One variable is a group of teachers response to a survey, the other will be the principal response. When I run a Pearson r I get a message about the one variable being constant. There is only one administrator per school. Any advice?
I am using likert scale data for the teacher and principal responses. However, the principal only has one response. There can be upward of 15 teachers per school. When I attempt to test for a relationship using Pearson r, I get the message above.
It sounds like you can use correlation. Are you just pooling the data ? That is, looking across all schools, but for each school using 15 observations from teachers and 15 observations for the principal (which are the same value). How many schools are there ? What software are you using ?
I don't know how much value that comparison is, as you are comparing in some ways apples with pears in that if I understand correctly you have 15 teachers, but 1 principle is that correct? so you'll have variation with the summary measures of the teachers, but only one principal measure with no variation