Estimating average causal effects under general interference, with application to a social network experiment

Peter M. Aronow, Cyrus Samii

    Research output: Contribution to journalArticle

    Abstract

    This paper presents a randomization-based framework for estimating causal effects under interference between units motivated by challenges that arise in analyzing experiments on social networks. The framework integrates three components: (i) an experimental design that defines the probability distribution of treatment assignments, (ii) a mapping that relates experimental treatment assignments to exposures received by units in the experiment, and (iii) estimands that make use of the experiment to answer questions of substantive interest. We develop the case of estimating average unit-level causal effects from a randomized experiment with interference of arbitrary but known form. The resulting estimators are based on inverse probability weighting.We provide randomization-based variance estimators that account for the complex clustering that can occur when interference is present. We also establish consistency and asymptotic normality under local dependence assumptions. We discuss refinements including covariate-adjusted effect estimators and ratio estimation. We evaluate empirical performance in realistic settings with a naturalistic simulation using social network data from American schools. We then present results from a field experiment on the spread of anti-conflict norms and behavior among school students.

    Original languageEnglish (US)
    Pages (from-to)1912-1947
    Number of pages36
    JournalAnnals of Applied Statistics
    Volume11
    Issue number4
    DOIs
    StatePublished - Dec 1 2017

    Fingerprint

    Causal Effect
    Social Networks
    Interference
    Randomisation
    Unit
    Assignment
    Inverse Probability Weighting
    Randomized Experiments
    Experiment
    Estimator
    Field Experiment
    Variance Estimator
    Experiments
    Experimental design
    Asymptotic Normality
    Covariates
    Refinement
    Probability Distribution
    Integrate
    Clustering

    Keywords

    • Causal inference
    • Interference
    • Networks
    • Potential outcomes
    • Randomization inference
    • SUTVA

    ASJC Scopus subject areas

    • Statistics and Probability
    • Modeling and Simulation
    • Statistics, Probability and Uncertainty

    Cite this

    Estimating average causal effects under general interference, with application to a social network experiment. / Aronow, Peter M.; Samii, Cyrus.

    In: Annals of Applied Statistics, Vol. 11, No. 4, 01.12.2017, p. 1912-1947.

    Research output: Contribution to journalArticle

    @article{55c0b719ea5446269af271c0f6fdcfda,
    title = "Estimating average causal effects under general interference, with application to a social network experiment",
    abstract = "This paper presents a randomization-based framework for estimating causal effects under interference between units motivated by challenges that arise in analyzing experiments on social networks. The framework integrates three components: (i) an experimental design that defines the probability distribution of treatment assignments, (ii) a mapping that relates experimental treatment assignments to exposures received by units in the experiment, and (iii) estimands that make use of the experiment to answer questions of substantive interest. We develop the case of estimating average unit-level causal effects from a randomized experiment with interference of arbitrary but known form. The resulting estimators are based on inverse probability weighting.We provide randomization-based variance estimators that account for the complex clustering that can occur when interference is present. We also establish consistency and asymptotic normality under local dependence assumptions. We discuss refinements including covariate-adjusted effect estimators and ratio estimation. We evaluate empirical performance in realistic settings with a naturalistic simulation using social network data from American schools. We then present results from a field experiment on the spread of anti-conflict norms and behavior among school students.",
    keywords = "Causal inference, Interference, Networks, Potential outcomes, Randomization inference, SUTVA",
    author = "Aronow, {Peter M.} and Cyrus Samii",
    year = "2017",
    month = "12",
    day = "1",
    doi = "10.1214/16-AOAS1005",
    language = "English (US)",
    volume = "11",
    pages = "1912--1947",
    journal = "Annals of Applied Statistics",
    issn = "1932-6157",
    publisher = "Institute of Mathematical Statistics",
    number = "4",

    }

    TY - JOUR

    T1 - Estimating average causal effects under general interference, with application to a social network experiment

    AU - Aronow, Peter M.

    AU - Samii, Cyrus

    PY - 2017/12/1

    Y1 - 2017/12/1

    N2 - This paper presents a randomization-based framework for estimating causal effects under interference between units motivated by challenges that arise in analyzing experiments on social networks. The framework integrates three components: (i) an experimental design that defines the probability distribution of treatment assignments, (ii) a mapping that relates experimental treatment assignments to exposures received by units in the experiment, and (iii) estimands that make use of the experiment to answer questions of substantive interest. We develop the case of estimating average unit-level causal effects from a randomized experiment with interference of arbitrary but known form. The resulting estimators are based on inverse probability weighting.We provide randomization-based variance estimators that account for the complex clustering that can occur when interference is present. We also establish consistency and asymptotic normality under local dependence assumptions. We discuss refinements including covariate-adjusted effect estimators and ratio estimation. We evaluate empirical performance in realistic settings with a naturalistic simulation using social network data from American schools. We then present results from a field experiment on the spread of anti-conflict norms and behavior among school students.

    AB - This paper presents a randomization-based framework for estimating causal effects under interference between units motivated by challenges that arise in analyzing experiments on social networks. The framework integrates three components: (i) an experimental design that defines the probability distribution of treatment assignments, (ii) a mapping that relates experimental treatment assignments to exposures received by units in the experiment, and (iii) estimands that make use of the experiment to answer questions of substantive interest. We develop the case of estimating average unit-level causal effects from a randomized experiment with interference of arbitrary but known form. The resulting estimators are based on inverse probability weighting.We provide randomization-based variance estimators that account for the complex clustering that can occur when interference is present. We also establish consistency and asymptotic normality under local dependence assumptions. We discuss refinements including covariate-adjusted effect estimators and ratio estimation. We evaluate empirical performance in realistic settings with a naturalistic simulation using social network data from American schools. We then present results from a field experiment on the spread of anti-conflict norms and behavior among school students.

    KW - Causal inference

    KW - Interference

    KW - Networks

    KW - Potential outcomes

    KW - Randomization inference

    KW - SUTVA

    UR - http://www.scopus.com/inward/record.url?scp=85037347959&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85037347959&partnerID=8YFLogxK

    U2 - 10.1214/16-AOAS1005

    DO - 10.1214/16-AOAS1005

    M3 - Article

    VL - 11

    SP - 1912

    EP - 1947

    JO - Annals of Applied Statistics

    JF - Annals of Applied Statistics

    SN - 1932-6157

    IS - 4

    ER -