BookmarkSubscribeRSS Feed
Choose Language Hide Translation Bar
Evan_Morris
Community Trekker

Accessing Control Chart constants/unbiasing constant

I'm doing a variance analysis and need to look at the unbiased variance estimates for a number of subgroups I'm looking at.  I can manually calculate them using a d2 table I have, but I was hoping there was a way to put that into a formula.    I know that it can be derived from the Chi-squared distribution, but I don't have the math for that.   

 

Any help on either a direct way to extract d2, or the math behind manually building it would be helpful.

 

Thanks

 

Evan Morris

0 Kudos
1 ACCEPTED SOLUTION

Accepted Solutions
ian_jmp
Staff

Re: Accessing Control Chart constants/unbiasing constant

Haven't seen that d2 thing in a while!

// https://v8doc.sas.com/sashtml/qc/chapc/sect9.htm

NamesDefaultToHere(1);

calculateD2 =
Function({n}, {Default Local},
	d2 = Round(Integrate(1 - Power(1 - NormalDistribution(x), n) - Power(NormalDistribution(x), n), x, ., .), 5);
	);

for(n=2, n<=25, n++, Print(n, calculateD2(n)));
6 REPLIES 6
Highlighted
gzmorgan0
Super User

Re: Accessing Control Chart constants/unbiasing constant

Are you looking for the variance of a subgroup?

https://en.wikipedia.org/wiki/Variance#Sample_variance

 

That will be the square of the coumns standard deviation the formulas is  (Col Std Dev( col, by1, by2, ...))^2

 

Run the script below.  You can also find this using Table> Summary.  

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/Big Class.jmp" );
dt << New Column( "Standard Variance for Each Age and Sex Group",
	Formula( (Col Std Dev( :height, :age, :sex ))^2 )
);
0 Kudos
Evan_Morris
Community Trekker

Re: Accessing Control Chart constants/unbiasing constant

No, I'm looking for the unbiased estimate of the population standard deviation (sigma-hat), that can be calculated from the subgroup sample standard deviation (s) and an unbiasing constant (either d2 or c4 depending on sample sizes and how I'm using them).

 

So lets say I have 30 subgroups of the same size (n=4).  I can calculate S-bar by averaging the within subgroup standard deviation.  I can then divide the s-bar by c4 and determine the unbiased estimate of the population standard deviation.

 

My problem is that I have varying subgroup sizes, so I can't just look it up once and put it in the table.  I'm also a bit squirelly on whether or not I need d2 or c4 (but that's my problem, I'll sort it out).  

 

Either way, I was just hoping those parameters were easily extractable by a function in JSL, they are pretty common in control charting/SQC work.

0 Kudos
Evan_Morris
Community Trekker

Re: Accessing Control Chart constants/unbiasing constant

Actually, let me update my question a bit.   Is the value reported int he variance components section of the variability gauge unbiased?  That may be sufficient for my work (although I would still prefer access to the constants)

0 Kudos
P_Bartell
Contributor

Re: Accessing Control Chart constants/unbiasing constant

Have you taken a look at the JMP documentation Statistical Details section for the specific control charts you are using? Generally, JMP documentation provides background detail for specific questions such as yours in the Statistical Details section. Here's the section for X bar and S charts in the JMP online documentation:

 

https://www.jmp.com/support/help/14-2/control-limits-for-xbar-and-s-charts.shtml#1149056

 

0 Kudos
ian_jmp
Staff

Re: Accessing Control Chart constants/unbiasing constant

Haven't seen that d2 thing in a while!

// https://v8doc.sas.com/sashtml/qc/chapc/sect9.htm

NamesDefaultToHere(1);

calculateD2 =
Function({n}, {Default Local},
	d2 = Round(Integrate(1 - Power(1 - NormalDistribution(x), n) - Power(NormalDistribution(x), n), x, ., .), 5);
	);

for(n=2, n<=25, n++, Print(n, calculateD2(n)));
Evan_Morris
Community Trekker

Re: Accessing Control Chart constants/unbiasing constant

This is exactly what I need, thanks.   

0 Kudos