Dear dr. Muthén, I'm trying to run a Monte Carlo simulation based on a multilevel VAR (1) model (Mplus 8), following the input of McNeish(2018). I'ts partially running and I receive the following message: THE GENERATED DATA FOR REPLICATION 3 HAS LARGE VALUES. THIS MAY BE DUE TO AUTOREGRESSIVE COEFFICIENT GREATER THAN 1. MODEL ESTIMATION WILL NOT BE INITIATED FOR THIS REPLICATION. Values are mostly based on the literature. Could you take a look at my input? Your help is greatly appreciated! Best, Ivonne
MONTECARLO: NAMES ARE binge na; NOBSERVATIONS = 5040; NREPS=100; NCSIZES = 1; CSIZES = 30(168); LAGGED ARE binge(1) na(1); SEED=1700; ANALYSIS: TYPE = TWOLEVEL RANDOM; ESTIMATOR = BAYES; BITERATIONS=(5000); MODEL POPULATION: %WITHIN% nana| na ON na&1; bb| binge ON binge&1; bna| binge ON na&1; nab| na ON binge&1; logv1|binge; logv2|na; %BETWEEN% [binge*0.16] [na*2.29] binge*0.21; na*1.91; [nana*0.30]; [bb*0.30]; [bna*0.20]; [nab*0.15]; [logv1*0.10]; [logv2*0.10]; nana*0.10; bb*0.10; bna*0.10; nab*0.10; logv1*0.10; logv2*0.10; MODEL:SAME AS ABOVE OUTPUT: TECH8 TECH9
Thank you very much for your reply! When I set the AR variances to 0.01, I still receive the same message for some replications. However, when I also adjust the variances of the crosslagged paths (nab and bna) to smaller values it seems to run correctly. Is it okay to proceed this way?