2C Multivariate Data Analysis
Alejandro Quiroz Flores (Essex University)
22 July - 2 August (two week course / 35 hrs)
Detailed Course Outline [PDF]
Course Content
This course presents important solutions to common problems in estimation. First and foremost, this involves inference and prediction, functional form and structural change, as well as model selection. The course then moves on to solutions to violations of the Gauss-Markov conditions—this includes endogenous covariates, non-spherical disturbances, and non-linearity. The course also covers solutions to other issues in estimation, such as heterogeneity, discrete choice, limited dependent variables, and survival models.
Course Objectives
This course prepares students to understand the limitations of the Gauss-Markov conditions and, most importantly, how to solve problems when the conditions are not met. For instance, the students will learn how to select models and specifications. They will also learn how to identify and deal with heteroscedasticity and autocorrelation, as well as what to do when their projects face non-linear parameters and non-normally distributed dependent variables. Among other things, participants will learn how to use instrumental variables and estimate models of simultaneous equations, truncated models, and survival models.
- Thoroughly understand the mechanics of Least Squares both mathematically and substantively.
- Correctly understand, interpret, and present empirical evidence as related to hypotheses testing in Least Squares.
- Identify violations of the Gauss-Markov conditions and potential solutions to these violations.
- Identify solutions to other problems in estimation not necessarily related to the Gauss-Markov Conditions.
- Comfortably use STATA to solve problems arising from violations of the Gauss-Markov conditions.
Course Prerequisites
Students should be quite familiar with Least Squares and the Gauss-Markov Conditions. Some knowledge of calculus (i.e. derivatives) and linear algebra is also important, as the course uses this framework to present key results. Students do not need to perform complicated linear algebra calculations, but should be able to understand what matrices are, how to transpose them, multiply them, and invert them, among some other simple operations.
Representative Background Reading:
Participants should be comfortable understanding Appendix E (The Linear Regression Model in Matrix Form) of Jeffrey Wooldridge's Introductory Econometrics. Familiarity with Appendix D (Summary of Matrix Algebra) in the same book will be very useful. Some experience with William Greene's Econometric Analysis, particularly Chapters 2-4 will give students a head on start with the course.
In terms of articles and applications, representative readings are the following:
Brambor, Thomas, William Roberts Clark, and Matt Golder. 2006. Understanding Interaction Models: Improving Empirical Analyses. Political Analysis 14 (1): 63-82.
Beck, Nathaniel, and Jonathan N. Katz. 1995. What to Do (and Not to Do) with Time-Series Cross-Section. American Political Science Review 89 (3): 634-647.
Miguel, Edward, Shanker Satyanath, and Ernest Sergenti. 2004. Economic Shocks and Civil Conflict: An Instrumental Variables Approach. Journal of Political Economy 112 (4): 725-753.
Philip, Paolino. 2001. Maximum Likelihood Estimation of Models with Beta-Distributed Dependent Variables. Political Analysis 9 (4): 325-346.
Beck, Nathaniel, Jonathan Katz and Richard Tucker. 1998. Taking Time Seriously: Time-Series-Cross-Section Analysis with a Binary Dependent Variable. American Journal of Political Science 42 (4): 1260-1288.
Required Reading:
The required textbook used for the course is:
Wooldridge, Jeffrey. 2003. Introductory Econometrics: A Modern Approach. Mason, OH: Thomson.
Students interested in learning more should try to get:
Greene, William. 2003. Econometric Analysis. New Jersey: Prentice Hall
Important: References for Wooldridge are for the 2nd Edition. References for Greene are for the 5th Edition