Experiments are often planned to collect data for a limited number of variables to answer a limited number of questions. In safety studies, we simply ask is the product harmful to the animal. From this simple question, we generate multiple hypotheses for multiple variables with multiple assumptions to test. From a theoretical perspective, we have many questions. What assumptions can be evaluated based on the typical behavior of similar data, e.g., transformations and covariance tests? What design components should be considered? Can design theory help answer the important questions for a safety study?