POLS W4912: Multivariate Political Analysis

POLS W4912: Multivariate Political Analysis
Prof. Gregory Wawro
Columbia University
Spring 2008

Description

The motivation for this course is first and foremost to give you the ability to use the classical linear regression model with confidence---both in theory and practice---to test causal models. Thus, by the end of the course, you should be aware of when ordinary least squares (OLS) yields desirable properties, the cases when it does not, how to test for these instances, and what procedures you should use for correct estimation and inference. In addition, you will have been exposed to a number of variants on the classical linear regression model that are applicable for the many different types of data you might encounter.

Additional information

Consult the syllabus for detailed description of course requirements, readings, schedule, and contact information.

Code and Data

linear_eq.r: Sample R code for solving a set of linear equations

makedata.ols1.r: Code for generating some data using multivariate normal random number generator

ols1.r: Code for reading in data and doing OLS

exampledta.txt: Data produced from makedata.ols1.r and used in ols1.r

multicollinearity.r: Monte Carlo code for exploring issues with multicollinearity

ml_ex.r: Illustration of maximum likelihood for a normally distributed random variable

Monte Carlo code exploring consistency of OLS (for question 1 on problem set 5)

Monte Carlo code exploring asymptotic distribution of OLS estimators (for question 2 on problem set 5)

Correct Monte Carlo code for heteroskedasticity problem (for question 3 on problem set 5)

garrum6.csv: ASCII file containing Garrett data from Partisan Politics in the Global Economy (Cambridge UP, 1998). Description of the variables in the data set is here ; Sample Stata code is here . Stata data set is here .

semdta.txt: Data for 2b and 2c on Problem Set #6.