|
Message/Author |
|
Jon Heron posted on Tuesday, February 27, 2007 - 4:24 am
|
|
|
I know p-value is a 4-letter word nowadays but I still decided to derive some for my current model. I'm assuming that all I need to do is examine the change in likelihood between c#1 c#2 c#3 on covariate; and c#1 c#2 c#3 on covariate@0; and derive a chi-square stat from this. Experimenting with a simple logistic model and comparing results with SPSS showed that the results were not in total agreement (although reasonably close) so i wonder if I'm not doing this exactly right. Many thanks Jon |
|
|
-2 times the loglikelihood difference is approximately equal to z square. |
|
Jon Heron posted on Thursday, March 01, 2007 - 12:59 am
|
|
|
Simple as that! Many thanks J |
|
Back to top |
|
|