# Measurement selection matrix

## Measurement selection matrix

### Highlights

Used to find the measures most strongly linked to customer needs

### To create and use a measurement system matrix …

1. Collect VOC data ( see Chapter 4) to identify critical-to-quality requirements. List down the side of a matrix.

2. Identify output measures (through brainstorming, data you're already collecting, process knowledge, SIPOC diagram, etc.) and list across the top of the matrix.

3. Work through the matrix and discuss as a team what relationshipa particular measure has to the corresponding requirement: strong, moderate, weak, or no relationship. Use numbers or symbols (as in the example shown here) to capture the team's consensus.

4. Review the final matrix. Develop plans for collecting data on the measures that are most strongly linked to the requirements.

## Stratification factors

### Highlights

Purpose is to collect descriptive information that will help you identify important patterns in the data (about root causes, patterns of use, etc.)

• Helps to focus the project on the critical few

• Speeds up the search for root causes

• Generates a deeper understanding of process factors

### To identify stratification factors …

Your team can identify stratification by brainstorming a list of characteristics or factors you think may influence or be related to the problem or outcome you're studying . The method described here uses a modified tree diagram (shown above) to provide more structure to the process.

1. Identify an Output measure (Y), and enter it in the center point of the tree diagram.

2. List the key questions you have about that output.

3. Identify descriptive characteristics (the stratification factors) that define different subgroups of data you suspect may be relevant to your questions. These are the different ways you may want to "slice and dice" the data to uncover revealing patterns.

Ex: You suspect purchasing patterns may relate to size of the purchasing company, so you'll want to collect information about purchaser's size

Ex: You wonder if patterns of variation differ by time of day, so data will be labeled according to when it was collected

Ex: You wonder if delays are bigger on some days of the week than on other days, so data will be labeled by day of week

4. Create specific measurements for each subgroup or stratification factor.

5. Review each of the measurements (include the Y measure) and determine whether or not current data exists.

6. Discuss with the team whether or not current measurements will help to predict the output Y. If not, think of where to apply measurement systems so that they will help you to predict Y.

## Operational definitions

### Highlights

• Operational definitions are clear and precise instructions on how to take a particular measurement

• They help ensure common, consistent data collection and interpretation of results

### To create operational definitions …

1. As a team, discuss the data you want to collect. Strive for a common understanding of the goal for collecting that data.

2. Precisely describe the data collection procedure.

• What steps should data collectors use?

• How should they take the measurement?

Ex: If measuring transaction time in a bank, what is the trigger to "start the stopwatch"? When a customer gets in line? When he or she steps up to a teller?

Ex: If measuring the length of an item, how can you make sure that every data collector will put the ruler or caliper in the same position on the item?

Ex: What counts as a "scratch" on a product finish? What counts as an "error" on a form? (Misspellings? missing information? incorrect information?)

• What forms or instruments will data collectors have to help them? Specifically how are these forms or instruments to be used?

• How will the data be recorded? In what units?

3. Test the operational definition first with people involved in Step 2 above and then again with people not involved in the procedure, and compare results. Does everyone from both groups get the same result when counting or measuring the same things? Refine the measurement description as needed until you get consistent results.

 Tip Develop visual guides to help people take the measurements correctly—such as photos with notes on what is to be measured or counted (and how ), "good" vs. "bad" standard examples, etc.