Control Charts Made Easy - Calculating d2 - How-to Guide for Excel and Minitab
Dennis Crommentuijn-Marsh
Director Consultant and Head of Data, Analytics and Insights Practice at Process Management International (PMI)
Note: This is the simplified How-to guide
The table shows d2 factors for n=1,2,3...10. You will notice that the d2 increases with each n, but that the difference between subsequent d2 values decreases.
Excel - How do I calculate d2?
Detailed steps in Excel
2. Enter the random number formula in cell A2: =NORMINV(RAND(), 0, 1)
Copy this formula down to A1000001
3. Enter the random number formula in cell B2: =NORMINV(RAND(), 0, 1)
Copy this formula down to B1000001
4. Enter the range formula in cell C2: =AVERAGE(C2:C1000001)
Copy this formula down to C1000001.
5. Calculate the average range in cell D1: =AVERAGE(C2:C1000001)
The results
See below a histogram displaying all the ranges of 1,000,000 datapoints, where n=2. In red the average of all these ranges = d2.
Using this Monte-Carlo simulation, I got for n=2; d2 = 1.12832
The published reference for d2 to 6 decimal places is 1.12838, thus the difference between my calculated and the published version is less than 0.006%. I think that is acceptable! ;)
I hope you have fun replicating this experiment!
Excel - distributions for different number of samples in your subgroup
Below are 3 additional histograms for n=3,4,5.
n=3, d2=1.693 (I got 1.694)
领英推荐
n=4, d2 = 2.059 (I got 2.057)
n=5, d2=2.326 (I got d2=2.326!) ;)
Minitab - How do I calculate d2?
2. Use the calculator function for n=2...10 to calculate the ranges
3. Finally, use the graphical summaries to assess d2 for each case
Here are the graphical summaries, have a look at the shape, centre and spread, then have a look at the mean value (d2), the p-value for the Anderson-Darling test for normality, the median and then the plots at the bottom. What conclusions do you draw from this?
Minitab - graphical results
If you liked this, read and watch these
If you like articles, webinars, podcasts or vodcasts, listen/watch/read these: