06-08-2016 10:13 AM
An HR company wants to see how much it costs for a new hired employee and its relationship between the cost and the scores of tests he took during the interview.
The cost is not normally distributed. Skewed. The correlation is low. Now I need to explore distribution of cost within each test quintile. Mean, median, inter-quartile ranges.
06-08-2016 02:01 PM
I guess that I would first divide the test scores into quintiles using PROC RANK and then do boxplots to describe the data. If you have different dimensions of test scores, you might be able to put together a two-dimensional array of box plots, but that might be hard to view.
06-08-2016 02:57 PM
I did rank but it is only one variable. Now I need to see how one variable distribution among groups of another variable. You are right, it is hard to see. And I do not know how to do box plot for two variables
I used KS method. Kolmogorov–Smirnov test . but I do not know how to interpret the SAS output. Anyone has advice? Thanks. Or is there a better way to see?