Long Story Short: Omitted Variable Bias in Causal Machine Learning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2021
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914812012789760 |
|---|---|
| author | Chernozhukov, Victor Cinelli, Carlos Newey, Whitney Sharma, Amit Syrgkanis, Vasilis |
| author_facet | Chernozhukov, Victor Cinelli, Carlos Newey, Whitney Sharma, Amit Syrgkanis, Vasilis |
| contents | We develop a general theory of omitted variable bias for a wide range of common causal parameters, including (but not limited to) averages of potential outcomes, average treatment effects, average causal derivatives, and policy effects from covariate shifts. Our theory applies to nonparametric models, while naturally allowing for (semi-)parametric restrictions (such as partial linearity) when such assumptions are made. We show how simple plausibility judgments on the maximum explanatory power of omitted variables are sufficient to bound the magnitude of the bias, thus facilitating sensitivity analysis in otherwise complex, nonlinear models. Finally, we provide flexible and efficient statistical inference methods for the bounds, which can leverage modern machine learning algorithms for estimation. These results allow empirical researchers to perform sensitivity analyses in a flexible class of machine-learned causal models using very simple, and interpretable, tools. We demonstrate the utility of our approach with two empirical examples. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2112_13398 |
| institution | arXiv |
| publishDate | 2021 |
| record_format | arxiv |
| spellingShingle | Long Story Short: Omitted Variable Bias in Causal Machine Learning Chernozhukov, Victor Cinelli, Carlos Newey, Whitney Sharma, Amit Syrgkanis, Vasilis Econometrics Machine Learning Methodology 62G We develop a general theory of omitted variable bias for a wide range of common causal parameters, including (but not limited to) averages of potential outcomes, average treatment effects, average causal derivatives, and policy effects from covariate shifts. Our theory applies to nonparametric models, while naturally allowing for (semi-)parametric restrictions (such as partial linearity) when such assumptions are made. We show how simple plausibility judgments on the maximum explanatory power of omitted variables are sufficient to bound the magnitude of the bias, thus facilitating sensitivity analysis in otherwise complex, nonlinear models. Finally, we provide flexible and efficient statistical inference methods for the bounds, which can leverage modern machine learning algorithms for estimation. These results allow empirical researchers to perform sensitivity analyses in a flexible class of machine-learned causal models using very simple, and interpretable, tools. We demonstrate the utility of our approach with two empirical examples. |
| title | Long Story Short: Omitted Variable Bias in Causal Machine Learning |
| topic | Econometrics Machine Learning Methodology 62G |
| url | https://arxiv.org/abs/2112.13398 |