Generally you should do the power analysis before collecting any data.
Since you have results it looks like you already have the data
collected.
But if you want to compute the power for a future study, one option is
to use simulation.
1. decide what the data will look like
2. decide how you will analyze the data
3. simulate data and analyze it based on 1 and 2
4. repeat step 3 a bunch of times, the proportion of "Significant"
results is your estimated power
Here is an example to use as a starting point:
simfun <- function(nblocks=10, means=c(0,0,0), within.sd=1, between.sd=1) {
g <- factor(rep(seq_len(nblocks), each=length(means)))
t <- rep(seq_along(means), nblocks)
y <- means[t] + rnorm(nblocks, 0, between.sd)[g] + rnorm(length(g),
0, within.sd)
friedman.test(y ~ t | g)$p.value
}
# test size of test
out <- replicate(1000, simfun())
hist(out)
mean(out <= 0.05)
# now for power
out2 <- replicate(1000, simfun(nblocks=25, means=c(10, 10.5, 11)))
hist(out2)
mean(out2 <= 0.05)
On Thu, Apr 18, 2019 at 1:40 PM George Karavasilis <gkaravas at
ee.duth.gr> wrote:>
> Hello,
> I am running a non parametric repeated measures experiment with
> Friedman?s test:
>
> Friedman rank sum test
>
> data: glikozi and week and subject
> Friedman chi-squared = 18.538, df = 3, p-value = 0.0003405
>
> How could I run a power analysis for this test in R?
> Thank you!
>
> --
> George Karavasilis
> Department of Business Administration &
> Serres, Greece
>
>
> ---
> ???? ?? e-mail ????????? ??? ???? ??? ?? ????????? Avast antivirus.
> https://www.avast.com/antivirus
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
--
Gregory (Greg) L. Snow Ph.D.
538280 at gmail.com