This paper deals with two problems confronting the applied researcher using a behavior analysis approach. The first is concerned with the need for independence of behaviors, subjects and settings. The effectiveness with which the researcher can solve this type of problem depends on the researcher's emphasis on the examination of procedures, and not experimental design. Similarly, the external validity problems involved in reactive experimental arrangements and multiple treatment interference are experimental procedure problems rather than design problems. Confronting those problems in this manner enables the applied researcher to add to the bank of empirical knowledge on human behavior. An extensive appendix which reviews the sources of invalidity proposed by Campbell and Stanley suggests that many are inappropriate to behavior analysis designs. Seven designs are presented graphically. (NG)