Pith. sign in

REVIEW 2 major objections 6 minor 56 references

Recursive Equations For Imputation Of Missing Not At Random Data With Sparse Pattern Support

T0 review · 2 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that the full data law in graphical missing data models can be identified by a constructive pattern-mixture recursion, and that the resulting imputer, MISPR, is consistent under both MAR and MNAR mechanisms even when…

desk verdict Solid constructive identification theory for MNAR with sparse pattern support; the MISPR consistency claim is asserted, not proven, and the simulations are too thin to back it. read the letter →

arxiv 2507.16107 v1 pith:GYJ5Z3RO submitted 2025-07-21 stat.ME cs.LG

Trung Phung , Kyle Reese , Ilya Shpitser , Rohit Bhattacharya This is my paper · ORCID
classification stat.MEcs.LG
keywords missingnotatrandomgraphicalmodelsofdatamultipleimputationpatternmixturefactorizationGibbssamplingpositivityviolationsm-graphsfulllawidentification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that the full data distribution of a graphical missing data model can be identified constructively, not just shown to be identifiable, and that this constructive identification directly yields a consistent imputation procedure. The method's payoff is practical: it imputes data that are missing not at random without invoking MAR or parametric approximations, and it continues to work when many missingness patterns never occur in the observed sample, a regime that defeats standard imputation packages. The authors claim that their identification step, PM-ID/PM-ID+, is sound for m-graphs satisfying a structural condition, and that their Gibbs-sampling imputer MISPR built on it matches MICE when data are MAR and achieves lower bias when data are MNAR.

What carries the argument

The central object is the pattern DAG, a directed acyclic graph whose vertices are missingness patterns and whose edge $\tilde{r} \to_i r$ records the equality $p(X^{(1)}_i \mid X^{(1)}_{-i}, O, R=r) = p(X^{(1)}_i \mid X^{(1)}_{-i}, O, R=\tilde{r})$, allowing a Gibbs factor of one pattern to be borrowed from another. This edge set encodes the recursive identification order: the complete-case pattern is the root, and every missing variable in every supported pattern must have a path from the root ending in an edge labeled with that variable. The construction also uses the pattern mixture factorization (interpolation vs. extrapolation densities) and the local Markov property of m-DAGs, m-ADMGs, and m-CGs to decide when such equalities hold; PM-ID+ selects candidate parents by the condition that no indicator differing between patterns lies in the variable's Markov blanket.

What would settle it

Simulate large samples from a known MNAR m-graph under (P1, P3), impute with MISPR using the true graph, and compare the estimated target law to the true one; separately, estimate a Gibbs factor directly from rows of a supported pattern and compare it with the factor MISPR copies from a parent pattern, since any systematic mismatch would refute the pattern-DAG equality.

Watch

Extended reading notes

Core claim

The central claim is that, under positivity assumptions that allow some missingness patterns to have zero support, the full law $p(X^{(1)}, O, R)$ is identified for any m-graph satisfying the structural condition (S1): no missingness indicator lies in the Markov blanket of the variable it indicates. The proof is constructive. Pattern mixture factorization splits each pattern's joint law into an interpolation density, which is observed and therefore identified, and an extrapolation density, whose Gibbs factors are matched one-by-one to Gibbs factors of already-identified patterns using conditional independences read off the graph. PM-ID+ repairs the recursion when a pattern has no support by re-routing each Gibbs factor through any earlier-identified pattern that the local Markov property makes equivalent, and the paper shows this succeeds on its examples. Equipped with the explicit identifying functionals, MISPR imputes each row by Gibbs sampling from the fitted extrapolation density of its pattern, with rows chosen for fitting according to the same pattern DAG.

Load-bearing premise

The load-bearing premise is that the analyst's missingness graph is correct and that the observed support of missingness patterns is compatible with that graph; if an edge in the true mechanism is omitted or spurious, the Gibbs-factor equalities that drive identification and imputation fail.

Editorial extensions

If this is right

  • If the full law is identified and MISPR is consistent, analysts can impute MNAR data and then run any downstream procedure as though missingness never happened, without MAR-based corrections.
  • For m-DAGs and m-ADMGs satisfying (S1), PM-ID matches the identification power of prior odds-ratio methods while also outputting explicit identifying functionals, which is what makes imputation feasible.
  • When some patterns have no support but the complete case is present and (P3) holds, PM-ID+ still identifies the full law, converting a common high-dimensional data complication from a barrier into a routine input.
  • In the paper's simulations, MISPR has lower bias than MICE under MNAR and in MAR settings with positivity violations, and it scales to a 20-variable problem with 476 supported patterns where MICE could not finish.
  • PM-ID+ is sound but the paper makes no completeness claim, so a returned 'fail' does not by itself prove non-identifiability.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A testable diagnostic follows that the authors do not develop: because each Gibbs factor may have several candidate parent patterns, discrepancies among the fitted copies of that factor across parents would signal misspecification of the m-graph before any downstream analysis.
  • The recursion's dependence on a user-supplied graph means the practical bottleneck is model specification, not estimation; automating pattern-compatible graph search would extend MISPR to settings where analysts cannot draw the graph.
  • If the constructive identification is as general as claimed, the same pattern-DAG recursion could be adapted to other missingness structures, such as longitudinal attrition, by replacing the pattern lattice with time-indexed support sets.
  • The paper leaves open whether PM-ID+ is complete; a characterization of when 'fail' coincides with genuine non-identifiability would settle the boundary of the method.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 6 minor

Summary. The paper develops a constructive pattern-mixture approach to identification of the full data law in graphical models of missing data, for m-DAGs, m-ADMGs, and m-CGs. It introduces PM-ID for strictly positive laws and PM-ID+ for settings where some missingness patterns have zero support, and it packages the resulting identifying functionals into an imputation algorithm, MISPR, that imputes each supported pattern using Gibbs factors recursively borrowed from previously identified patterns. Theorems 2 and Corollaries 3-5 give the identification results; Algorithm 3 implements the imputation; Section 6 compares MISPR to MICE on synthetic DGPs. The paper claims in the abstract that MISPR is consistent under both MAR and MNAR settings and that it handles unsupported patterns without assumptions beyond the missing data model.

Significance. The identification contribution is substantial. Unlike the earlier odds-ratio identification approach, PM-ID provides explicit identifying functionals for each pattern, which is a genuine practical advance and is the key enabler of the imputation algorithm. The extension to sparse pattern support via PM-ID+ is novel, and the pattern-DAG formulation unifies DAG, ADMG, and chain-graph missing data models in a single recursive construction. The paper also gives an explicit completeness statement for PM-ID in m-DAGs and m-ADMGs. These contributions are likely to be useful beyond the specific imputation algorithm. However, the paper's headline statistical claim that the imputation algorithm MISPR is consistent is not backed by a theorem, and the numerical evidence is too thin to establish it. The identification results are sound; the missing consistency theory is the main barrier.

major comments (2)
  1. [Abstract; Section 5, Algorithm 3] The abstract and Section 5 describe MISPR as 'consistent under both MAR and MNAR settings', but no theorem or proof of statistical consistency is provided anywhere in the paper. In Algorithm 3, line 11, each Gibbs factor is fit using rows from parent patterns pa(r, i); by the induction built into lines 13-16, those rows have had their missing values filled in from estimated conditionals in earlier iterations. The pattern-DAG equalities in Eq. (5), (6), and (8) are population-level statements about the true law; they do not, by themselves, imply that the recursive estimator computed from imputed rows converges to the true extrapolation density. This is exactly the feedback mechanism that can bias chained-equation imputation when the imputation conditionals are misspecified, and footnote 9 concedes the risk of incongenial specifications. The numerical evidence in Section 6 and Table 4 consists of a single dataset per DGP with no Monte Carlo standard errors and no shared code, so it cannot substitute for a proof. Please either state and prove a consistency theorem for MISPR under explicit conditions on the graphical model, the conditional estimators, and Gibbs convergence, or revise the abstract and Section 5 to claim identification of the target law and algorithmic validity, rather than consistency.
  2. [Section 4.2, Eq. (8) and Algorithm 2] The condition defining Z(r, i) in Eq. (8) is asserted to be sufficient for the Gibbs-factor equality p(X_i | X_{-i}, O, R=r) = p(X_i | X_{-i}, O, R=er), but no proof is given for any of the three graph classes, and for ADMGs and CGs the claim is not immediate from the text. The equality appears to follow from the local Markov property with the Markov blankets defined in Tables 2 and 3, because the conditioning set in Eq. (5) contains mbG(X_i). Since Eq. (8) is the engine of PM-ID+ and underlies Corollary 5, please add a short lemma with proof establishing that if R_j notin mbG(X_i) for all j in diff(r, er), then the equality holds for every distribution satisfying the relevant local Markov property. This will also clarify whether the ordered Markov blanket used for ADMGs needs any additional qualification.
minor comments (6)
  1. [Section 1; Section 5] The acronym MISPR is expanded inconsistently: the abstract says 'Multivariate Imputation via Supported Pattern Recursion', while Section 1 says 'Pattern mixture Imputaton with Gibbs Sampling (MISPR)'; the latter also contains a typo ('Imputaton').
  2. [Section 3.2.1] The notation R12 in the display for the odds-ratio example is not defined; write OR({R1,R2}, R3 | ...) or define R12 explicitly.
  3. [Section 6, Experiment 3] The text says that the full data law for Figure 7(e) has 24 out of 25 possible patterns and then repeats 'the one for Figure 7(e)' for the 476-out-of-2^20 case; the second reference should be to Figure 7(f).
  4. [Section 6, Table 4] Rows without a '+' report both MISPR with PM-ID+ and MISPR with PM-ID, but Algorithm 3 as written selects PM-ID whenever all patterns have positive support; please clarify how the PM-ID+ column was produced in those rows.
  5. [Section 6] The numerical comparison reports no Monte Carlo repetitions, standard errors, hardware details, or code, which makes it difficult to assess whether the large timing and error differences in Table 4 are stable.
  6. [Appendix A] The no-complete-case extension PM-ID* is presented only as an example; if this is intended as a contribution, a formal statement and proof analogous to Corollary 5 would be needed.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the constructive identification proofs are self-contained, and the only self-citation (completeness of (S1)) is an independent theorem rather than an unverified load-bearing premise.

full rationale

The paper's derivation chain is: PM factorization (Eq. 4) + Markov-blanket independences (3) -> pattern DAG equality (5) -> Theorem 2 -> Corollaries 3/5 -> MISPR in Algorithm 3. Each identification step is a genuine functional identity derived from stated graphical and positivity assumptions. For example, Eq. (6) rewrites the independence X_i ⟂ R_i | X_{-i}, R_{-i}, O implied by (S1); Algorithm 1 adds an edge only when R_i ∉ mb(X_i), which is a sufficient condition for the equality required by Definition 1. Algorithm 2's set Z(r,i) in Eq. (8) similarly uses local Markov independences, not fitted parameters. No estimated quantity is used to define the identified full-law functional, so no fitted input is renamed as a prediction. MISPR estimates the identified Gibbs factors from data and imputes by Gibbs sampling; the abstract's claim that MISPR is 'consistent under both MAR and MNAR settings' is asserted without a theorem, and the recursive use of imputed values could cause feedback bias, but that is an unproven statistical correctness claim, not an equivalence-by-construction. The paper also relies on prior completeness results (Bhattacharya et al. 2019; Nabi et al. 2020) for the claim that (S1) is necessary and sufficient for m-DAGs and m-ADMGs; this is a self-citation by overlapping authors in Corollary 4, but the cited theorem is a parameter-free result with stated assumptions and does not assume the present paper's target conclusion. PM-ID's constructive identification is independently derived from (3). No equation or algorithm step reduces to its own input, so there is no circularity beyond a minor non-circular self-citation.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The central claim rests on substantive domain assumptions about correct graph specification and pattern support compatibility, plus standard graphical model mathematics. No fitted free parameters are introduced into the identification derivation itself, though the algorithm requires a consistent learner for the Gibbs factors.

assumptions (6)
  • domain assumption The analyst-provided m-graph G is correctly specified for the missing data mechanism.
    All identification and imputation rests on G being correct; misspecification breaks the equality in Eq. (5) and biases MISPR, as the paper itself notes in Section 4.3.
  • domain assumption Positivity conditions (P1) and (P2), or (P1) and (P3) in the relaxed setting, hold.
    Strict positivity of the target law and support conditions on missingness patterns are needed for the Gibbs factors and pattern mixture factorization to be well defined, as stated in Sections 3.1 and 4.1.
  • standard math The local Markov property with the Markov blanket definitions in Tables 1-3 holds for m-DAGs, m-ADMGs, and m-CGs.
    The paper uses these to induce the conditional independences in Eq. (3) and the PM-ID+ condition in Eq. (8). They are standard results for the respective graphical model classes.
  • domain assumption The observed set of supported patterns R+ is P3-compatible with the graph factorization of G.
    Compatibility is required for the pattern DAG construction in Section 4.1 to be valid; without it, zero-support patterns cannot be justified from the graphical model.
  • ad hoc to paper The nonparametric estimator used for the Gibbs factors (e.g., random forests) consistently estimates the conditional distributions.
    The consistency claim for MISPR in Section 5 and the abstract assumes the chosen learner converges to the true Gibbs factors, but no regularity conditions or proof are given.
  • standard math If R_j is not in the Markov blanket of X_i, then X_i is conditionally independent of R_j given all other variables, as used in PM-ID+.
    This separation property is standard for DAGs, but the paper does not prove it for the ADMG and CG Markov blanket definitions in Tables 2 and 3, where it is less immediate.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Recursive Equations For Imputation Of Missing Not At Random Data With Sparse Pattern Support." pith.science (2026). https://pith.science/paper/GYJ5Z3RO

@misc{pith2026250716107,
  author       = {Pith},
  title        = {Pith review of: Recursive Equations For Imputation Of Missing Not At Random Data With Sparse Pattern Support},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GYJ5Z3RO}},
  note         = {Machine review of arXiv:2507.16107}
}
read the original abstract

A common approach for handling missing values in data analysis pipelines is multiple imputation via software packages such as MICE (Van Buuren and Groothuis-Oudshoorn, 2011) and Amelia (Honaker et al., 2011). These packages typically assume the data are missing at random (MAR), and impose parametric or smoothing assumptions upon the imputing distributions in a way that allows imputation to proceed even if not all missingness patterns have support in the data. Such assumptions are unrealistic in practice, and induce model misspecification bias on any analysis performed after such imputation. In this paper, we provide a principled alternative. Specifically, we develop a new characterization for the full data law in graphical models of missing data. This characterization is constructive, is easily adapted for the calculation of imputation distributions for both MAR and MNAR (missing not at random) mechanisms, and is able to handle lack of support for certain patterns of missingness. We use this characterization to develop a new imputation algorithm -- Multivariate Imputation via Supported Pattern Recursion (MISPR) -- which uses Gibbs sampling, by analogy with the Multivariate Imputation with Chained Equations (MICE) algorithm, but which is consistent under both MAR and MNAR settings, and is able to handle missing data patterns with no support without imposing additional assumptions beyond those already imposed by the missing data model itself. In simulations, we show MISPR obtains comparable results to MICE when data are MAR, and superior, less biased results when data are MNAR. Our characterization and imputation algorithm based on it are a step towards making principled missing data methods more practical in applied settings, where the data are likely both MNAR and sufficiently high dimensional to yield missing data patterns with no support at available sample sizes.

Figures

Figures reproduced from arXiv: 2507.16107 by the authors.

Figure 1
Figure 1. Different kinds of missing data graphs: (a) A missing data DAG; (b) A missing [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Example demonstrating PM-ID compatibility of pattern DAGs: (a) An m-graph [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. The pattern DAG H obtained by applying the PM-ID pattern DAG construction method in Algorithm 1 to any of the m-graphs shown in [PITH_FULL_IMAGE:figures/full_fig_p019_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: (a) An m-graph G used to illustrate pattern mixture identification with positivity violations on the patterns R = 101 and R = 100. (b) Pattern DAG obtained by applying the PM-ID construction algorithm. The red strikethrough indicates patterns with no support. (c) Patte…
Figure 5
Figure 5. Figure 5: (a) An m-graph used to demonstrate that PM-ID+ can borrow information from [PITH_FULL_IMAGE:figures/full_fig_p026_5.png]
Figure 6
Figure 6. Figure 6: (a) A sparse m-graph used to demonstrate the relative data efficiency of PM-ID+ [PITH_FULL_IMAGE:figures/full_fig_p026_6.png]
Figure 7
Figure 7. Figure 7: Missing data models used in our numerical experiments. [PITH_FULL_IMAGE:figures/full_fig_p030_7.png]
Figure 8
Figure 8. Figure 8: (a) An m-graph with a sparse structure for variables in [PITH_FULL_IMAGE:figures/full_fig_p039_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 53 canonical work pages

  1. [1]

    M. J. Azur, E. A. Stuart, C. Frangakis, and P. J. Leaf. Multiple imputation by chained equations: W hat is it and how does it work? International Journal of Methods in Psychiatric Research, 20 0 (1): 0 40--49, 2011

    work page 2011

  2. [2]

    Berchtold

    A. Berchtold. Treatment and reporting of item-level missing data in social science research. International Journal of Social Research Methodology, 22 0 (5): 0 431--439, 2019

    work page 2019

  3. [3]

    Bhattacharya, R

    R. Bhattacharya, R. Nabi, I. Shpitser, and J. M. Robins. Identification in missing data models represented by directed acyclic graphs. In Proceedings of the 35th Conference on Uncertainty in Artificial Intelligence. AUAI Press, 2019

    work page 2019

  4. [4]

    Bhattacharya, D

    R. Bhattacharya, D. Malinsky, and I. Shpitser. Causal inference under interference and network uncertainty. In Uncertainty in Artificial Intelligence, pages 1028--1038. PMLR, 2020

    work page 2020

  5. [5]

    H. Y. Chen. A semiparametric odds ratio model for measuring association. Biometrics, 63: 0 413--421, 2007

    work page 2007

  6. [6]

    J. M. Chen, D. Malinsky, and R. Bhattacharya. Causal inference with outcome-dependent missingness and self-censoring. In Uncertainty in Artificial Intelligence, pages 358--368. PMLR, 2023

    work page 2023

  7. [7]

    Y.-C. Chen. Pattern graphs: A graphical approach to nonmonotone missing data. The Annals of Statistics, 50 0 (1): 0 129--146, Feb. 2022. ISSN 0090-5364, 2168-8966. doi:10.1214/21-AOS2094

    work page doi:10.1214/21-aos2094 2022

  8. [8]

    R. M. Daniel, M. G. Kenward, S. N. Cousens, and B. L. De Stavola. Using causal diagrams to guide analysis in missing data problems. Statistical Methods in Medical Research, 21 0 (3): 0 243--256, 2012

    work page 2012

Show all 56 references
  1. [9]

    Didelez and M

    V. Didelez and M. J. Stensrud. On the logic of collapsibility for causal effect measures. Biometrical Journal, 64 0 (2): 0 235--242, 2022

    2022

  2. [10]

    Didelez, S

    V. Didelez, S. Kreiner, and N. Keiding. Graphical models for inference under outcome-dependent sampling. 2010

    2010

  3. [11]

    Eekhout, R

    I. Eekhout, R. M. de Boer, J. W. Twisk, H. C. De Vet, and M. W. Heymans. Missing data: A systematic review of how they are reported and handled. Epidemiology, 23 0 (5): 0 729--732, 2012

    2012

  4. [12]

    R. J. Evans. Margins of discrete B ayesian networks. The Annals of Statistics, 46 0 (6A): 0 2623--2656, 2018

    2018

  5. [13]

    Gain and I

    A. Gain and I. Shpitser. Structure learning under missing data. In International Conference on Probabilistic Graphical Models, pages 121--132. Pmlr, 2018

    2018

  6. [14]

    Geman and D

    S. Geman and D. Geman. Stochastic relaxation, G ibbs distributions, and the B ayesian restoration of images. IEEE Transactions on pattern analysis and machine intelligence, 0 (6): 0 721--741, 1984

    1984

  7. [15]

    Honaker, G

    J. Honaker, G. King, and M. Blackwell. Amelia II : A program for missing data. Journal of Statistical Software, 45: 0 1--47, 2011

    2011

  8. [16]

    Karvanen and S

    J. Karvanen and S. Tikka. Multiple imputation and full law identifiability. arXiv preprint arXiv:2410.18688, 2024

    2024 arXiv

  9. [17]

    Kyono, Y

    T. Kyono, Y. Zhang, A. Bellot, and M. van der Schaar. MIRACLE : C ausally-aware imputation via learning missing data mechanisms. Advances in Neural Information Processing Systems, 34: 0 23806--23817, 2021

    2021

  10. [18]

    S. L. Lauritzen. Graphical Models, volume 17. Clarendon Press, 1996

    1996

  11. [19]

    R. J. Little. Pattern-mixture models for multivariate incomplete data. Journal of the American Statistical Association, 88 0 (421): 0 125--134, 1993

    1993

  12. [20]

    Liu and A

    Y. Liu and A. C. Constantinou. Greedy structure learning from data that contain systematic missing values. Machine Learning, 111 0 (10): 0 3867--3896, 2022

    2022

  13. [21]

    Malinsky, I

    D. Malinsky, I. Shpitser, and E. J. Tchetgen Tchetgen. Semiparametric inference for nonmonotone missing-not-at-random data: the no self-censoring model. Journal of the American Statistical Association, pages 1--9, 2021

    2021

  14. [22]

    Mathur, T

    M. Mathur, T. Vanderweele, and I. Shpitser. A common-cause principle for eliminating selection bias in causal estimands through covariate adjustment. Annals of Statistics, 2025

    2025

  15. [23]

    M. B. Mathur and I. Shpitser. Imputation without nightMARs : G raphical criteria for valid imputation of missing data. Technical report, Center for Open Science, 2024

    2024

  16. [24]

    M. B. Mathur and I. Shpitser. Pitfalls of imputing using incomplete auxiliary variables. American Journal of Epidemiology, 194 0 (6): 0 1801--1802, 2025

    2025

  17. [25]

    P. E. McKnight, K. M. McKnight, S. Sidani, and A. J. Figueredo. Missing Data: A Gentle Introduction . Guilford Press, 2007

    2007

  18. [26]

    W. Miao, L. Liu, Y. Li, E. J. Tchetgen Tchetgen, and Z. Geng. Identification and semiparametric efficiency theory of nonignorable missing data with a shadow variable. ACM/JMS Journal of Data Science, 1 0 (2): 0 1--23, 2024

    2024

  19. [27]

    Mohan and J

    K. Mohan and J. Pearl. On the testability of models with missing data. In Artificial Intelligence and Statistics, pages 643--650. PMLR, 2014

    2014

  20. [28]

    Mohan and J

    K. Mohan and J. Pearl. Graphical models for processing missing data. Journal of the American Statistical Association, 116 0 (534): 0 1023--1037, 2021

    2021

  21. [29]

    Mohan, J

    K. Mohan, J. Pearl, and J. Tian. Graphical models for inference with missing data. Advances in Neural Information Processing Systems, 26, 2013

    2013

  22. [30]

    Nabi and R

    R. Nabi and R. Bhattacharya. On testability and goodness of fit tests in missing data models. In Uncertainty in Artificial Intelligence, pages 1467--1477. PMLR, 2023

    2023

  23. [31]

    R. Nabi, R. Bhattacharya, and I. Shpitser. Full law identification in graphical models of missing data: Completeness results. In Proceedings of the 37th International Conference on Machine Learning, pages 7153--7163. PMLR, 2020

    2020

  24. [32]

    R. Nabi, R. Bhattacharya, I. Shpitser, and J. Robins. Causal and counterfactual views of missing data models. arXiv preprint arXiv:2210.05558, 2022

    2022 arXiv

  25. [33]

    E. L. Ogburn, I. Shpitser, and Y. Lee. Causal inference, social networks and chain graphs. Journal of the Royal Statistical Society Series A: Statistics in Society, 183 0 (4): 0 1659--1676, 2020

    2020

  26. [34]

    J. Pearl. Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference. Morgan Kaufmann, 1988

    1988

  27. [35]

    J. Pearl. Causality. Cambridge University Press, 2009

    2009

  28. [36]

    J. M. Pe \ n a. Unifying Gaussian LWF and AMP chain graphs to model interference. Journal of Causal Inference, 8 0 (1): 0 1--21, 2020

    2020

  29. [37]

    N. J. Perkins, S. R. Cole, O. Harel, E. J. Tchetgen Tchetgen, B. Sun, E. M. Mitchell, and E. F. Schisterman. Principled approaches to missing data in epidemiologic studies. American Journal of Epidemiology, 187 0 (3): 0 568--575, 2018

    2018

  30. [38]

    T. D. Pigott. A review of methods for missing data. Educational Research and Evaluation, 7 0 (4): 0 353--383, 2001

    2001

  31. [39]

    T. E. Raghunathan, J. M. Lepkowski, J. Van Hoewyk, and P. Solenberger. A multivariate technique for multiply imputing missing values using a sequence of regression models. Survey Methodology, page 85, 2001

    2001

  32. [40]

    B. Ren, S. R. Lipsitz, R. D. Weiss, and G. M. Fitzmaurice. Multiple imputation for non-monotone missing not at random data using the no self-censoring model. Statistical Methods in Medical Research, 32 0 (10): 0 1973--1993, 2023

    1973

  33. [41]

    T. S. Richardson. Markov properties for acyclic directed mixed graphs. Scandinavian Journal of Statistics, 30 0 (1): 0 145--157, 2003

    2003

  34. [42]

    T. S. Richardson, R. J. Evans, J. M. Robins, and I. Shpitser. Nested Markov properties for acyclic directed mixed graphs. The Annals of Statistics, 51 0 (1): 0 334--361, Feb. 2023. ISSN 0090-5364, 2168-8966. doi:10.1214/22-AOS2253

    2023 doi

  35. [43]

    J. M. Robins and R. D. Gill. Non-response models for the analysis of non-monotone ignorable missing data. Statistics in Medicine, 16 0 (1): 0 39--56, 1997

    1997

  36. [44]

    Saadati and J

    M. Saadati and J. Tian. Adjustment criteria for recovering causal effects from missing data. In Joint European Conference on Machine Learning and Knowledge Discovery in Databases, pages 561--577. Springer, 2019

    2019

  37. [45]

    Sadinle and J

    M. Sadinle and J. P. Reiter. Itemwise conditionally independent nonresponse modelling for incomplete multivariate data. Biometrika, 104 0 (1): 0 207--220, 2017

    2017

  38. [46]

    Shpitser

    I. Shpitser. Consistent estimation of functions of data missing non-monotonically and not at random. Advances in Neural Information Processing Systems, 29: 0 3144--3152, 2016

    2016

  39. [47]

    Shpitser

    I. Shpitser. The L auritzen- C hen likelihood for graphical models. In International Conference on Artificial Intelligence and Statistics, pages 4181--4195. PMLR, 2023

    2023

  40. [48]

    Sinharay, H

    S. Sinharay, H. S. Stern, and D. Russell. The use of multiple imputation for the analysis of missing data. Psychological methods, 6 0 (4): 0 317, 2001

    2001

  41. [49]

    P. L. Spirtes, C. N. Glymour, and R. Scheines. Causation, Prediction, and Search. MIT Press, 2000

    2000

  42. [50]

    J. A. Sterne, I. R. White, J. B. Carlin, M. Spratt, P. Royston, M. G. Kenward, A. M. Wood, and J. R. Carpenter. Multiple imputation for missing data in epidemiological and clinical research: P otential and pitfalls. Bmj, 338, 2009

    2009

  43. [51]

    Y. Sun, J. Li, Y. Xu, T. Zhang, and X. Wang. Deep learning versus conventional methods for missing data imputation: A review and comparative study. Expert Systems with Applications, 227: 0 120201, 2023

    2023

  44. [52]

    E. J. Tchetgen Tchetgen, I. R. Fulcher, and I. Shpitser. Auto-g-computation of causal effects on a network. Journal of the American Statistical Association, 116 0 (534): 0 833--844, 2021

    2021

  45. [53]

    R. Tu, C. Zhang, P. Ackermann, K. Mohan, H. Kjellstr \"o m, and K. Zhang. Causal discovery in the presence of missing data. In The 22nd International Conference on Artificial Intelligence and Statistics, pages 1762--1770. PMLR, 2019

    2019

  46. [54]

    Van Buuren

    S. Van Buuren. Multiple imputation of discrete and continuous data by fully conditional specification. Statistical Methods in Medical Research, 16 0 (3): 0 219--242, 2007

    2007

  47. [55]

    Van Buuren and K

    S. Van Buuren and K. Groothuis-Oudshoorn. mice: Multivariate imputation by chained equations in R . Journal of statistical software, 45: 0 1--67, 2011

    2011

  48. [56]

    Verma and J

    T. Verma and J. Pearl. Equivalence and synthesis of causal models. In Proceedings of the Sixth Annual Conference on Uncertainty in Artificial Intelligence, pages 255--270, 1990

    1990

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.