SMLP2018

Statistical Methods for Linguistics and Psychology, University of Potsdam, Germany, 10-14 September 2018

Project maintained by vasishth Hosted on GitHub Pages — Theme by mattgraham

Welcome to The Second Summer School on Statistical Methods for Linguistics and Psychology, 2018, 10-14 September

Summer School Location

Griebnitzsee Campus, University of Potsdam, Germany

The summer school will be held at the Griebnitzsee campus of the University of Potsdam; this is about 15-20 minutes away from Berlin zoo station by train. Lectures will be held in Haus 6, Rooms S17 (Frequentist stream), S18 (Bayesian stream). Invited lectures will be held in Hoersaal H02. Campus map: download from here. Please use bvg.de for planning your travel (by train or bus).

Fees

There will be a 30 Euro fee to cover expenses for the two evenings (Tuesday and Thursday) of snacks, and for coffee and snacks during the breaks. International participants can pay on arrival in cash. Others have the option to pay electronically in advance to the following account: Landeshauptkasse Potsdam, IBAN: DE09 3005 0000 7110 4028 44, BIC: WELADEDDXXX.

Participants will pay for their own accommodation and travel, and will have to pay for their lunch themselves.

Code of conduct

All participants will be expected to follow the code of conduct, taken from StanCon 2018. In case a participant has any concerns, please contact any of the following people: Audrey Bürki, Shravan Vasishth, Bruno Nicenboim, Daniel Schad, or Reinhold Kliegl (see instructors’ list below).

About the summer school

The Second Summer School on Statistical Methods for Linguistics and Psychology will be held in Potsdam, Germany from September 10-14 2018. Note that this summer school follows the annual international AMLaP conference, which will be held in Berlin September 6-8 2018.

Like the first edition, this summer school will have a Bayesian and frequentist stream. Bayesian data analysis will be taught (first four days) by Michael Betancourt, who is one of the lead developers of Stan; he will be assisted by Bruno Nicenboim and Shravan Vasishth. The frequentist stream will be led by Reinhold Kliegl, Audrey Bürki, and Daniel Schad. Both tracks will be aimed at linguists and psychologists, but are relevant for cognitive scientists working in many different areas. We will cover the necessary theoretical background for both statistical frameworks, and participants will get hands-on practice in learning to analyze real data sets.

The pace of this summer school will be intense! Participants should expect to be working hard during this one week, and to go home and review the materials after the course is over.

Curriculum

The summer school is intended for participants who have data analysis experience (especially, experience with linear mixed models is assumed), but want to work with state-of-the-art Bayesian and frequentist methods. We will assume that participants should be comfortable with the material presented in this article and these introductory lecture notes.

We will mainly focus on modeling repeated measures data. We will discuss hierarchical modeling in detail, using the generalized linear mixed modeling framework as a starting point. We will use R, the lme4 package, and Stan. Participants will be assumed to have a working knowledge of R; Stan will be taught from scratch. Everyone is expected to bring their own laptop computer.

The summer school will consist of lectures followed by hands-on exercises.

Hour-by-hour schedule

The schedule can be downloaded here.

Slides+code+data

Slides etc. will appear early Sept 2018.

Bayesian stream: syllabus

Taught by Michael Betancourt (10-13 Sept), Bruno Nicenboim, Shravan Vasishth (14 Sept)

This course is designed for people who want a fast entry into Bayesian data analysis and modeling using Stan. The participants should be fairly familiar with linear modeling theory and should be experienced in fitting frequentist linear mixed models.

Mailing list: click here.

Frequentist stream: syllabus

Taught by Audrey Bürki, Reinhold Kliegl, Daniel Schad

This course is designed for researchers who are interested in fitting linear mixed models using the lme4 package in R, but are unsure about how to proceed with advanced topics such as model selection, contrast coding, and visualizing model fit.

Invited talks

We will have two invited talks during the summer school.

Preparation for summer school

We are assuming that participants will have a very good working knowledge of R and have considerable experience in fitting linear mixed models. Participants are expected to bring their own laptops. Wifi access will be available.

Please make sure you have the current versions of R and RStudio installed on your computer by the time you start the summer school, and that you have the R packages rstan, brms, and rstanarm installed. To install rstan, follow the instructions here.

Background Material

Overview and methodological papers

Good textbooks and articles covering fundamental ideas (mostly for Bayesian track, but also helpful for frequentist track)

This list will be extended.

Papers relating to linear mixed modeling

Examples of Bayesian data analysis and Bayesian modeling from our lab

Lecture notes

These lecture notes serve as reference materials for the summer school. You are not expected to read them during the summer school, but they may help for review later on.

Video lectures and tutorials on Bayes using Stan

Contact details

For any questions regarding this summer school, please contact Shravan Vasishth.

Summer school organizers and instructors

Organisers:

Audrey Bürki, Bruno Nicenboim, Reinhold Kliegl, Daniel Schad, Shravan Vasishth

Instructors:

Michael Betancourt, Audrey Bürki, Bruno Nicenboim, Reinhold Kliegl, Daniel Schad, Shravan Vasishth

Funding

This summer school is funded by the DFG and is part of the SFB “Limits of Variability in Language”.