Bob and I’ve posted another Excel workbook to the Library under the title of Cost Data Study. This time it concerns what are called Process Behavior Charts or otherwise referred to as Statistical Process Control (SPC) Charts.

This concept was first put forth by Walter Shewhart of Bell Laboratories in the 1920’s, and popularized by W. Edwards Deming throughout the world in the 80’s and 90’s. In recent years, Dr. Donald Wheeler has added the name of Process Behavior Charts, since the charts aren’t so much about “controlling” a process as they are for understanding it’s behavior.

The first chart is about installing some siding. I asked Bob Kovacs to aid me in collecting several times for installing siding as they might be listed in various estimating manuals. We came up with a listing of seven that have average times ranging from 1.70 to 4.10. Since there’s no time sequence involved in the collection of this data, all the individual times were randomly placed.

The corresponding chart is plotted to the right. Along with the average time of 2.9257, there are two other reference lines posted – one at 4.5835 and another at 1.2679. These are the Natural Upper and Lower Control Limits. What the chart is saying is that a process with this set of data can produce individual times that range anywhere between the Lower and Upper Control Limits (LCL & UCL). There is nothing unusual about this fact so the variation is called Common Cause Variation. If you were using the average time to develop your bid as in Sample Job #1, and the actual time came in at the still acceptable range of the Upper Control Limit, then you stand to lose $1,657.80! Remember, the process behavior chart is merely showing you the “natural” range you can expect. If you were to do this same task repeatedly under the same conditions, you could reasonably expect the time to average anywhere between 1.27 and 4.58 man/hrs per square.

The Second chart and Sample Job #2, use the same data only I have thrown out the previous high and low value of 1.70 and 4.10. These could be aberrations. If you plot the remaining data, you still get a similar average of 2.936, but now the Upper Control Limit is a tighter 3.428 and the Lower 2.444. If you look at Sample Job #2 you’ll see that it’s still possible to generate a 16.76% variance and loss of $492.00 and be within perfectly acceptable limits!

This is part of the danger in using an average to forecast your future performance. And what Jerry and I are referring to when we are talking about Common Cause Variation vs. Special Cause Variation.

There are rules for determining the Upper and Lower Control Limits as well as for reading the chart. If you like, we can discuss all of this at length and you’ll begin to understand how all of this can aid you in your estimating techniques.

The second page of my workbook involves the hanging of a pre-hung door. For the data here, I took the values from my Historical Database (also in the Library) and added two more values #5 and #6. Ideally, you’d like to have a minimum of 5-6 data points to do a chart, but you can get by with 4. The more data points you use, the more accurate the chart.

It’s important to note that these times are chronologically ranked. Along with the data plotted in the Individuals (X) chart, I’ve added another chart called the Moving Range (mR). The Moving Range chart is another good indicator about the performance of a process as long as the data points are time ordered. To plot this data, you find the difference between a data point and the one preceding it. If you subtract 1.50 from 1.60 you get .10. You do this for all the available times, which yield the data in the green shaded area. This is what’s plotted on the mR chart.

Notice again, the difference between the Average and the Upper Control Limit (UCL). If you were bidding this job, what time value would you use to base your estimate upon?

The XmR chart is a powerful tool you can use in your estimating procedures. It tells you what you can expect from your current process. If you don’t like what it’s telling you, if you don’t like the parameters of the range, you have little recourse other than to change the process. If you’re not going to change the process, you need to realize that there are times when your performance can approach the UCL!

This concept was first put forth by Walter Shewhart of Bell Laboratories in the 1920’s, and popularized by W. Edwards Deming throughout the world in the 80’s and 90’s. In recent years, Dr. Donald Wheeler has added the name of Process Behavior Charts, since the charts aren’t so much about “controlling” a process as they are for understanding it’s behavior.

The first chart is about installing some siding. I asked Bob Kovacs to aid me in collecting several times for installing siding as they might be listed in various estimating manuals. We came up with a listing of seven that have average times ranging from 1.70 to 4.10. Since there’s no time sequence involved in the collection of this data, all the individual times were randomly placed.

The corresponding chart is plotted to the right. Along with the average time of 2.9257, there are two other reference lines posted – one at 4.5835 and another at 1.2679. These are the Natural Upper and Lower Control Limits. What the chart is saying is that a process with this set of data can produce individual times that range anywhere between the Lower and Upper Control Limits (LCL & UCL). There is nothing unusual about this fact so the variation is called Common Cause Variation. If you were using the average time to develop your bid as in Sample Job #1, and the actual time came in at the still acceptable range of the Upper Control Limit, then you stand to lose $1,657.80! Remember, the process behavior chart is merely showing you the “natural” range you can expect. If you were to do this same task repeatedly under the same conditions, you could reasonably expect the time to average anywhere between 1.27 and 4.58 man/hrs per square.

The Second chart and Sample Job #2, use the same data only I have thrown out the previous high and low value of 1.70 and 4.10. These could be aberrations. If you plot the remaining data, you still get a similar average of 2.936, but now the Upper Control Limit is a tighter 3.428 and the Lower 2.444. If you look at Sample Job #2 you’ll see that it’s still possible to generate a 16.76% variance and loss of $492.00 and be within perfectly acceptable limits!

This is part of the danger in using an average to forecast your future performance. And what Jerry and I are referring to when we are talking about Common Cause Variation vs. Special Cause Variation.

There are rules for determining the Upper and Lower Control Limits as well as for reading the chart. If you like, we can discuss all of this at length and you’ll begin to understand how all of this can aid you in your estimating techniques.

The second page of my workbook involves the hanging of a pre-hung door. For the data here, I took the values from my Historical Database (also in the Library) and added two more values #5 and #6. Ideally, you’d like to have a minimum of 5-6 data points to do a chart, but you can get by with 4. The more data points you use, the more accurate the chart.

It’s important to note that these times are chronologically ranked. Along with the data plotted in the Individuals (X) chart, I’ve added another chart called the Moving Range (mR). The Moving Range chart is another good indicator about the performance of a process as long as the data points are time ordered. To plot this data, you find the difference between a data point and the one preceding it. If you subtract 1.50 from 1.60 you get .10. You do this for all the available times, which yield the data in the green shaded area. This is what’s plotted on the mR chart.

Notice again, the difference between the Average and the Upper Control Limit (UCL). If you were bidding this job, what time value would you use to base your estimate upon?

The XmR chart is a powerful tool you can use in your estimating procedures. It tells you what you can expect from your current process. If you don’t like what it’s telling you, if you don’t like the parameters of the range, you have little recourse other than to change the process. If you’re not going to change the process, you need to realize that there are times when your performance can approach the UCL!

## Comment