It is not at all practical to change ##r##, I agree. You are missing that the coil separation ##d## must be part of the expression for the B-field at the midpoint. The expression that you are given is valid for only for a fixed ratio ##\beta = d/r##. Thus, if you want to test the relationship, you must vary both ##r## and ##d## in a way that keeps the ratio constant. In other words, the relationship you are given appears to have ##r## as an independent variable, but it is not independent. The numerical factor 0.72 implicitly contains the ratio ##\beta##. I strongly recommend that you derive a general expression for the B-field at the midpoint when the separation is ##d## and the radius is ##r##. It will help you see what's going on and see where the factor 0.72 is coming from.
One thing you could do experimentally is to test the given expression is to measure the B-field at the midpoint as a function of coil separation ##d## and find for what value the given relationship is valid. However, I am not sure that this is what the exercise expects you to do.