Topics in machine learning for causal inference with applications in social science
Abstract/Contents
- Abstract
- Random forests are a powerful method for non-parametric regression, but are limited in their ability to fit smooth signals. Taking the perspective of random forests as an adaptive kernel method, we pair the forest kernel with a local linear regression adjustment to better capture smoothness. The resulting procedure, local linear forests, enables us to improve on asymptotic rates of convergence for random forests with smooth signals, and provides substantial gains in accuracy on both real and simulated data. We prove a central limit theorem valid under regularity conditions on the forest and smoothness constraints, propose a computationally efficient construction for confidence intervals, and discuss an extension to local linear causal forests for learning heterogeneous treatment effects. Following this deep dive into local linear forests, we discuss two applications of machine learning for causal inference. The first is a retirement reform in Denmark, in which shifting eligibility ages for an early retirement program provide an opportunity to analyze heterogeneous treatment effects of the age of retirement eligibility. The second is a randomized controlled trial in Nairobi, Kenya, aiming to lower rates of gender-based violence against adolescent students living in informal settlements. In the latter example, we explore how local linear causal forests help to uncover and emphasize trends in marginalized student responses to the intervention. In both cases, we address how machine learning and causal inference are powerful tools to discover patterns in individual treatment effects, and to advocate for marginalized groups when estimates reveal troubling patterns in the data
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2020; ©2020 |
| Publication date | 2020; 2020 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Friedberg, Rina Siller |
|---|---|
| Degree supervisor | Athey, Susan |
| Thesis advisor | Athey, Susan |
| Thesis advisor | Hastie, Trevor |
| Thesis advisor | Owen, Art B |
| Degree committee member | Hastie, Trevor |
| Degree committee member | Owen, Art B |
| Associated with | Stanford University, Department of Statistics. |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Rina Siller Friedberg |
|---|---|
| Note | Submitted to the Department of Statistics |
| Thesis | Thesis Ph.D. Stanford University 2020 |
| Location | electronic resource |
Access conditions
- Copyright
- © 2020 by Rina Siller Friedberg
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...