the same results in doing bootstrap PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 li weidong posted on Wednesday, September 14, 2011 - 8:45 pm
Dear Mr.Muthen,
I use Mplus 6.1DEMO to test the significance of multiple mediation. I set the bootstrap number as 1000,2000,5000. However,the Mplus outputs always report the same results as in doing normal analysis which not use bootstrap method except the S.E.is different.
Could you tell me what"s wrong with my analysis?
Thank you very much!
Best Regards!
li weidong
2011-9-15
 Linda K. Muthen posted on Thursday, September 15, 2011 - 5:36 am
We only bootstrap the standard errors. So only the standard errors will be different.
 li weidong posted on Friday, September 16, 2011 - 5:57 pm
Thank you,Linda.
It is very nice of you!
Can I report the parameter bootstrap results with other SEM software(i.e. AMOS)and corresponding standard errors with Mplus?
 Linda K. Muthen posted on Saturday, September 17, 2011 - 10:22 am
I don't think the Mplus bootstrapped standard error would be appropriate for the bootstrapped parameter estimate.
 li weidong posted on Sunday, October 16, 2011 - 6:48 pm
Hi!Linda,
I have a question which puzzled me so far.
As in Bollen & Stine(1990),the bootstrapped standard errors were generated with the average estimation of the bootstrapped unstandardized or standardized indirect effects.How can Mplus only bootstrap the standard errors without bootstrap the unstandardized or standardized indirect effects?
I do want to know how Mplus cannot report the bootstrapped unstandardized indirect effects or the standard errors of standardized indirect effects? if I report the results from Mplus as above,in which degree I can trust the results?
Many thanks to you and your team for developing this powerful analytic tool!
best regards!
liweidong
2011-10-17
 Tihomir Asparouhov posted on Monday, October 17, 2011 - 10:18 am
There are several different options in Mplus about how the bootstrap SE are computed. The basic version is this: for each bootstrap sample we estimate the model then we compute the standard deviation of these estimates. This approach is used for the model estimates, the standardized estimates and the indirect effect estimates.
 Tihomir Asparouhov posted on Monday, October 17, 2011 - 11:24 am
A nice introduction to the bootstrap methodology can be found here
http://www.amazon.com/Introduction-Bootstrap-Monographs-Statistics-Probability/dp/0412042312

The basic version of bootstrap does not affect the point estimates, only the standard errors.
 li weidong posted on Monday, October 17, 2011 - 5:48 pm
Thank you very much!
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: