GAMs

One of the preferred statistical modeling methods in the lab is the generalized additive model (GAM), as well as its extension, the generalized additive mixed model (GAMM). Briefly, it’s a powerful statistical approach to model both linear and non-linear associations between a group of independent variables (for instance: age, sex, and other covariates) and a dependent variable (for instance, some brain measure you’re interested in). It’s a frequently used tool in the lab when looking at development, as brain and other developmental measures often show non-linear changes across childhood and adolescence.

Here is an awesome tutorial on GAMMs written by lab alumnus & GAM expert Bart Larsen.

For support from lab experts on GAMs, post your questions in the #gams Slack channel.

There is also a really useful GAM workshop taught by Gavin Simpson at Physalia. It usually happens remotely once or twice a year.