Would you trust more in the asymptotic distribution of the test statistic (in red) or in the other one (in black). The answer to this question is quite important, since the observed test statistic takes a particular value (green vertical line) which makes the asymptotic and the exact p-values to be very different.
Think about it...
Permutation power.
Checca
i am puzzled why your simulated null takes negative values?
ReplyDeleteBustra
Bustra, it just the estimate of the null distribution obtained with plot(density(x)) which takes negative values... come on...
ReplyDeleteI miss you all!!!!!!!!!!!!
Checca
So it is not permutations power but kernel density estimation power....
ReplyDeletethink about it!
davide.
.... no comment ....
ReplyDeleteChecca
Hi checca,,,
ReplyDeletethis is my topic,
I think in order to compare b\w test statistics you need to draw what so-called ROC curve.
Have a nice work
I miss you too
Ciao
Thanks Monjed!!! What could we do without you!
ReplyDeleteCiao!!!
Checca
I think the last dot should be a question mark and not an exclamation one... anyway, the more interesting question is: does anyone know how to have as output of "density(x)" a function and not a list of points?
ReplyDeleteThanks!
Hi Checca,
ReplyDeleteSometimes I have mind, I think.
Good work
no go capio un casso de biass...
ReplyDeletebrava checca! linux a cannone!! passami la pswd del calculus belga!! :-)