POL S 15 Lecture Notes - Lecture 9: Dependent And Independent Variables, Linear Model, Linear Probability Model
Document Summary
Calculate conditional distributions: for each category of iv, calculate the percentage of cases in each category of dv. Then compare conditional distributions across categories of iv. Association = conditional distributions differ from iv to iv. Where expected is the frequency we woud expect if the two variables were independent (not associated) Minimum value is 0, the value indicating no association. Values > 0 indicate association (some kind of relationship) T-test for difference in means (quantitative dv, qualitative iv) A dummy variable: nominal variable with only two categories (binary coding, like 0 or 1) Calculate mean value of dependent variable for each group and compare; difference in mean quantifies association. Instead of a t-test, can use linear regression. In linear regression, the iv can be nominal scale: create a dummy variable for each category save one and add those variables to model.
