Reply To: Ordering categories within categorical variables

#722
Andee Rubin
Participant
4 pts

I’d like to weigh in on this, since I actually disagree with the decision – and didn’t realize that Fathom did it this way, either.  I think there should be consistency in how we think about the Y-axis and it should ALWAYS be that lower values are further down.  If not, we get the attached counter-intuitive graph, in which a “perfect” correlation is represented as a line with slope -1, rather than 1.  This seems wrong to me.

Attachments: