Message/Author 

KK posted on Monday, July 16, 2007  10:52 am



Hello. For many models 90% confidence intervals for the population RMSEA are printed. However, I cannot find an option/specification for modifying the confidence interval coverage. Is there a way to obtain a 95% and/or 99% confidence interval coverage for the RMSEA? Thanks. 


When we have knowledge of how to compute the confidence intervals, they are printed automatically. When they are not printed, it means the theory for them has not yet been developed. 

KK posted on Monday, July 16, 2007  11:34 am



Actually I'm asking about having Mplus print a 95% or a 99% confidence interval (rather than a 90% confidence interval). I'm specifically talking about cases in which the theory of confidence interval construction for the RMSEA does exist and is already implemented in Mplus (just at a different level than I would prefer). Thanks. 


No, there is no option for that. 

QianLi Xue posted on Friday, March 30, 2012  12:27 pm



To followup on the previous question, why is it that you decided to give out 90% confidence interval instead of the conventional 95% CI? 


Some of the recommendations in the literature are based on the 90% confidence interval, see for example http://davidakenny.net/cm/fit.htm "A confidence interval can be computed for the RMSEA. Ideally the lower value of the 90% confidence interval includes or is very near zero (or no worse than 0.05) and the upper value is not very large, i.e., less than .08. " 

KK posted on Thursday, April 05, 2012  7:25 am



To followup, you can obtain a confidence interval for the population RMSEA using the MBESS R package with the ci.rmsea function as follows: require(MBESS) ci.rmsea(rmsea=.07, df=42, N=225, conf.level=.95) which returns the following: $Lower.Conf.Limit [1] 0.04498848 $RMSEA [1] 0.07 $Upper.Conf.Limit [1] 0.09416656 Obviously, by adjusting conf.level you can obtain any desired confidence interval. For more flexibility, alpha.lower and alpha.upper can be used in place of conf.level. 

Back to top 