RE: [R] Power w/ unequal sample sizes

From: Bashir Saghir (Aztek Global) <>
Date: Fri 20 May 2005 - 23:36:03 EST

It seems although your are trying to do a retrospective power calculation - not something to be encouraged.

I don't think that power.t.test was designed to work on observed data in the way you seem to want to use it. You could use power.t.test to do a prospective calculation where you know the difference (delta) and the sd...


-----Original Message-----
[] On Behalf Of Jim BRINDLE Sent: Thursday, May 19, 2005 16:39
Subject: [R] Power w/ unequal sample sizes


I am hoping someone could shed some light on power calculations for me. I have two small data sets of unequal sample size after NA removal (m = 5, f =


m <- c(2.0863, 2.1340, 2.1008, 1.9565, 2.0413, NA, NA) f <- c(1.8938, 1.9709, 1.8613, 2.0836, 1.9485, 2.0630, 1.9143)

In a R help message/reply from Sep 30, 2001, it was noted that the "power.t.test" function assumes equal group sizes and that the groups have the same theoretical standard deviation. In "analyzing" this data, I ran a Welch Two Sample t-test and a Wilcoxon Rank Sum test on the data sets and both tests reveal a slight statistical difference for alpha = 0.05 (Welch Two Sample t-test p-value = 0.045 and Wilcoxon Rank Sum test p-value = 0.048).

I suspect that the "power" of these tests will be quite low but I am trying to quantify it. Based on the insight in the R help message from 2001, I am not sure how to go about this with R. Is this feasible with R or is there another approach I should be considering altogether?

Any insight would be most appreciated. Thanks a million.... mailing list PLEASE do read the posting guide!

Legal Notice: This electronic mail and its attachments are i...{{dropped}} mailing list PLEASE do read the posting guide! Received on Fri May 20 23:52:38 2005

This archive was generated by hypermail 2.1.8 : Fri 03 Mar 2006 - 03:31:57 EST