Greg Kochanski

Hilary Term 2006, weeks 1-8 (Jan-Apr 2006) University of Oxford. The course meets Tuesdays at 11:00, in the Centre for Linguistics and Philology.

This course parallels "A Cartoon Guide to Statistics" (Larry Gonick and Wollcott Smith, HarperCollins 1993, ISBN 0-6-273102-5). It covers (mostly) basic statistics with an emphasis on understanding the connection between statstical tests and hypotheses. It uses the open source "R" package for demonstrations, which can be obtained at http://www.r-project.org . (R is available both for Windows and Linux machines.)

• Why one needs statistics and quantitative models: a historical view of goofs in other fields.
• We will cover basic statistical estimation of answers and the standard hypothesis tests that are useful for Linguistics and related fields.
• Statistical tests are based on the mathematics of probabilities. Ultimately, they tell you the probability that a hypothesis is wrong, or they tell you the range into which an answer is likely to fall. We will review probability, conditional probabilities and Bayes' Theorem.
• At the end of the course, we will briefly look into a few statistical techniques that your mother didn't teach you.

The primary goal is to give the student the tools to figure out what data and theories are reliable. Secondarily, to provide tools for his or her own work.

The course will demonstrate statistics on R, and will discuss some examples of the use of statistical techniques. Examples may include spam filtering, attribution of authorship and style.

Prerequisites: General understanding of algebra. Some familiarity with computers is useful, programming experience may be helpful but is not essential. This is a survey course intended for a broad audience.