Plotting the data and inspecting it visually. Should be done every time if possible before doing any analyses.
Yeah fam. Visualise it. Not visualising is like a doctor who doesn’t read your notes.
LCA, LPA, and LTA are all underutilized. I think the perception is that they’re no different than cluster analysis, so why not use the easier method (like k means cluster analysis). I’ve seen so many people conflate IRT and CFA with them as well. I think a general lack of understanding about the approach is definitely a contributing factor.
Thinking about and modeling the generative process of your data.
coarsened exact matching for causal inference.
Came to see this