Skip to main content
1:00 PM - 4:00 PM
B110G Wells Hall

Determining the Right Standard Errors & Statistical Tests Using Mixed Models

Many studies that are created to infer upon treatment effects or other effects of primary interest also involve factors whose elements (e.g. animals, days, locations, schools) would be best designated as having random effects. That is, these random elements would not necessarily be re-used from one repetition of the same experiment to the next. When classical ANOVA methods (based on ordinary least squares) are used to analyze such studies, these random effects are not appropriately modeled, and the resulting standard errors of treatment or group means can be substantially understated. Mixed model analysis will be presented in this workshop as a way to appropriately account for this uncertainty and obtain correct standard errors and tests of hypotheses. In addition, the utility of mixed model analysis to discern true experimental replication from pseudo-replication will be demonstrated. Several applications based on the use of SAS PROC MIXED/GLIMMIX will be presented.