The traditional approach used by formulators of agricultural and pesticide formulations relies heavily on experience, and has been quite valuable in generating acceptable formulations that may or may not be at a global optimum. However, it can be time consuming, and it generally lacks the ability to provide a global view of the contributions of the different components to the performance of the whole. Additionally, the approach does not permit the use of the collected information to investigate or predict the behavior of formulations within the composition space without carrying out further experimentation.
In the statistical design approach, one begins with the definition of objectives and constraints, then sets up the experiment with the appropriate response factor measurements. After data collection, a best-fit mathematical model is generated to describe the results and can be displayed graphically in various ways for various purposes. Among other things, model analysis typically reveals how each component contributes to the properties being measured. The model is used to locate optimum performance regions that balance the properties of the different components and satisfy any other constraints that are imposed. Scenario analysis can then be used to predict the performance as component levels are changed. In this work, we will discuss the conceptual framework of the statistical design approach, and exemplify its usefulness with an emulsifiable concentrate formulation.