Bootstrapping: Which values should be used for cases/samples
Posted: Fri Mar 15, 2013 8:40 am
Hi all,
I'm actually working for my master thesis on a study with N=2947.
In some papers I read that in relation to Bootstrapping sample size should be as high as valid observations, but better 5.000 and that cases size should at least equal N.
Is this right? So I should use sample: 5.000 and cases 2.900? Problem is it takes my computer a lot of time to work... but I think it is important because I experimented a little bit (sample 200, cases 100 and sample 1000 cases 500) and in the latter version 3 more path coefficients were significant so that exerts quite an influence on the results...
Can somebody help me solve that out? Thanks a lot ahead...
I'm actually working for my master thesis on a study with N=2947.
In some papers I read that in relation to Bootstrapping sample size should be as high as valid observations, but better 5.000 and that cases size should at least equal N.
Is this right? So I should use sample: 5.000 and cases 2.900? Problem is it takes my computer a lot of time to work... but I think it is important because I experimented a little bit (sample 200, cases 100 and sample 1000 cases 500) and in the latter version 3 more path coefficients were significant so that exerts quite an influence on the results...
Can somebody help me solve that out? Thanks a lot ahead...