Very small sample size

#1
Hi there,

I have a very small sample size n=8 in which I have the estimated survival date of each individual and 5 behavioural traits which I would like to correlate with their survival. Can I run an regression model with bootstrap? Is it possivel to run a cox regression?

Thanks,
Silvana
 
#3
Is there any analysis that will make sense with this data? Even if I select one of the variable that I hypothesize as most strong for the model or run a PCA? Or shall I throw this data away as it will be useless for any kind of analysis...
 
Last edited:

noetsi

Fortran must die
#7
I think a basic issue, which tends to get ignored in a lot of analysis since it is not a formal statistical requirement is whether you can generalize to anything with just 8 cases. This deals with external not internal validity.