Re: null hypothesis always false?

S. F. Thomas (sthomas@decan.com)
Tue, 2 Jul 1996 13:34:44 +0200


John Whittington wrote:

> Yes, but as I often say, there is some inherent 'fuzziness' built into
> real-world applications of hypothesis tests, even when they have 'point
> nulls'. For a real-word test, with finite alpha and beta, there will be a
> *range* of values of an effect estimate which will result in failure to
> reject the null. Does this not effectively produce the 'fuzziness' that is
> needed?

But the fuzziness rears its head prior to that, at the measurement stage,
precisely where the paradigm in use asserts the fiction of point
measurement correct to an infinite number of decimal places. Sensible
people of course make the correction for the ideal approximation, but
*outside* of the formalisms actually in use. If one changes the
paradigm at the start to reflect the fuzziness in the data, the fundamental
changes required in the whole inferential setup can be addressed at
the outset. We may find that we can do without the whole notion of
null hypothesis testing in the first place.

> John
>
> -----------------------------------------------------------
> Dr John Whittington, Voice: +44 1296 730225
> Mediscience Services Fax: +44 1296 738893
> Twyford Manor, Twyford, E-mail: johnw@mag-net.co.uk
> Buckingham MK18 4EL, UK CompuServe: 100517,3677
> -----------------------------------------------------------

Regards,
S. F. Thomas