Fall 2010 - 61245 - PA397 - Introduction to Empirical Methods for Policy Analysis
|Instructor(s):|| Matwiczak, Kenneth
|Day & Time:||M 2:00 - 5:00 pm|
|Waitlist Information:||For LBJ Students: UT Waitlist Information|
This course helps students develop an understanding of how basic quantitative tools are used in policy analysis. The major concepts discussed include modeling, optimization, sensitivity analysis, statistical inference, estimation, and prediction. These concepts are covered in the context of applications such as constrained decisionmaking based on calculus and on linear programming; policy choices with probabilistic information; evaluating and updating information with Bayesian techniques; estimating the impact of policy factors using regression models; and practical methods for forecasting. As the first course in the quantitative sequence, the emphasis is on broad exposure of techniques and appreciation of their contributions as well as their limitations in policymaking. Students must have fulfilled prerequisites in college-level algebra, calculus, and statistics before enrolling in this course. It is usually taken during the fall semester of the first year.
This course develops basic competence and skills in problem solving and quantitative methods applied to public policy analysis. It emphasizes the art and skill of converting problem descriptions into quantitative models, and the analysis and interpretation of these models. This course develops basic competence and skills in problem solving and quantitative methods applied to public policy analysis. It emphasizes the art and skill of converting problem descriptions into quantitative models, and the analysis and interpretation of these models.
We will review some basic concepts of probability, probability distributions, and descriptive statistics as a means of communicating and describing data of all types. Following a brief discussion of data sources and sampling methods, we will study the use of sample data to make estimates of, and inferences about, the parameters of larger populations. Using these inference skills, we develop and test linear regression models to describe relationships between a characteristic dependent variable (such as income) that describes a sample or population, and one or more other explanatory variables (such as gender, age, etc.), based on sample data. Building on these regression skills, we use historical data to estimate and forecast trends and seasonality.
We will also look at how quantitative methods can be applied to support decision-making. In our look at decision analysis, we will model decisions using decision matrices and decision trees. With these tools, we will evaluate decisions made under conditions of certainty, risk, and uncertainty. We will also consider the value and cost of obtaining additional information when making decisions, and how to incorporate the decision maker's value judgments into the decision model. In all our decision analysis work we will consider the sensitivity of the decision to changes in the problem conditions. After a brief look at making choices involving several different, diverse characteristics, (i.e. multi-criteria decisions), we introduce the idea of "mathematical optimization" as a way to search for the best possible solution to a decision problem which may be constrained or unconstrained. This will include using linear programming techniques as well as calculus-based techniques for non-linear functions.
The course is an applications course rather than a rigorous theoretical or mathematical development. Emphasis in the course is on the application and interpretation of quantitative modeling and analysis methods in policy evaluation and decision-making. Students will be required to make extensive use of Microsoft Excel computer spreadsheets for homework assignments, applications exercises, and take-home exams. Students will also participate in a phased semester project on a pseudo-real world problem of their own choosing, making an oral presentation on the results of their work at the end of the semester. Two take-home exams will complement the four or five homework problem sets.
- Successful completion of a "typical" college-level mathematics course, including algebra / linear algebra. Plus, one semester of introductory statistics, or equivalent. Calculus "validation."
- Familiarity with computer spreadsheets, or willingness to devote out of class time to become familiar with using spreadsheets such as MS-Excel. "Familiarity" includes recording data in a spreadsheet, performing simple calculations, and using the graphing functions.