University of St Andrews

Key information

SCOTCAT credits : 15

ECTS credits : 7

Level : SCQF level 11

Semester: 1

Availability restrictions: Not automatically available to General Degree students

Planned timetable: Lecture: 3 – 4pm Mon, Tues, Thurs, Fri + practical Tues 4-5 + tutorial (one of 2-3 or 4-5 Thurs)

This applied statistics module covers the main aspects of linear models (LMs) and generalized linear models (GLMs). In each case the course describes model specification, various options for model selection, model assessment and tools for diagnosing model faults. Common modelling issues such as collinearity and residual correlation are also addressed, and as a consequence of the latter the Generalized Least squares (GLS) method is described. The GLM component has emphasis on models for count data and presence/absence data while GLMs for multinomial (sometimes called choice-based models) are also covered for nominal and ordinal response outcomes. The largest part of the course material is taught inside an environmental impact assessment case study with reality-based research objectives. Political and medical examples are used to illustrate the multinomial models.

Relationship to other modules

Pre-requisite(s): Undergraduates must have passed at least one of MT4113, MT4527, MT4528, MT4530, MT4531, MT4537, MT4539, MT4606, MT4608 MT4609, MT4614

Anti-requisite(s): You cannot take this module if you take MT4607 or take MT5753

Learning and teaching methods and delivery

Weekly contact: 4 lectures & 1 practical & 1 tutorial (x 5 weeks)

Scheduled learning hours: 30

Guided independent study hours: 117

Assessment pattern

As used by St Andrews: 2-hour Written Examination = 50%, Coursework = 50%

As defined by QAA
Written examinations : 50%
Practical examinations : 0%
Coursework: 50%

Re-assessment: Oral examination = 100%

Personnel

Module coordinator: Dr H Worthington
Module teaching staff: Dr H Worthington and Dr B Baer
Module coordinator email hw233@st-andrews.ac.uk

Intended learning outcomes

• Understand Generalised Least Squares (GLS) models and how they relate to Linear regression Models (LMs), and be able to apply GLS models appropriately
• Understand how Generalised Linear Models (GLMs) extend LMs and GLS models, recognise problems whose solution requires GLMs, and be able to apply GLMs appropriately to address such problems
• Learn to apply GLMs for continuous response data, count data, binary data, ordinal data and categorical data
• Be able to perform objective model selection and conduct diagnostics for GLS models and GLMs, to check models' adequacy for the data at hand
• Be able to interpret fitted GLS model and GLMs output to draw appropriate conclusions about the problem being addressed using these models
• Become competent in using the statistical programming language R to solve problems using GLS models and GLMs