Practical Issues in Data Analysis
Out in the big bad world of practical data analysis, lots of issues can occur that make analysis more difficult and interpretation more complex. In the next two lectures, we will introduce and explain a number of the most common complexities along with common approaches to dealing with them. These will include topics of causality vs. prediction, confounding, suppression, Simpson's paradox, range restriction and missing data. We will also discuss extensions to the linear model that has been described in this course that handle different types of data.
David Hume Tower, Lecture Theatre C