Остання редакція: 2026-06-25
Анотація
This paper investigates the influence of causal model identifiability on missing data recovery and causal effect estimation. A series of computational experiments using classical imputation methods and structural causal models was conducted. Both identifiable and non-identifiable settings were analyzed within the framework of causal inference theory. The results obtained are consistent with established theoretical principles of causal analysis and illustrate that identical observable data may correspond to different complete-data models, leading to ambiguity in missing value reconstruction and causal interpretation.
Ключові слова
Посилання
1. Rubin D. B. Inference and missing data // Biometrika. – 1976. – Vol. 63, no. 3. – P. 581–592.
2. Little R. J. A., Rubin D. B. Statistical analysis with missing data. – Hoboken: Wiley, 2019. – 450 p.
3. Pearl J. Causality: Models, reasoning, and inference. – 2nd ed. – Cambridge: Cambridge University Press, 2009. – 480 p.
4. Pearl J., Glymour M., Jewell N. P. Causal inference in statistics: A primer. – Hoboken: Wiley, 2016. – 150 p.
5. Hernán M. A., Robins J. M. Causal inference: What if. – Boca Raton: Chapman & Hall/CRC, 2020. – 624 p.
6. Shpitser I., Pearl J. Identification of conditional interventional distributions // UAI Proceedings. – 2006. – P. 437–444.
7. Spirtes P., Glymour C., Scheines R. Causation, prediction, and search. – 2nd ed. – MIT Press, 2000. – 543 p.
8. Mohan K., Pearl J., Tian J. Graphical models for inference with missing data // UAI. – 2013. – P. 1–10.
9. Ding P., Li F. Causal inference: A missing data perspective // Statistical Science. – 2018. – Vol. 33, no. 2. – P. 214–237.
10. Tian J., Pearl J. A general identification condition for causal effects // AAAI. – 2002. – P. 567–573.