549 Gale Lemerand Drive
Gainesville, FL 32611-6400
Textbooks promote the idea that “statistics” is concerned only with data analysis. Without statistical grounding, formulation and implementation of research problems are often performed by trial and error, with tweaks to the system made one factor at a time. These strategies are time-consuming, inefficient, miss potential interactions, and provide little insight. In contrast, statistical design of experiments (DOE) is rigorous, highly efficient, and informative. DOE relies on iterative and structured inputs of multiple input factors and settings, with clearly defined and measurable outputs. Calculations involve simple arithmetic and quadratic regression models, allowing rapid and informative interpretation and visualization of multivariate interactions and complex responses. Results from successive tests are used to inform and refine next steps in product or process development. Benefits include more informed decision-making at each stage of the problem-solving process, more rapid convergence on solutions, lower costs, and better product quality. I provide examples of DOE applied to testing biomaterials, processes, and simulations.