Hi Maya,
This is a response to your clarification for calculation of first differences.
When we use setx ( ) and sim ( ) for calculating first differences, we can explicitly set "contact" to a preferred value. Since you mentioned that zelig uses mean (for numerical vars) and median (for categorical vars) as default values on other covariates, I got a bit confused. I guess we do not need to go beyond using setx for anything else but contact. Can you confirm that?
Sim ( ) will give us a slightly different value each time we run it, but that's due to the fact that the function calculates simulated first differences. Does that make sense? Do we need to do something more for calculating first differences in 1b. Also, in 1d, you wrote "Simulate first differences in levels of attention to the campaign when going from the lowest to the highest educational category". Do you imply calculating one first difference for two observations where educ is min in one and max in other, or do you in fact imply calculating first differences between all educational levels, moving from the lowest to the highest level in the increment of one level?
Your help is appreciated.
Nino Malekovic
MPA Candidate, Class 2011
Harvard Kennedy School
________________________________________
From: gov2001-l-bounces at lists.fas.harvard.edu [gov2001-l-bounces at lists.fas.harvard.edu] On Behalf Of gov2001-l-request at lists.fas.harvard.edu [gov2001-l-request at lists.fas.harvard.edu]
Sent: Wednesday, April 07, 2010 8:00 PM
To: gov2001-l at lists.fas.harvard.edu
Subject: gov2001-l Digest, Vol 57, Issue 8
Send gov2001-l mailing list submissions to
gov2001-l at lists.fas.harvard.edu
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
or, via email, send a message with subject or body 'help' to
gov2001-l-request at lists.fas.harvard.edu
You can reach the person managing the list at
gov2001-l-owner at lists.fas.harvard.edu
When replying, please edit your Subject line so it is more specific
than "Re: Contents of gov2001-l digest..."
Today's Topics:
1. Regression variables (Gavinlertvatana, Poj)
2. clarification for first differences (Maya Sen)
3. Re: Regression variables (Iain Osgood)
4. how many leaders, really? (Mark Brewster)
5. Re: how many leaders, really? (Mark Brewster)
6. Re: how many leaders, really? (Maya Sen)
----------------------------------------------------------------------
Message: 1
Date: Wed, 7 Apr 2010 15:36:14 -0400
From: "Gavinlertvatana, Poj" <pgavinlertvatana at hbs.edu>
Subject: [gov2001] Regression variables
To: Class List for Gov 2001/E-2001 <gov2001-l at lists.fas.harvard.edu>
Message-ID:
<704295F116B34E47A0DFC59ADF333D31014E49638788 at WINHBSMAIL01.hbs.edu>
Content-Type: text/plain; charset="us-ascii"
Hi all,
I'm test two models in linear regression, and I get this situation:
* When I add variable X as a covariate, it is significant.
* When I add variable X*Y interaction, X*Y is significant but X becomes insignificant
The fit (R-squared) is the equal for both models (c. 0.9).
How would I choose one over another? The former is more parsimonious, but the second is just as valid, isn't it? I want to find arguments to choose model 2 over model 1, but can't really find a justification.
Best regards,
Joseph
Joseph Poj Gavinlertvatana
Doctoral student, Marketing
Harvard Business School
Wyss Hall, Soldiers Field, Boston, MA 02163
Ph 617.230.5907
Fx 617.496.4397
Txt/Vm 617.910.0563
Em pgavinlertvatana at hbs.edu
Hi
Instead of the outcome name, y, as.factor(y) will work. Please try!
Akihiro
>>> Meryl Federman 04/06/10 8:52 PM >>>
For 1C - using the zelig with just saying "oprobit" instead of
regular probit throws "response must be a factor".
I've poked around a bit but haven't found the right thing yet...
Any thoughts?
~Meryl
_______________________________________________
gov2001-l mailing list
gov2001-l at lists.fas.harvard.eduhttp://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
Hello,
The assignment says there are 2228 leaders, but when I unique()-afied them,
there are only 1838 left.
Should we use unique leaders (1838), or just # observations (2228)?
Mark
Hi folks,
Thomas Kane, from the GSE, is speaking today at the applied stats
workshop on some experimental studies that he's working on. I think it
might be particularly interesting to folks in Gov 2001, since we've
spent the past week talking about research design and experiments.
The talk is today at 12:15 in CGIS 354 (come at noon if you want to get lunch).
best, Maya
---------- Forwarded message ----------
From: Matt Blackwell <mblackwell at iq.harvard.edu>
Date: Thu, Apr 1, 2010 at 5:33 PM
Subject: [gov3009-l] Kane on "Teacher Impacts on Student Achievement"
To: gov 3009 <gov3009-l at fas.harvard.edu>
Hello,
We hope you will join us next Wednesday, April 7th at the Applied
Statistics workshop when we will be happy to have Thomas Kane from the
Graduate School of Education and currently at the Gates Foundation.
Details, abstract and a link to the paper are below. In addition,
Thomas asked that I include a link to a contrasting paper that comes
to a different conclusion than his own. Part of the talk will be about
the reconciliation of these two lines of research. A light lunch will
be served. Thanks!
"Estimating Teacher Impacts on Student Achievement: An Experimental Approach"
Thomas J. Kane
Graduate School of Education
April 7th, 2010, 12 noon
K354 CGIS Knafel (1737 Cambridge St)
Abstract:
We used a random-assignment experiment in Los Angeles Unified School
District to evaluate various ?non-experimental methods for estimating
teacher effects on student test scores. ?Having estimated ?teacher
effects during a pre-experimental period, we used these estimates to
predict student achievement ?following random assignment of teachers
to classrooms. While all of the teacher effect estimates we considered
were significant predictors of student achievement under random
assignment, those that ?controlled for prior student test scores
yielded unbiased predictions and those that further controlled ?for
mean classroom characteristics yielded the best prediction accuracy.
In both the experimental and non-experimental data, we found that
teacher effects faded out by roughly 50 percent per year in the two
years following teacher assignment.
Thomas's paper:
http://isites.harvard.edu/fs/docs/icb.topic646669.files/kane%20and%20staige…
Rothenstein paper:
http://isites.harvard.edu/fs/docs/icb.topic646669.files/rothstein_teacher_e…
Cheers,
matt.
~~~~~~~~~~~
Matthew Blackwell
PhD Candidate
Institute for Quantitative Social Science
Department of Government
Harvard University
email: mblackwell at iq.harvard.edu
url: http://people.fas.harvard.edu/~blackwel/
_______________________________________________
gov3009-l mailing list
gov3009-l at lists.fas.harvard.eduhttp://lists.fas.harvard.edu/mailman/listinfo/gov3009-l
Hey folks,
Just a quick clarification -- on Problems 1(B) and 1(D), you should
calculate the first differences with the rest of the covariate values
set at their default (which in Zelig is the mean of numerical vars and
the median of categorical vars). It's not worded super clearly, so
apologies for the confusion.
Maya
this is antually so nice for our problem set this time!
http://gking.harvard.edu/zelig/docs/static/Zcommands.pdf
>>> Meryl Federman 04/06/10 9:09 PM >>>
Thanks so much! I shall try!
~Meryl
On Apr 6, 2010, at 8:59 PM, Akihiro Nishi wrote:
> Hi
>
> Instead of the outcome name, y, as.factor(y) will work. Please try!
>
> Akihiro
>
> >>> Meryl Federman 04/06/10 8:52 PM >>>
> For 1C - using the zelig with just saying "oprobit" instead of
> regular probit throws "response must be a factor".
>
> I've poked around a bit but haven't found the right thing yet...
>
> Any thoughts?
>
> ~Meryl
> _______________________________________________
> gov2001-l mailing list
> gov2001-l at lists.fas.harvard.edu
> http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
> _______________________________________________
> gov2001-l mailing list
> gov2001-l at lists.fas.harvard.edu
> http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
For 1C - using the zelig with just saying "oprobit" instead of
regular probit throws "response must be a factor".
I've poked around a bit but haven't found the right thing yet...
Any thoughts?
~Meryl
I'm having trouble loading the RData file - is there something
funkier than load that we need to use for it? Did anyone else have
trouble loading it?
~Meryl
in case someone feels like doing a nice survey experiment...
Gary
---
http://gking.harvard.edu
---------- Forwarded message ----------
From: ANES <anes at electionstudies.org>
Date: Mon, Apr 5, 2010 at 4:34 PM
Subject: ANES Announcement: Only 30 days left to propose questions
To: ANES Mailing List <nesusers at listserver.itd.umich.edu>
There are only 30 days left to propose questions for inclusion on the
first two waves of The 2010-2012 Evaluations of Government and Society
Study. ?Proposals will be accepted until 3:00p.m. EDT, May 5, 2010. The
deadline for members of the Online Commons community to comment on
proposals will be May 12, 2010. The deadline for revisions to proposals
will be end at 3:00p.m. EDT on May 19, 2010. For additional information
about how to submit a proposal, please visit:
http://www.electionstudies.org/
The following describes the goals of this study and provides greater
detail about the opportunity to include questions on it.
About The 2010-2012 Evaluations of Government and Society Study
The overarching theme of the surveys we propose for the period of
2010-2012 is citizen attitudes about government and society. These
Internet surveys represent the most cost-effective way for the ANES user
community to gauge political perceptions during one of the most momentous
periods in American history. Aside from the historic nature of the current
administration and the almost unprecedented economic crisis facing the
country, we believe it is imperative that researchers assess attitudes
about politics and society in the period leading up to the 2012 national
elections. Specifically, we have in mind here attitudes about the
performance of the Obama administration on the major issues of the day,
evaluations of Congress and the Supreme Court, identification with and
attitudes about the major political parties, and levels of interest in and
engagement with national politics. This is primarily because these
perceptions are unmistakably correlated with both presidential vote choice
and levels of political participation. We intend to measure each of these
topics at multiple points throughout the two-year period preceding the
2012 elections. However in addition to these subjects, we envision that
each of these surveys would explore a particular aspect of these political
perceptions.
This Study will include five rolling cross-section waves that will allow
us the opportunity to pilot new items for possible inclusion on the 2012
time series. By offering multiple opportunities for the user community to
place their items on one or more surveys, we will provide the capacity to
survey on a diverse set of topics that are relevant to a wide set of
research communities. Lastly, the flexibility of these surveys as to both
content and timing will allow the ANES to respond promptly to emerging
political issues in this volatile period in our country?s history.
About the Online Commons
The design of the questionnaires for The 2010-2012 Evaluations of
government and Society Study will evolve from proposals and comments
submitted to the Online Commons (OC). The OC is an online system designed
to promote communication among scholars and to yield innovative proposals
about the most effective ways to measure electorally-relevant concepts and
relationships. The goal of the OC is to improve the quality and scientific
value of ANES data collections, to encourage the submission of new ideas,
and to make such experiences more beneficial to and enjoyable for
investigators. In the last study cycle, more than 700 scholars sent over
200 proposals through the Online Commons.
Proposals for the inclusion of questions must include clear theoretical
and empirical rationales. All proposals must also clearly state how the
questions will increase the value of the respective studies. In
particular, proposed questions must have the potential to help scholars
understand the causes and/or consequences of turnout or candidate choice.
For more information about the criteria that will be used to evaluate
proposals, please see
http://www.electionstudies.org/studypages/2010_2012EGSS/2010_2012EGSScriter…
This opportunity is open to anyone who wants to make a constructive
contribution to the development of ANES surveys. Feel free to pass this
invitation along to anyone (e.g., your colleagues and students) who you
think might be interested. We hope to hear from you.
Hi everyone,
Here is a quick question. When we derive the standard error of MLE analytically, is that just the square root of the inverse of the second derivative of MLE with respect to the parameter, adjusted for N?
Any help is appreciated.
Nino Malekovic
MPA Candidate, Class 2011
Harvard Kennedy School
________________________________________
From: gov2001-l-bounces at lists.fas.harvard.edu [gov2001-l-bounces at lists.fas.harvard.edu] On Behalf Of gov2001-l-request at lists.fas.harvard.edu [gov2001-l-request at lists.fas.harvard.edu]
Sent: Saturday, April 03, 2010 12:00 PM
To: gov2001-l at lists.fas.harvard.edu
Subject: gov2001-l Digest, Vol 57, Issue 3
Send gov2001-l mailing list submissions to
gov2001-l at lists.fas.harvard.edu
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
or, via email, send a message with subject or body 'help' to
gov2001-l-request at lists.fas.harvard.edu
You can reach the person managing the list at
gov2001-l-owner at lists.fas.harvard.edu
When replying, please edit your Subject line so it is more specific
than "Re: Contents of gov2001-l digest..."
Today's Topics:
1. simulating first differences using Zelig (Viral Gandhi)
2. Re: simulating first differences using Zelig (Ariel Dora Stern)
3. Re: simulating first differences using Zelig (Viral Gandhi)
4. Re: simulating first differences using Zelig (Ariel Dora Stern)
----------------------------------------------------------------------
Message: 1
Date: Fri, 2 Apr 2010 15:47:46 -0400
From: Viral Gandhi <viral_gandhi at hksphd.harvard.edu>
Subject: [gov2001] simulating first differences using Zelig
To: "Class List for Gov 2001/E-2001" <gov2001-l at lists.fas.harvard.edu>
Message-ID:
<r2qbbeedc781004021247u4fa216f8l8aa294de18656fba at mail.gmail.com>
Content-Type: text/plain; charset="iso-8859-1"
Hi,
Is there a quick way using Zelig to simulate first differences as a variable
is increased multiple times in increments of 1 unit?
Viral