I appreciate your interest in my article and I would like to clarify your concerns raised in above comment.

Issue of Normality: The assumption of normal data is relevant only to individual moving (I-MR) range control chart while control charts which have sample size greater than 10 such as Xbar and S does not mandate that the data should be normal. Hence I have deliberately suggested to utilize Xbar & S control chart to plot CR (critical ratio) so that even if the CR data is not normal there is no impact on control limits.

Further Walter Lipkes article does make references to SPI,CPI but it does not talk about the behavior of critical ratio (CR) .In Walters article the Cost Ratio is referred to as CR while in this article CR refers to critical ratio and not to cost ratio.

Sample Size: The sample size for the Xbar and S chart depicted is not 1.The article suggests that a set of critical ratios should be collected for each phase of the project (Please refer the following line from the article One way to group the project performance data is by project phase (another option would be to group the data by month or quarter). Use Xbar-S control charts to plot the CRs (continuous data) over the project duration)

Data point for each phase in control chart does not represent individual critical ratio but represents central value for each sample of critical ratios ( i.e for a set of critical ratios collected for a particular project phase) . Example the single data point in the Xbar chart for design phase represents average value of critical ratios collected for design phase and similarly data point in S chart represents standard deviation of critical ratios collected for design phase.

I hope the above explanation would have addressed your concerns.

Regards,

Abhishek Soni ]]>

Dear Taha,

I appreciate your interest in my article and I would like to clarify your concerns raised in above comment.Issue of Normality: The assumption of normal data is relevant only to individual moving (I-MR) range control chart while control charts which have sample size greater than 10 such as Xbar and S does not mandate that the data should be normal. Hence I have deliberately suggested to utilize Xbar & S control chart to plot CR (critical ratio) so that even if the CR data is not normal there is no impact on control limits.Further Walter Lipkes article does make references to SPI,CPI but it does not talk about the behavior of critical ratio (CR) .In Walters article the Cost Ratio is referred to as CR while in this article CR refers to critical ratio and not to cost ratio.

Sample Size: The sample size for the Xbar and S chart depicted is not 1.The article suggests that a set of critical ratios should be collected for each phase of the project (Please refer the following line from the article One way to group the project performance data is by project phase (another option would be to group the data by month or quarter). Use Xbar-S control charts to plot the CRs (continuous data) over the project duration) Data point for each phase in control chart does not represent individual critical ratio but represents central value for each sample of critical ratios ( i.e for a set of critical ratios collected for a particular project phase) . Example the single data point in the Xbar chart for design phase represents average value of critical ratios collected for design phase and similarly data point in S chart represents standard deviation of critical ratios collected for design phase.I hope the above explanation would have addressed your concerns.

Regards,

Abhishek Soni

Hello Chris,

Thanks for your interest in my article. The project with longer duration will certainly be in better position to employ this tool for project tracking because of the abundance of data points .But in case projects of very short durations I would leave the decision of employing this tool for project tracking to the wisdom of individual project managers. One needs to keep a balance between the net project worth vis a vis cost of tracking the project.

Regards,

Abhishek Soni

Taha, I am interested in hearing about your recommendations for use of statistical process control to improve EVM for project management.

]]>Hello,

Thank you for providing the paper entitled “Manage Project Performance with EVM and Control Charts”. I read the paper carefully, and I would like to draw your attention to a few points.

Using statistical process control to improve earned value project management is an attractive and practical area, so, this aspect of your work is appreciated. But some significant problems is seen in your paper as follows:

1. Normality

As we know, Xbar-S control chart uses for data that have a Normal distribution. If this is not taken into account, control limits are strongly influenced. Many studies (especially Walter Lipke’s studies) have shown that EVM indices including SPI, CPI, SPI(t), CR and etc. do not follow a normal distribution and have a right-skewed distribution! For more detail you can refer to Earnedschedule.com. Therefore we can

2. Sample Size

As we know, projects are finite, they have a start and an end. During the project control using earned value management method, for each period we have only an index! i.e. n=1, so, how did you use Xbar-S control chart, while there is only one observation, for example at period 2 we have just one SPI, one CPI, and etc.

Of course there are other significant problems. Therefore it is not really simple to incorporate SPC and EVM.

I am a graduate student working on my thesis entitled “Using statistical process control to improve earned value project management”, pleases feel free to contact me if I am not correct.

Regards,

Taha Mortaji

Mortaji@ymail.com