Analyze relationships between variables
PAIR will ensure that the variables selected are not highly correlated with each other. For instance, it would not be appropriate to use “Number of Enrolled Undergraduate Students” and “Total Enrolled Students” because total enrolled will likely be highly related to the number of enrolled undergraduates. If two or more of the variables are highly correlated with each other (>=0.9), it could cause undesired weighting in the analysis. Basically, if two variables are highly correlated then using both would be the same as using only one and weighting it twice in the analysis. Once the initial variables are identified, PAIR staff will review the data to identify any potential issues with highly correlated variables and can advise you on how to identify variables that may better address your needs.
Once the selected variables have been reviewed for completeness (Step 5) and correlations (Step 6), the final step before analysis is to decide whether any variables are “more important” in the analysis. This is the second point in the process where constituent feedback can be helpful.