Re: Help, please ...

slippymississippi_at_yahoo.com
Date: 01/28/05 

Date: 28 Jan 2005 08:41:28 -0800

> It is a beta distribution with parameters 20 and 13.
> The probability that p > 0.5 for this distribution is 0.8923

That's the number I arrived at using the assertion that "if p hat is
used as an estimate of p, we can be at least (1-alpha) confident that
the error will not exceed a specified amount E when the sample size is
n = [z(alpha/2) ^ 2] / (4 * E^2)" (page 240 in Walpole/Myers "Stats for
Dummies" err... I mean "Probability and Statistics for Engineers and
Scientists").

Given a maximum error of 0.11 with 31 samples, if you solve for z and
look up the result on the table, you get ~.89 ... however, since z is
calculated for 1/2 alpha, then your actual confidence is 78%, not 89%,
right?

Sorry for all the mental cobwebs, I've been away from stats for nearly
20 years.

> Also, Usenet newsgroup is different from a chat room.

I'm considering making the switch to barbless hooks. Thanks for your
help!