Good evening,

Thanks for a good session today. Before we start our final day I wanted to give you a few things to keep in mind pertaining to what we covered today: 

1. The concept of nested models is very important for model comparison approaches. Remember that two models (e.g., a "restricted" and "full" model) are nested if you can switch between them by *either* adding or subtracting variables (and two models are not nested when when variables are *both* added and subtracted at the same time). The concept of nested models is important for any model comparison (in regression, SEM, HLM, and more) because specifying nested models allows you to directly compare models. 

2. Mean-centering variables will eliminate nonessential collinearity (but not all collinearity), but that may come at the cost of interpretability. This comes into play in both moderation and polynomial regression models. The concept of centering is also very important in more advanced statistics (e.g., HLM). 

3. In logistic regression models (with dichotomous or polytomous criterion variables), we compute and interpret *pseudo* R-squared values, can use odds ratios [exp(B)] as effect sizes, and can calculate a number of indices using the confusion matrix (e.g., specificity, sensitivity, positive predictive value, negative predictive value, percentage accuracy in classification). 

Let me know if you have any questions (here or on Zoom tomorrow) and I will see you tomorrow for the last day of the short course.

-Justin