Beyond conditional averages: Estimating the individual causal effect distribution
In recent years, the field of causal inference from observational data has emerged rapidly. The literature has focused on (conditional) average causal effect estimation. When (remaining) variability of individual causal effects (ICEs) is considerable, average effects may be uninformative for an indi...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
De Gruyter
2025-05-01
|
| Series: | Journal of Causal Inference |
| Subjects: | |
| Online Access: | https://doi.org/10.1515/jci-2024-0007 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850271441406132224 |
|---|---|
| author | Post Richard A. J. van den Heuvel Edwin R. |
| author_facet | Post Richard A. J. van den Heuvel Edwin R. |
| author_sort | Post Richard A. J. |
| collection | DOAJ |
| description | In recent years, the field of causal inference from observational data has emerged rapidly. The literature has focused on (conditional) average causal effect estimation. When (remaining) variability of individual causal effects (ICEs) is considerable, average effects may be uninformative for an individual. The fundamental problem of causal inference precludes estimating the joint distribution of potential outcomes without making assumptions. In this work, we show that the ICE distribution is identifiable under (conditional) independence of the individual effect and the potential outcome under no exposure, in addition to the common assumptions of consistency, positivity, and conditional exchangeability. Moreover, we present a family of flexible latent variable models that can be used to study individual effect modification and estimate the ICE distribution from cross-sectional data. How such latent variable models can be applied and validated in practice is illustrated in a case study on the effect of hepatic steatosis on a clinical precursor to heart failure. Under the assumptions presented, we estimate that 20.6% (95% Bayesian credible interval: 8.9%, 33.6%) of the population has a harmful effect greater than twice the average causal effect. |
| format | Article |
| id | doaj-art-9d8add95f1434d8784d5efc96acbdfe1 |
| institution | OA Journals |
| issn | 2193-3685 |
| language | English |
| publishDate | 2025-05-01 |
| publisher | De Gruyter |
| record_format | Article |
| series | Journal of Causal Inference |
| spelling | doaj-art-9d8add95f1434d8784d5efc96acbdfe12025-08-20T01:52:14ZengDe GruyterJournal of Causal Inference2193-36852025-05-01131190210.1515/jci-2024-0007Beyond conditional averages: Estimating the individual causal effect distributionPost Richard A. J.0van den Heuvel Edwin R.1Department of Biostatistics, Erasmus University Medical Center, Rotterdam, The NetherlandsDepartment of Mathematics and Computer Science, Eindhoven University of Technology, Eindhoven, The NetherlandsIn recent years, the field of causal inference from observational data has emerged rapidly. The literature has focused on (conditional) average causal effect estimation. When (remaining) variability of individual causal effects (ICEs) is considerable, average effects may be uninformative for an individual. The fundamental problem of causal inference precludes estimating the joint distribution of potential outcomes without making assumptions. In this work, we show that the ICE distribution is identifiable under (conditional) independence of the individual effect and the potential outcome under no exposure, in addition to the common assumptions of consistency, positivity, and conditional exchangeability. Moreover, we present a family of flexible latent variable models that can be used to study individual effect modification and estimate the ICE distribution from cross-sectional data. How such latent variable models can be applied and validated in practice is illustrated in a case study on the effect of hepatic steatosis on a clinical precursor to heart failure. Under the assumptions presented, we estimate that 20.6% (95% Bayesian credible interval: 8.9%, 33.6%) of the population has a harmful effect greater than twice the average causal effect.https://doi.org/10.1515/jci-2024-0007causal inferenceheterogeneity of treatment effectsprecision medicinebayesian analysisrandom effects models62d2062f1562p10 |
| spellingShingle | Post Richard A. J. van den Heuvel Edwin R. Beyond conditional averages: Estimating the individual causal effect distribution Journal of Causal Inference causal inference heterogeneity of treatment effects precision medicine bayesian analysis random effects models 62d20 62f15 62p10 |
| title | Beyond conditional averages: Estimating the individual causal effect distribution |
| title_full | Beyond conditional averages: Estimating the individual causal effect distribution |
| title_fullStr | Beyond conditional averages: Estimating the individual causal effect distribution |
| title_full_unstemmed | Beyond conditional averages: Estimating the individual causal effect distribution |
| title_short | Beyond conditional averages: Estimating the individual causal effect distribution |
| title_sort | beyond conditional averages estimating the individual causal effect distribution |
| topic | causal inference heterogeneity of treatment effects precision medicine bayesian analysis random effects models 62d20 62f15 62p10 |
| url | https://doi.org/10.1515/jci-2024-0007 |
| work_keys_str_mv | AT postrichardaj beyondconditionalaveragesestimatingtheindividualcausaleffectdistribution AT vandenheuveledwinr beyondconditionalaveragesestimatingtheindividualcausaleffectdistribution |