0
$\begingroup$

This question is about the difference between R's pwr::pwr.f2.test and G*Power regarding the estimation of a required sample size based on given power (a priori power analysis).

In G*Power, you can perform the power analysis for the whole model with Test family = F tests and Statistical test = Linear multiple regression: Fixed model, R2 deviation from zero.

enter image description here

The equivalent in R is the pwr.f2.test function:

# u = number of predictors
u = 3

# determine v
library(pwr)
(v <- pwr.f2.test(u = u, v = , f2 = 0.25, sig.level = .05, power = .80)$v)
[1] 43.70444

# Determine required sample size
ceiling(v + u + 1)
[1] 48

Same results as G*Power. So far so good.

But if, instead of testing a whole model, I want to test a single predictor, in G*power I would go with Test family = F tests and Statistical test = Linear multiple regression: Fixed model, R2 increase. Notice that there now is a new box called Total number of predictors.

enter image description here

So what is the equivalent R function for G*Power's R2 increase variant? I don't see how to specify Total number of predictors in pwr.f2.test. Is it even possible? What workaround do you guys use? Does it even matter from a statistical point of view?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.