MODULE 1: June 8 (noon) – June 11 (noon)

Bayesian Latent Class Modelling in Animal Health

Instructors: Wes Johnson (lead instructor), Ian Gardner and Henrik Stryhn

Wes Johnson: University of California, Irvine

Ian Gardner and Henrik Stryhn: CVER, AVC, University of PEI

This 3-day course will provide the theoretical background to Bayesian modeling with applications to animal health including evaluation of test accuracy, prevalence estimation and logistic regression. Participants will obtain practical hands-on experience using OpenBUGS software and other tools (Betabuster and sliders) to apply the methods to real-life examples. Specific topics covered on the first day 1 include: introduction to probability using diagnostic testing examples (perfect and imperfect test examples), laws of probability, Bayes’ theorem, probability distributions, prior distributions likelihood functions, and introduction to simulation and Markov-chain Monte Carlo methods for simulating priors and posteriors. On days 2 and 3, models for estimating accuracy of single and multiple independent and dependent tests, prevalence estimation including hierarchical models, logistic regression with random effects and logistic regression when imperfect tests are used to measure outcomes will be discussed.

MODULE 2: June 13 – June 17

Introduction to Multilevel Modelling (schedule)

Instructors: Ian Dohoo, Henrik Stryhn and Javier Sanchez

CVER, AVC, University of PEI

The 5-day course is designed to provide participants with the knowledge and skills required to successfully fit multilevel models to both continuous data (linear models) and discrete data (emphasis on logistic and Poisson models). The presentation of theoretical background material is limited to that which is required for a reasonable understanding of the methods employed. Specific topics covered in the course include: introduction to multilevel/hierarchical data, mixed models for continuous data, mixed models for discrete data, model evaluation (diagnostics), analysis of repeated measures and alternative approaches to dealing with clustered data (including Bayesian methods). The main software used for the instruction is MLwiN, but code for fitting models in additional software packages (including Stata and R) will also be provided.

REGISTRATION

For each module, the following registration fees apply (amounts are in $Cdn and include taxes and administration fees):

MODULE 1 - Bayesian Latent Class Modelling in Animal Health:

Participant Up to April 29th After April 29th

Student* 500 600

Non-student 750 850

MODULE 2 - Introduction to Multilevel Modelling:

Participant Up to April 29th After April 29th

Student* 700 800

Non-student 1100 1200

*Proof of enrollment in a graduate program at a university (not necessarily UPEI) required

A discount of 10% will be offered for participants taking both Module 1 and Module 2. Refunds will be allowed before May 27th subject to 20% of administration fee.