I'm a graduate student, and I am about to work on analysis of the ECLS-K longitudinal data. These data were collected in a fashion using multilevel sampling units. These sampling units are nonrandom and oversampled for Asian Pacific Islanders. Because Mplus assumes randomness, do you have any suggestions on how I can make necessary accommodations without violating any assumptions, or does anyone have any useful information after working with this specific dataset?
Mplus has many options for complex survey data. They take into account weighting, non-independence of observations, and stratification. See Chapter 9 and pages 455-460 of the Mplus User's Guide for more information.