Forum: open-discussion
Monitor Forum | Start New ThreadLR test of nested model [ Reply ] By: Jeevan Patel on 2017-02-01 10:19 | [forum:43831] |
Hi, i Need a litlle help about how to test the Fitness of variable in a sfa model. I have a variable RD_Gov, i want to test whether it fits better in the production function or ineffiency part. so i made three models : model 1, model1.1, model 1.2. Model 1.1, model 1.2 are nested in model 1. As you can see below, Model 1.1 has RD_Gov in the production function part and modeel 1.2 has RD_Gov in inefficiency part. The Problem is when i test model 1 with model 1.1 using lrtest(model1, model 1.1) and lrtest(model1,model1.2) the degree of freedoms is negative. I couldnt figure whats wrong with it. Model1= sfa(log(pat+1)~lag(log(RD_Firm+1),1)+lag(log(RD_Gov+1),1) |lag(log(RD_Gov+1),1) -1 , data = datineff) Model1.1= sfa(log(pat+1)~lag(log(RD_Firm+1),1)+lag(log(RD_Gov+1),1) |-1 , data = datineff) Model1.2= sfa(log(pat+1)~lag(log(RD_Firm+1),1) |lag(log(RD_Gov+1),1) -1 , data = datineff) lrtest(Model1.1) lrtest(Model1.2) lrtest(Model1) lrtest(Model1,Model1.1) lrtest(Model1,Model1.2) did i nested the models in wrong way? looking Forward for your help |