Choose Language Hide Translation Bar

## Confidence in Std Deviation

Is there a way to graph Confidence (say .9) in a bunch of Std Deviations other than digging through each of their distributions?  Maybe I'm just brain-dead this morning.

*Edit*  Right now I'm just doing this but I feel like there is a better way.

x = current report();

est = (x << xpath("//OutlineBox[text() = 'Confidence Intervals']/TableBox/NumberColBox/NumberColBoxItem/text()"));

LCI = (x << xpath("//OutlineBox[text() = 'Confidence Intervals']/TableBox/NumberColBox/NumberColBoxItem/text()"));

UCI = (x << xpath("//OutlineBox[text() = 'Confidence Intervals']/TableBox/NumberColBox/NumberColBoxItem/text()"));

New Table("Sigma GRR",

new column("Estimates", Character, SetValues(est)),

new column("LCI", Character, SetValues(LCI)),

new column("UCI", Character, SetValues(UCI)),

);

Vince Faller - Predictum
1 ACCEPTED SOLUTION

Accepted Solutions

## Re: Confidence in Std Deviation

Hi, Vince!

Not sure this is what you're after, since there are many, many ways to do this, but you could always "roll your own"...

Names Default to Here(1);

// Open the dataset in question.

BigClass_dt = Open( "\$SAMPLE_DATA/Big Class.jmp" );

// Summarize the standard deviation by age.

BigClassSummary_dt = BigClass_dt << Summary(

Group( :age ),

Std Dev( :height ),

output table name("Big Class Summary"));

// Set alpha as a table variable.

BigClassSummary_dt << Set Table Variable("alpha", 0.10);

// Calculate the Confidence Intervals

BigClassSummary_dt << New Column( "LCI",

numeric, continuous,

Formula( Root(((:N Rows - 1) * :Name( "Std Dev(height)" ) ^ 2) / ChiSquare Quantile( 1 - :alpha / 2, :N Rows - 1 ), 2 )),

eval Formula);

BigClassSummary_dt << New Column( "UCI",

numeric, continuous,

Formula( Root( ((:N Rows - 1) * :Name( "Std Dev(height)" ) ^ 2) / ChiSquare Quantile( :alpha / 2, :N Rows - 1 ), 2 ) ),

eval Formula);

// Revel in your fly statistical swerve!

5 REPLIES 5

## Re: Confidence in Std Deviation

Hi, Vince!

Not sure this is what you're after, since there are many, many ways to do this, but you could always "roll your own"...

Names Default to Here(1);

// Open the dataset in question.

BigClass_dt = Open( "\$SAMPLE_DATA/Big Class.jmp" );

// Summarize the standard deviation by age.

BigClassSummary_dt = BigClass_dt << Summary(

Group( :age ),

Std Dev( :height ),

output table name("Big Class Summary"));

// Set alpha as a table variable.

BigClassSummary_dt << Set Table Variable("alpha", 0.10);

// Calculate the Confidence Intervals

BigClassSummary_dt << New Column( "LCI",

numeric, continuous,

Formula( Root(((:N Rows - 1) * :Name( "Std Dev(height)" ) ^ 2) / ChiSquare Quantile( 1 - :alpha / 2, :N Rows - 1 ), 2 )),

eval Formula);

BigClassSummary_dt << New Column( "UCI",

numeric, continuous,

Formula( Root( ((:N Rows - 1) * :Name( "Std Dev(height)" ) ^ 2) / ChiSquare Quantile( :alpha / 2, :N Rows - 1 ), 2 ) ),

eval Formula);

// Revel in your fly statistical swerve!  Steven_Moore
Super User

## Re: Confidence in Std Deviation

Why not use the "bootstrap" capability for the statistic?

Steve

## Re: Confidence in Std Deviation

Yeah, smoore2, bootstrapping is one of the "many, many ways" to do this, and it has some advantages.

I wish JMP had implemented a better approach to their bootstrap estimation.  I ask Santa for this every Christmas!

I had to write an addin to get the Bias Corrected accelerated (BCa) estimations from the R "boot" package, which is a more defensible method, in my opinion.

Maybe if more individuals lifted their voice in unison, Santa would hear us??

Highlighted

## Re: Confidence in Std Deviation

Please see Efron, B. (1987), "Better Bootstrap Confidence Intervals", Journal of the American Statistical Association, 82:397, 171-185

## Re: Confidence in Std Deviation

Most of our licenses here are not JMP Pro.

Vince Faller - Predictum