On Wed, 31 Oct 2001, Glen Barnett wrote, in response to my comment:
> > On Sun, 28 Oct 2001, Melady Preece wrote:
> >
MP> Hi. I want to compare the percentage of correct identifications (taste
MP> test) to the percentage that would be correct by chance 50%? (only two
MP> items being tasted). Can I use a t-test to compare the percentages?
MP> What would I use for the s.d. for by chance percentage? (0?)
>
DB> Standard comparison would be the formal Z-test for a proportion; see
DB> any elementary stats text. If you have a reasonably large sample size,
DB> use the normal approximation to the binomial; if you have a small
DB> sample, it may be necessary to use the binomial distribution itself,
DB> which is considerably more tedious unless you have comprehensive tables.
>
DB> Sounds as though you'd wish to test H0: P = .50 vs. H1: P <> .50.
>
GB> I'd kind of expect them to want this one to be one tailed - it would
GB> seem strange to be interested in the circumstance where tastebuds do
GB> worse than chance (well, it'd be kinky and fun, but would it change
GB> your action from no difference? I can conceive of it, but I'd bet not.)
You may be right; but this surely depends on a number of details not
supplied by the querent in the original post. If the thing(s) tasted are
items with a familiar taste (chocolate, strawberry, vinegar, ...) I might
be inclined to agree with your predisposition. But if the tastes to be
encountered are unfamiliar ones for which one has no prior referent(s),
there might be no reason whatever to expect one's tasting mechanism(s) to
perform better than "chance" (whatever that is in context -- I commented
on this point earlier), at least not until after they'd been trained on
the new tastes. In any case, on a preliminary analysis, I would
routinely recommend (and carry out) two-sided tests. Unreasonable, in
my view, to deliberately blind oneself to the unexpected before the
latter has had an opportunity to leap out of the undergrowth where it's
been lurking and bite one's ankle...
-- Don.
------------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
184 Nashua Road, Bedford, NH 03110 603-471-7128
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
