|
|
 |
| The chi-square statistic is negative |
 
|
| Message/Author |
|
|
| Rob Gommans posted on Tuesday, August 23, 2016 - 5:03 am
|
|
|
I ran this Group Actor-Partner Interdependence Model (GAPIM) in a ML framework, but since ICC was very low, I removed the ML clustering.
I am now running a pretty straightforward path model I think, and get very similar estimates and SABIC value. However, I also get an error message now:
THE CHI-SQUARE STATISTIC IS NEGATIVE.
THE LOGLIKELIHOOD VALUES MAY NOT BE RELIABLE.
What does this mean and how to solve this?
Could it have anything to do with collinearity? Some predictors are highly correlated (.8).
ANALYSIS:
ESTIMATOR IS MLR;
MODEL:
Y1 ON kX kXprime kI@0 kIprime@0;
Y1 ON nX nXprime nI@0 nIprime@0;
Y1 ON gX gXprime gI@0 gIprime@0;
Y1 ON cX cXprime cI@0 cIprime@0; |
|
|
| Rob Gommans posted on Tuesday, August 23, 2016 - 5:55 am
|
|
|
In the mean time, I managed to get rid of this message by removing kIprime, rather than constraining it to zero. kI and kIprime were highly correlated (-.77).
Still wondering though how come than chi-square was negative... |
|
|
|
|
| Please send the output to Support along with your license number. |
|
|
| Back to top |
|
|
