The main method to characterize and operationalize neighborhood social and economic change in my dissertation are finite mixture models.

These are a generalization of latent class analysis. Intuitively what these models do is look for subdistributions and find classes (clusters) of subdistributions across several variables that are assigned in a stochastic way to each unit of analysis.

So, in essence, the model input are several variables (without regards to the distribution of each variable, but normally these would be normally distributed or discrete variables). The model output are a posterior probability of belonging to each class (hence the stochastic clustering).

Association analysis will include a mix of either generalized estimating equations or generalized linear mixed models.