[Gretl-users] Degrees of freedom in panel data
Dear all,
Let's say I have three balanced cross sections with 20 observations
each. If I were to choose a single cross-section and perform a time series regression on
two explaining variables; admitting a constant, the corresponding degrees of freedom
should be: 20-2-1(n-k-1)=17.
Ok, what if I decide to consider the other two cross-sections for a panel
regression?. Do the degrees of freedom become: 57?(N-k-1).
Thanks
--- On Sat, 6/4/11, Allin Cottrell <cottrell(a)wfu.edu> wrote:
From: Allin Cottrell <cottrell(a)wfu.edu>
Subject: Re: [Gretl-users] how to delete observations and merge databases
To: "Gretl list" <gretl-users(a)lists.wfu.edu>
Date: Saturday, June 4, 2011, 2:28 AM
On Fri, 3 Jun 2011, Leandro Zipitria wrote:
I have a panel dataset and I want to delete the first 31 observations
of
each unit. I have read that one possible way is to restrict the sample and
then save it.
That's right.
genr time
smpl time > 31 --restrict
store smaller.gdt
The second one is a little more tricky. I have one dataset as a
panel, and
another that has additional information of some of the variables, but just
once. The two databases are of different length. Is there any way that gretl
merge this two databases and repeat each value of ?
Yes. Given a panel dataset of dimension N x T you can append data
with a series length of N or T and the added data will be repeated
as appropriate to complete the panel.
Allin Cottrell
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
Attachments:
- attachment.html (text/html — 2.4 KB)