I am conducting a series of GMM analyses and am running across a similar issue each time. When deciding on the best fit for a model, I use visual criteria as one of my decision points. This usually involves looking at the estimated mean and observed individual values for each latent class. Recently, I have also started looking at the estimated mean and estimated individual values. Every time I do so, the results are confusing. For example, I have an estimated mean that is flat (starts around 90 points and stays around 90 points across time). Yet the estimated individual values start with a range of 75-104 and show a large decrease over time (ranging from 55 to -20). I am puzzled as to how it is possible that the estimated mean is flat, while the estimated individual values are changing.