Brianna H posted on Monday, December 10, 2012 - 7:42 am
Hello, I created a zero-inflated Poisson model to examine growth in a count outcome over four equidistant time points. The model has one time-invarying covariate, ethnicity. There are three ethnicity codes in this sample: non-Hispanic White, Hispanic (English-speaking), and Hispanic (Spanish-speaking). When I create a data file for feeding the data to Mplus, can I keep the meaningless numeric ethnicity codes of 1, 2, 3, or do I need to create contrast codes to reduce the number of categorical groups to two? In other words, can a categorical covariate with more than two levels be interpreted meaningfully in this context? If so, would I be interpreting the effects of the covariate on the intercepts and slopes in relation to a reference group? Thank you very much for your help.