An effect that is seen consistently in replications can only reasonably be
explained as being caused by the manipulation of the experimental
variables. With an observational study the same consistency of results
might be obtained because all the data are affected in the same way by
some unknown and measured variable.
