Time: August 11, 2015 at 9am to August 12, 2015 at 4pm
Location: Hotel Birger Jarl Conference
Street: Tulegatan 8
City/Town: Stockholm, Sweden
Website or Map: http://statisticalhorizons.co…
Event Type: training, seminar
Organized By: Statistical Horizons LLC
Latest Activity: Jul 19, 2015
Taught by Paul Allison, Ph.D.
The most common type of longitudinal data is panel data, consisting of measurements of predictor and response variables at two or more points in time for many individuals. Such data have two major attractions: the ability to control for unobservables, and the determination of causal ordering.
However, there is also a major difficulty with panel data: repeated observations are typically correlated and this invalidates the usual assumption that observations are independent. There are four widely available methods for dealing with dependence: robust standard errors, generalized estimating equations, random effects models and fixed effects models. This course examines each of these methods in some detail, with an eye to discerning their relative advantages and disadvantages. Different methods are considered for quantitative outcomes, categorical outcomes, and count data outcomes.
This is a hands-on course with ample opportunity for participants to practice the different methods.
This workshop will focus on analysis of dichotomous, ordinal and nominal multilevel outcomes. Both clustered and longitudinal data will be considered, and the following models will be described: multilevel logistic regression for dichotomous outcomes, multilevel logistic regression for nominal outcomes, and multilevel proportional odds and non-proportional odds models for ordinal outcomes. The latter models are useful because the proportional odds assumption of equal covariate effects across the cumulative logits of the model is often unreasonable.