How can i find the correlation between a categorical independant variable and a continuous dependant variable? Traditional correlation metrics dont work in this scenario, should I use any statistical test? If so which test ?
$\begingroup$
$\endgroup$
1
-
$\begingroup$ Short answer is you cannot (it does not make sense), unless these categories have some kind of ordering to it (ordinal variable). Otherwise you can use a t-test or something. $\endgroup$– user2974951Commented Dec 21, 2018 at 8:51
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
If you don't have too many different values for the categorical variable, you could encode each value as a binary variable (e.g. for colour of car: one variable: 1 if car is green, 0 else). For these variables, regular statistical tests can be applied. But be careful. If you test all binary variables for significant correlation, the probability that one of them is significant by chance increases (multiple testing problem). You need to choose a lower p-value threshold to compensate for that.
But maybe this is also helpful to you: Correlations between continuous and categorical (nominal) variables
