I am having little difficulty with understanding the difference between, testing association between two variables and modelling them.
Say I have binary outcome x(sold, not sold), and i have all other variables, such as the price, the age, and few other continuous variables and some discrete variables.
When looking for association between two discrete variables, one uses chi sq test. But, Now if I want to test association between x and price which is continuous variable, should I use linear model then carry out anova test or model it as GLM and look at see if the variable is significant.
What is the difference between the two approach?