QIPSR will be offering a 3-day introductory workshop on Bayesian analysis and multilevel modeling in Stata from May 15th-17th. Please register below for updates on location and reading material.
Note: On Tuesday 5/15 and Thursday 5/17, the workshop will be held in Gatton Classroom 199, and in the Marksbury Theater on Wednesday 5/16. The workshop will run from 9:00-4:00 each day with breaks throughout.
(We will have dedicated labs using Stata 15 each day. If you do not have a copy of Stata 15 then please download a temporary copy
here. If you have a copy of Stata 15 already then please ignore the link. Note that the file is large and may take some time to download depending on your internet connection. I've attached license information to this email. Please put "QIPSR workshop" as the organization and DO NOT register the license when you install it. This license will expire automatically on June 1.
If you do not have install access on your computer or simply do not wish to install it then you may check out a laptop for the labs with it preinstalled.)
Stata has recently made great strides in allowing users to employ Bayesian models. However, these techniques are very different from the traditional frequentist approaches that are familiar to Stata users. This workshop will walk participants through some of the primary differences between frequentist and Bayesian models, the underlying techniques used to solve Bayesian problems, how to specify models in Stata, and how torun and interpret diagnostics for these models. While we will focus on multilevel modelling applications, this workshop will be generally useful for anyone interested in understanding the Bayesian options in Stata 15.
We will use the canned ‘bayes’ suite of commands as well as the StataStan program. Registered participants will be provided with a temporary license for Stata 15MP and links to files for Stan. Users will need to have both programs pre-installed on their machines before the workshop begins or will otherwise be able to check out a laptop while they last.
Prior exposure to Bayesian methods is useful but not expected. The minimum requirement to find this workshop useful is a background in linear modeling and some exposure to maximum likelihood. We will focus on multilevel modeling in particular so some background exposure to mixed effects or multilevel models would also be beneficial. A set of optional (strongly recommended) background readings as well as the lecture outline can be found on the syllabus linked at the bottom of this page
Supplementary Readings
Lecture Files
Lab 1 Files
Lab 2 Files
