Locked lesson.

Upgrade

About this lesson

The purpose of the Measure phase of a Lean Six Sigma project is to collect complete, accurate, and meaningful data. There is a simple data collection approach that can be used by the team to ensure this is accomplished.

Quick reference

Data Collection

The purpose of the Measure phase of a Lean Six Sigma project is to collect complete, accurate, and meaningful data.  There is a simple data collection approach that can be used by the team to ensure this is accomplished.

When to use

The Data Collection approach should be used in the Measure phase to ensure that data collected is accurate and meaningful.  The approach may also need to be used in other succeeding stages if new or additional data is needed.

Instructions

Lean and Six Sigma both rely on data for the analysis and ultimate improvement of the process or product.  Process data needs to be collected at each step of the process. The reliance on data is why an entire Phase of a Lean Six Sigma project is devoted to Measure.  The following simple process provides guidance for managing the data collection process.

  1. Determine the data collection goals.  Decide what data you want to collect.  Consider using existing databases and records.  Determine the attribute categories to be collected with the data in order to facilitate sorting and segmenting the data for analysis.
  2. Create the data collection procedure.  The procedure is a reference document for anyone involved in the data collection process.  It should clarify the data definition of what data is to be collected.  Describes how the data is to be collected and recorded.  Identify when to start and stop collecting data.  And, if appropriate, describe the sampling plan for determining which process points or products to measure.
  3. Ensure data accuracy and stability.  Variation in measured items is due either to true variation from part to part or process to process or it is due to variation within the measurement system.  For accurate analysis, the measurement system variation must be minimized.  The best technique for ensuring that the measurement system had little effect on the data is to conduct a measurement systems analysis.  You should also be certain that everyone involved in collecting data has been trained on your procedure.
  4. Begin collection of data.  When the system is in place and validated with a measurement system analysis, start collecting data.  Collect data from finished items and historical data when available.  Data can be collected during a process operation, but be careful to minimize the impact of the data collection on the process.  The data collection can become a special cause to make things must better or worse than the normal process.

Hints & tips

  • It may feel like a wasted bureaucracy to create a data collection procedure, but the first time you find that all the data collected during the past two weeks is worthless because the person collecting the data didn’t understand what was wanted, you will understand the importance.
  • When collecting in-process data, explain to the operators what you are doing and why.  Otherwise, they may change their behavior and not perform the process the way they normally do.  This is referred to as the Hawthorne Effect based on research done years ago in Hawthorne, Illinois.  During this research, the presence of the researchers collecting data had a far more profound effect on the operator performance than the attribute that was under study.  Because of the presence of the researchers, the operators significantly changed their operating practices.
  • Make sure you have measures that directly correlate with the big “y” of the project. You want to understand how that is changed by changing conditions within the process.
  • If you are not familiar with the MSA methodology, take our course on that subject.
Login to download
  • 00:05 Hi, I'm Ray Sheen.
  • 00:06 I've been discussing various attributes of data and its implications, but
  • 00:10 haven't actually talked about collecting data.
  • 00:13 And since that is the point of the measure stage,
  • 00:16 let's spend just a few minutes on that topic.
  • 00:19 Let me highlight the principle the data collection is done at the process step
  • 00:24 level, not at the full process level.
  • 00:27 So create your flow for the process, either a process map or
  • 00:30 a value stream map.
  • 00:32 Often, you will want to measure the time at each step, and if its focus is a lean
  • 00:37 waste reduction, you'll want to measure the value added for each step.
  • 00:41 If cost is a major concern, then look at the actual cost spent at each step, and
  • 00:46 if quality defects were the issue,
  • 00:48 measure the number of defects created at each step.
  • 00:52 In recent years, I've been seeing projects where environmental waste or
  • 00:56 opportunities were being measured.
  • 00:58 And the point is, whatever is a meaningful measure for
  • 01:02 your particular problem, measure that at each step in the process.
  • 01:07 You don't need to measure everything, instead,
  • 01:10 focus the measurements on the ones that are related to the nature of the problem.
  • 01:15 Also, if possible, use measurements that are easy to observe or measure.
  • 01:20 So rather than measuring motor torque which may be hard to measure in your
  • 01:24 process, measure the amount of current draw for the motor.
  • 01:28 Finally, unless I'm measuring a variance within a step,
  • 01:31 I want to measure each step in a consistent manner.
  • 01:35 So I usually measure at the beginning of the step.
  • 01:39 Now, I want to walk through the four step process for
  • 01:42 ensuring that your data collection activities are successful.
  • 01:46 The first step is to determine your goals.
  • 01:49 In this first stage,
  • 01:51 you need to answer several questions to clarify your data collection goals.
  • 01:56 We don't want to waste time and effort collecting unnecessary data,
  • 02:00 we do want to focus all of our effort on the critical data needs.
  • 02:04 The first question is an easy one,
  • 02:06 what is the process attribute that you're investigating?
  • 02:10 You need to collect data about those attributes in order to both quantify
  • 02:14 the magnitude of the problem now and to be able to demonstrate when we're all done
  • 02:18 that your solution actually made things better.
  • 02:22 So based upon those attributes, what data is needed?
  • 02:26 Is it time data, quality data, cost data?
  • 02:29 Another important question to ask is, what data is already available?
  • 02:34 Think about the business systems, the quality records, or the test systems.
  • 02:38 Often there's a wealth of data already available and
  • 02:41 there are only a few other items that need to be collected.
  • 02:45 Finally, consider what categories of data you would like to collect.
  • 02:49 The category data is often critical for helping you segment and
  • 02:52 sort the data in order to isolate the contributing causes of the problem.
  • 02:58 Things like customers, product, the location of the events,
  • 03:01 the time of the events, the location of the process, process operator or
  • 03:06 the shift that occurs, the magnitude of the event or the type of defect.
  • 03:11 The next step is the process to create the data collection procedure that
  • 03:15 will be used.
  • 03:16 Now, you may be thinking this is just useless bureaucracy, but
  • 03:20 if you will have multiple people involved in data collection,
  • 03:24 this is a very important element.
  • 03:26 If not done well, everyone will collect the data a little differently, and
  • 03:30 the next thing you know, you're comparing apples to oranges.
  • 03:34 So take a few minutes to create a good procedure.
  • 03:37 In this procedure, precisely define the data you want collected.
  • 03:41 Where on the product or in the process will the measurement or observation occur?
  • 03:46 What are the units to use?
  • 03:48 And what are the associated attributes that need to be recorded?
  • 03:51 Specify how the data is collected, what measurement system is used,
  • 03:55 and any special factors associated with the collection process.
  • 04:00 Also, clarify how the data is to be recorded, what system or form is used, and
  • 04:05 what fields are completed, what attributes are being measured or inspected.
  • 04:11 Next, determine when you want the data collection to start and
  • 04:13 when it should stop.
  • 04:15 Also, if multiple shifts are involved, clarify which shifts.
  • 04:20 Finally if a sampling plan is used,
  • 04:22 you should spell out how many samples and how the samples are selected.
  • 04:29 Now onto the Stage 3, data accuracy and stability.
  • 04:33 There are two components that are sources of variation in any
  • 04:36 data collection process.
  • 04:38 This is the variation in the product or process that's being measured,
  • 04:42 we want to find this out, but
  • 04:44 there's also the variation that occurs within the measurement system itself.
  • 04:48 I already introduced the concept of a measurement system analysis in
  • 04:52 an earlier lesson.
  • 04:53 This is sometimes called a gauge R&R analysis.
  • 04:56 We have a short course on how to conduct a measurement system analysis.
  • 05:01 If you're not experienced in this area, I strongly suggest you take the course.
  • 05:05 In addition to the MSA, there are several other things that you may need to do.
  • 05:09 One is to provide training for everyone who will be collecting the data.
  • 05:13 This training may only be an informal five-minute meeting to review
  • 05:16 the procedure and the data collection form.
  • 05:19 Or it may need to be a more formal process if complex equipment is to be used.
  • 05:24 Also conduct a pilot run or a test trial of the data collection process.
  • 05:29 Make sure everything works, and I like to do that pilot run with known good and
  • 05:33 known bad parts or processes so that I am certain that the measurement system is
  • 05:38 able to discriminate between the two.
  • 05:41 The fourth stage of the data collection is really pretty obvious, collect the data.
  • 05:47 In addition to the in-process data collection,
  • 05:49 collect data on items that have finished going through the process.
  • 05:52 That way you know that you are seeing what the process customers are seeing.
  • 05:57 If there is historical data in databases or other systems, retrieve that data.
  • 06:02 Be careful to ensure the data definitions are consistent with what you
  • 06:05 are measuring.
  • 06:07 If applicable, collect data at various points within the process.
  • 06:11 Just be aware your data collection is creating a special cause effect.
  • 06:16 Because you are watching the operator may be doing things differently.
  • 06:21 This is called the Hawthorne effect and
  • 06:23 it's based upon research done in Hawthorne, Illinois, years ago.
  • 06:26 The research shows that the presence of the researchers has a significant impact
  • 06:31 on the process performance.
  • 06:35 One way to minimize this is to talk with the process operators.
  • 06:38 Explain to them what you're doing.
  • 06:41 Discuss how you're measuring and what you're measuring in the process or
  • 06:45 product.
  • 06:46 Answer their questions and reinforce that you want them to
  • 06:49 do everything in their normal way so that you can understand what is happening.
  • 06:54 Often I find I need to reassure them that,
  • 06:57 we're not out to find a problem and a blame on them, rather,
  • 07:01 we want to understand and improve the customer experience.
  • 07:06 So we need to know what is really happening.
  • 07:09 So go collect data, we're going to need that.
  • 07:12 Find the root cause or
  • 07:13 causes in the analyze phase of our Lean Six Sigma project.

Lesson notes are only available for subscribers.

PMI, PMP, CAPM and PMBOK are registered marks of the Project Management Institute, Inc.

Gift this course