Misconceptions about Using Design of Experiments, Part 1

A common reason for not using Design of Experiments is “I don't have a project DOE applies to.” While this is true occasionally, this reason is usually the result of a misconception. This and the next two articles will deal with the three most common misconceptions about the use of DOE, namely, the belief that DOE doesn't apply when all of the factors are discrete (categorical), thinking that responses are factors, and believing that factor constraints prevent the use of DOE. Design of Experiments most definitely does apply when all of the factors are discrete. (Remember, discrete factors have no in-between levels. For the factor “beverage” there is no level between “coffee” and “tea.”) Sir Ronald Fisher invented design of experiments to study discrete factors. It is true that you must run all combinations of the factors levels, so you won't be able to reduce the number of trials. However, since you have to collect all this data anyway, why not get the maximum amount of information from your work? DOE will let you find the best trial and the interactions among the factors. This could lead to a better theoretical understanding, benefiting future work. Here's an example: You want to study your guest's preferences for refreshment at your annual holiday party. For hot beverages you have coffee, tea, cider, and hot cocoa. For snacks you have cookies, cake, and pie. You would need to ask your guests to try all the possibilities and rank their preferences: coffee and cookies coffee and cake coffee and pie tea and cookies tea and cake tea and pie cider and cookies cider and cake cider and pie hot cocoa and cookies hot cocoa and cake hot cocoa and pie As you can see, all possibilities must be studied, so there is no reduction in the number of trials. Through a DOE analysis you discover that coffee is the most desired hot beverage. However, you discover that if people are eating cookies, they prefer hot cocoa. (This would show up in your analysis as a strong main effect for coffee and a strong interaction between hot cocoa and cookies.) So, you would generally want to supply coffee at your parties in the future. However, if you have a party with only cookies available, you would want to supply hot cocoa instead. (This is only a hypothetical example — you would need to run a real experiment to draw real conclusions.) So Design of Experiments is useful even when all of the factors are discrete. It helps you extract the maximum amount of information from your work. You see main effects and interactions. Next time let's look at how easy it is to confuse factors and responses, and how you can sort them out. Objective Design of Experiments workshops will teach you to use DOE in your work. Design of Experiments is a fundamental technique for industrial experimentation. You will learn to apply DOE easily without excessive math and theory. We will help you be even more successful!