I now realize that LTA is possible with only 2 time points. My issue is that I only have a single variable (math score) at each time point. I want to identify patterns of change from T1 to T2; where kids start at T1 is an important component of my research question, so I'm not just interested in change from t1 to t2 (but rather the pattern of change).
If I only have mathT1 and mathT2, would LCA be more appropriate than LTA? Or can I estimate latent classes with a single variable and then conduct the LTA using those classes?
LTA can be done with a single indicator at each time point - see handout on our web site for Topic 8. It would tell you who goes from low to high, high to low, low to low, and high to high. LCA would probably give similar results but would not work with 2 latent class variables, but 1. LCA would also not require invariant measures at the 2 time points; LCA does not know this is longitudinal data.