All material posted here is subject to common sense copyright ;). That is, I'd appreciate if you cite if something is useful, and if you ask me before using materials in your classes (fjaeger@bcs.rochester.edu). This will also help to reduce damage whenever I (or someone else) discovers mistakes in any of the material. Feedback welcome. Thanks. Whenever I incorporated materials from others, this is mentioned in the slides.
[http://prosodylab.org/?p=407 Two day statistics tutorial at McGill] (5/2010) with Maureen Gillespie (Northeastern) and Peter Graff (MIT). The workshop was organized by Michael Wagner, Kris Onishi, and Aparna Nadig and sponsored by CRLMB, prosody.lab, the Mcgill Infant Development Cluster, the PoP lab, the Auditory Cognitive Neuroscience training grant, Digging into Data (SSHRC/NSF), and BRAMS. My own lectures were updated versions of the tutorial at Penn State: [attachment:lecture1McGill.pdf Lecture 1 "Introduction to Generalized Linear and Generalized Linear Mixed Models"] and [attachment:lecture2McGill.pdf Lecture 2 "Common issues and solutions in regression analyses"]. In addition, Maureen Gillespie lead a [attachment:gillespie-tutorial.pdf hands-on tutorial on Implementing Hypotheses: Coding], along with a [attachment:gillespie-script.R R script] using a [attachment:fakedata.txt fake data set]. Peter Graff gave a [attachment:graff-tutorial.pdf tutorial on model comparison "Comparing linguistic theories using logistic regression"], also along with
[attachment:graff-script.R R script] and the [attachment:PluralComparisonMontreal.csv linguistic data data set described in the slides].
- Two 3h-lectures held at Penn State (2/2010). [attachment:Day1.pdf Lecture 1] provides an introduction to generalized linear models and generalized linear mixed models. The slides contain a lot of R code, but see also this [attachment:Day1.R documented R script]. I incorporated (and modified) a handful of slides from Roger Levy (with his permission) from the 2009 Workshop on Multilevel Modeling (WOMM) held at CUNY 2009, UC Davis. Most of the slides are introducing linear regression and linear mixed models. The goal of this lecture was to provide researchers unfamiliar with these models with an intuition about what it means to use a generalized linear (mixed) model to analyze your data. Since the audience was particularly interested in comparisons to ANOVA, you'll find a lot of that, too. [attachment:Day2.pdf Lecture 2] is an updated version of the presentation that Victor Kuperman and I gave at WOMM. The lecture covers common problems and solutions with regression models. There also might be a podcast available to go along with these lectures (soon).
- [wiki:/CUNY09MiniWorkshop Materials for workshop on common issues and standard in ordinary and multilevel regression modeling in psycholinguistic data analysis] (Florian Jaeger and Victor Kuperman; CUNY, March, 2009): assumes some prior familiarity with regression modeling. This is is not an introduction, but rather anattempt of a concise summary of common issues in regression modeling, some standards for evaluating and visualizing models, and some open questions about multilevel models.
[wiki:/DenmarkMiniCourse 2-day crash-course on issues in model fitting for ordinary and multilevel models] (Florian Jaeger; Royal Academy of Denmark, November 2008]: assumes background in probability theory and regression fitting, focuses on common issues and the steps involved in model fitting.
- [wiki:/HLPMiniCourse 7-session introduction to ordinary and multilevel regression methods] (Austin Frank and Florian Jaeger; HLP lab, June-July 2008): assumes no background and goes all the way to simple linear and logit multilevel models, but is very reading intensive.