Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Finale
Doshi-Velez*, Assistant Professor of Computer Science at Harvard
University. She will be presenting work entitled *Cross-Corpora Learning of
Trajectories in Autism Spectrum Disorders**.* Please find the abstract
below and on the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/presentations…>
.
As usual, we will meet in CGIS Knafel Room 354 from noon to 1:30pm, and
lunch will be provided. See you all there! To view previous Applied
Statistics presentations, please visit the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/videos>.
-- Aaron Kaufman
Title: Cross-Corpora Learning of Trajectories in Autism Spectrum Disorders
Abstract: Patients with developmental disorders, such as autism spectrum
disorder (ASD), present with symptoms that change with time even if the
named diagnosis remains fixed. For example, a child may have delayed speech
as a toddler and difficulty reading in elementary school. Characterizing
these trajectories is important for early treatment. However, deriving
these trajectories from observational sources is challenging: electronic
health records only reflect observations of patients at irregular intervals
and only record what factors are clinically relevant at the time of
observation. Meanwhile, caretakers discuss daily developments and concerns
on social media. I will present ongoing work on a fully unsupervised
approach for learning disease trajectories from incomplete medical records,
including records with only a single observation of each patient, combined
with disease descriptions from alternate data sources. In particular, we
use a dynamic topic model with efficient inference via polya-gamma
augmentation. We learn disease trajectories from the electronic health
records of 13,435 patients with ASD and the forum posts of 13,743
caretakers of children with ASD, deriving interesting clinical insights as
well as good predictions.
Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Jessica
Myers Franklin*, Assistant Professor at Harvard Medical School
and Biostatistician at Brigham & Women's Hospital. She will be presenting
work entitled *Comparing Marginal Estimators of Propensity-Adjusted
Treatment Effects in Studies With Few Observed Outcome Events.* Please
find the abstract below and on the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/presentations…>
.
As usual, we will meet in CGIS Knafel Room 354 from noon to 1:30pm, and
lunch will be provided. See you all there! To view previous Applied
Statistics presentations, please visit the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/videos>.
-- Aaron Kaufman
Title: Comparing marginal estimators of propensity-adjusted treatment
effects in studies with few observed outcome events
Abstract: Nonrandomized studies of treatments from electronic healthcare
databases are critical for producing the evidence necessary to making
informed treatment decisions, but often rely on comparing rates of events
observed in a small number of patients. In addition, a typical study
constructed from an electronic healthcare database, for example,
administrative claims data, requires adjustment for many, possibly
hundreds, of potential confounders. Despite the importance of maximizing
efficiency when there are many confounders and few observed outcome events,
there has been relatively little research on the performance of different
propensity score methods in this context. In this talk, I will describe and
compare a wide variety of propensity-adjusted estimators of the marginal
relative risk. In contrast to prior research that has focused on specific
statistical methods in isolation of other analytic choices, I instead
consider a method to be defined by the complete multi-step process from
propensity score modeling to final treatment effect estimation. I evaluate
methods via a “plasmode” simulation study, which creates simulated data
sets based on a real cohort study of 2 treatments constructed from
administrative claims data. Our results suggest a reconsideration of the
most popular approaches to propensity score adjustment in this context.
Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Jann
Spiess*, PhD Candidate in Economics at Harvard University. He will be
presenting work joint work with Alberto Abadie entitled *Robust
Post-Matching Inference**.* Please find the abstract below and on the
website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/presentations…>
.
As usual, we will meet in CGIS Knafel Room 354 from noon to 1:30pm, and
lunch will be provided. See you all there! To view previous Applied
Statistics presentations, please visit the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/videos>.
-- Aaron Kaufman
Title: Robust Post-Matching Inference
Nearest-neighbor matching (Cochran, 1953; Rubin, 1973) is a popular
nonparametric tool to create balance between treatment and control groups
in non-experimental data. As a preprocessing step for regression analysis,
it reduces the dependence on parametric modeling assumptions (Ho et al.,
2007). In this paper, we show how to obtain valid standard error estimates
for linear regression after nearest-neighbor matching without replacement.
We show that standard error estimates that ignore the matching step are not
generally valid if the regression model is misspecified, and can lead to
severe over- or underestimation of the asymptotic variance of the
post-matching estimator. As a remedy, we offer two easily implementable
tools for inference that are robust to misspecification: First, we show
that standard errors clustered at the match level are valid. Second, we
show that a simple nonparametric block bootstrap procedure yields a valid
approximation of the distribution of the post-matching estimator. A
simulation study and an empirical example demonstrate the importance and
usefulness of our results.
Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Hanna
Wallach*, Senior Researcher in the Microsoft Research NYC office and
Associate Professor of Computer Science at UMass Amherst. She will be
presenting work entitled *Modeling Topic-Partitioned Network Structure**.*
I will post her abstract to the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/presentations…>
as
soon as she writes it..
As usual, we will meet in CGIS Knafel Room 354 from noon to 1:30pm, and
lunch will be provided. See you all there! To view
previous Applied Statistics presentations, please visit the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/videos>.
-- Aaron Kaufman
Title: Modeling Topic-Partitioned Network Structure
Abstract: TBA
Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Nicole
Immorlica*, Researcher in the Microsoft Research New England Theory Group,
where she studies algorithmic game theory. She will be presenting work
entitled *The Degree of Segregation in Social Networks**.* Please find the
abstract below and on the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/presentations…>
.
As usual, we will meet in CGIS Knafel Room 354 from noon to 1:30pm, and
lunch will be provided. See you all there! To view previous Applied
Statistics presentations, please visit the website
<http://projects.iq.harvard.edu/applied.stats.workshop-gov3009/videos>.
-- Aaron Kaufman
Title: The Degree of Segregation in Social Networks
In 1969, economist Thomas Schelling introduced a landmark model of racial
segregation in which individuals choose residences based on the racial
composition of the corresponding neighborhoods. Simple simulations of
Schelling's model suggest this local behavior can cause segregation even
for racially tolerant individuals. In this talk, we provide rigorous
analyses of the degree of segregation in Schelling's model on
one-dimensional and two-dimensional lattices. We see that if agents refuse
to live in neighborhood in which their type constitutes a strict minority,
then the outcome is nearly integrated: the average size of an
ethnically-homogeneous region is independent of the size of the society and
only polynomial in the size of the neighborhood. A natural question arises
regarding how tolerance impacts segregation. We show the surprising result
that tolerance can actually increase segregation: the average size of an
ethnically-homogeneous region is now exponential in the size of the
neighborhood.
--
Aaron R Kaufman
PhD Candidate, Harvard University
Department of Government
818.263.5583