are clustered SEs the same as "random effects by cluster."? Wondering, bc if so, there might be some guidance on how to do this in the Gellman reference on multilevel models . . .
EXL
Hi all,
I want to add a whole bunch of coefficients to my regression. There are 44
in total and they are labelled
yearadj2, yeardadj3... yearadj46.
In stata you can do:
logit churnlatter yearadj2 - yearadj46
In R, I could write out:
glm(churninglatter ~ yearadj2 + yearadj3 + yearadj4 ... yearadj46)
but does anyone know of a more elegant way to do it that does not involve
writing out 44 variable names?
Thanks,
Viral
Hi all,
I know a number of people are struggling with replicating their standard
errors. I need both hetero robust and clustered SE.
When I use *robcov *with the following syntax after running the ols command
I get absolutely enormous SE( like *10^15):
robcov(my.ols1, cluster_variable, method=('efron')) *
'huber' gives similarly large estimates
When I try to simply use *vcovHC* or *hccm* with the lm function to correct
for at least heter. I am getting an error that says:
"Error in V %*% t(X) : non-conformable arguments" with *hccm*
and
"Error in bread. %*% meat : non-conformable arguments" with *vcovHC*
I don't know how to correct this or why the regression will run but then
these corrections for heter. will not.
Thanks,
Matt
--
Matthew Kraft
Doctoral Candidate
Quantitative Policy Analysis
Harvard Graduate School of Education
Hi everyone,
I am trying to use the optim function in ps4 and I am always getting the following error:
Error in optim(par = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, :
(list) object cannot be coerced to type 'double'
Could anyway tell me what I am doing wrong???
Thanks
Chiara
Hey all,
I've been at this regression for a while and I'm still confused so I decided
to turn to you for help. I need to replicate this STATA code in R:
xtregar zFDlaborfemale zFDlogGDPcap_1 zFDlogGDPcapSQ_1 zFDage_1, fe
Any ideas?
-Ashley Anderson
Hi all,
I am getting the following error when I try to use Zelig to implement a
proportional harazards model:
*Error in `[.data.frame`(d, , all.vars(as.expression(formula))) :
undefined columns selected*
Can anyone help with what this error means?
I have tried two different variants on the input but I get the same error
both times:
*Input data*
*data <- read.dta("Kroenig_ImportingtheBomb.dta")*
*Method 1*
*z.out <- zelig(Surv(data$risk0, data$risk1, data$acquire)~data$nucass,
robust = TRUE, cluster = "cowcc",
+ model = "coxph", data)
*
*Method 2*
Reading in the data as as dataframe and explicitly defining each covariate
as a matrix:
*toy <- as.data.frame(data)
start <- as.matrix(data$risk0)
stop <- as.matrix(data$risk1)
eventdefine <- as.matrix(data$acquire)
nucassmat <- as.matrix(data$nucass)
cowccmat <- as.matrix(data$cowcc)
z.out <- zelig(Surv(start,stop,eventdefine)~nucass, robust = TRUE, cluster =
"cowccmat",
model = "coxph", data=toy)*
Much appreciated,
Abdul
Hey folks,
So, per our announcement in lecture today, you now have your choice of which
of your colleagues' papers you want to re-replicate! All of the papers are
posted on the course website, and there is a little survey. Simply go to the
website, browse through your colleagues' papers, fill out the survey by
listing your top three choices, and Iain and I will do our best to
accommodate your preferences.
You have until Wednesday by 6pm to specify your preferences. If you don't
fill out the survey, Iain and I will assign a paper to you.
Note: There will be no problem set going out on Thursday; instead, your
assignment for the week is to re-replicate the paper you are assigned and
write a memo with constructive suggestions and feedback. This assignment
will be due Thursday, April 1.
Maya
I'd strongly encourage you all to apply to this. Yes, you're ready.
No, its not too early. Yes, I'm going. its a terrific, highly
informal but intense experience. Definitely worth it.
Gary
---
http://gking.harvard.edu
---------- Forwarded message ----------
From: Jeff Gill <jgill at wustl.edu>
Date: Tue, Mar 16, 2010 at 3:58 PM
Subject: [POLMETH] Summer Meeting Application Reminder
To: POLMETH at artsci.wustl.edu
? ? ? ? ? ? ? ?REMINDER, REMINDER, REMINDER!
The application deadline for The 27th Annual Summer Meeting of the
Society for Political Methodology at the University of Iowa, July
22-24, 2010 is March 22, this
? ? ? ? ? ? ? ?TUESDAY, TUESDAY, TUESDAY!
We invite proposals to attend or participate in the 27th annual Summer
meeting of the Society for Political Methodology (a group separate from
but nearly coincident with the APSA Political Methodology Section). This
year we are generously hosted by the University of Iowa and expect to have
a very stimulating program. ?Past participants often describe these
meetings as the most lively and intellectually engaging academic
experience in Political Science. ?Graduate students and faculty interact
closely together over the continuous formal and informal sessions,
including all meals. ?Also, nearly all of the papers presented at past
conferences have been published in leading journals of the field.
Who Should Apply Now?
? ? ? ? - Faculty who want to present papers or posters
? ? ? ? - Faculty volunteering to discuss
? ? ? ? - Faculty who simply want to attend
? ? ? ? - Junior faculty who want to be considered for funding
? ? ? ? - Graduate students who want to attend or a present poster
(in other words: everybody). ?The DEADLINE for applications is March 22,
and unlike with other Political Science conferences this date does not get
extended.
How Does One Apply?
Complete the simple forms at the SPM webpage:
http://polmeth.wustl.edu/conferences/methods2010/register/apply.php
What Can Be Proposed?
Faculty applicants are invited to propose a poster, a paper, a complete
panel; to act as a discussant; or simply to attend the meeting. Graduate
students are strongly encouraged to propose a poster presentation, which
has a long history of being career-enhancing, but may also apply just to
attend. ?All research topics relevant to political methodology are
welcome. ?Courtesy of the National Science Foundation, we are able to fund
30-40 graduate students whose poster proposals are selected as the most
promising. ?All graduate students need to supply a letter from a faculty
sponsor. ?In addition, we fund a limited number of junior faculty at
departments with highly restricted travel funds. ?Other faculty are
expected to seek support from their home institutions to cover their
travel and hotel expenses.
Feel free to direct any procedural questions to me: jgill at wustl.edu.
**********************************************************
? ? ? ? ? ?Political Methodology E-Mail List
?Editors: Xun Pang ? ? ? ?<xpang at wustl.edu>
? ? ? ? ? Jon C. Rogowski <rogowski.jon at gmail.com>
**********************************************************
? ? ? Send messages to polmeth at artsci.wustl.edu
?To join the list, cancel your subscription, or modify
? ? ? ? ?your subscription settings visit:
? ? ? ? http://polmeth.wustl.edu/polmeth.php
**********************************************************
Hey all,
for all of you STATA wonks out there . . .
when you ask STATA to run a regression with lots of dummies (say, an indicator variable for each year in a dataset with multiple years covered), there is a good chance that STATA will kick some out because there are perfectly collinear with some other dummies. When multiple dummies are kicked out STATA is making some decisions about what to keep and what to kick out. Does anyone know how to force STAT to keep a certain dummy variable in, so that the program needs to look to other dummies to drop in order to avoid perfect collinearity?
EXL
--
ERIC LIN
Technology and Operations Management
Harvard Business School
Boston, MA 02163
elin at hbs.edu
mobile: +1.216.225.2545
Hey,
I do not have much experience with Stata, but what does e(sample) do?
I looked online but could not find a satisfactory answer. The code is:
Table 3 */
tab code, gen (ca);
logit reelect2 ddef3_n ddef1 gdppc_gr2 nd maj ca* if dev==1 ;
/* Equation 1 */
logit reelect2 ddef3_n ddef1 gdppc_gr2 nd maj if dev==1 & e(sample)==1 , robust;
I am trying to model what is under Equation 1.
Thank you so much!
Han
Harvard College Class of 2013
614-329-1324