Retrospective causal inference with machine learning ensembles

An application to anti-recidivism policies in Colombia

Cyrus Samii, Laura Paler, Sarah Zukerman Daly

    Research output: Contribution to journalArticle

    Abstract

    We present new methods to estimate causal effects retrospectively from micro data with the assistance of a machine learning ensemble. This approach overcomes two important limitations in conventional methods like regression modeling or matching: (i) ambiguity about the pertinent retrospective counterfactuals and (ii) potential misspecification, overfitting, and otherwise bias-prone or inefficient use of a large identifying covariate set in the estimation of causal effects. Our method targets the analysis toward a well-defined "retrospective intervention effect" based on hypothetical population interventions and applies a machine learning ensemble that allows data to guide us, in a controlled fashion, on how to use a large identifying covariate set. We illustrate with an analysis of policy options for reducing ex-combatant recidivism in Colombia.

    Original languageEnglish (US)
    Pages (from-to)434-456
    Number of pages23
    JournalPolitical Analysis
    Volume24
    Issue number4
    DOIs
    StatePublished - Oct 1 2016

    Fingerprint

    Colombia
    learning
    assistance
    regression
    trend

    ASJC Scopus subject areas

    • Sociology and Political Science

    Cite this

    Retrospective causal inference with machine learning ensembles : An application to anti-recidivism policies in Colombia. / Samii, Cyrus; Paler, Laura; Daly, Sarah Zukerman.

    In: Political Analysis, Vol. 24, No. 4, 01.10.2016, p. 434-456.

    Research output: Contribution to journalArticle

    Samii, Cyrus ; Paler, Laura ; Daly, Sarah Zukerman. / Retrospective causal inference with machine learning ensembles : An application to anti-recidivism policies in Colombia. In: Political Analysis. 2016 ; Vol. 24, No. 4. pp. 434-456.
    @article{0bcf1394a3cd43aca512d8bfdf20449b,
    title = "Retrospective causal inference with machine learning ensembles: An application to anti-recidivism policies in Colombia",
    abstract = "We present new methods to estimate causal effects retrospectively from micro data with the assistance of a machine learning ensemble. This approach overcomes two important limitations in conventional methods like regression modeling or matching: (i) ambiguity about the pertinent retrospective counterfactuals and (ii) potential misspecification, overfitting, and otherwise bias-prone or inefficient use of a large identifying covariate set in the estimation of causal effects. Our method targets the analysis toward a well-defined {"}retrospective intervention effect{"} based on hypothetical population interventions and applies a machine learning ensemble that allows data to guide us, in a controlled fashion, on how to use a large identifying covariate set. We illustrate with an analysis of policy options for reducing ex-combatant recidivism in Colombia.",
    author = "Cyrus Samii and Laura Paler and Daly, {Sarah Zukerman}",
    year = "2016",
    month = "10",
    day = "1",
    doi = "10.1093/pan/mpw019",
    language = "English (US)",
    volume = "24",
    pages = "434--456",
    journal = "Political Analysis",
    issn = "1047-1987",
    publisher = "Oxford University Press",
    number = "4",

    }

    TY - JOUR

    T1 - Retrospective causal inference with machine learning ensembles

    T2 - An application to anti-recidivism policies in Colombia

    AU - Samii, Cyrus

    AU - Paler, Laura

    AU - Daly, Sarah Zukerman

    PY - 2016/10/1

    Y1 - 2016/10/1

    N2 - We present new methods to estimate causal effects retrospectively from micro data with the assistance of a machine learning ensemble. This approach overcomes two important limitations in conventional methods like regression modeling or matching: (i) ambiguity about the pertinent retrospective counterfactuals and (ii) potential misspecification, overfitting, and otherwise bias-prone or inefficient use of a large identifying covariate set in the estimation of causal effects. Our method targets the analysis toward a well-defined "retrospective intervention effect" based on hypothetical population interventions and applies a machine learning ensemble that allows data to guide us, in a controlled fashion, on how to use a large identifying covariate set. We illustrate with an analysis of policy options for reducing ex-combatant recidivism in Colombia.

    AB - We present new methods to estimate causal effects retrospectively from micro data with the assistance of a machine learning ensemble. This approach overcomes two important limitations in conventional methods like regression modeling or matching: (i) ambiguity about the pertinent retrospective counterfactuals and (ii) potential misspecification, overfitting, and otherwise bias-prone or inefficient use of a large identifying covariate set in the estimation of causal effects. Our method targets the analysis toward a well-defined "retrospective intervention effect" based on hypothetical population interventions and applies a machine learning ensemble that allows data to guide us, in a controlled fashion, on how to use a large identifying covariate set. We illustrate with an analysis of policy options for reducing ex-combatant recidivism in Colombia.

    UR - http://www.scopus.com/inward/record.url?scp=84995768861&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84995768861&partnerID=8YFLogxK

    U2 - 10.1093/pan/mpw019

    DO - 10.1093/pan/mpw019

    M3 - Article

    VL - 24

    SP - 434

    EP - 456

    JO - Political Analysis

    JF - Political Analysis

    SN - 1047-1987

    IS - 4

    ER -