I have a cross-lagged model with 2 observed variables at 2 time points. It's a saturated model and one cross-lag is significant and the other one is not. I wanted to test if these two where different in strength, so I constrained both and got a non-significant drop in chi2. My question is, should I keep them constrained and conclude that they are not significantly different from each other?
Also I have heard that one could ask for critical ratios of differences and look at the t-value that one gets in the output to see which paths are significantly different from each other. I couldn't find any information about this in the Mplus manual. What should I ask for in the output part?
You can do a 2-group analysis if the 2 samples are independent (not the same subjects at 2 time points for instance).
Rhyan posted on Wednesday, May 23, 2018 - 11:57 am
Thank you. I did a multi group analysis and got the results (and questions) below.
The unconstrained model had acceptable (but not great) fit with RMSEA 0.051 and CFI 0.902. The model with all paths and covariates constrained had unacceptable fit with RMSEA 0.047 and CFI 0.877. Does it mean something when the model fit changes from acceptable to unacceptable?
Also, the Chi2 comparison shows the models in the 2 groups (Group A and Group B)are significantly different. When looking at the parameter estimates in the 2 groups, we see that the estimates are all in the same direction, but the estimates are higher in Group B. Is it okay to state that the parameters are different but for the purpose of assessing the model in a manuscript, we only present Group B results (as opposed to presenting the results as a single sample or presenting both Group A and Group B results)?
Hi there. I have a question about comparing path estimates. Basically, I have 8 models with the same outcome variable across the 8 models, but different predictors in each model. There are 5 time points of data for both variables in each model. I have constrained the cross-lagged path from the predictor to the outcome variable to be the same at each time point, and so I essentially have a cross-lagged path estimate of interest from each model.
In summary, I have 8 cross-lagged path estimates from 8 different models for the same outcome variableÖ.and I want to compare them. Could you explain how best to go about doing so? Iíve seen two approaches in the literature which donít actually seem to be widely used: 1) The Cumming approach of testing for 50% overlap in the confidence intervals of the standardized regression coefficients, and 2) The Clogg et al. (1995) approach and calculate z scores from the standardized regression coefficients and their standard errors.