Targeting for long-term outcomes

Yang, Jeremy J.; Eckles, Dean; Dhillon, Paramveer S.; Aral, Sinan

doi:10.48550/arxiv.2010.15835

Cited by 5 publications

(6 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following Athey et al [2019], some recent literature also combine experimental and observational data, and rely on the statistical surrogate criterion, either to estimate cumulative treatment effects in dynamic settings [Battocchi et al, 2021] or learn optimal treatment policies [Yang et al, 2020, Cai et al, 2021b. Chen and Ritzwoller [2021] derive the efficiency lower bound for average longterm treatment effect in settings of Athey et al [2019] and Athey et al [2020].…”

Section: Data Combination For Long-term Causal Inference and Decision...mentioning

confidence: 99%

“…Empirical researchers and decision-makers are often interested in learning the long-term treatment effects of interventions. For example, economists are interested in the effect of early childhood education on lifetime earnings [Chetty et al, 2011], marketing practitioners are interested in the effects of incentives on customers' long-term behaviors [Yang et al, 2020], IT companies are interested in the effects of webpage designs on users' long-term behaviors [Hohnhold et al, 2015]. Since a long-term effect can be quite different from short-term effects [Kohavi et al, 2012], accurately evaluating the long-term effect is crucial for comprehensively understanding the intervention of interest.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Long-term Causal Inference Under Persistent Confounding via Data Combination

Imbens¹,

Kallus²,

Mao³

et al. 2022

Preprint

View full text Add to dashboard Cite

We study the identification and estimation of long-term treatment effects when both experimental and observational data are available. Since long-term outcome is observed only after a long delay, it is not measured in the experimental data, but only recorded in the observational data. However, both types of data include observations of some short-term outcomes. In this paper, we uniquely tackle the challenge of persistent unmeasured confounders, i.e., some unmeasured confounders that can simultaneously affect the treatment, short-term outcomes and the long-term outcome, noting that they invalidate identification strategies in previous literature. To address this challenge, we exploit the sequential structure of multiple short-term outcomes, and develop three novel identification strategies for the average long-term treatment effect. We further propose three corresponding estimators and prove their asymptotic consistency and asymptotic normality. We finally apply our methods to estimate the effect of a job training program on long-term employment using semi-synthetic data. We numerically show that our proposals outperform existing methods that fail to handle persistent confounders.

show abstract

Section: Data Combination For Long-term Causal Inference and Decision...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Long-term Causal Inference Under Persistent Confounding via Data Combination

Imbens¹,

Kallus²,

Mao³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In such cases, it is common to use the effect of an intervention on surrogate outcomes as a proxy for its effect on long-term outcomes. Recently, Yang al. (2020) propose how to use surrogate outcomes to impute the missing long-term outcomes and use the imputed long-term outcomes to optimize a targeting policy; they demonstrate their approach in the context of proactive churn management.…”

Section: Leveraging Alternative Target Variablesmentioning

confidence: 99%

Causal Decision Making and Causal Effect Estimation Are Not the Same…and Why It Matters

Fernández-Loría

Provost

2022

INFORMS Journal on Data Science

View full text Add to dashboard Cite

Causal decision making (CDM) at scale has become a routine part of business, and increasingly, CDM is based on statistical models and machine learning algorithms. Businesses algorithmically target offers, incentives, and recommendations to affect consumer behavior. Recently, we have seen an acceleration of research related to CDM and causal effect estimation (CEE) using machine-learned models. This article highlights an important perspective: CDM is not the same as CEE, and counterintuitively, accurate CEE is not necessary for accurate CDM. Our experience is that this is not well understood by practitioners or most researchers. Technically, the estimand of interest is different, and this has important implications both for modeling and for the use of statistical models for CDM. We draw on recent research to highlight three implications. (1) We should carefully consider the objective function of the causal machine learning, and if possible, optimize for accurate “treatment assignment” rather than for accurate effect-size estimation. (2) Confounding affects CDM and CEE differently. The upshot here is that for supporting CDM it may be just as good or even better to learn with confounded data as with unconfounded data. (3) Causal statistical modeling may not be necessary at all to support CDM because a proxy target for statistical modeling might do as well or better. This third observation helps to explain at least one broad common CDM practice that seems “wrong” at first blush—the widespread use of noncausal models for targeting interventions. The last two implications are particularly important in practice, as acquiring (unconfounded) data on both “sides” of the counterfactual for modeling can be quite costly and often impracticable. These observations open substantial research ground. We hope to facilitate research in this area by pointing to related articles from multiple contributing fields, most of them written in the last five years.

show abstract

“…Recently, Yang et al (2020) propose how to use surrogate outcomes to impute the missing long-term outcomes and use the imputed long-term outcomes to optimize a targeting policy; they demonstrate their approach in the context of proactive churn management. The use of short-term proxies to estimate long-term treatment effects is discussed in more detail by .…”

Section: Leveraging Alternative Target Variablesmentioning

confidence: 99%

Causal Decision Making and Causal Effect Estimation Are Not the Same... and Why It Matters

Fernández-Loría¹,

Provost²

2021

Preprint

View full text Add to dashboard Cite

Over the past two decades, causal decision making (CDM) at scale has become a routine part of business, and increasingly CDM is based on statistical models and machine learning algorithms. For example, businesses target offers, incentives, and recommendations algorithmically with the goal of affecting consumer behavior.Not surprisingly, recently we have seen an acceleration of research related to CDM and to causal effect estimation (CEE) using machine learned models. This article highlights an important perspective: CDM is not the same as CEE, and counterintuitively, accurate CEE is not necessary for accurate CDM. Our experience is that this is not well understood by practitioners nor by most researchers. Technically, the estimand of interest is different, and this has important implications both for modeling and for the use of statistical models for CDM. We draw on recent research to highlight three of these implications. (1) We should consider carefully the objective function of the causal machine learning, and if possible we should optimize for accurate "treatment assignment" rather than for accurate effect-size estimation. (2) Confounding does not have the same effect on CDM as it does on CEE. The upshot here is that for supporting CDM it may be just as good or even better to learn with confounded data as with unconfounded data. Finally, (3) causal statistical modeling may not be necessary at all to support CDM, because there may be (and perhaps often is) a proxy target for statistical modeling that can do as well or better. This observation helps to explain at least one broad common CDM practice that seems "wrong" at first blush-the widespread use of noncausal models for targeting interventions. The last two implications are particularly important in practice, as acquiring (unconfounded) data on both "sides" of the counterfactual for modeling can be quite costly and often impracticable. Our perspective is that these observations open up substantial fertile ground for future research. Whether or not you share our perspective completely, we hope we facilitate future research in this area by pointing to related articles from multiple contributing fields, including two dozen articles published the last three to four years.

show abstract

Targeting for long-term outcomes

Cited by 5 publications

References 41 publications

Long-term Causal Inference Under Persistent Confounding via Data Combination

Long-term Causal Inference Under Persistent Confounding via Data Combination

Causal Decision Making and Causal Effect Estimation Are Not the Same…and Why It Matters

Causal Decision Making and Causal Effect Estimation Are Not the Same... and Why It Matters

Contact Info

Product

Resources

About