Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am aware that multivariate non-normality in SEM can inflate the chi-square statistic and deflate standard errors of parameter estimates. I can deal with the latter (in AMOS) using bootstrapping procedures. But here's my specific question: while I can't correct for chi-square inflation in AMOS (as far as I am aware), does this inflation matter if I am mostly interested in comparing several models using the chi-square change, each of which will presumably be inflated in the same manner, rather than simply testing whether a single model is well-fitting? Moreover, my best fitting model fits well by all sensible criteria, so the chi-square inflation per se is not a major concern.

As this query implies, I use AMOS and haven't currently got access to M-Plus or another similar package to apply the correction to chi-square inflation they provide.

share|improve this question

This won't work, the models rarely get inflated in the same way. The proper analysis should involve Satorra-Benter scaled difference. I am not sure as to what the bootstrap analogues would be, although I am sure something can be constructed along the lines of the Bollen-Stine bootstrap (which should have become known as Beran-Srivastava bootstrap). I don't know if AMOS allows enough flexibility to construct the appropriate bootstrap scheme for the nested model testing, although I am sure it can run the bootstrap for the overall test.

share|improve this answer
Hi @StasK: Thanks for those comments. I didn’t realize that a Bollen-Stine bootstrap could be produced by AMOS (good to learn that). So, can one use the mean chi-square value produced by the bootstrapping for each model to then compare the models? For instance, one could then calculate the probability of the chi-square change in R to distinguish models? And is the mean chi-square value from the bootstrapping - along with the corresponding p-value - the figure to report in a manuscript (i.e. in the same way one would for a regular chi-square)? – Gary Sep 12 '12 at 17:33
I am afraid you are all mixed up about what the Bollen-Stine bootstrap does. Did you read the actual paper? – StasK Sep 12 '12 at 20:34

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.