Hi everyone!
This week is the final workshop of the semester and we will be welcoming *James
Robins*, Professor of Epidemiology at the Harvard School of Public Health.
He will be presenting work entitled *Confidence Intervals for Causal
Effects with Propensity Score and Outcome Regression Estimated with Machine
Learning: When are They Valid?*. Please find the abstract below and on the
Applied Stats website here
<https://projects.iq.harvard.edu/applied.stats.workshop-gov3009>.
As usual, we will meet at noon in CGIS Knafel Room 354 and lunch will be
provided. See you all there!
-- Dana Higgins
*Title:* *Confidence Intervals for Causal Effects with Propensity Score and
Outcome Regression Estimated with Machine Learning: When are They Valid?*
*Abstract:* In estimation of causal effects (such as the average causal
effect or the variance weighted average causal effect) in the presence of
high dimensional covariates sufficient to control confounding, an
increasingly popular procedure is to estimate the causal effect using
doubly robust estimators with the propensity score and outcome regression
estimated by machine learning and then to construct Wald confidence
intervals based on a estimator of the standard error. The validity of these
intervals depends critically on the assumption that the bias is less than
the standard error. If the latter assumption is wrong, the intervals will
undercover, perhaps dramatically. Can anything be done about this problem
since the bias of the estimator is unknown. Recently a number of approaches
to this problem have been offered. I will discuss these and then offer my
own approach which generally greatly improves upon alternatives..
Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Rebecca
Betensky*, Professor of Biostatistics at the Harvard School of Public Health.
She will be presenting work entitled *Nonidentifiability in the presence of
factorization for truncated data*. Please find the abstract below and on
the Applied Stats website here
<https://projects.iq.harvard.edu/applied.stats.workshop-gov3009>.
As usual, we will meet at noon in CGIS Knafel Room 354 and lunch will be
provided. See you all there!
-- Dana Higgins
*Title:* *Nonidentifiability in the presence of factorization for truncated
data*
*Abstract:* Truncation is a structured form of selection bias that arises
often in cohort studies. A time to event, X, is left truncated by T if X
can be observed only if T < X. This often results in over sampling of large
values of X, and necessitates adjustment of estimation procedures to avoid
bias. Simple risk-set adjustments can be made to standard risk-set based
estimators to accommodate left truncation as long as T and X are
“quasi-independent,” i.e., independent in the observable region. Through
examination of the likelihood function, we derive a weaker factorization
condition for the conditional distribution of T given X in the observable
region that likewise permits risk-set adjustment for estimation of the
distribution of X (but not T). Quasi-independence results when the
analogous factorization condition for X given T holds, as well, in which
case both distributions of X and T are easily estimated. While we can test
for factorization, if the test does not reject, we cannot identify which
factorization condition holds, or whether both (i.e., quasi-independence)
hold. Importantly, this means that we must ultimately make
an unidentifiable assumption in order to estimate the distribution of X
based on truncated data. This contrasts with common understanding that
truncation is distinct from censoring in that it does not require any
unidentifiable assumptions. We illustrate these concepts through examples
and a simulation study.
Hi everyone!
This week at the Applied Statistics Workshop we will be welcoming *Kenneth
Bollen*, Professor of Psychology and Neuroscience at the University of
North Carolina at Chapel Hill. He will be presenting work entitled *Model
Implied Instrumental Variables*. Please find the abstract below and on the
Applied Stats website here
<https://projects.iq.harvard.edu/applied.stats.workshop-gov3009>.
As usual, we will meet at noon in CGIS Knafel Room 354 and lunch will be
provided. See you all there!
-- Dana Higgins
*Title:* *Model Implied Instrumental Variables: An Alternative Orientation
to Structural Equation Models*
*Abstract:* It is hardly controversial to say that our models are
approximations to reality. Yet when it comes to estimating structural
equation models (SEMs), we use estimators that assume true models (e.g.,
ML) and that can spread bias through estimated parameters when the model is
approximate. This talk presents the Model Implied Instrumental Variable
(MIIV) approach to SEMs originally proposed in Bollen (1996). The MIIV
estimator using Two Stage Least Squares (2SLS) or MIIV-2SLS has greater
robustness to structural misspecifications and the conditions for
robustness are better understood than other estimators. In addition, the
MIIV-2SLS estimator is asymptotically distribution free. Furthermore,
MIIV-2SLS has equation based overidentification tests that can help
pinpoint errors in specification. Beyond these features, the MIIV approach
has other desirable qualities (e.g., a new test of dimensionality). MIIV
methods apply to higher order factor analyses, categorical measures, growth
curve models, dynamic factor analysis, and nonlinear latent variables.
Finally, it permits researchers to estimate and test only the
latent variable model or any other subset of equations. Despite these
promising features, research is needed to better understand its performance
under a variety of conditions that represent real world empirical examples.
In addition, other MIIV estimators beyond 2SLS are available. This
presentation will provide an overview of this new orientation to SEMs and
illustrate MIIVsem, an R package that implements it.