Dear Applied Statistics Workshop Community,
Our next meeting of the semester will be on November 2 (12:00 EST). Edward
Kennedy will present "Doubly robust capture-recapture methods for
estimating population size."
Please note that this meeting will take a hybrid format. Professor Kennedy
will join us on Zoom, but we will gather in CGIS K354 (I will use the
projection screen in K354 to display the Zoom window). Lunch will be
provided.
<Where>
Hybrid: CGIS K354 or Zoom
Bagged lunches are available for pick-up at 11:40 (CGIS K354).
Zoom:
https://harvard.zoom.us/j/99181972207?pwd=Ykd3ZzVZRnZCSDZqNVpCSURCNnVvQT09
<Abstract>
Estimation of population size using incomplete lists (also called the
capture-recapture problem) has a long history across many biological and
social sciences. For example, human rights and other groups often construct
partial and overlapping lists of victims of armed conflicts, with the hope
of using this information to estimate the total number of victims. Earlier
statistical methods for this setup either use potentially restrictive
parametric assumptions, or else rely on typically suboptimal plug-in-type
nonparametric estimators; however, both approaches can lead to substantial
bias, the former via model misspecification and the latter via smoothing.
Under an identifying assumption that two lists are conditionally
independent given measured covariate information, we make several
contributions. First, we derive the nonparametric efficiency bound for
estimating the capture probability, which indicates the best possible
performance of any estimator, and sheds light on the statistical limits of
capture-recapture methods. Then we present a new estimator, and study its
finite-sample properties, showing that it has a double robustness property
new to capture-recapture, and that it is near-optimal in a non-asymptotic
sense, under relatively mild nonparametric conditions. Next, we give a
method for constructing confidence intervals for total population size from
generic capture probability estimators, and prove non-asymptotic
near-validity. Finally, we study our methods in simulations, and apply them
to estimate the number of killings and disappearances attributable to
different groups in Peru during its internal armed conflict between 1980
and 2000.
<2022 Schedule>
GOV 3009 Website:
https://projects.iq.harvard.edu/applied.stats.workshop-gov3009
Calendar:
https://calendar.google.com/calendar/embed?src=c_3v93pav9fjkkldrbu9snbhned8…
Best,
Shusei
Dear Applied Statistics Workshop Community,
Our presenter this week, Professor Himabindu Lakkaraju, canceled the talk
due to a recent teaching schedule change at HBS.
She kindly shared her presentation video with us:
https://drive.google.com/drive/folders/1Jxmrwd8JJ8meDcUvT7eLzGK_3gHvKy3u?us…
or
https://hbs.zoom.us/rec/share/8y6TM5TjMfEFHGyDEwMYLBctw52FVAC2RveazP3qz_Y6O…
Passcode: GCM?1&!9
Given this change in format, we will have an asynchronous Q&A this week (no
meeting at CGIS K354). If you have any questions, please send them to me at
shuseieshima(a)g.harvard.edu by the end of Wednesday. I will send the list of
questions to Professor Lakkaraju and share a reply.
Best,
Shusei
Dear Applied Statistics Workshop Community,
Unfortunately, our next meeting on October 19th got postponed due to the
presenter's availability.
Our next meeting will be next week, October 26th.
Best,
Shusei
Dear Applied Statistics Workshop Community,
This is a friendly reminder that *we're meeting on Zoom* today. Tom Leavitt
will present "Model selection for Decreasing Dependence on Counterfactual,
Identification Assumptions in Controlled Pre-Post Designs."
Please use this link:
https://harvard.zoom.us/j/99181972207?pwd=Ykd3ZzVZRnZCSDZqNVpCSURCNnVvQT09
Best,
Shusei
Dear Applied Statistics Workshop Community,
Our next meeting of the semester will be on October 12 (12:00 EST). Tom
Leavitt will present "Model selection for Decreasing Dependence on
Counterfactual, Identification Assumptions in Controlled Pre-Post Designs."
*Please note that we will meet on Zoom this week.*
*<Where (Zoom)>*
*Zoom Link*:
https://harvard.zoom.us/j/99181972207?pwd=Ykd3ZzVZRnZCSDZqNVpCSURCNnVvQT09
<Abstract>
Researchers often draw causal leverage from measures of outcomes before and
after treatment in both a treated group and an untreated, comparison group.
Such controlled pre-post designs, e.g., Difference-in-Differences and
Comparative-Interrupted-Time-Series, differ in terms of their predictive
models and associated counterfactual assumptions to identify the treated
group’s average effect (ATT). This paper derives a general, one-stop shop
counterfactual assumption — and associated sensitivity analysis — that
unifies the differently named assumptions of each design. While the
definition of our one-stop shop assumption is general, its validity depends
on the specific predictive model. Existing best practice in light of this
model dependence is to choose the model that is most plausible. However,
this practice is not especially useful when reasonable people disagree
about the plausibility of competing models. We instead propose a
cross-validation procedure that anticipates the results of a sensitivity
analysis to violations of our one-stop shop assumption and then chooses the
model that yields the least sensitivity. We formally show that our
procedure maximizes robustness to identification assumption violations
among a large class of predictive models, and then apply our procedure to
the debate about the effect of concealed-carry laws on violent crime. Our
method contributes to this debate, which has been stymied by the
sensitivity of researchers’ findings to minor changes in model
specifications, by choosing not the model that makes a counterfactual
assumption most plausible, but instead the model that makes our causal
conclusions least dependent on this counterfactual assumption.
<2022 Schedule>
GOV 3009 Website:
https://projects.iq.harvard.edu/applied.stats.workshop-gov3009
Calendar:
https://calendar.google.com/calendar/embed?src=c_3v93pav9fjkkldrbu9snbhned8…
Best,
Shusei
Dear Applied Statistics Workshop Community,
Our next meeting of the semester will be on October 5 (12:00 EST). Nima
Hejazi will present "Evaluating treatment efficacy in vaccine clinical
trials with two-phase designs using stochastic-interventional causal
effects."
<Where>
In-person: CGIS K354
Bagged lunches are available for pick-up at 11:40 (CGIS K354).
Zoom:
https://harvard.zoom.us/j/99181972207?pwd=Ykd3ZzVZRnZCSDZqNVpCSURCNnVvQT09
<Abstract>
In clinical trials randomizing participants to active vs. control
conditions and following study units until the occurrence of a primary
clinical endpoint, evaluating the efficacy of a quantitative exposure
(e.g., drug dosage, drug-induced biomarker activity) is often challenging,
as statistical innovations in causal inference have historically focused on
estimands compatible only with binary or categorical exposures.
Stochastic-interventional effects, which measure the causal effect
attributable to perturbing the exposure's natural (i.e., observed) value,
provide an interpretable solution. Unfortunately, their use in vaccine
efficacy trials requires extra care, for such trials measure immunologic
biomarkers – useful for understanding the mechanisms by which vaccines
confer protection or as surrogate endpoints in future trials – via
outcome-dependent two-phase sampling (e.g., case-cohort) designs. These
biased, outcome-dependent sampling designs have earned their popularity:
they circumvent the administrative burden of collecting potentially
expensive biomarker measurements on all study units without limiting
opportunities to detect important biomarkers that may be mechanistically
informative of the disease or infection process. We outline a
semiparametric biased sampling correction that allows for asymptotically
efficient inference on a causal vaccine efficacy measure defined by
contrasting assignments of study units to active vs. control while
simultaneously hypothetically shifting biomarker expression in the active
condition, yielding a causal dose-response analysis informative of
next-generation vaccine efficacy and useful for transporting efficacy from
a source pathogen strain (e.g., SARS-CoV-2 at outbreak) to variants of
concern (e.g., Omicron BA.4/BA.5). We present the results of applying this
approach in an analysis of the U.S. Government / COVID-19 Prevention
Network’s COVE (Moderna) COVID-19 vaccine efficacy clinical trial.
<2022 Schedule>
GOV 3009 Website:
https://projects.iq.harvard.edu/applied.stats.workshop-gov3009
Calendar:
https://calendar.google.com/calendar/embed?src=c_3v93pav9fjkkldrbu9snbhned8…
Best,
Shusei