Measurement Models (Factor Analysis and Item Response Models)
Key Facts
Module Code |
POLI4122 |
Module Convenor |
Cees van der Eijk |
Teaching Pattern |
Intensive Block
Two full days of computer-room-based instruction which contain both lecture-like elements and hands-on practical training. |
Semester Taught |
Spring Semester (usually in February)
|
Method of Assessment |
Written assignment |
Pre Requisites |
NURS4048 Fundamentals of Quantitative Analysis, or equivalent to be approved by the convenor.
Psychology students are strongly advised to have POLI4130 Intermediate Quantitative Analysis in addition, in order that they can be adequately prepared with the software required for this module.
|
Module Administrator |
Rosemary McCabe |
Teaching on this module uses STATA software.
This module focuses on the use of methods commonly used to assess whether a set of observed variables can be assumed to measure the same underlying phenomenon (often referred to as a latent factor, trait, or dimension). If so, the information from the separate variables can be combined into a composite measure (multiple item measurement), which yields important benefits for conceptualisation and further analysis.
Common instances in which these questions arise can be found in education and psychology (where multiple items are used to construct tests of achievement, proficiency and various kinds of psychological traits); in survey research in sociology, political science and economics (where, e.g., a set of agree/disagree statements or Likert items are used to determine people’s level of attitudes or orientations); and in macro-economics and social-indicator research (where a large number of conceptually and empirically related measures is available for macro-units such as regions or countries).
The module focuses in particular on two broad approaches that are very frequently used in the social and behavioural sciences: factor analysis (including exploratory and confirmatory factor analysis) and item-response scaling (including Mokken and Rasch scaling). The module covers both kinds of measurement models, practical considerations in actual applications, empirical examples from different disciplines, and hands-on training.
Please note, this is an interdisciplinary module that is open to students from across the social sciences. If you have any doubts about whether the course is suitable for your needs or level of study, please contact the module convenor before registering.
Some kinds of measurement modelling (particularly CFA) are strongly linked to the tradition of structural equation modelling (SEM), which is the topic of a separate module (POLI4123 Structural Equation Modelling also taught by Prof van der Eijk and usually takes place in March). Although there is no requirement to do so, there is therefore a distinct benefit in taking both modules.