Intro Statistics and Linear Models in R

Alison E. Post, Assistant Professor of Political Science, University of California, Berkeley
Ryan T. Moore, Assistant Professor of Political Science, Washington University in St Louis

These interactive teaching materials were written to accompany lectures for Government 2000, the introductory graduate quantitative methods course in Harvard University's Department of Government. They demonstrate the use of R programming techniques that implement concepts discussed in John Fox's Applied Regression Analysis, Linear Models, and Related Methods (Sage Publications, 1997), An R and S-Plus Companion to Applied Regression (Sage Publications, 2002), William Cleveland's Visualizing Data (Hobart Press, 1993), and Professor Kevin Quinn's lecture notes. They also provide extra mathematical intuition. Data analysis examples primarily employ the "car" library in R, which can be obtained (along with other necessary libraries) from the R-project website.

These documents were originally written as interactive section presentations. For this reason, it is best to open the files in R and enter example code at the command line. One way to do this is to open the .R files and R itself in XEmacs or a similar editor and toggle between the code and the command prompt. This will yield calculation results and graphical illustrations of the concepts discussed. The topics covered with direct links to .R files:

• Introduction to R and Emacs
• Probability Distributions in R
• Notes on moments and central moments
• Exploratory Data Analysis: Kernel Smoothing and Loess Fitting in R
• Data Visualization: Lattice Graphics in R
• QQ plot referred to in discussion
• Bivariate Linear Regression in Scalar Form
• Vector Geometry and Matrix Algebra
• Multiple Regression
• Statistical Inference in Matrix Form
• F distribution simulation sourced in above notes (from Kevin Quinn)
• Assessing Model Adequacy
• Detecting Outliers
• Robust Regression
• Dummy Variables, Interaction Terms, and Polynomial Regression

For those interested in learning more about R, additional resources can be found at the R-project website. We would like to acknowledge Professor Quinn for the simulations of sampling distributions called in some of the section presentations.