## Abstract

It is common to compare the difference of two means as discussed in Procedure 39 (*t*-test). Sometimes it is desirable to compare the *ratio* of the means; for example in computing potency ratios. Discussed here is the computation of confidence limits for the ratio of two independent means (mathtype) and(mathtype). The observed ratio is \({{{{\bar x}_1}} \mathord{\left/
{\vphantom {{{{\bar x}_1}} {{{\bar x}_2}}}} \right.
\kern-\nulldelimiterspace} {{{\bar x}_2}}}\), where the numerator is derived from *N*_{1} values of *x* and the denominator from *N*_{2} values of *x*. The samples are assumed to be independent and the values *x* normally distributed. The means \({\bar x_1}\) and \({\bar x_2}\) are also normal with variances *V*_{ 1 } and *V*_{ 2 } and covariance of zero.

### Keywords

Covariance## Preview

Unable to display preview. Download preview PDF.