What is DOE?

Design of Experiments (DOE) is a systematic, mathematical approach used to understand and optimize the relationship between input factors and output responses [1]. It is a branch of applied statistics that allows researchers to change multiple variables at once in a controlled way to determine which factors are truly significant and how they interact with one another [3].

Core Components of DOE

To build a DOE model, you must define the following four elements:

• Factors: The independent variables you intentionally change (e.g., Temperature, Pressure) [1].
• Levels: The specific settings or values assigned to each factor (e.g., 180°C and 200°C) [3].
• Response: The output or result you are measuring (e.g., Product Yield, Car Speed) [2].
• Noise: Uncontrollable variables that might cause variation in the results (e.g., ambient humidity) [2].

The 5 Phases of DOE

A standard DOE framework follows a specific lifecycle to move from initial curiosity to a fully optimized process [2, 5]:

1. Planning: Defining the objective, identifying potential factors, and ensuring the measurement system is accurate [5].
2. Screening: If there are many potential factors (usually >5), screening experiments like Plackett-Burman or Fractional Factorial are used to narrow them down to the “vital few” [1].
3. Modeling: Once significant factors are known, a mathematical relationship (regression) is built to map how inputs affect the output [3].
4. Optimizing: Using methods like Response Surface Methodology (RSM) to find the exact “sweet spot” or peak performance [5].
5. Verifying: A final “confirmation run” is performed at the optimized settings to ensure the results match the model’s predictions [1].

Why DOE is Better than Traditional Testing

Traditional testing often uses the One-Factor-at-a-Time (OFAT) method. DOE is statistically superior for several reasons:

Feature	Traditional (OFAT)	DOE Framework
Interactions	Cannot see if two factors depend on each other [3].	Specifically designed to find and map interactions [1].
Efficiency	Requires many more runs to cover the same ground [3].	Finds the best answer with the minimum number of tests [2].
Predictive Power	Only tells you about the specific points you tested [3].	Creates a mathematical “surface” to predict results at any setting [5].

[Image comparing one factor at a time vs factorial design]

References

• [1] American Society for Quality (ASQ): Design of Experiments (DOE) Overview.
• [2] JMP Statistics Knowledge Portal: The DOE Workflow and Analysis.
• [3] Czitrom, V. (1999). One-Factor-at-a-Time Versus Designed Experiments. The American Statistician.
• [4] NIST/SEMATECH e-Handbook of Statistical Methods: Process Improvement via DOE.
• [5] Montgomery, D. C. (2019). Design and Analysis of Experiments, 10th Edition. Wiley.

---