Bayesian CFA: ERROR in MODEL PRIORS ... PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
Message/Author
 Justin Wiegand posted on Wednesday, October 07, 2015 - 11:09 am
Hello all,

I am experiencing difficulty running a Bayesian Bi-factor CFA. Specifically, after assigning labels to crossloadings for specific factors and then using them in the "MODEL PRIORS" statement, they are not recognized. I receive the following error:
*** ERROR in MODEL PRIORS command
Unknown parameter label(s): F1X1-F3X21


Below is a copy of the relevant input:

fConn BY PA1-BC5*;
fConn@1;

!Like;
f1 BY PA1-ES2R*;
f1@1;
RICR1-BC5 (f1x1-f1x16);
!Drive;
f2 BY RICR1-IO1*;
f2@1;
PA1-ES2R (f2x1-f2x15);
PS1-BC5 (f2x16-f2x25);
!Influence;
f3 BY PS1-BC5*;
f3@1;
PA1-IO1 (f3x1-f3x21);

MODEL PRIORS:
f1x1-f3x21 ~N(0,0.01);

*** ERROR in MODEL PRIORS command
Unknown parameter label(s): F1X1-F3X21
 Linda K. Muthen posted on Wednesday, October 07, 2015 - 12:05 pm
Instead of

F1X1-F3X21

try

F13X1-F3X21
 Justin Wiegand posted on Wednesday, October 07, 2015 - 2:04 pm
Still getting the same error. Just so I'm following, what's the intuition with your prior suggestion? I'm worried there's some problem with the naming of items.

For the record, my input items have various names as follows:

VARIABLE: NAMES = PA1 PA2 PA3 NA1 NA2 Ag1 Ag2R Ex1 Ex2R Co1 Co2R OTE1 OTE2R
ES1 ES2R RICR1 RICR2 RICR3 RICR4 RICR5 IO1 PS1 PS2 PS3 PS4 SM1
BC1 BC2 BC3 BC4 BC5 RICRC1 RICRC2 RICRC3 RICRC4 RICRC5 Study2 Attn;

USEVARIABLES = PA1-BC5;
 Justin Wiegand posted on Wednesday, October 07, 2015 - 2:07 pm
For the record, I also tried a more general labeling of crossloadings and listed all variables as follows (same error produced):

fConn BY PA1-BC5*;
fConn@1;

!Like;
f1 BY PA1-ES2R*;
RICR1 RICR2 RICR3 RICR4 RICR5 IO1 PS1 PS2 PS3 PS4 SM1 BC1 BC2 BC3 BC4 BC5 (x1-x16);

!Drive;
f2 BY RICR1-IO1*;
PA1 PA2 PA3 NA1 NA2 Ag1 Ag2R Ex1 Ex2R Co1 Co2R OTE1 OTE2R ES1 ES2R (x17-x31);
PS1 PS2 PS3 PS4 SM1 BC1 BC2 BC3 BC4 BC5 (x32-x41);

!Influence;
f3 BY PS1-BC5*;
PA1 PA2 PA3 NA1 NA2 Ag1 Ag2R Ex1 Ex2R Co1 Co2R OTE1 OTE2R ES1 ES2R (x42-x56);
RICR1 RICR2 RICR3 RICR4 RICR5 IO1 (x57-x62);

MODEL PRIORS:
x1-x62~N(0,0.01);
 Justin Wiegand posted on Wednesday, October 07, 2015 - 3:07 pm
Below is (as far as I can tell) a replication of Example 5.31 (except with 3 specific factors and y1-y31 items). This still produces the same error. Please unpack what's going on with the assigning of labels to crossloadings as that seems to be the problem.

MODEL:
fg BY y1-y31;
fg@1;

!Like;
f1 BY y1-y15;
y16-y31 (f1x16-f1x31);
!Drive;
f2 BY y17-y22;
y1-y15 y23-y31 (f2x1-f2x24);
!Influence;
f3 BY y23-y31;
y1-y22(f3x1-f3x22);

MODEL PRIORS:
f1x1-f3x21~N(0,0.01);

*** ERROR in MODEL PRIORS command
Unknown parameter label(s): F1X1-F3X21
 Justin Wiegand posted on Wednesday, October 07, 2015 - 3:42 pm
Solved: Semi-colons go after the entire specification of a factor (something most users know), INCLUDING the labeling of crossloadings (something users may not know). Thus,

CORRECT:
f1 BY y1-y15
y16-y31 (f1x16-f1x31);


INCORRECT:
f1 BY y1-y15;
y16-y31 (f1x16-f1x31);
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: