Growth Curve Analysis: A Hands-On Tutorial on Using Multilevel Regression to Analyze Time Course Data

Objectives and Scope Growth curve analysis (multilevel regression) offers a statistical framework for analyzing longitudinal or time course data and for quantifying differences between individuals in the context of a model of the overall group effects. These methods have been known to statisticians for many years (e.g., Wishart, 1938), but they have only recently become a prominent statistical tool in the cognitive sciences. Interest in using these statistical methods to analyze cognitive science data has outpaced cognitive scientists’ knowledge of how to implement them. This one-day tutorial will provide a hands-on introduction to using multilevel regression to analyze longitudinal or time course data and individual differences in the cognitive sciences. The focus will be on practical aspects of implementation and common pitfalls, rather than statistical theory. The tutorial will begin with the challenges of time course data, focusing on why traditional methods such as bin-bybin t-tests and repeated-measures ANOVA are sub-optimal. It will then describe how growth curve analysis (GCA) addresses those challenges and demonstrate how to use GCA to analyze simple linear data, non-linear data, and categorical outcome data (using logistic and quasi-logistic GCA). The final section will focus on using GCA to quantify and analyze individual differences. Throughout, example data sets will be provided for hands-on exercises that will give tutorial participants practice implementing the analyses, interpreting the results, and plotting data with model fits. There will also be time for tutorial participants to work on analyzing their own data with the instructor’s support.