Written by Cheryl Pammer, one of Minitab’s resident statisticians.
Way back in my statistical infancy, I was working with a research group on an interesting problem. The challenge was to design an experiment to understand how environmental factors—such as humidity levels, carpet materials, and cleaning frequency—influence dust mite populations. Dust mites are tiny, invisible creatures that thrive in household dust and are a major source of allergens.
Studying dust mite content in indoor environments is important because dust mites are a leading cause of allergic reactions and asthma, especially in children and individuals with respiratory conditions. This research helps public health officials, allergists, and homeowners make data-based decisions to create healthier living spaces. Additionally, the findings from studies like this one contribute to building codes, HVAC design standards, and cleaning recommendations aimed at minimizing indoor allergens.
When I was working on this problem, I was still in graduate school and had not yet taken my first course in Design of Experiments (DOE). So, DOE buzz words like 2k, central composite and screening had no meaning to me. Heck, I didn’t even know there were better ways to design an experiment other than using every combination of factor settings multiple times. So, while the experiment I designed adequately did the job so to speak, it wasn’t the most efficient way to collect the necessary information.
THE EXPERIMENT
The objective of the experiment was to determine the effect of humidity level, carpet type, and cleaning frequency on dust mite content in household environments. Test chambers were set up to mimic household environments with employees regularly using the test chamber to eat and watch television for roughly the same amount of time in each chamber. (Hey! Mimicking real world conditions can be challenging!)
After an 8-week period under these somewhat controlled conditions, dust samples were collected and analyzed for dust mite allergen content using standardized testing methods. The results provided a rich dataset for understanding how each factor, alone and in combination, affected allergen levels.
Prefer to watch instead? Check out our step-by-step video tutorial where Cheryl walks through this exact experiment using Quick Designs in Minitab. Watch the video here.
If only there had been Quick Designs back then. With the new Quick Designs in Minitab Statistical Software, you can quickly design a reasonable experiment without being an expert in DOE methodology. Let’s see how Quick Designs would have worked for me when designing this experiment.
Start by choosing Stat > DOE > Quick Designs. Your first step is simply to select the number of factors you want to consider.
Okay. This isn’t bad. I know I have three factors, so I’ll select a three-factor design.
My next choice is about what type of factors I have. Specifically, are my factors categorical, continuous or components of a mixture? Here, carpet type is categorical because carpet types are distinct groups while humidity and cleaning frequency are continuous because they could take on various numbers along a continuum.
From the above choices, I know I need to select Create an experiment with two or three continuous factors. That’s simple enough.
My next set of choices is based on the type of model I need to fit. I don’t have any factors that will be hard to change, so I need to decide whether I expect to see any curved effects on the response or only main effects and interactions. Because I’ve just begun my experimentation process, I’ll focus only on main effects and interactions and select Estimate main and interaction effects from this screen.
Finally, all I need to do is determine the low and high levels of each of my three factors and whether I want to run each setting more than once (i.e. replicate).
Pro Tip: While it is possible to set all three of these factors at several different settings, the experiment will be much more efficient if you can start by limiting each factor to just two levels. Minitab will automatically add a few points in the center for the continuous factors.
To be honest, the most difficult part of this process is related to a scientific problem, not a statistical one. I don’t want to trivialize the importance of selecting the right low and high factor levels to study. Choosing the correct factor levels involves selecting ranges that are realistic, relevant, and likely to influence the outcome based on prior research or practical considerations. For example, humidity levels should span values commonly found in homes (e.g., 30%–70%), while carpet types should represent materials widely used in residential settings. Cleaning frequency levels should reflect typical household routines. In this case 1 to 4 times per month. Ideally, levels should be far enough apart to detect meaningful differences but still within a range of practical values.
Now that all our design choices are behind us, click OK and Minitab has set up an appropriate experiment to run.
Full disclosure. This is not the experiment I ran. Without the benefit of guidance from a tool such as Minitab Statistical Software’s Quick Designs, I ran an experiment with way too many factor levels leaving me with not enough runs available to replicate. Lesson learned.
Using Quick Designs allowed for efficient investigation of multiple factors and their interactions. A well-structured DOE like this ensures that the effects of humidity, carpet type, and cleaning frequency on dust mite content can be accurately isolated and analyzed, leading to reliable conclusions.
Ultimately, a robust DOE enhances the scientific credibility of your experiments and supports data-driven recommendations for improving products and processes. With Minitab’s new Quick Designs, nothing should hold you back from obtaining the insights that only a carefully designed experiment can provide!
Design of Experiments doesn’t have to be complicated—especially when tools like Minitab’s Quick Designs simplify every step of the process. Whether you're studying allergens, improving a product, or solving real-world quality challenges, a well-structured DOE gives you clear, actionable results. Get started with Minitab today.
Ready to streamline your experiments and start getting insights faster?