Hi all,
I'm sorry to let you know that we won't be having Applied Statistics
this week due to speaker illness. We will, however, be back and ready
for action next Wednesday, 3/2, when we have Jean-Baptiste Michel and
Erez Lieberman Aiden presenting their exciting recent paper in
Science, "Quantitative Analysis of Culture Using Millions of Digitized
Books." If you haven't seen this work, you can take a look at their
website for culturomics to learn more:
http://www.culturomics.org/
Be sure to check out the Google Ngram Viewer, which is based on their work:
http://ngrams.googlelabs.com/
Cheers,
matt.
~~~~~~~~~~~
Matthew Blackwell
PhD Candidate
Institute for Quantitative Social Science
Department of Government
Harvard University
url: http://people.fas.harvard.edu/~blackwel/
Hello,
We hope you can join us tomorrow, Wednesday, February 16th at the
Applied Statistics Workshop when we will be happy to have Cassandra
Wolos Pattanayak presenting joint work with Jim Greiner. Their paper
focuses on an experiment where offers of legal representation were
randomized to claimants. You will find an abstract and a link to the
paper below. As usual, the workshop will begin at 12 noon with lunch
and wrap up at 1:30.
"What Difference Representation?” (with Jim Greiner)
Cassandra Wolos Pattanayak
Harvard Department of Statistics
February 16th, 2011, 12 noon
K354 CGIS Knafel (1737 Cambridge St)
Paper: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1708664&download=yes
Abstract:
We report the results of the first in a series of randomized control
trials designed to measure the effect of an offer of, and the actual
use of, legal representation. The results are unexpected. In the
context of administrative litigation to determine eligibility for
unemployment benefits, a service provider’s offer of representation to
a claimant had no statistically significant effect on the claimant’s
probability of a victory, but the offer caused a delay in the
proceeding. Because a substantial percentage of the provider’s client
base consisted of claimants who were initially denied benefits but who
would have that initial denial reversed as a result of the litigation,
the offer of representation inflicted a harm upon such claimants in
the form of an additional waiting time for benefits to begin, this
with no discernible increase in the probability of a favorable
outcome. In other words, within the limits of statistical uncertainty,
these claimants would have been better off without the offer of
representation. The size of the delay (a median effect of about 16
days, depending on how measured) was not large in absolute terms, and
would be negligible in many other legal settings, but was relevant in
the context of this particular administrative and legal framework, one
in which speed has remained an extraordinary concern for decades.
Moreover, in a small number of cases with a certain profile, the delay
caused the unemployment system to continue paying benefits erroneously
for a longer period of time, potentially imposing costs on the
financing of the unemployment system. We were also able to verify a
delay effect due to the actual use of (as opposed to an offer of)
representation; we could come to no firm conclusion on the effect of
actual use of representation on win loss.
We caution against both over- and under-generalization of these study
results. Stepping back, we use these results as a springboard for a
comprehensive review of the quantitative literature on the effect of
representation in civil proceedings. We find that this literature
provides virtually no credible information, excepting the results of
two randomized evaluations occurring in different legal contexts and
separated by over three decades. We conclude by advocating for, and
describing challenges associated with, a large program of randomized
evaluation of the provision of representation, particularly by legal
services providers.
Cheers,
matt.
~~~~~~~~~~~
Matthew Blackwell
PhD Candidate
Institute for Quantitative Social Science
Department of Government
Harvard University
url: http://people.fas.harvard.edu/~blackwel/
Hi all,
Please join us for the Applied Statistics Workshop next Wednesday,
February 9th, when we are excited to have Dustin Tingley from the
Department of Government presenting joint work with Kosuke Imai. You
will find a link to the paper below. As usual, the workshop will begin
with a light lunch at 12 noon, followed by the presentation at 12:15.
“A Statistical Method for Empirical Testing of Competing Theories”
(with Kosuke Imai)
Dustin Tingley
Assistant Professor, Departmtent of Government
Wednesday, February 9th, 12 noon
CGIS K354 (1737 Cambridge St.)
Paper: http://imai.princeton.edu/research/files/mixture.pdf
Cheers,
matt.
~~~~~~~~~~~
Matthew Blackwell
PhD Candidate
Institute for Quantitative Social Science
Department of Government
Harvard University
url: http://people.fas.harvard.edu/~blackwel/
Hi all,
Due to the poor weather and shut down of non-essential/non-critical
Harvard activities, we are cancelling tomorrow's Applied Statistics
Workshop. We have rescheduled Patrick Perry for April 13th, so we
won't miss his talk. We will return to normal programming next week
(2/9) with Dustin Tingley (as along as there is not another two feet
of snowfall that day).
Stay warm,
matt.
~~~~~~~~~~~
Matthew Blackwell
PhD Candidate
Institute for Quantitative Social Science
Department of Government
Harvard University
url: http://people.fas.harvard.edu/~blackwel/
Hi there,
We hope that you can join us for the Applied Statistics Workshop this
Wednesday, February 2nd when we will be happy to have Patrick Perry
from the Statistics and Information Sciences Laboratory here at
Harvard. Patrick will be presenting joint work with Patrick Wolfe on
directed interaction networks. You will find an abstract and link to
the paper below. As always, we will serve a light lunch and the talk
will begin around 12:15p.
“Point process modeling for directed interaction networks”
Patrick Perry and Patrick Wolfe
Statistics and Information Sciences Laboratory, Harvard University
CGIS K354 (1737 Cambridge St.)
Wednesday, February 2nd 12 noon
Abstract:
Network data often take the form of repeated interactions between
senders and receivers tabulated over time. Rather than reducing these
data to binary ties, a model is introduced for treating directed
interactions as a multivariate point process: a Cox multiplicative
intensity model using covariates that depend on the history of the
process. Consistency and asymptotic normality are proved for the
resulting partial-likelihood-based estimators under suitable
regularity conditions, and an efficient fitting procedure is
described. Multicast interactions--those involving a single sender but
multiple receivers--are treated explicitly. A motivating data example
shows the effects of reciprocation and group-level preferences on
message sending behavior in a corporate e-mail network.
Paper: http://arxiv.org/abs/1011.1703
Cheers,
matt.
~~~~~~~~~~~
Matthew Blackwell
PhD Candidate
Institute for Quantitative Social Science
Department of Government
Harvard University
url: http://people.fas.harvard.edu/~blackwel/