|
|
 |
| LGC with Positive Log likelihood |
 
|
| Message/Author |
|
|
| Chuang Chen posted on Thursday, March 29, 2018 - 11:43 am
|
|
|
| I run a latent growth curve model. The time is base on tscore, individual time. My outcome is a continuous variable. When I got the result, the log likelihood is a positive number. I'm very curious about it. Because based on my knowledge, it should be a negative number. I'd like to ask why it happens? Thanks a lot! |
|
|
|
|
There is no rule that says that the normal log-likelihood has to stay negative.
If you divide your variable by a constant C you add N*log(C) to the likelihood i.e. you can make any likelihood positive simply by changing the scale of the variable. Similarly If you multiply your variable by a constant C you subtract N*log(C) from the likelihood i.e. you can make any likelihood negative by changing the scale of the variable. |
|
|
| Chuang Chen posted on Thursday, March 29, 2018 - 5:26 pm
|
|
|
| Thanks Tihomir!! |
|
|
| Back to top |
|
|
