Hi all,
I'll be around the basement computer lab (probably in one of the small
rooms) between 11 and noon today, in case you have any last-minute
questions.
Holger
--
Holger Lutz Kern
Ph.D. cand.
Department of Government
Cornell University
Institute for Quantitative Social Science
Harvard University
1737 Cambridge Street N350
Cambridge, MA 02138
www.people.cornell.edu/pages/hlk23
Title: Heterogenous Treatments and Noncompliance in a Randomized
Class-Size Experiment
Abstract: Using data from the Tennessee STAR experiment, Krueger and
Whitmore (2001) find that students randomly assigned to small classes in
the early grades are more likely to take a college entrance exam. STAR
was actually a series of four randomized experiments (corresponding to
the four points in time at which students entered a participating
school), so we estimate treatment effects separately for each entry
wave. We demonstrate that the estimated effect of attending a small
class on the probability of taking a college entrance exam is
essentially zero among students who entered the experiment in
Kindergarten (and comprise more than half of all students in the STAR
data). There is suggestive evidence that the effect is larger for black
students. These findings hold when we restrict our analysis to students
who complied with their assigned treatment (i.e., attended a small or
regular-size class for four consecutive years). Treatment effects across
the later three treatment waves vary widely (in no discernible pattern)
and are too imprecisely estimated to allow for meaningful interpretation.
Hey Everyone,
We know that you all are busy completing your papers, but we'd like to
encourage you to take a quick time out to evaluate the course.
We want to hear about what went well, what failed, and how we can make
this course better for future students.
To reach the class evaluation go to my.harvard.edu and then click on the
courses tab.
Cheers,
Justin
OK, so now that we've gone through most of the abstracts, and you're
working hard to find the angle, the contribution, etc., of your work,
what's next? when you find it, you change the title to make your point;
you change the abstract to make your point, and you change the intro to
make your point.
the rest of the paper then has to be organized. you're writing for a
journal. every word you write costs expensive journal space, so keep it
concise. don't just fill pages because you think you need more. you
don't. we'd be happy with a 5 page paper if you can do everything in that
space. AND even more importantly, the paper needs to be rigorously
organized. if we read ONLY the titles of your sections and subsections we
ought to know the point of the paper and most everything in it. the
Organization of the paper is really important and should follow directly
from the point. nothing in the paper belongs in there unless it supports
the key point you're making, which ideally should be this angle,
contribution, etc.
Gary
hey guys,
does anyone know if treating a rare outcome event as a non-rare outcome event in
a multinomial logit bias coefficients? i have a 4-category logit (and a 5th for
the
baseline), and 2 of the categories have very few observations. i am
considering consolidating 1 of the problematic categories with another (which is
something the authors do for one of their tables as well so i
think it's kosher) and taking out the other problematic category altogether
(it's
the baseline in my model). substantively, i think the transformation would be
ok, but my question is whether it makes sense methodologically. specifically,
is it: A) worthwhile - ie are rare-event outcomes a problem (in terms of
causing
bias etc), and B) does it sense given that there is no rare events *multinomial*
logit?
thanks for any suggestions,
anya
Analyst-based Predictions of Future Stock Returns
In this paper, we investigate the findings of Karl B. Diether, Christopher
J. Malloy, and Anna Scherbina in their paper \emph{Differences of Opinion}.
We validate their claim that greater differences of opinion in analyst
predictions of future stock returns tracks poorer future returns of stocks
and that small and poorly performing stocks are especially affected. We
extend this project by exploring the relationship between a stock's returns
and the number of analysts covering the stock (coverage). We also use the
normalized difference between a stock's predicted expected returns and its
current returns (potential) to improve upon the original paper's
dispersion-based portfolio strategy. Surprisingly, we found that stocks
covered by the median number of analysts performed best, and portfolios of
stocks with the greatest positive potential outperformed stocks with the
greatest negative potential by a large margin.
Here is our abstract. Sorry for the delay. We appreciate any feedback.
How the expansion of health insurance changes self-medication practice for the elderly? The case of Mexico
How much self-medication can be reduced by extending health insurance? Who will benefit more from having health insurance? Self-medication is a common response to illness in many countries, especially among uninsured and those under poverty and Mexico is no exception. Pag?n et al, (2001) found that health insurance coverage as well as higher socio-economic status reduce self-medication in the elderly population in Mexico; however their results show no interaction effect between insurance coverage and income level. To estimate the potential effect of the current health reform aiming to extend health insurance primarily for poor households, we extend the analysis to estimate this interaction by using imputation and propensity score matching. Our results indicate that health insurance has a differential effect on reducing self-medication by income level ranging from 10% for the poorest quintile to 5.0% for the richest quintile. This indicates that the poor elderly may benefit the most of a health insurance extension.
Improving Predictions of Civil Wars: Inequality Matters
Jill Goldenziel and Sana Nourani
Collier and Hoeffler's seminal article "Greed and Grievance in Civil War"
(2004) posits the so-called "greed explanation" for civil war: variables
tied to opportunity costs
for attacking the government are significantly correlated with civil war
occurrence. On the other hand, "grievance" variables such as income
inequality and ethnic fractionalization are uncorrelated with civil wars. We
replicate the author's findings and offer an improved model for civil war
prediction by using multiple imputation, controlling for rare events, and
adding alternative measures of ethnic fractionalization. We find that
increased social inequality and its interaction with measures of ethnic
stratification have a significant effect on the probability of civil war.
Conversely, a number of "greed" variables identified by Collier and Hoeffler
no longer have explanatory power.
Hi all,
I'm having regular office hours today, in case you have any last minute
questions before the paper deadline on Monday.
Holger
--
Holger Lutz Kern
Ph.D. cand.
Department of Government
Cornell University
Institute for Quantitative Social Science
Harvard University
1737 Cambridge Street N350
Cambridge, MA 02138
www.people.cornell.edu/pages/hlk23
Thanks in advance!
Best,
Miya and Porsha
----------------------------------------------------------------
Title: Does Descriptive Representation Alienate Black Conservatives? The
Differential Effect of Ideology on Black Turnout
Abstract: In response to inconclusive research on descriptive representation
and African American turnout, Griffen and Keane (2006) examine how
ideological orientation conditions the effect of descriptive representation
on turnout. They find that while descriptive representation mobilizes
liberal African Americans it depresses turnout among conservative blacks. A
replication of Griffen's and Keane's analysis and simulations reflecting
uncertainty about African American liberal and conservative voting
propensities reveals that ideology *does not* condition the effect of
descriptive representation on conservative black turnout. Conservative
blacks are *no less likely* to participate when represented by African
American members of congress. These findings counter Griffen's and Keane's
claim that descriptive representation increases the political alienation of
conservative blacks.
--
K. Miya Woolfalk
Graduate Student
Harvard University
Government and Social Policy