On 28 Oct 2003 20:50:42 -0800, [EMAIL PROTECTED] (Greg) wrote:

> I want to use the bootstrap to test H0: g(x) = 0 vs Ha: g(x) > 0,
> where g() is a very coplicated non-negative function of the data. 

So, you are saying that g()  can never be negative, and
you want to test whether it is exactly zero, always.
Don't you have to reject the hypothesis whenever you 
have any single data point that is not zero?


Now, if you have an 'error function'  that tells you that 
sometimes g()  is erroneously scored as not-zero, then 
you could test against how often  *that*  should occur.
But that might be more like a binomial test.

> Using a one-sided percentile (nonparametric) confidence interval seems
> very easy, i.e. for an alpha = 0.05 test, reject H0 if g-hat > the
> 95th percentile of the bootstrap distribution of g.
> 
> My question is, am I overlooking any problems or subtleties stemming
> from the fact the g is strictly non-negative?

Not  subtle, if I am grasping it right.

-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
"Taxes are the price we pay for civilization." 
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to