Hi, everyone.
As I mentioned in section, problem set 6 will be due in lecture
on Monday, Nov. 4th. Problem set 7 will be due in section on
Thursday, Nov. 18th. These problem sets aren't longer than the
other problem sets -- this is intended to give you more time to
work on your papers.
About last week's section and next week's section:
Gary covered two really important concepts in lecture last week:
1) maximum likelihood and 2) interpreting regression output via
simulation. I covered (1) in section this week and will cover
(2) this coming week in section (in other words, I'll present
the Zelig deck from APSA).
As I mentioned to those who showed up, these two sections are
the most important in the course because we're going over how to
write a new model, and a generalized framework for interpreting
output from all statistical models, and irrespective of the
methodology of the original paper, we expect a few things in
your final paper:
1) You should choose a model compatible with the hypothesized
event generating mechanism (and if the model hasn't been
programmed, you should program it yourseif via maximum
likelihood, and if you don't know what I'm talking about, you
need to read the handout for 10/28...)
and
2) You should interpret the regression results in your final
paper by simulating quantities of interest (which is why next
week is important).
So if you couldn't make it to section last Thursday, read the
handout and let me know if you have questions. And you should
come to section this Thursday, because I don't think Gary is
going to go back and cover simulation again.
Yours,
Olivia Lau
Hi, everyone.
I've had several requests for help on loading data and other
basic R stuff related to your papers. If you need help, please
come to OH -- Tuesdays from 3-5pm, or by appoinment. I can't do
this stuff over email.
Also, those of you with large datasets have realized that you
can't use Windows R for your replication assignment. You'll
need to use the servers if you have a lot of data. If your data
is so large that you exceed your unix account quota, you can ask
FAS directly for an increase to 200MB and can ask your
department administrators (Thom Wall or the equivalent) for an
increase to 400MB.
If you have data that's tab-delimited, space-delimited, comma
separated, SPSS or STATA data, you can read your data directly
into R following the directions in Zelig Chapter 3, section 2.2.
If your data is in another format, you need to convert your data
using DBMS copy, which is available in the HMDC and CBRSS
computer labs.
Yours,
Olivia Lau
Hi,
Does anybody know...
1) How to clean up a R console?
2) In LaTeX, how to show the date other than \today by \maketitle?
3) When I do coauthor work,
I highlight what parts I revise on Word.
How can we do this by XEmacs?
4) what is the natural conjugate prior distribution
of negative binomial (poisson * gamma) or beta?
(I do not ask beta is conjugate for what distribution
(actually, for bernoulli related distributions.))
Thank you in advance!
Kentaro
In case anyone else was confused by that log of gamma thing, Olivia just
confirmed that it is the SAME thing in R to do:
log (gamma(a + b))
or
lgamma(a+b)
A good question, Sara.
For question 1 and 2(a), you should show some math. For questions 2(b) and
and 3, you should do it in R.
(You can check your math with R and your R with math, if you want, but that
won't be graded.)
----- Original Message -----
From: "Sara Bleich" <bleich(a)fas.harvard.edu>
To: "Olivia Lau" <olau(a)fas.harvard.edu>
Sent: Wednesday, October 27, 2004 5:03 PM
Subject: Clarification
> Olivia,
>
> For the problem set, I just wanted to clarify whether we should be writing
> down the likelihood functions for the various questions in R or by hand.
> If
> it is in R, does that apply to the analytic solution for the Poisson?
> Thanks.
>
> Sara
>
just found it myself: prod() works fine
> -----Ursprungliche Nachricht-----
> Von: Jens Hainmueller [mailto:jens_hainmueller@ksg05.harvard.edu]
> Gesendet: Wednesday, October 27, 2004 3:48 PM
> An:
> Betreff: parameterization of product symbol (uppercase Pi) in R
>
>
> Any hints on a smart way to parameterize the product symbol
> (upper case "Pi") in R?
>
> Thanks!
> Jens
Hi, people.
Per Kentaro and Paul's comments this weekend, I've posted a revised
problem set -- the typos weren't major so I hope this hasn't held anyone
up.
And as a reminder: I do have OH from 3-5 on Tuesday afternoon. Please
come by if you need help with your papers as well.
Yours,
Olivia.
this is entirely voluntary of course, but if you are interested in
academia, I would strongly recommend that you all attend these talks.
You'll be in the same position as them one day soon, and having a good
feel for how these job talks go -- and what works and what doesn't -- will
be exceptionally useful for you. there are some principles of what works,
but its mostly clinical knowledge and so seeing it in action can help a
lot.
and quite separately, i'm chair of the cttee, and so if you have any
views about the candidates, I'd appreciate if you let me know.
Gary
---------- Forwarded message ----------
Date: Mon, 25 Oct 2004 08:48:28 -0400
From: Jaronica Fuller <jfuller(a)latte.harvard.edu>
To: Gov_Senior(a)latte.harvard.edu, Gov_Junior(a)latte.harvard.edu
Subject: Formal Political Theory Job Talks
Dear Government Faculty,
Enclosed is the list of junior faculty recruitment candidates for the
Formal Political Theory junior faculty search. Also included are the
dates for each candidate visit, job talk titles, presentation times and
room locations. Please feel free to attend the job talk presentations.
If you would like to schedule a 30 minute interview with any of the
candidates or if you have any other questions, please feel free to
contact me at 496-1512 or jfuller(a)latte.harvard.edu
Sincerely,
Jaronica C. Fuller
Gerard Padro Miquel (Ph.D. candidate, MIT)
Date of Full Day Visit, Tuesday, October 26
Title of Job Talk: "The control of Politicians in Divided Societies: The
Politics of Fear"
Location: Center for Basic Research in the Social Sciences (34 Kirkland
Street)
Time: 9-10:30 a.m.
John Patty (Ph.D. California Institute of Technology 2001, currently
Assistant Professor Carnegie Mellon)
Date of Full Day Visit, Tuesday, October 26
Title of Job Talk: TBA
Location: Center for Basic Resarch in the Social Sciences (34 Kirkland
Street)
Time: 2:00 - 3:30 p.m.
Jaehon Kim (Ph.D. Rochester 2004, currently Visiting Scholar Kellogg
School of Management, Northwestern University)
Date of Full Day Visit, Wednesday, October 27
Title of Job Talk: "Legislative Organization and Rule Choice: Monopoly
Agenda Setting Power and the Status Quo"
Location: Center for Basic Research in the Social Sciences (34 Kirkland
Street)
Time: 10:30 - 12:00
Elizabeth Penn (Ph.D. California Institute of Technology 2003, currently
Assistant Professor Carnegie Mellon)
Date of Full Day Visit, TBA
Title of Job Talk: "Farsighted voting and agenda formation"
Location: TBA
Time: TBA
_____________________________________________________________
gov_senior mailing list served by Harvard-MIT Data Center
List Address: gov_senior(a)latte.harvard.edu
Hi,
In ps#5, 2(b), should we plot the likelihood function
with \lambda, not $y$ as indicated,
on the x-axis?
Also, as for solution set #4, p. 7,
should l. 4 read
sample.sd[i] <- (N*prob*(1-prob))/i
?
Kentaro
My study group and I are confused about the relationship between the
student t distribution and what we create in part 1. In particular,
what values of nu and s^2 should make our distribution look like a
t-distribution of n-degrees of freedom?
Andy