2022
DOI: 10.1073/pnas.2108097119
|View full text |Cite
|
Sign up to set email alerts
|

Universal adaptability: Target-independent inference that competes with propensity scoring

Abstract: The gold-standard approaches for gleaning statistically valid conclusions from data involve random sampling from the population. Collecting properly randomized data, however, can be challenging, so modern statistical methods, including propensity score reweighting, aim to enable valid inferences when random sampling is not feasible. We put forth an approach for making inferences based on available data from a source population that may differ in composition in unknown ways from an eventual target population. W… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
12
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
2
2
2
1

Relationship

1
6

Authors

Journals

citations
Cited by 7 publications
(12 citation statements)
references
References 25 publications
0
12
0
Order By: Relevance
“…3. Furthermore, we apply HappyMap to problems in target-independent learning that lie beyond the statistical estimation problems considered in [30], obtaining target-independent statistical inference and uncertainty quantification. Our approach also yields a fruitful new perspective on analyzing missing data, giving new solutions to this problem (Section 5).…”
Section: :3mentioning
confidence: 99%
See 2 more Smart Citations
“…3. Furthermore, we apply HappyMap to problems in target-independent learning that lie beyond the statistical estimation problems considered in [30], obtaining target-independent statistical inference and uncertainty quantification. Our approach also yields a fruitful new perspective on analyzing missing data, giving new solutions to this problem (Section 5).…”
Section: :3mentioning
confidence: 99%
“…Definition 1 (Multicalibration [22] as presented in [30]). Let C ⊆ {[0, 1] × X → R} be a collection of functions.…”
Section: Introductionmentioning
confidence: 99%
See 1 more Smart Citation
“…socio-demographic attributes of individuals or make and type of digital devices). While applying pre-processing techniques such as re-weighting may not be feasible in all ADM contexts, recent work on post-processing predictions exemplifies how ideas from survey research (mass imputation; Yang and Kim, 2020) and computer science (multi-calibration; Hebert-Johnson et al, 2018) can be combined to tackle misrepresentation in training data (Kim et al, 2022).…”
Section: Sources Of Bias and Social Impacts Along The Adm Processmentioning
confidence: 99%
“…Our approach would be classified by Kim et al (2022) as a form of "universal" estimation of any "target" distribution from a "source" since the inclusion probabilities we estimate encode the sampling design such that they may be used with any response variables from the target population to produce a population survey-based estimator.…”
Section: Introductionmentioning
confidence: 99%