Course info
GLBH0043: Quantitative Analysis of Observational Data: Theory, Design and Execution (19/20)
This module will provide you with the tools to conduct sound data analysis
based on careful combination of theory and statistical methods to arrive at
supportable conclusions. The course will first cover finding and adapting
an appropriate existing conceptual framework based on theories of health
production, and generating from this a meaningful, falsifiable research
hypothesis. It will then explain core concepts in building directed
acyclical graphs (DAGs) to represent the causal processes reflected in
research hypotheses, and converting these DAGs into a statistical analysis
plan. Finally, it will cover practical statistical analysis to test
hypotheses using both standard regression and marginal structural model
approaches. The core of the module will be student practical implementation
of methods presented through lectures, with examples drawn from a wide
range of topics but with a focus on infectious diseases. After taking this
module you should be able to: Describe key theories of how health and
illness are patterned across populations.Outline the principles behind, and
key features of, directed acyclical graphs (DAGs).Construct a causal DAG
that reflects assumptions about how factors relate to one-another,
including confounding and selection bias.Explain the difference between
standardization and stratification for managing confounding and bias.Use an
appropriate theory and conceptual framework to build a DAG for a testable
hypothesis about health outcomes.Build an approach to address confounding
or bias in design or analysis of data.Discuss how theories and DAGs for
infectious disease may differ from, and be similar to, those for
non-communicable health outcomes.Implement an analysis using marginal
structural models in a statistical package.Compare and contrast results
from regression and marginal structural models.Discuss the strengths and
weaknesses of various quasi-experimental methods for finding causal
effects. The module is open to students on the MSc/PG Dip Global Health and
Development, other UCL MSc/PG Dip students, TropEd students, Taster course
students and Short course students. Prerequisites for this module include
experience working in Stata and an understanding of different study designs
and linear/logistic regression methods. Some understanding of theories of
health would be helpful. You will have a mix of teaching, supervised
computer lab sessions and self-directed reading and learning. Interactive
lectures will include group discussion based on lectures and readings, and
a mid-module class presentation of final project ideas for peer review. The
module will also prioritize time for private reading of materials and group
and self-study for the final project – while the assessment will be
individual the implementation of standard methods will provide opportunity
for group-work. Moodle will be used to give students access to reading
materials. The module will be assessed by a 2000-word individual essay
(100%) conducting a quantitative analysis of a research question justified
with reference to the literature. This will be based on a dataset provided
to you. There will be a brief formative classroom presentation midway
through the course to provide feedback on your planned topic.
Course contacts
Tutor
KB
Course Administrator
GH