[gradsusr] P-value of scorr & tcorr

[tcavazos at cicese.mx]

Hi,

A simple way for tcorr: look at the statistically significant values for
correlation in a statistics book; then according to your n-1 observations 
get the threshold values for the level of interest (p<0.05 or p<0.01).

In grads you you can black out the values lower than those thresholds to
display only the significant ones. Also you can set a minimum value to
plot. For example, if your significant correlation at the 95% level is
0.45:

"set cmin 0.45"

By the way how do you calculate the scorr, do you only average the spatial
window after calculating the tcorr?

The spatial significance is much more complicated...I am also interested
in that, if anyone has it. It may require a bootstrap...

Tereza Cavazos


> Dear gradsusrs,
>
> I have a question about scorr and tcorr functions. Is there a way to
> obtain
> the p-value in this cases? i.e. one value for scorr and map value for
> tcorr
>
> Or anything of the sort?
>
> Regards,
>
> James
> _______________________________________________
> gradsusr mailing list
> gradsusr at gradsusr.org
> http://gradsusr.org/mailman/listinfo/gradsusr
>