Triple Changes Estimator for Targeted Policies
Authors: Sina Akbari, Negar Kiyavash
ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Subsequently, we empirically evaluate the proposed framework and apply it to a study examining the impact of Medicaid expansion on children s preventive care. We conduct an empirical evaluation of our estimator on both synthetic and real datasets. |
| Researcher Affiliation | Academia | Sina Akbari 1 Negar Kiyavash 1 1EPFL, Switzerland. Correspondence to: Sina Akbari <sina.akbari@epfl.ch>. |
| Pseudocode | No | The paper provides mathematical formulas for its estimator (e.g., Eq. 13) but does not include any pseudocode or algorithm blocks. |
| Open Source Code | Yes | The code for reproducing the simulation results of this paper are accessible at https://github.com/Sina Akbarii/Triple-Changes. |
| Open Datasets | Yes | We utilized publicly available anonymous NSCH data for the years 2016 (t0) and 2017 (t1) to assess the impact of Medicaid expansion on children s access to preventive healthcare... https://mchb.hrsa.gov/national-survey-childrens-healthquestionnaires-datasets-supporting-documents |
| Dataset Splits | No | The paper describes generating synthetic data and using publicly available NSCH data, but it does not specify any training, validation, or test splits. It mentions using bootstrapping for confidence intervals, which is a resampling method, not a data split for model development. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware (e.g., CPU, GPU models, memory, or cloud instances) used for conducting the experiments. |
| Software Dependencies | No | The paper discusses the use of empirical and maximum likelihood estimators but does not specify any software packages, libraries, or their version numbers used for implementation. |
| Experiment Setup | Yes | For producing Figure 4(a), we sampled the latent variable U given s, d from a Gaussian distribution with mean νs,d and variance 1, where νs,ds were chosen as follows: ... We also generated the Y 0(t) counterfactuals as Y 0(t) = hs,d(u; t), where hs,d was defined as a linear function: hs,d(u; t) := 2u + 1 + s + 2t. For producing Figure 4(b), the model above was slightly modified as follows. First, we sampled U from a Gaussian distribution with the following means and variances given each group: ... Second, to add non-linearities to the model, two of the production functions were modified as: ... Applying the triple difference and triple changes estimators on the frequency of doctor visits for preventive purposes with 1000 bootstraps resulted in means of 0.170 and 0.145, respectively... |