I’m trying to run an autoregressive cross-lagged model with two time-series variables measured at 3 timepoints. One is continuous(M1-M3) and the other(TR1-TR3) is binary(I set TR2 and 3 to be categorical), and both variables have missing at some points. The model is as below.
ANALYSIS: TYPE is general; Estimator is MLF; integration = montecarlo;
Model: M_1 on TR1; M_2 on TR2; M_3 on TR3; M_2 on M_1; M_3 on M_2; TR2 on TR1; TR3 on TR2; TR2 on M_1; TR3 on M_2;
1. Mplus do not provide fit indices such as chi-square, TLI, CFI, and RMSEA for this model.In this case, how can I judge the goodness of fit of this model? Is there any alternative way to see its fit, instead? (not comparison with other models)
2. Second, I was not able to obtain the standardized coefficients, as well.It seems that Mplus suggests of models with categorical variables(using TECH10).But I got a message like this.
TECH10 OUTPUT FOR CATEGORICAL VARIABLES IS NOT AVAILABLE FOR MODELS WITH COVARIATES.
Is there no way to compare the effect of two variables? for example, TR2 on TR1 and M_1(like logistic reg) Thank you for reading.
steve posted on Wednesday, March 07, 2018 - 9:02 am
I have a question concerning the interpretation of coefficients in a CLPM with binary variables at two timepoints.
X1-X2 are continuous, Y1-Y2 are a binary variables; Z is a covariate.
X2 on X1 Y1 Z; Y2 on Y1 X1 Z;
Y1 on Z; X1 on Z;
X1 with Y1; X2 with Y2;
My question: Do both "with" statements reflect the correlated residuals between X and Y at Time 1 and Time 2, respectively (because the variables are endogenous at both measurements)? And indicates the correlation between X2 and Y2 the common change in the variables? Or am I wrong since Y1-Y2 are binary variables?
Q2: I don't know what you mean by "common change" but residual covariance don't reflect change.
Q3: See answer under Q1.
steve posted on Thursday, March 08, 2018 - 12:56 am
Thank you very much for your quick response.
Common change or correlated change is a concept in personality research. It indicates the degree to which changes in one measure (e.g., trait) are related to changes in another measure. According to the literature, this is indicated from the correlation between the residuals of variables at Time 2 (when controlling for autoregressive and cross-lagged effects, e.g., Klimstra et al., 2013; Neyer & Asendorpf, 2001).