|
|
| Message/Author |
|
|
|
|
when I ran LCA I got:
> # of classes AIC BIC Adjusted BIC Log-likelihood p-values of LMR test
> 2 43629.67 43802.05 43694.050 -21780.835 0.0000
> 3* 41773.57 42006.79 41860.674 -20840.786 0.0003
> 4 41088.91 41382.96 41198.735 -20486.456 0.2605
> 5 40380.13 40735.02 40512.679 -20120.066 0.0956
> 6 40117.31 40533.04 40272.580 -19976.656 0.0000
> 7 39806.56 40283.13 39984.551 -19809.280 0.0123
I have two questions:
1. Are the values of AIC and BIC too large? Do large AIC/BIC values indicate some issues?
2. In my run, Tech14 never gave clear indication of # of classes, but tech11 did . In your 2007 paper in SEM (14, p535-569), it said BLRT (from Tehc 14) does better job than BIC and LMR (from tech11). Not sure why in my case LMR works better than BLRT. |
|
|
| Back to top |
|
|
