The dangers of conditioning on the time of occurrence of one demographic process in the analysis of another