I am running a growth model. I measured student achievement scores in fall and spring for 3.5 years but the data are time-unstructured. I want separate trajectories for the months in school and for the months in summer.
Here is an excerpt from my code Model: iw scw| RIT_0-RIT_6 AT tsc_0 tsc_1 tsc_2 tsc_3 tsc_4 tsc_5 tsc_6; suw| RIT_0-RIT_6 AT tsu_0 tsu_1 tsu_2 tsu_3 tsu_4 tsu_5 tsu_6;
(iw- intercept, scw- school year slope, suw- summer slope, RIT_1-RIT_6 dv, tsc_#- time in school, tsu_#- time in summer).
I got this error message:
THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THIS IS OFTEN DUE TO THE STARTING VALUES BUT MAY ALSO BE AN INDICATION OF MODEL NONIDENTIFICATION. CHANGE YOUR MODEL AND/OR STARTING VALUES.PROBLEM INVOLVING PARAMETER 14.
Parameter 14 is the covariance between the intercept and the summer slope.
First, can MPlus handle unstructured data and estimate two different slopes?
If so, how can I best address the above error message?
If not, is it possible to model this in Mplus using a hybrid: estimated time scores that has individually varying times of observations?
My students were not all measured at the same time, so the "distance" between each observation varies as a function of students.
And, yes, you are right. I am trying to capture the summer drop off by measuring time in school separately from time in the summer. So, I am attempting to get two slopes (one for school time and one for summer time).
Can I use tscores to estimate two different slopes? The first 6 tscores (in my code in my second post) would estimate the school slope and the final 6 would measure the summer slope.
Is it possible to use tscores to estimate a school linear growth slope and a separate linear summer slope?
I don't see how that captures the drop due to the summer break. Perhaps a piecewise approach would be better, except you have only 2 timepoints per piece. Or, can you somehow work with a time-varying covariate, a summer dummy that influences the Fall outcome?