Dear Applied Statistics Community,
Please join us this Wednesday when Arthur Spirling, Department of
Government, will present ``Bargaining Power in Practice: US Treaty-making
with American Indians, 1784--1911". Arthur provided the following overview
for his talk:
I will discuss a new data set of treaties signed 1784--1911 between the
United States government and American Indian tribes, and comment on some
early findings using kernel methods to analyze these texts. I particularly
welcome feedback and suggestions from the ASW on the appropriateness of the
techniques given the problem at hand.
Arthur also provided the following abstract for a paper that is the basis
for his talk:
Native Americans are unique among domestic actors in that their relations
with the United States government involve treaty-making, with almost 600
such documents signed between the Revolutionary War and the turn of the
twentieth century. We obtain and digitize all of these treaties for textual
analysis. In particular, we employ new 'kernel methods' to study the
evolution of their nature over time and show that the Indian Removal Act of
1830 represents a systematic shift in language. We relate our findings to a
bargaining model with the parties---government and tribes---varying in power
according to contemporary political and economic events. With a mind to
earlier historical and legal literatures, we also show that the 'broken'
treaties do not form their own cluster in the data, and that the post-1871
'agreements' represent a straightforward continuation of earlier treaty
policy in both style and substance.
The Applied Statistics Workshop meets each Wednesday at 12 noon in K-354
CGIS-Knafel (1737 Cambridge St). The workshop begins with a light lunch and
presentations usually start around 1215 and last until about 130 pm.
All are welcome--
Justin Grimmer
Dear Applied Statistics Workshop Participants,
Please join us this Wednesday when Gabriel Lenz, MIT Department of Political
Science, will present ``Getting Rich(er) in Office? Corruption and Wealth
Accumulation in Congress", work that is joint with Kevin Lim. Gabe provided
the following abstract:
How corrupt is Congress? We provide an indirect test by comparing wealth
accumulation from 1995 to 2005 among members of the U.S. House of
Representatives and members of the public. Data on representatives are from
Personal Financial Disclosure forms and data on the public are from the
Panel Study of Income Dynamics (PSID). To test whether representatives
accumulate wealth at a faster rate than expected, we construct
counterfactuals based on the PSID with two approaches. We first use
statistical models, conditioning on asset distribution over stocks, bonds,
businesses, and land, as well as demographic variables. These models find
representatives accumulating wealth about 20 percent faster than expected.
Second, we employ matching. Unlike the modeling approach, matching finds an
almost identical rate of wealth accumulation among both groups. Further
analysis reveals that matching reduces bias from several incorrect
functional form assumptions in the statistical models. We thus conclude that
representatives report accumulating wealth at a rate consistent with similar
non-representatives, suggesting no aggregate corruption. Besides examining
overall wealth accumulation, we also test for effects of committee
assignments, safe seats, career trajectories, and campaign contributions
The Applied Statistics Workshop meets each Wednesday at 12 noon in K-354
CGIS-Knafel (1737 Cambridge St). The workshop begins with a light lunch and
presentations usually start around 1215 and last until about 130 pm.
Hope you can make it
Cheers
Justin Grimmer
Dear Applied Statistics Workshop,
Please join us this Wednesday when Dan Hopkins, Post-Doctoral Fellow at
Harvard University (and soon to be Assistant Professor at Georgetown), will
present "Making Credible Inferences about the Effects of Local Contexts".
Dan provided the following abstract for his presentation:
In the last decade, there has been an explosion of social science research
exploring the influence of local contexts on attitudes and behavior. Yet
such studies face methodological hurdles, including the endogeneity of
individuals' moving decisions, significant measurement error, and ambiguity
about their causal interpretation. This presentation reconceptualizes the
effects of local contexts as an interaction between the local context and
salient national issues. It then uses panel or time-series cross-sectional
data to explore the impact of exogenous changes in the salience of national
issues on local contextual effects. Across three empirical examples on
attitudes toward immigration drawn from two countries, we observe that local
contexts only correlate with attitudes when immigration is a nationally
salient issue. The effects of local contexts vary in predictable ways with
the topics of national politics. All politics might not be local after all.
Dan provided the following paper as background for his talk:
http://people.iq.harvard.edu/~dhopkins/immpap75tot.pdf
The Applied Statistics Workshop meets each Wednesday in room K354, 1737
Cambridge St (CGIS-Knafel). A light lunch is served at 12 noon, with
presentations usually beginning at 1215 pm and the workshop usually
concludes by 130 pm. All are welcome!
Cheers
Justin Grimmer
Dear Applied Statistics Workshop,
Please join us this Wednesday, March 4th when Jamie Robins will present ``A
Bold Vision of Artificial Intelligence and Philosophy: Finding Causal
Effects Without Background Knowledge or Statistical Independences", a
project that is joint with Thomas Richardson, Ilya Shpitser, and Steffen
Lauritzen. Jamie provided the following abstract:
I describe a statistical methodology based on philosophy, causal directed
acyclic graphs, and a pinch of magic and miracle that holds the promise of
making a silk purse of causal knowledge out of the sow's ear of an
observational data set with no obvious structure. In 10 years or so, for
better or worse,
this methodology may become part of mainstream genomics.
The workshop will meet at 12 noon in room K-354, CGIS-Knafel (1737 Cambridge
St) with a light lunch served. The presentation will begin at 1215 and
usually ends around 130 pm. All are welcome
Cheers,
Justin Grimmer