

Nested model comparison  gender diff... 

Message/Author 

Gerine L posted on Thursday, October 23, 2014  3:11 am



I am trying to compare 2 models to see if there are gender differences. In Model 1, I let Mplus freely estimate all paths for girls and for boys separately. This model has 12 df. In Model 2, I restrict all paths to be the same between boys and girls. This model has 8 df. If I look at this website: http://statmodel.com/chidiff.shtml it says: "The nested model is the more restrictive model with more degrees of freedom than the comparison model. " I don't understand this. How can a more restrictive model have more degrees of freedom? How does this translate to my model as described above? 


Degrees of freedom are the number of H1 parameters minus the number of free parameters in the H0 model. If there are fewer free parameters, the degrees of freedom will be larger. 

Gerine L posted on Thursday, October 23, 2014  7:04 am



Thank you very much for your response. I thought I understood the basics of degrees of freedom. However, I fail to grasp it here, it seems. For me, it seems like if I put more restrictions on the data (e.g., confine 2 paths to be the same), that means there is less room, thus fewer degrees of freedom. I.e., if I know path 1, I know path 2, hence path 2 cannot be freely estimated once path 1 is known. Is this untrue? 


See my answer above. 

Back to top 

