Selection of an appropriate model must be based on extensive familiarity with the problem or system to be studied, so that the researcher can determine the range of biological responses necessary to the experimental design. Once this familiarity is developed, either by extensive review of the published literature or from pilot studies, the researcher can proceed to select an appropriate model, whether a whole animal, animal-derived material or a non-animal model.
Whole animal models are usually chosen when the system being studied can be best (or only) understood in the context of its interactions with other systems in the organism (e.g., sexual differentiation in embryonic development) or when it is the organism as a whole which is the system to be studied (e.g., the ontogeny of aggressive behavior). Some systems are better studied in isolation in animal cells, tissue or organs. For example, a number of biochemical and cellular processes can be studied in tissue or organ cultures derived from animal material. For other kinds of studies, biostatistical or computer models may be appropriate. It should be obvious, however, that the data generated from such non-animal models are only as good as the data upon which the models are based; thus, animal studies of some kind are prerequisite for developing and verifying all models.