08-03-2017 05:53 PM
I am currently working on the second part of an analysis. In the first part I identified a latent variable of problem behaviour with 4 categories, at three timepoints. My goal is to conduct an LTA to look at the transitions between these categories over time. My syntax is as follows:
PROC LTA DATA=EXAMPLE OUTPOST=EXAMPLE_POST; NSTATUS 4; NTIMES 3; ITEMS Problem_Behaviour_T1 Problem_Behaviour_T2 Problem_Behaviour_T3; CATEGORIES 4; MEASUREMENT TIMES; SEED 100; RUN;
My questions are as follows:
Is it okay to not include the indicator variables that went into indicating the latent variable: ProblemBehaviour? My indicators were continuous so I ran the latent class/latent profile analysis in MPlus.The analysis does seem to be working, but some weird stuff is happening when I look at the OUTPOST data. It appears that the categories of Problem_Behaviour are being presented in an incorrect order (i.e. 1=2, 2=1, 3=4, and 4=3). Any ideas why this might be happening?
I know that PROC LTA is not built by SAS -- it's a download from here. But some of you might know enough about it to answer the question or lend some advice. I'll be happy to summarize any responses back on Reddit.