Sample size for linear regression?

#1
I have two groups of participants: one group with eight people and one group with nine people.

Can I run a linear regression?

I am looking to see which group is more likely to complete a study intervention.
 

obh

Active Member
#2
Hi QFD,

What do you want to check?
Do you have a model?
How many predictors? what are the predictors?

For example, the Green's rule of thumb for a medium effect to test the entire model: (under the assumption you know the model)
sample size = 50 + 8*predictors
 

Karabiner

TS Contributor
#3
A linear regression requires an interval scaled dependent variable.
"Completion of study intervention" yes/no is a categorical variable.
Association between group membership and a binary categorical
variable can be analysed using Fisher's exact test (because sample
size is very low, Chi² does not seem suitable).

With kind regards

Karabiner