likelihood ratio tests

[Anya hewertson]

Am i correct in thinking that likelihood‐ratio tests can be used to assess whether extra variables jointly add explanatory power to the model(regression) and how are the results from such tests interpreted?

[Carlo Lazzaro]

Anya:
if you refer to -regress- the that job of

assess whether extra variables jointly add explanatory power to the model

is accomplished by adjusted squared-R.
The effect each predcitor exerts (when adjusted for the other ones) on the regessand is proved by p-values and, even more important, (usually 95%) CIs.

[Anya hewertson]

Thanks Carlo, however i am using a tobit regression and from what i understand, i don't believe the turns a pseudo Rsq is as informative for the tobit as the adjusted r2 is for OLS.
If i ran a tobit regression without the extra explanatory variables and then one with the explanatory variables, would it then be possible to carry out the likelihood ratio tests?

[Carlo Lazzaro]

Anya:
if you are using -tobit- -lrtest- is the way to go to compare nested or composite regression models (see -help lrtest-).