[Limdep Nlogit List] Fixed-effect logit Vs OLS

[William Greene]

Twyeafur: The FE logit is based on sets of observations in which there
is variation in the dependent variable.  When the LHS variable is the same
in every period, that group gets dropped.  The "linear probability model"
does not drop observations this way.
/B. Greene

On Fri, Mar 4, 2016 at 10:19 AM, Md Twyeafur Rahman <m.t.rahman at strath.ac.uk
> wrote:

> Hi
>
> I am looking to predict factors affecting the branch placement decision of
> a bank (binary dependent variable). To do this I have collected information
> (3020 villages information) on villages where there is a branch and where
> there is no branch. The immediate higher level administrative unit of
> village is called Thana. The total number of observations came from 515
> Thanas. However, I have observations on the branch placed from 438 Thanas.
> I have both the information (villages where is no branch and villages where
> is a branch) from the rest of the Thanas (77).
>
> Now, I am losing observations (observations from those 438 Thanas) when I
> run fixed-effect logit model (Fixed-effect at Thana Level). However, fixed
> effect linear regression (linear probability model) does NOT drop any
> observations. Why this is happening? What are the underlying statistical
> assumptions behind these two model. Could you please explain me this?
>
> I would highly appreciate your response.
>
> Many thanks in advance.
>
> Kind regards
> Twyeafur
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
>



-- 
William Greene
Department of Economics
Stern School of Business, New York University
44 West 4 St., 7-90
New York, NY, 10012
URL: http://people.stern.nyu.edu/wgreene
Email: wgreene at stern.nyu.edu
Ph. +1.212.998.0876