Identify Responses

After clarifying the objective, the next step is to figure out which responses (quality characteristics) to measure and how to measure them. Identifying the critical quantifiable responses is an important step for a successful DOE.

Numeric (continuous): A measured response is the best type of response. During the designed experiment, there should be a range of response values, from good to bad.

Rating System: For subjective measurements, a rating system is a workable substitute for a continuous response. For example, product quality may be rated as 1-very bad to 5-okay to 9-very good. Likewise, taste ratings may be 1-sour to 5-neutral to 9-sweet. It is better to have several people rate the product. The response should be the average rating for each product. The standard deviation or range of the ratings often makes a useful secondary response.

Ranking: Another type of response that is okay to use is a ranking. Take the results from all the runs and rank them from worst to best. Then assign a number to each result starting at 1 and increasing to reflect the rank. Results that are functional ties get an average (and identical) rank.

Follow the work on responses by choosing the factors then determining the number of runs.