Although studies of performance are found in many research traditions, they share
the basic approach of "natural experimentation." Because it is generally infeasible to
establish true experimental controls in studying financial performance, authors typically
estimate the impact of a particular factor on performance, using statistical techniques to
hold other causal factors constant. Most statistical tests of the effects of individual explanatory
variables continue to be against the null hypothesis of "no effect," even though
this null should often be replaced by comparison of results with the work of others in a
'compare and contrast" framework.