Tip:
Highlight text to annotate it
X
Hello. I'm Mike Berbaum. I'm the instructor for the longitudinal analysis course.
I come from the University of Illinois at Chicago, where I head a small biostatistical consulting unit
and advise on biomedical health research, generally.
This course is pitched at the track 2 level, which means it's assuming that
you've had a good solid course in regression analysis,
such as the ones offered by ICPSR in the Summer Program.
But it's not at the highest level, Track 3,
where all of the tools of calculus and matrix algebra and so on can be used ad libidum.
So this is a course where we try to focus on the conceptual understanding of the techniques,
as well as the practical implementation in computer software,
as well as the practical implementation in computer software,
as well as the practical implementation in computer software,
so that by the end of it, you should be able to run many of the standard analyses
that occur in longitudinal analysis with software we've introduced you to.
The course starts with the linear model, and then moves to the general mixed model,
which is a linear model that includes random effects: random intercepts and random slopes.
From there we jump over to the generalized linear model,
which handles categorical data from a linear model's perspective.
That allows to take into account logits and probits and Poisson variables and ordinal variables.
From there, we combine the two sides of this story.
We put the mixed model, the mixed random effects terms, with the categorical data analysis,
and we cover the generalized linear mixed model.
Along the way, we'll cover generalized estimating equations and some additional techniques
that are relevant for the analysis of repeated measures or longitudinal data.
There are a number of exercises. I provide the data, and typically sample runs
that have been done in the stat packages, and participants are asked to extend those analyses
or put a new twist on them and explain what they've done.
So by the time you finish up, you will have run three or four rather substantial exercises
with data provided, and you'll know your way around all of these techniques.