Re: variance across 2 independent experiments

Thanks Rich

That is close.


>>>>Are you saying that b is not a replicate of a? or
that a(2) is not a replicate (or paired value) of a(1)?

>>>>This *seems* to show that the data are paired, in which
case you can look at the correlation, or (equivalently)
the variation between the paired items.



That's correct a(1) - a(2) ..... (j1) - (j2) are replicates across
the two experiments and the variances are unequal for
a's, b's .....

So from what you say, ANOVA does not suit my purpose and
that I need some sort of transformation. The next thing that bites
me is how do I find which transformation I need and what to to do next.

Many Thanks
josh

.

Relevant Pages

  • Re: Formal test for pooled variance?
    ... >> reason to assume equal variances. ... >If the variances are equalized by a rational transformation, ... Herman Rubin, Department of Statistics, Purdue University ...
    (sci.stat.edu)
  • Re: Formal test for pooled variance?
    ... equality of variances is not all that matters. ... transformation is needed, either ... Both t-tests then perform badly in one-tailed testing, ... a little bit worse for the pooled test than the "separate" test. ...
    (sci.stat.edu)