differrence between standard error and standar deviation

chinasun · Post by **chinasun** » Wed Feb 18, 2009 2:19 pm

hi all
who can explain the difference between bootstrap standard error and standard deviation ? in the outcome of bootstrap ,standard error have the same value with standard deviation. which one i should get into research outcome?
thanks in advance!

cringle · Post by **cringle** » Wed Feb 18, 2009 6:40 pm

Hi,

these bootstrapping results are equivalent (I quoted a good link, when this question was asked previously but cannot find it right away).

Best regards
Christian

chinasun · Post by **chinasun** » Thu Feb 19, 2009 7:28 am

thank you professor!!

pufall · Post by **pufall** » Mon Apr 06, 2009 12:51 pm

Hi chinasun - in case you have found the link could you pleaes quote it here? I didnt find the link unfortunately.

Post by **admin** » Mon Apr 06, 2009 1:04 pm

Well,

try harder next time ;) ! viewtopic.php?t=151

These links may provide the information you are looking for:

http://www.uvm.edu/~dhowell/StatPages/R ... pping.html

http://www.stat.wisc.edu/~larget/math496/bootstrap.html

Best regards
Christian

pufall · Post by **pufall** » Tue Apr 07, 2009 9:41 am

Thanks, although this is not answering the question; refer to the comment which I have found under the link you quoted:

"Theory tells us that the standard error of the sample mean equals the population standard deviation divided by the square root of the sample size. In cases in which the population size is unknown, we may use the sample standard deviation instead of the population standard deviation."

So why is the calculation of the std error not taking the sample size into account (which should be the number of cases selected for bootstrap - as this represents the sample size)?

Could this be an error of the smartPLS SW?

cringle · Post by **cringle** » Tue Apr 07, 2009 10:02 am

In bootstrapping, the standard deviation of the distribution would be the standard error.

As you can see in the SmartPLS results tables, this kind computation was implemented into SmartPLS to get the t-value:
w / se

If you do not "like" this kind of "standard" computation for any reason, you can just take the bootstrap results for each subsample (just look into the results tables) and compute the t-value as you prefer