🚀 go-pugleaf

OOps!  Please ignore my previous message...

On looking again John Wendel's query I see that the first time I misread it
completely.  Sorry.

However the right way is not too dissimilar.  By sufficiency you can view
it as a sample of 200 vouchers, of which 10 are in error, and then find the
value of p that puts your test statistic at the appropriate point of its
null distribution.  I get p = .0772 or p = .0833 as the one sided 95% upper
limit, depending on how you define the rejection region.

If you assume that  p  is fixed conditional on the batch, but random between
batches, then that's a different problem, and one for which to my knowledge
there is no exact procedure in any sense.  The odd looking data, with
nearly all 0 or 1 but also a 4 and a 5, gives just a hint that something
like this might be going on, though.

Sorry for the wasted message.  I'm so embarrassed you can be sure I'll be
more careful next time.

--
Bill Venables, Dept. of Statistics, | Email: venables@stats.adelaide.edu.au
Univ. of Adelaide, South Australia. | Tel: +61 8 228 5412  Fax: ...232 5670