Alyssa Carlson
University of Missouri-Columbia, MO·Social Science·Early-career
Measuring whether a policy actually worked is harder than it sounds. People who enroll in a program, or states that adopt a law, usually differ from those that don't in ways no dataset captures.
- Fields
Social Science
Mathematics & Statistics
- Topics
econometrics
causal inference
endogeneity
- Grants
- 1 known grant, through May 2029
What they work on
- Establish identification in triangular random coefficient models with continuous endogenous variables using a general control function that does not require scalar heterogeneity, the assumption of a single hidden first-stage factor.
- Extend the general control function approach to binary endogenous variables, the policy-evaluation case, and recover local average treatment effects without the first-stage monotonicity assumption the current literature depends on.
- Quantify how sensitive control function estimates are to misspecified nonlinearity and heteroskedasticity, and derive guidance for empirical researchers on building specifications flexible enough to avoid these problems.
- Build tests for whether the identifying assumptions behind these methods are realistic, and distribute implementations as code for popular statistical software in public repositories.
More about the research
Economists call this endogeneity, and if it is ignored, estimated policy effects can be badly off. Statistical fixes exist, but each rests on assumptions about the data that may be unrealistic in real economic settings. This project builds tools for estimating causal effects under weaker, more believable assumptions, plus tests that check whether the standard assumptions hold at all.
Why it matters
Policy decisions lean on empirical estimates of what programs do. When those estimates depend on assumptions that fail in practice, conclusions can mislead. Methods valid under weaker assumptions, together with tests of whether standard assumptions are credible, could make causal estimates more trustworthy for economists, program evaluators and policymakers, and the publicly posted software would let applied researchers use them directly.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Caleb Miles
Columbia University, NY·Mathematics & Statistics
Develop simple parametric and semiparametric efficient estimators of the average treatment effect in multi-study, multi-outcome data fusion, using related follow-up measurements as proxies for outcomes missing in some trials.
Matias Cattaneo
Princeton University, NJ·Mathematics & Statistics
Demonstrate the empirical and theoretical limitations of binned regression models applied to mixed-frequency panel data in climate economics, showing when they can lead to incorrect conclusions.
Nicole Pashley
Rutgers University-New Brunswick, NJ·Mathematics & Statistics
Develop randomization-based methods to analyze incomplete block designs, where each site runs only a random subset of the treatments, using the potential outcomes framework that relies on fewer assumptions than model-based approaches.
The school
University of Missouri-Columbia: 21 labs on PI Pages, $509M a year on research, 260 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.