This workshop will cover how to use the Stata command -margins- to explore the results of
linear models. When models include many categorical predictors or interaction terms, the
reported estimates of the model coefficients are difficult to interpret.
Estimated marginal means are model predictions based on a set of combinations of predictor
variables. This workshop will cover how to use estimated marginal means to interpret model
output in tabular or graphical form. We will also cover post hoc comparisonsm such and the
Tukey test for all pairwise comparisons, as well as custom contrasts and difference-in-
differences. The -margins- command also allows for testing and comparison of slopes by group
in an ANCOVA model, and aids in interpretation of output when the response has been
transformed, or for generalized linear models (such as logistic or Poisson regression).
This workshop assumes familiarity with linear models (regression and ANOVA) as well as
logistic regression.
Methods will be demonstrated in a hands-on fashion using Stata.