Lecture 10 - Correlation (Part 1)

Meta

Key Topics

Correlation Handouts knitr Public Polling Scatterplots

Resources

Open on   Open on   Equations Handout Example Lab-09

Lecture Slides

Lecture Prep

Instructions

The instructions for this week’s lecture prep are available in the Week-10-Deprecated GitHub repository. Make sure you submit your work and answers by the beginning of class on Monday!

Replication

Correlation with Binary Variables

There is one point missing from today’s lecture that comes up on the lab. I mentioned that the assumptions of correlation are that x and y are continous. Given some of the details we’ve already covered with the binomial distribution, we can extend Pearson’s r to binary variables as well. These are interpretd slightly differently, since there is not “increase” in the binary variable in practical terms. We interpret them in relation to the 1 or TRUE category:

The correlation (r = .54, p < .001) between x, a binary variable, and y, suggests that the TRUE category is associated with an increase in y.