Bootstrapping and small samples sizes

This forum is closed, and read-only.
Locked
bergsorensen
PLS User
Posts: 13
Joined: Wed Jun 25, 2008 10:40 am
Real name and title:

Bootstrapping and small samples sizes

Post by bergsorensen »

Iwe done a survey for a company with a small customer sample (56 observations).

Running the bootstrap procedure (cases = observations (55) and samples 500) gives me no significant relations between my latent variables (using the ECSI framework).

Is there a way to go arround this, assuming that the relations are in fact significant? Raising the case number (for example to 100 which is the standard value in smartpls) gives me significant relations, but i suppose this aint a valid way arround...

How do i test/prove the significance of my relations with such a small sample size?

Thanks in advance!
Last edited by bergsorensen on Tue Jul 01, 2008 12:02 pm, edited 1 time in total.
bergsorensen
PLS User
Posts: 13
Joined: Wed Jun 25, 2008 10:40 am
Real name and title:

Post by bergsorensen »

Anyone?
Locked