Psychometrics with small datasets?


I am developing a new measurement scale for cannabis withdrawal, but only have 50 people in my dataset (they filled the scale in every day for 3 weeks - 1 week was a baseline smoking as usual phase, the other two were abstinence from cannabis).

I want to do a full psychometric development study - but dont think I can do factor analysis with such small numbers....but what can I do to validate and test the reliability of my new scale with these small numbers?