I wrote out A two-level SEM program (factors:t1sc, t2sc and t1lsc; t1sc and t2sc are level-1 factors and t1lsc is a level-2 factor) as follows: VARIABLE: MISSING ARE ALL (-999); NAMES = id sex groupid class pa1 psn1 tsc1 forl1 pa2 psn2 tsc2 forl2 lpa1 lpsn1 ltsc1 lforl1 ; USEVARIABLES = groupid pa1 psn1 tsc1 forl1 pa2 psn2 tsc2 forl2 lpa1 lpsn1 ltsc1 lforl1 ; WITHIN = pa1 psn1 tsc1 forl1; BETWEEN = lpa1 lpsn1 ltsc1 lforl1; CLUSTER = groupid; ANALYSIS: TYPE = TWOLEVEL RANDOM; ESTIMATOR = MLR; MODEL: t1lsc BY lpa1 lpsn1 ltsc1 lforl1; %WITHIN% t1sc BY pa1 psn1 tsc1 forl1; t2sc BY pa2 psn2 tsc2 forl2; t2sc ON t1sc; edslope | t2sc ON t1sc; %BETWEEN% t2sc ON t1lsc; edslope ON t1lsc; t1lsc WITH t1sc; But I got a warning saying that "*** ERROR in MODEL command The labels %BETWEEN% or %WITHIN% must be specified." I don't know how to fix it and also wonder if it is because I use wide form rather than long form data. I don't think I need to reshape the dataset into a long set because I only have two time occasions here. Indeed, variables at time1 are indepent while same variables measured at time2 are dependent.
Thank you for your prompt response, Linda! In this case, where should I put this statement in the program so as to represent the CFA part? In addition, a general question, is long format specific to longitudinal data? I am confused here that although my data is longitudinal (measured at two occasions), it is actually not that time-repeated as most of the examples illustrated in the Mplus User's Guide.