How can I find the correlation between a categorical (dependent) variable and a continuous-scale (independent) variable? Is a Kruskal Wallis test appropriate? I'm a little confused as my independent variable is not nominal but scale!
5
-
$\begingroup$ Is the categorical variable ordered or not? $\endgroup$– highBandWidthNov 18, 2017 at 13:37
-
$\begingroup$ No it is not ordered $\endgroup$– AchNov 18, 2017 at 14:00
-
1$\begingroup$ Did you just make a try to search an answer on the site? $\endgroup$– ttnphnsNov 19, 2017 at 1:03
-
$\begingroup$ Yes I have tried to search the answer. A possible solution to this problem is to convert my scale independent variable into a categorical one and conduct a chi square test between my 2 nominal variables. $\endgroup$– AchNov 19, 2017 at 7:40
-
$\begingroup$ stats.stackexchange.com/q/308897/3277, stats.stackexchange.com/q/73065/3277 $\endgroup$– ttnphnsNov 20, 2017 at 6:07
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
Correlation notions do not care about the direction of causality (if there is any). So it shouldn't matter which one is dependent and which independent. If assumptions of normality are not a problem, then one way ANOVA should be ok. If you want it to be non-parametric, then Kruska-Wallis should be appropriate.
answered Nov 20, 2017 at 3:53