[Limdep Nlogit List] generalised linear model

[A.M.Zanni at lboro.ac.uk]

Thank you very much for your reply Prof. Greene
alberto


On Wed, 16 Dec 2009 10:22:36 -0500 (EST)
 William Greene <wgreene at stern.nyu.edu> wrote:
> Mr. Zanni. That model is simply a LOGIT model in LIMDEP.
> LIMDEP inspects your data for your logit model. If the 
>data
> are "fractional" it treats them exactly as described in 
>the
> Papke/Wooldridge paper. It also allows for some data to 
>be
> exactly 0 or 1, and the treatment is, once again, 
>exactly as
> described in the paper. The estimator has always been 
>structured
> that way; it is not a new feature in the program. (It 
>was there
> long before the JAE paper was published.) Do use it, 
>just use
> the LOGIT verb and specify the LHS and RHS as usual. If 
>you want
> a sandwich estimate for the variance matrix, just add 
>;ROBUST
> to the command. The program will figure out internally 
>what 
> the data look like.
> /B. Greene
> 
> 
> ----- Original Message -----
>From: "A M Zanni" <A.M.Zanni at lboro.ac.uk>
> To: limdep at limdep.itls.usyd.edu.au
> Sent: Wednesday, December 16, 2009 4:32:09 AM GMT -05:00 
>US/Canada Eastern
> Subject: [Limdep Nlogit List] generalised linear model
> 
> Dear all,
>> 
>> two questions for you:
>> 
>> I would like to use Limdep to estimate a Generalised 
>>Linear Quasi Likelihood regression model for proportion 
>>data as dependent variable. This is the model developed 
>>by Papke and Wooldridge in 1996 (Econometric methods for 
>>fractional response variables with an application to 
>>401(K) plan participation rates. Journal of Applied 
>>Econometrics, 11, 619-632). In STATA it corresponds to 
>>the command "glm dependent independent family (binomial) 
>>link(logit) robust". It seems to me that this model 
>>(which allows for consideration of the values 0 and 1 of 
>>the dependent variable, and not only those within the 
>>interval) is not among the 
>>options of generalised models in Limdep. Could somebody 
>>please help?
>> 
>> Second question is about the constant. A senior 
>>colleague has suggested to me to remove the constant from 
>>a regression model (which contains among the independent 
>>variable a tax level) because by keeping it the model 
>>would predict change even when the tax is zero). I have 
>>always thought removing a constant is not good. Any ideas 
>>on that?
>> 
>> Thank you very much for your attention.
>> 
>> Best
>> Alberto
>> 
>> Dr Alberto M Zanni
>> Transport Studies Group
>> Department of Civil and Building Engineering 
>> Loughborough University
>> Loughborough LE11 3TU
>> United Kingdom 
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au