Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Nonlinear Panel Data

From   sazas <[email protected]>
To   [email protected]
Subject   st: Nonlinear Panel Data
Date   Thu, 28 Oct 2010 13:38:58 -0700 (PDT)

I am studying the the effectiveness of a policy (policy effect) using panel
data (T=5). The dependent outcome is dichotomous (binary). The panel data am
using have two waves before the policy was implemented and three waves post
policy implementation. My policy dummy takes the value of 0 pre policy
implementation and 1 otherwise.

I have estimated my model using random effect probit/logit model and
conditional fixed model. 
The policy effect variable which is of interest to me is quite different
between the random and fixed effect model. In the fixed effect model, more
than 70% of the observations are lost due to lack of within variation in the
data. I am not comfortable reporting just the random effect model due to the
unpalatable assumptions.

Please any advice?

View this message in context:
Sent from the Statalist mailing list archive at
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index