[Limdep Nlogit List] Pseudo R-squared

Marc Goergen m.goergen at shef.ac.uk
Thu Jun 15 23:55:51 EST 2006


Dear all,

I am trying to estimate the McKelvey and Zavoina pseudo R-squared for an ordered probit model.

I am following the procedure as stated in the online help file. However, when I do the calculations, Limdep overwrites the first 12 rows of my data file. Does anybody know what is going on?

This is the procedure I am using:

?===================================================================
?  Compute a Pseudo Rsquared for a probit model based on the       
?  McKelvey and Zavoina's formula in their paper on the ordered     
?  probit model.  The formula suggested is                                 
?                                                                          
?         E[y*|y]  =  yf  =  beta'x  +  lambda                                
?                                                                          

?         Rý = N * [ var(yf) / (1 + var(yf) ]                              
?                                                                          
?  where  lambda is the inverse Mill's ratio usually kept for SELECT.      
?  (To get the MZ result exactly, we'd use 1/N not 1/(N-1) in var(yf).)    
? User must define the data matrix with the following line:        
?                                                                  
?        NAMELIST         ; X = ... $                              

?                                                                  
? User's dependent variable is Y.  Change the following line       
?                                                                  
?        CREATE           ; Y = the dependent variable $           
?                                                                          
?===================================================================
   PROBIT   ; lhs = y ; rhs = X ; hold(imr=lambda) $                       

   CREATE   ; yf = x'b + lambda $                                     
   CALC     ; list ; zm = (n-1)*var(yf) / (n + (n-1)*var(yf)) $                   



More information about the Limdep mailing list