Continuous Treatment Effects with Surrogate Outcomes

Authors: Zhenghao Zeng, David Arbour, Avi Feller, Raghavendra Addanki, Ryan A. Rossi, Ritwik Sinha, Edward Kennedy

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive simulations show our methods enjoy appealing empirical performance.In this section we use simulations to evaluate the performance of the proposed methods.In this section we apply the proposed method to the Job Corps study
Researcher Affiliation Collaboration 1Department of Statistics and Data Science, Carnegie Mellon University, Pittsburgh, PA, USA 2Adobe Research, San Jose, CA, USA 3Goldman School of Public Policy and Department of Statistics, University of California, Berkeley, Berkeley, CA, USA.
Pseudocode Yes Algorithm 1 Doubly Robust Estimation
Open Source Code No The paper does not contain any explicit statements about releasing source code for the described methodology, nor does it provide a link to a code repository.
Open Datasets Yes We re-analyze their dataset publically available on Harvard dataverse (Huber, 2020)
Dataset Splits No The paper states We split the sample into two parts D, T and randomly separate the sample into two parts D, T for its estimators, but it does not provide specific split percentages or absolute sample counts.
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory) used for running its simulations or real data analysis.
Software Dependencies No The paper mentions software components like superlearner, generalized linear models, random forests, and kernel density estimator, but it does not specify any version numbers for these software dependencies.
Experiment Setup Yes For a fixed α we let ϵ1, . . . , ϵ4 N(n α, n 2α) and set bλ(V) = λ(V)+ϵ1... For sample size n {500, 2000} and convergence rate α {0.1, 0.13, . . . , 0.4}, we repeat the data generation and estimation process M = 500 times.We estimate the treatment effects θ(a) for each of a {100, 150, 200, . . . , 2000}