ABSTRACT
This chapter introduces the distinction between fixed effects and random effects. Multilevel regression provides a way to explictly model the nested structure of time course data. A core aspect of multilevel regression methods is that they simultaneously quantify both group-level and individual-level patterns within a single analysis framework. At the most general level, the goal of regression analysis is to find the parameters that best describe the data. Fixed effects are those factors that the analyst believes to be reproducible, fixed properties of the world, and their parameters are estimated independently. Random effects correspond to observational units that the analyst believes to be random samples from some population to which (s)he wishes to generalize. It is always a good idea to plot the model fit with the observed data. The log-likelihood model fit statistic provides relative goodness of fit information, but this doesn’t tell us how good the fit actually was.
