Mark C Long
PB AF 528
Second quarter of a two-quarter sequence aimed at helping students become informed users and critical consumers of research and statistical analysis. Combines material on research design and data collection methods with tools for multivariate analysis. The multivariate analysis methods include correlation and an introduction to multivariate regression. Prerequisite: PB AF 527.
This course is the second in a two-course sequence aimed at helping Evans School MPA students become informed users and critical consumers of research and statistical analyses. This course introduces the application of probability, hypothesis testing, and confidence intervals to multivariate models in the context of policy and management research. We strive to isolate and measure the effects of one factor (an independent variable – often the introduction of a policy) on an outcome (a dependent variable) while controlling any other factors. We begin with the linear regression model in its basic form and move on to modeling techniques. Along the way we will consider some of the limitations and potential problems associated with using regression models and alternative models. Students will develop a first hand appreciation of these topics through in-class exercises and homework problems.
Student learning goals
Understand how complex policy analysis can be conducted using multivariate regression analysis.
Be aware of the conditions necessary to establish causal relationships on a given outcome, emphasizing the need to disentangle the effects of multiple factors.
Select appropriate univariate, bivariate, or multivariate analytic techniques to answer a given policy or management question.
Understand the mechanics, assumptions, and interpretation of regression models to policy or management questions, how to use regression models for both prediction and hypothesis testing, and the assumptions behind and possible "fixes" for problems with models.
Learn how to read and analyze empirical studies. Produce a useful multivariate empirical analysis for a non-statistician, including clear data presentation and the graphical display of data.
Recognize how policy analysis, program evaluation, and performance measurement employ research methods and statistical techniques. Be exposed to nonlinear models and understand their purposes.
General method of instruction
Class assignments and grading
The course requirements include six homework sets, two in-class exams, and a final data analysis exam. The purpose of the two exams is to help diagnose your progress in learning the mechanics and interpretation of regression. [These exams will be closed book, but you will be allowed to use 2 pages (4 sides) of notes.] The data analysis exam allows you to consolidate your learning about regression models, apply what you are learning to a policy context, and learn to communicate your results to a nontechnical audience.