topic Re: 95% confidence intervals in Discussions

95% confidence intervals

Mickyboy — Wed, 20 Nov 2019 05:04:57 GMT

Hi All,

l am trying to get a summary from a data table, l would like the mean of a variable l10t, the standard error, and 95% confidence interval. It all seemed fairly straight forward, but am having trouble with the 95% confidence intervals, have tried different ways to no avail, currently have

dt2<< Summary(
	Group( :Sample ),
	Mean( :dilution factor, :l10t ),
	Freq( "None" ),
	Weight( "None" ),
	Output Table( "log10t" ),
	Std Err( :l10t ),
	Confidence Interval( :l10t, 0.95),
	Link to original data table(0)

l am obviously doing something incorrectly, can anyone offer some advice.

Thanks Heaps

Re: 95% confidence intervals

Mark_Zwald — Wed, 20 Nov 2019 06:30:13 GMT

You can accomplish something similar like this:

Names Default To Here( 1 );
dt = Open( "$SAMPLE_DATA/Big Class.jmp" );
platform = dt << Oneway(
	Y( :height ),
	X( :sex ),
	Means and Std Dev( 1 ),
	Box Plots( 1 ),
	Mean Diamonds( 1 ),
	Mean Error Bars( 1 ),
	Std Dev Lines( 1 ),
	Connect Means( 1 ),
	Points Jittered( 1 ),
	Grand Mean( 0 )
);
Wait( 0 );
Report( platform )[Outline Box( "Oneway Analysis of height By sex" )][
Outline Box( "Means and Std Deviations" )][Table Box( 1 )] << Make Into Data Table;
Report( platform ) << Close Window;

Place your group variable in the X() role.

Re: 95% confidence intervals

Mickyboy — Thu, 21 Nov 2019 00:51:18 GMT

Hi mzwald,

Thanks very much for your help and your suggestion, i tried this code

platform = dt2 << Oneway(
	Y( :l10t ),
	X( :fold dilution ),
	Means and Std Dev( 1 ),
	Box Plots( 1 ),
	Mean Diamonds( 1 ),
	Mean Error Bars( 1 ),
	Std Dev Lines( 1 ),
	Connect Means( 1 ),
	Points Jittered( 1 ),
	Grand Mean( 0 )
);
Wait( 0 );
Report( platform )[Outline Box( "Oneway Analysis of height By sex" )][
Outline Box( "Means and Std Deviations" )][Table Box( 1 )] << Make Into Data Table;
Report( platform ) << Close Window;

and received the following error message

Cannot subscript Display Box in access or evaluation of 'Report(platform)[Outline Box("Oneway Analysis of height By sex")]' , Report( platform )[/*###*/Outline Box( "Oneway Analysis of height By sex" )]

however, l am almost there as to what i want to achieve for my data table with this,

dt2<< Summary(
	Group( :Sample ),
	Mean( :dilution factor, :l10t ),
	Freq( "None" ),
	Weight( "None" ),
	Output Table( "log10t" ),
	Std Err( :l10t ),
	Confidence Interval( :Name ("Mean(l10t)")),
	Link to original data table(0)
);

Loq = Data Table( "log10t" );

Loq << New Column( "Back T'sform Mean", Numeric, Continuous, Formula( Round(10^:Name ("Mean(l10t)"),0)));

Loq << New Column( "Back T'sform Std error", Numeric, Continuous, Formula(10^:Name ("Std Err(l10t)"), Format( "fixed dec", 6, 0 ) ) );

giving me this,

so l am almost there, l was hoping the confidence intervals would be a single simple line of code.

Thanks

Mickyboy

Re: 95% confidence intervals

Mark_Zwald — Fri, 22 Nov 2019 19:22:08 GMT

Well it looks like Summary doesn't currently support CI's, but you can always make a formula for them yourself.
The formulas for a 95% confidence interval (assuming a normal distribution) would be:
Upper95CI = mean + 1.96*stdev
Lower95CI = mean - 1.96*stdev
You can generate those two columns from the fields returned by the Summary platform. Those formulas will get you pretty close to what you are looking for.

Re: 95% confidence intervals

Mickyboy — Tue, 26 Nov 2019 05:29:25 GMT

Hi mzwald,

i have used the following and has worked a treat

New Column( "Lower 95% CI", Numeric, Continuous, Formula ( Round(:Name ("Mean(l10t)") - (:Name("Std Err(l10t)") * t Quantile (0.975,23)), 4)));

if i have used the following to create a variable

dt2<< Summary(
	Group( :Sample ),
	Mean( :dilution factor, :l10t ),
	Freq( "None" ),
	Weight( "None" ),
	Output Table( "log10t" ),
	Std Err( :l10t ),
	Link to original data table(0)
);

and i would like to round the variable Mean(l10t), to four decimal places, can you do that from the above block of syntax???

thanks

Mickyboy