ExcelBanter - View Single Post

Ian Smith[_2_]

(Heather Rabbitt) wrote in message . com...
Thanks - just wondering how do I account for scores of zero i.e. (0%)
using this formula:

lower bound formula would be
= A1-1.96*sqrt(((100%-A1)*A1)/(B1-1))
upper bound formula would be
= A1+1.96*sqrt(((100%-A1)*A1)/(B1-1))

Regardless of the sample size I always get lower bound and upper bound
scores of zero?

Thanks again,

Heather

[snip]

Tom's formulae approximate the exact 97.5% lower/upper confidence
bounds, assuming the data is binomially distributed. The accuracy of
the approximations depends on the sample size and the percentage. They
are less accurate when the sample size is small or the percentage is
close to 0 or 1.

To get exact 97.5% lower/upper confidence bounds, assuming the data is
binomially distributed you could use
100*lcb_binomial(B1,B1*A1/100,0.025) and
100*ucb_binomial(B1,B1*A1/100,0.025).

VBA versions of these functions can be found at
http://members.aol.com/iandjmsmith/Examples.xls

Ian Smith