The Six Sigma process improvement methodology relies heavily on data. A critical aspect of working with data is effectively defining the measures so that the information gathered is reliable and valid. Several steps can be taken to ensure that the measures for a Six Sigma project are defined appropriately. These steps should be followed for any metric that is used for the project, and the definitions must remain consistent throughout the project for before-after comparisons to be of value.
Instructions
1. Clarify your objective and parameters. Be sure that everyone involved is clear on the purpose of each measure that has been selected for use in the project. Also make sure that all agree on the parameters for the measure. This may include clarifying the timeframe, the part of the process or other factors. Put these decisions in writing.
2. Create an operational definition. It is not as simple as it seems to define a measure in a way that the definition will be interpreted unambiguously. A good operational definition is very specific and descriptive, and will produce the same results no matter who is doing the measure. Be sure to test yours by having different people apply it to a set of information, and confirm that the results are identical (or at least within tolerable limits).
3. Design your data collection procedure. For each measure, document the specific methods that must be used to collect data. The instructions should be so specific that all individuals following them will use the same process and obtain the same results. This may involve creating a specific database querying, describing an observation or monitoring process, or creating visual aids and other guidelines for categorizing data.
4. Confirm the validity and reliability of your measure. Validity refers to the extent to which a metric actually measures what it was designed to measure. Reliability is the extent to which the metric remains unchanged regardless of who does the measuring or when it is done, using the same data set. In some cases an informal assessment of reliability is sufficient, such as comparing categorization results for two different groups working with the same data. In other cases, a formal tool such as a Gage R&R test should be used to confirm reliability (see Resources below).
5. Document your definition and process to ensure standardization. It is critical that your operational definition and data collection procedure be documented so that the measure is standardized for use throughout your project and even beyond. Create a document that specifies the operational definition, the data collection process and any issues that were addressed in creating them.
