Interpretation of Final Likelihood values

[ikaufhold]

Hello!

Thank you for developing this great tool!
I have two questions regarding the interpretation of the likelihood value in the results:

1. The Final Likelihood that is given in the output is a -lnL value. Is it correct, that because it is the negative log likelihood, a lower value means a higher likelihood?

2. Related to my first question: I first did multiple runs with the Base model and determined the "best" lambda value based on the lowest -lnL value. Then I used the lambda value and error model from that particular run with the Gamma model and tried out different numbers of Gamma categories.
   These are the likelihood values from the different runs:

k=1: 67523.6
k=2: 66345
k=3: 65967.1
k=4: 65840
k=5: 65778.3

Do these results tell me that a higher number of Gamma categories is a better fit for my data, since the -lnL value decreases?

Best regards!

[benfulton]

1. That's correct.

2. No, but it probably indicates that your gamma curve is a good match for your data. See the answer in How to determine gamma value (k) in CAFE5 #184 below.

[ikaufhold]

Thank you very much!