|
|
 |
| Odds Ratio Effects for Rare Binary Ou... |
 
|
| Message/Author |
|
|
|
|
I am testing the following mediation model:
MODEL: ATT_SUIC !rare binary outcome
ON T_OVRALL (B1)!count mediator
GLS_G_D (B2)
BI_D (B3)
PAN_D (B4)
QQ_D (B5) !B2-B5 dummy-coded exposure
CISMEN_D (B6)
TRANS_D (B7) !B6-B7 dummy-coded moderator
MZ1 (B8)
MZ2 (B9)
DSBLTY (B10)
RACE (B11)
SAFETY (B12)
LFDPRSSN (B13); !B10-B13 controls
T_OVRALL ON GLS_G_D (G1)
BI_D (G2)
PAN_D (G3)
QQ_D (G4)
CISMEN_D (G5)
TRANS_D (G6)
DSBLTY (G7)
RACE (G8)
SAFETY (G9)
LFDPRSSN (G10);
[T_OVRALL] (G0);
Am I calculating the odds ratio effects correctly?
MODEL CONSTRAINT:
NEW (TNIE PNDE_GLS PNDE_BI PNDE_PAN PNDE_QQ);
TNIE = EXP((B1*G1 + B8*G1*1 + B9*G1*1) + (B1*G1 + B8*G2*1 + B9*G2*1) +
(B1*G1 + B8*G3*1 + B9*G3*1) + (B1*G1 + B8*G4*1 + B9*G4*1));
PNDE_GLS = EXP((B2);
PNDE_BI = EXP(B3);
PNDE_PAN = EXP(B4);
PNDE_QQ = EXP(B5);
Thank you. |
|
|
|
|
| A count variable as a mediator is not well-defined; there is no research to draw on for that case. Mplus treats it as count for the equation where it is the DV but as continuous in the equation where it is the predictor (IV). Indirect effects are therefore not defined. You can try to instead treat the count variable as categorical. |
|
|
| Back to top |
|
|
