Red Bead Experiment: Part 5

This is Part 5 of 6 of the Deming Red Bead Video Tutorial. In this video, we learn tricks to facilitating such an event. Watch below and be sure to also view Part 6, the final video in the Red Bead Experiment Series.

Video Transcript

Steve Prevette: On to QA we go.

Male Voice: Six dismissed.

Steve Prevette: Well, okay. Put that back in. Now, the key here seems to be the second dip is always better than the first one. So do the second dip first. [laughter] Oh, a little modification here. Okay. Except that modification didn’t work too good. You left a hole. Can’t sell holes. Got to be sure of that count. I’m glad I have my ethical QA department here. Except you weren’t dismissed. Hold on.

Male Voice: Sixteen dismissed.

[audience groan]

Steve Prevette: Oh, sixteen. You see what happened when you changed side? Okay. Up. Forward. You just¦ I kept my best workers and¦ Yeah, we’re just trying to record here. I’ve got to make these beads and get them out the door. Just put them down there. [laughter] Now, you’re not dismissed yet. We’re slipping here.

Male Voice: Thirteen.

Steve Prevette: Thirteen. Well please total up the number of beads for the day, 29, 38, 42, something like¦ 57. Are you serious, 57?

Male Voice: You got me on the wrong line.

Steve Prevette: Fifty-seven.

Male Voice: You lost it. You haven’t got it.

Steve Prevette: Ah, it doesn’t matter. Fifty-seven. [laughter] Because we kept our best workers, and I had my worst day yet. So nothing more from you. Mr. President, we completed our fourth day of work with our best workers and we had 57 red beads. Our worst day to date.

Steve Byers: Well, all I can say is that the management team is stunned and disappointed. This is certainly not what we thought would happen keeping our best, above average workers. We have bet the company on this contract, and so because it has been cancelled, because of our poor performance, I’ll ask you to please dismiss the remaining workers as well as the QA team, recorder, and so on. We’re closed for business.

Steve Prevette: Here’s your severance pay and¦ [laughter] your severance pay. And please give our work force a round of applause.

[clapping]

Steve Prevette: Well that was the red bead experiment. Now this was a tool that Dr. Deming used during his four day seminars. And we’re using it here to demonstrate some ISMS principles and we’re going to critique what just happened. By the way, I’m Steve Prevette. I’m from Fluor Hanford, ES and H, and the president is Steve Byers, representing the American Society for Quality.

Well our first step here is, we used a lot of feedback here. ISMS sure includes a lot of feedback. But maybe we could do something different here and plot some dots. We’re just going to put a little graph up. If I can ask the president to help by reading off some points as we go, let us go forth.

Steve Byers: Ready?

Steve Prevette: Yep. Ready.

Steve Byers: Eight.

Steve Prevette: Eight.

Steve Byers: Ten.

Steve Prevette: Ten.

Steve Byers: Ten.

Steve Prevette: Ten.

Steve Byers: Nine.

Steve Prevette: Nine. Now, I came here to work early on with Tank Farms and they were wrapping one fiscal year after the other. Only producing first fiscal year data, and they put a nice curve function through on Harvard Graphics, and fed a curve like this and said, Notice the fine, improving trend to zero. [laughter] We won’t do that, of course. Okay. What was willing worker number five?

Steve Byers: Eleven.

Steve Prevette: Eleven. Hmm.

Steve Byers: Eight.

Steve Prevette: Eight.

Steve Byers: Second day.

Steve Prevette: What happened to that dip down to zero? Hmm. Oh, well. Second day.

Steve Byers: Ten.

Steve Prevette: Ten.

Steve Byers: Seven.

Steve Prevette: Seven.

Steve Byers: Thirteen.

Steve Prevette: Thirteen.

Steve Byers: Seven.

Steve Prevette: Seven.

Steve Byers: Ten and nine.

Steve Prevette: Ten and nine. Boy, it’s just bouncing all over the place, isn’t it?

Steve Byers: Seven.

Steve Prevette: Seven.

Steve Byers: Eleven.

Steve Prevette: Eleven.

Steve Byers: Seven.

Steve Prevette: Seven.

Steve Byers: Four.

Steve Prevette: Oh, four. Hmm.

Steve Byers: Eleven and nine.

Steve Prevette: Eleven and nine. And the last day with our best workers.

Steve Byers: Sixteen.

Steve Prevette: Sixteen.

Steve Byers: Thirteen.

Steve Prevette: Thirteen.

Steve Byers: Nine.

Steve Prevette: Nine.

Steve Byers: Four.

Steve Prevette: Four.

Steve Byers: Nine.

Steve Prevette: Look at that. Linear regression right on down. Four. Okay. what was after four?

Steve Byers: Nine and six.

Steve Prevette: Nine and six. Boy, this is all over the place, isn’t it? And, of course, if we’re going to produce this as a chart, we’d probably put bars under it, and three dimensions, and colors, and the whole works, and there’s our performance chart, right? What does it mean?

Well, let’s offer something here. We’re going to make what is called a control chart, which is Hanford procedure 4294. When you work performance indicators and when you work ISMS, we use these control charts. Now the first thing you’ve got to do is determine what was the average. And we total up all the days, 56 and 56 and 49 and 50. I’m going to do this without a calculator even. Everybody thinks control charts are hard to do. I’m doing this without a calculator, without a computer. Twenty-one, 28, 2, 7, 12, 16, 21. Two hundred and eighteen¦ 218 total red beads, for 24 people pulled the red beads. And that gives me 9, 18, 36, looks like the average is nine.

So we draw a line across the screen at nine. An amazing thing occurs, does it not? Half the data’s above the average, and half the data’s below the average. That’s amazing isn’t it? Yet how many people are grading our facilities as above average, below average? How many people’s children get graded as above average or below average at school? We do performance appraisals. Who’s above average, who’s below average? We’re fating 50% of the population to be below average. And it’s always going to happen, right? Doesn’t say anything, but we do it.

Well, what we do that should be better is the next thing I’ve got to do is express the spread of this data, which is to put three standard deviation control limits on the line. Now, in this case, this is a plus on distribution because I’m counting defects. This works well when you’re counting defects. The standard deviation is the square root of the average. The square root of nine is three. Nice round number. Three times three, three standard deviations, is nine again. So the upper control limit is 18, and the lower control limit is zero. UCL average, LCL. There we have a control chart. No computer, no calculators, just a control chart.

Now we interpret this. What is happening is I actually have an extremely predictable process. We couldn’t understand why every worker was changing from day to day, from shift to shift. But I really have a predictable process. I average nine, I get no more than 18, I get no less than one, because zero is on the lower control limit. So anywhere from one to 17 red beads is what I get. And that’s what we got. We have no seven in a row above average, or any other rules for a control chart. So I have a stable process. A stable process.

Yet, what was I doing? What was the effect of having the $30 on the table? Nothing. Do we have bonus pay here? What’s its effect? What was the effect of doing performance appraisals? Do we do performance appraisals? Interesting. What was the effect of the procedure?

Male Voice: Controlled system.

Steve Prevette: Controlled system?

Male Voice: Controlled results.

Steve Prevette: Yeah, controlled results, but vary widely. It certainly did not meet the customer’s demands, did it? How about this numerical goal? Three. Weren’t you finally motivated when you had this number three out there? Yet in the MYPPs our numerical targets for OSHA recordable case rates. What good does the numerical target do? Interesting. So what can we do different? Well let’s take this in an ISMS context.

Get instructions on how to conduct the Red Bead Experiment >>>