From: Dennis Shea <shea_at_nyahnyahspammersnyahnyah>

Date: Tue Feb 22 2011 - 15:17:12 MST

Date: Tue Feb 22 2011 - 15:17:12 MST

You do not put the significance level directly

into the incomplete beta func

First you calculate the t-statistic

t = r*sqrt((Nr-2)/(1-r^2))

See for example the Description section of:

http://www.ncl.ucar.edu/Document/Functions/Built-in/rtest.shtml

NCL follows the same approach outlined in "Numerical Recipies"

in the "Statistical Description of Data" [Linear Correlation]

On 02/22/2011 03:03 PM, eddycarl wrote:

*> Hi there,
*

*>
*

*> I tried to use the betainc function in NCL to determine the significance
*

*> level of a linear regression analysis. My question is that: what the
*

*> inputs a (the first beta distribution parameter) and b (the second beta
*

*> distribution parameter) mean? And, how do they affect the significance
*

*> level? More specifically, in the example 2 of
*

*> http://www.ncl.ucar.edu/Document/Functions/Built-in/betainc.shtml, where
*

*> the significance level is 5%, b is set as 0.5. If I want 1% significance
*

*> level, do I need to change parameter b?
*

*> Thanks!
*

*>
*

*> -Eddy
*

_______________________________________________

ncl-talk mailing list

List instructions, subscriber options, unsubscribe:

http://mailman.ucar.edu/mailman/listinfo/ncl-talk

Received on Tue Feb 22 15:17:17 2011

*
This archive was generated by hypermail 2.1.8
: Wed Feb 23 2011 - 16:47:57 MST
*