personalised list)
Title:Advanced Multivariate Statistical Methods in Psychology
Credit value:.5
Division:Division of Psychology and Language Sciences
Module organiser (provisional):Prof Chris McManus
Organiser's location:Division of Psychology and Language Sciences
Available for students in Year(s):3
Module prerequisites:PSYC2204 The Design and Analysis of Psychological Experiments. Students without Maths at A-level or equivalent should consider carefully the difficulty level of this course, which uses matrix algebra quite extensively. “Non-Psychology students (including non-Psychology affiliates) must visit the link below to establish whether they can take this module. Students should have an appropriate background in the subject area and must complete the ‘Module Request Form’, failure to do so will result in rejection on Portico.”  
Module outline:Most psychological datasets are inherently multivariate, and proper analysis requires that the subtleties of the interrelationships between multiple measures are taken into account. The advent of cheap computing power and sophisticated computer packages in the past couple of decades has transformed psychological statistics, and this module introduces a range of techniques which once were only for specialists and now are increasingly expected of all psychologists. The first half of the module concentrates on multiple regression, and the problems that can arise in what is effectively a paradigmatic case for all multivariate analysis, and the second half extends the analysis into properly multivariate techniques such as factor analysis, MANOVA, canonical correlation and path analysis. The examples classes are an integral part of the course, not only providing practical experience, but also supporting the lecture material. 
Module aims:To understand a range of modern techniques used in the statistical analysis of multivariate data in psychology. 
Module objectives:By the end of the module you should: • Understand the principles of basic matrix algebra, including the concepts of an eigenvalue and a determinant, be able to manipulate simple matrix equations, recognise the applications of matrix algebra to multivariate statistics, and be able to carry out matrix calculations using Matlab. • Understand the basics of multiple regression, including its matrix formulation, the problems of multicollinearity, suppressor variables and missing values, the use of dummy variables, polynomial terms, interaction terms, the relationship to ANOVA, and the differences between forward and backwards stepwise models, and hierarchical models. • Be able to use R to carry out multiple regression analyses, and be able to carry out a Monte Carlo analysis using R to generate random variables of known distributions. • Understand the basics of exploratory factor analysis, including its matrix representation, the use of principle component analysis for data reduction, the advantages and disadvantages of the eigenvalue>l and scree-slope criteria for the numbers of factors, and the nature of rotation, including Varimax and oblique rotations. • Understand the difference between confirmatory and exploratory factor analysis and the advantages of each, the application of path analysis to understanding simple problems of test-retest correlations, etc., and the relationship of path analysis to matrix representations. • Understand how factor scores can be used for reducing a complex set of dependent variables to orthogonal measures which can be used as dependent variables in regression or ANOVA, and how multiple regression can be generalised to multiple dependent variables as canonical correlation and similarly ANOVA can be generalised as MANOVA. • Understand the basic form of structural equation modelling in the specific form of the LISREL model, and be able to describe the purposes of the various matrices in the formulation. • Be able to use SPSS for Windows to carry out multiple regression, factor analysis, canonical correlation and MANOVA. Be able to interpret the various forms of output and to recognise the common options available for the various programs. • Be able to use LISREL for Windows to carry out a simple confirmatory factor analysis, and recognise the various parts of the output from the program.  
Key skills provided by module: 
Module timetable: 
Module assessment:In-course assessment (3,000 words plus appendices) 50.00%
Unseen two-hour written examination 50.00% 
Notes:Optional Course for BSc Psychology 3rd years. Students without Maths at A-level or equivalent should consider carefully the difficulty level of this course, which uses matrix algebra quite extensively. 
Taking this module as an option?:Not without pre-requisites and approval, see above 
Link to virtual learning environment (registered students only)
Last updated:2017-07-14 19:49:57 by ucjtscm