As researchers, what is our primary goal in relation to

As researchers, what is our primary goal in relation to behavioral variability?  When there is variability that is extraneous to the independent variable, what can we do to manage the variability so that we produce data that describe the phenomena that we are interested in describing?