Difference between Bootstrapping and jackknife resampleing

[rpalrecha]

Hello,

What is the difference between bootstrap and jackknife / blindfold?

I read that jacknife re-sampling has been largely replaced by bootstrap.

What criteria should be discussed to decide between these two techniques?
Does sample size matter?

Thanks
Rita

[viswadatta]

Both are methods to find out path significances.
Assume sample size is 50
In boot strap
We select any one case., record it and replace it back into the data set. Then we pick another and do the same. We pick 55 cases after replacing each case. Some cases get repeated. This constitutes one resample case. We resample at least 200 times. According to central limit theorem the mean of means is normally distributed about the population mean. This is used to do a t-test to decide path significances

In jack knife
Each resample is decided by deleting 'k' cases from the original sample. We take at least 200 resamples. The procedure for path sgnificances is then similar to bootstrap. The process is most accurate when k=1

[rpalrecha]

Thanks.

I have two more questions. Is there a criteria list for choosing bootstrapping or jackknifing procedure? My question is specifically related to small sample size data sets (n=86). What will be the implications for formative, reflective, and mixed models (with both formative and reflective LVs).

Thanks in advance.

Rita

[viswadatta]

If you want to check path significances, you definitely need either bootstrap or jack knife. It is the resample size(at least 200) that matters and not raw sample size. Raw sample size should be determined using g-power software.