Hi All,
* First, to better line up with our problem set schedule, I will shift my
office hours to Friday, April 21st from 4-6 PM in room N-018. I won't
have Wednesday office hours, but Ian will hold his regular Tuesday office
hours this upcoming week (April 18th), and we will see how that works for
the final few weeks.
* Also, the notes and R code from tonight's section are now posted.
Good weekends to all!
Best,
Dan
----
Ph.D. Student
Department of Government
Harvard University
Tutor, Currier House
dhopkins(a)fas.harvard.edu
http://www.danhopkins.org
Hi List,
A question of clarification for the problem set: what exactly do you mean by
"predicted values" in the case of the probit model? i.e. in a continuous
variable case it makes sense to predict the outcome variable, but this may not
be very interesting here- should we be concentrating instead on the probability
of seeing a particular outcome?
thanks,
Lucy
Hi all,
I find the on-line Zelig manual unclear regarding how to set a set of
variables to values other than the default mean in 'setx'.
What is the syntaxs to set 'all the other' variables to their median?
Juan
Greetings,
I attempted some of the Whatif commands in R and they subsequently
crashed my R session. The R program seems to kill the session after
letting the code run for about an hour. My model had about 15,000
observations and about 15 variables and interactions. Does anyone
have any intuition as to why this is occuring and has anyone run
Whatif successfully? In addition, I attempted to use the "by" command
in my Zelig models and got an error message informing me that I cannot
create a vector of size 121 kb. Has anyone encountered this error,
and does anyone have a suggestion of how to proceed?
Regards,
Sheldon
Hi everyone,
A few announcements/reminders to pass along:
1. *Room change: starting tomorrow, lecture will be held in
CGIS N354*
2. A reminder that your memo offering suggestions on another group's
replication is due at the beginning of lecture (2 PM EST for distance
students). Please be on time, so that we can exchange memos and get them
back to you as quickly as possible. Details about what we expect in the
memos are available at the end of this handout:
http://www.courses.fas.harvard.edu/~gov2001/Assignments/repassign.pdf
3. Gary will discuss model dependence and matching tomorrow, and Dan will
build on these topics in section. Therefore, please be sure to read the
previously assigned articles, available at the links below:
http://gking.harvard.edu/files/matchp.pdfhttp://gking.harvard.edu/files/counterft.pdf
4. As previously announced, problem set 6 is now due Monday, April 17th at
the beginning of lecture (2 PM EST for distance students).
Hope you are having a nice weekend...
Best,
Ian
programming the gec involves a bit more specialized plumbing than usual. If you want to give it a try, I'd start with the Gauss version, which has it all programmed in.
(right, you probably don't know gauss, but the secret is that now that you've got the hang of R you also will be able to decipher most other stat'l programming languages). If you do it, let me know and we'll include it in Zelig with your name on it -- same for others who implement, or wrap existing, methods not already in Zelig.
Gary
-----Original Message-----
From: "Jacqueline Chattopadhyay" <jchattop(a)fas.harvard.edu>
Date: Saturday, Apr 8, 2006 6:33 pm
Subject: Re: [gov2001-l] GEC Estimator
Hi all,
I'm wondering if anyone has worked with the Generalized Event Count (GEC) estimator. I see that the log-likelihood function is given on page 775 of Gary's article, "Variance Specification in Event Count Models." I assume that it's possible to enter the GEC log-likelihood function into R and then use ML estimation, as we do for other distributions. It looks, however, like one would need to find a way to program the three different forms that C_{i} could take into R and instruct the program to choose the appropriate one of those three when running ML. I don't know how to do this. It doesn't look like Zelig is pre-programmed to run models that assume the GEC distribution. I'm not sure where to begin with this and would much appreciate any
insights.
Thanks,
Jacqueline
------=_Part_32691_31885762.1144535594252
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
<div>Hi all,</div>
<div>I'm wondering if anyone has worked with the Generalized Event Count (GEC) estimator. I see that the log-likelihood function is given on page 775 of Gary's article, 'Variance Specification in Event Count Models." I assume that it's possible to enter the GEC log-likelihood function into R and then use ML estimation, as we do for other distributions. It looks, however, like one would need to find a way to program the three different forms that C_{i} could take into R and instruct the program to choose the appropriate one of those three when running ML. I don't know how to do this. It doesn't look like Zelig is pre-programmed to run models that assume the GEC distribution. I'm not sure where to begin with this and would much appreciate any insights.
</div>
<div>Thanks,</div>
<div>Jacqueline</div>
------=_Part_32691_31885762.1144535594252--
Re: [gov2001-l] GEC Estimator"Jacqueline Chattopadhyay" <jchattop(a)fas.harvard.edu>To: gov2001-l(a)lists.fas.harvard.edu Reply-To: gov2001-l(a)lists.fas.harvard.edu
Hi all,
I'm wondering if anyone has worked with the Generalized Event Count (GEC)
estimator. I see that the log-likelihood function is given on page 775 of
Gary's article, "Variance Specification in Event Count Models." I assume
that it's possible to enter the GEC log-likelihood function into R and then
use ML estimation, as we do for other distributions. It looks, however, like
one would need to find a way to program the three different forms that C_{i}
could take into R and instruct the program to choose the appropriate one of
those three when running ML. I don't know how to do this. It doesn't look
like Zelig is pre-programmed to run models that assume the GEC distribution.
I'm not sure where to begin with this and would much appreciate any
insights.
Thanks,
Jacqueline
Hi everyone,
Given that you are working on the replications, we have decided to extend
the deadline for Problem Set 6 until Monday, April 17th at the beginning
of lecture (2 PM EST for distance students).
Hopefully this will make your weekends a bit better...
Best,
Ian
An article from The Chronicle of Higher Education was forwarded to you by:
king(a)harvard.edu
The following message was enclosed:
Given some of the questions last Monday, I thought you might
be interested in this.
Gary
This article, "Sense of Injustice Can Lead Scientists to Act
Unethically, Study Finds" is available online at this
address:
http://chronicle.com/temp/email2.php?id=jm5JNTYdhXKVyDmHjX95vcyFJR8HrQby
This article will be available to non-subscribers of The
Chronicle for up to five days after it is e-mailed.
The article is always available to Chronicle subscribers at this
address:
http://chronicle.com/daily/2006/04/2006040704n.htm
_________________________________________________________________
Finding it hard to keep up with all that's happening in academe?
The Chronicle's e-mailed Daily Report keeps you up-to-date in a
matter of minutes by quickly summarizing current events in higher
education while providing links to complete coverage on our
subscriber-only Web site. The Daily Report and Web access come
with your Chronicle subscription at no extra cost. Order your
subscription now at http://chronicle.com/4free?es
_________________________________________________________________
Visit us on the Web -- http://chronicle.com
Careers & advice -- http://chroniclecareers.com
Opinions & critique -- http://chroniclereview.com
Arts & Letters Daily -- http://aldaily.com
_________________________________________________________________
If you have other problems or questions, please send a message to
help(a)chronicle.com
_________________________________________________________________
Copyright (c) The Chronicle of Higher Education, Inc.