[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: RE: RE: distribution of the kernel density
Lee Sieswerda <Lee.Sieswerda@tbdhu.com>
> Nick is bang on about the unsuitability (non-suitability?
> of testing the kernel density for the purposes of making conclusions
> the original variable. After all, the kernel density is a
> construct based on a host of assumptions that determine its form.
> to suppose that a test of normality on the values of the kernel
> would have much to say substantively on the subject of the normality
> original variable would be erroneous. But, suppose, as I do, that
> Alejandro's reason for testing the kernel density isn't to make
> about the normality of the original variable, but wants to determine
> close the kernel density approaches the normal distribution. Then,
> he be able to gather some useful information from -qnorm-
Alejandro can -- indeed should -- answer for himself, but I doubt that
is what he wanted. If it is, then -qnorm- might be useful graphically,
but for any significance testing, there are further difficulties
those I mentioned.
What is the sample size? The number of original values, or the number
points at which the density is estimated? Even more fundamentally, the
estimates of the density are themselves not independent, as in
be useful the kernels around each data point must overlap. That alone
would seem to invalidate any P-values from (say) -sktest-.
* For searches and help try: