This document provides guidance on planning and conducting a measurement system analysis (MSA). It discusses determining what to measure, selecting MSA tools, identifying participants, acquiring sample products, and developing a study plan. Key steps include identifying critical outputs and inputs from a cause-and-effect matrix, using 3 operators to test reproducibility, selecting 10-20 random samples as a "master" dataset, and distributing samples and collection instructions to operators without calibrating the measurement system during the MSA. The goal is to reflect the true "as-is" measurement process capabilities.

1. Section & Lesson #:
Pre-Requisite Lessons:
Complex Tools + Clear Teaching = Powerful Results
MSA – Planning & Conducting the MSA
Six Sigma-Measure – Lesson 29
An extended review on the series on building a MSA that covers the first
two steps on how to plan and conduct the MSA.
Six Sigma-Measure #28 – MSA - Overview
Copyright © 2011-2019 by Matthew J. Hansen. All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted by any means
(electronic, mechanical, photographic, photocopying, recording or otherwise) without prior permission in writing by the author and/or publisher.

2. Overview of Building the MSA
o Remember, there are 4 general steps for building the MSA:
Copyright © 2011-2019 by Matthew J. Hansen. All Rights Reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted by any means (electronic, mechanical, photographic,
photocopying, recording or otherwise) without prior permission in writing by the author and/or publisher.
2
1. Plan
the MSA
2. Conduct
the MSA
3. Analyze
the Results
4. Improve
measurement
system
This involves determining:
• What needs to be measured
• What MSA tools will be used
• Who will be part of the study
• What samples/master to use
• How to conduct the study
This involves:
• Scheduling the study
• Leading the study
• Collecting data from the study
This involves:
• Compiling all collected data
• Processing data in MSA tools
• Interpreting analysis results
This involves determining:
• If MS needs improvement
• How to improve the MS
• Redoing MSA (as needed)

3. 1. Planning the MSA
o What needs to be measured?
• Begin with the project Y (output).
 Since the Y is the most critical metric for the
project, it should be included in the MSA.
• Include potentially high impacting X’s that were identified in the Cause & Effect (C&E) Matrix.
• Include any additional metrics that are considered essential for analysis.
• Do measurements from a system need to be included in the MSA?
 Maybe. System-generated measurements generally have better precision than manually-based
measurements. Many prudently managed systems normally undergo accuracy & precision validations.
However, if there is a potential risk in the system’s integrity, the MSA can help measure that risk.
o What MSA tools will be used?
• There are two basic tools we’ll review that will depend on the type of data you’re analyzing.
 Discrete Data = Attribute Agreement Analysis or Multiple Attribute R & R
 Continuous Data = Gage R & R Study
• If you require a MSA on both data types, then both types of tools will be necessary.
 The difference may affect how you collect the data and eventually set it up for analysis in the tools.
o Who will be part of the study?
• You will need to decide who is responsible for recording and collecting the data.
• A general rule for testing reproducibility is to use 3 different operators.
 The operators should represent a typical skill level (if possible, exclude strong experts and novices).
Copyright © 2011-2019 by Matthew J. Hansen. All Rights Reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted by any means (electronic, mechanical, photographic,
photocopying, recording or otherwise) without prior permission in writing by the author and/or publisher.
3
1. Plan
the MSA
2. Conduct
the MSA
3. Analyze
the Results
4. Improve
measurement
system

4. 1. Planning the MSA (cont’d)
o What samples/masters will be used?
• “Masters” refer to a known set of accurate
items to measure. They will serve as the
standard for validating accuracy & precision.
 In a manufacturing environment, it can be sample products or customer production requirements.
 In a transactional environment, it can be a set of existing transactions/measures approved as accurate.
• The “master” process outputs should be randomly selected so they represent process variation.
• A general rule for testing is to use 10 to 20 different process outputs.
 If that’s not possible, then select enough samples so that the (# of samples) x (# of operators) ≥ 14.
o How will you conduct the study?
• A plan should be developed that outlines the following:
 When will the MSA begin?
 Who are the operators that are part of the MSA?
 How will the process outputs to test be distributed to each operator?
 When will the process outputs be distributed to the operators for each review trial?
 When are the operators expected to return the results for each review trial?
 What processes are the operators expected to follow for their review process?
– Is there a specific SOP that should be referenced?
– Are there specific criteria or requirements they are expected to follow?
 Will calibration be done before the MSA begins? If so, when will it begin?
– Calibration should NOT be performed during the MSA. It should only be done before the MSA, if at all.
– If no calibration is done before the MSA, the MSA results will more likely reflect the true As-Is process.
• Build the document or template that the operators will use for collecting the data.
 This will probably require understanding what data is collected and what MSA tools will be used.
– This will be reviewed in more detail in step #3 – Analyze the Results.
Copyright © 2011-2019 by Matthew J. Hansen. All Rights Reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted by any means (electronic, mechanical, photographic,
photocopying, recording or otherwise) without prior permission in writing by the author and/or publisher.
4
1. Plan
the MSA
2. Conduct
the MSA
3. Analyze
the Results
4. Improve
measurement
system

5. 2. Conduct the MSA
o Schedule the MSA
• Begin to execute the MSA plan.
 Begin to coordinate and schedule the MSA
with all operators and analysts.
 Pull the sample that will be used as the master dataset.
 Distribute to the operators the document or method the operators will use to record the data.
– Ensure the operators have full understanding of what data to record for the study and how to record it.
– Ensure the operators understand the deadlines for providing the data.
o Leading and collecting data from the study
• If observing the collection of the data, try to observe and document the following:
 Process deviations
 Environmental changes
 Differences in skills or behaviors between operators
 Observable errors in data entry
 Observable errors in use of the measurement tool
• Important: Do NOT calibrate nor make any process changes during the MSA!
Copyright © 2011-2019 by Matthew J. Hansen. All Rights Reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted by any means (electronic, mechanical, photographic,
photocopying, recording or otherwise) without prior permission in writing by the author and/or publisher.
5
1. Plan
the MSA
2. Conduct
the MSA
3. Analyze
the Results
4. Improve
measurement
system

6. Practical Application
o Due to the nature of a MSA, there isn’t a practical application for the first two steps of
planning and conducting the MSA.
• Despite this, please review the suggested steps for planning and conducting the MSA.
 How do these steps differ from other methods you may have seen for validating the reliability of data?
Copyright © 2011-2019 by Matthew J. Hansen. All Rights Reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted by any means (electronic, mechanical, photographic,
photocopying, recording or otherwise) without prior permission in writing by the author and/or publisher.
6