Pith. sign in

REVIEW 3 major objections 4 minor 51 references

The paper claims saddlepoint likelihood fitting can be automated for any model built from a short menu of structural operations, with a diagnostic that quantifies the difference from the exact MLE.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 17:51 UTC pith:6T5YS2DS

load-bearing objection A genuinely useful unified saddlepoint-likelihood framework, but the flagship capture-recapture validation leans on an error diagnostic that is only weakly checked in-paper. the 3 major comments →

arxiv 2607.17464 v1 pith:6T5YS2DS submitted 2026-07-20 stat.CO stat.ME

A general framework for computation and estimation using the saddlepoint approximation

Godrick Oketch , Rachel M. Fewster , Jesse Goodman This is my paper
classification stat.CO stat.ME
keywords saddlepoint approximationsaddlepoint likelihoodcumulant generating functionmodel-building operationslatent identity modelstatistical linear inverse modelconstrained optimizationdiscrepancy diagnostic
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper makes saddlepoint-likelihood estimation a practical, almost turn-key operation for a wide class of statistical models. Its premise is that many difficult models share a small set of structures—sums, thinning, randomly stopped sums, compound distributions, and linear maps of latent counts—each of which preserves access to the moment generating function. The framework encodes these operations as composable software objects, so the user specifies only the model's high-level shape, and the software assembles the cumulant generating function, its derivatives, the saddlepoint, and the optimized estimates. It also provides a diagnostic that approximates the difference between the saddlepoint maximum likelihood estimate and the exact one, even when the exact likelihood cannot be computed. If the framework works as claimed, researchers who fit latent-count models gain fast estimates, standard errors, and a built-in error check without bespoke derivations.

Core claim

The central claim is that the entire saddlepoint likelihood workflow—constructing the cumulant generating function of the observable variable, solving the saddlepoint equation, and maximizing the approximated likelihood—can be automated for any model assembled from a short menu of structural operations. The paper's building blocks include sums of independent draws, binomial thinning, randomly stopped sums, compound distributions, inhomogeneous sums, and linear transformations Y = AX that turn latent counts into observed counts. Each operation is implemented as a function on CGF code objects, so operations chain together the way model structures do. The same code objects carry the first and s

What carries the argument

The central object is the CGF code object: a software representation of a parametric cumulant generating function KY(t; θ) together with its first two derivatives, its saddlepoint solver, and automatic handling of domain constraints. Model-building operations—summing, thinning, random stopping, compounding, linear mapping—are functions that take one CGF code object and return another, so access to the moment generating function is preserved by construction. An adaptor function h(θ) connects model parameters to the distributional parameters of each base distribution. A constrained optimizer maximizes the saddlepoint log-likelihood simultaneously in (t, θ), subject to the saddlepoint equation

Load-bearing premise

The load-bearing premise is that the closed-form discrepancy formula imported from the authors' companion paper approximates the true difference between saddlepoint and exact maximum likelihood estimates for every model in the framework; in this paper it is validated only on the multivariate Poisson example, with the capture-recapture validation deferred to a forthcoming simulation study.

What would settle it

Simulate the capture-recapture latent-identity model at small population size, where exact likelihoods can be computed by enumerating latent identities, and compare the diagnostic's estimated discrepancy against the true discrepancy, and both against standard errors; if the diagnostic systematically mis-states the sign or magnitude of the difference, the method's validation collapses. The multivariate Poisson case already provides a template for this comparison.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • For models with intractable likelihoods—capture-recapture with latent identities, partial-sum contingency tables, aggregate insurance claims—parameter estimates and standard errors become available from just a few lines of model-facing code.
  • The discrepancy diagnostic turns approximation error into a reported number, so users can judge whether the saddlepoint error matters relative to sampling error.
  • In exponential-family settings, the framework reproduces the theoretical result that saddlepoint and exact maximum likelihood estimates coincide, giving a discrepancy of zero up to round-off.
  • Because CGF operations compose, models built from several blocks, such as a linear map inside a randomly stopped sum, need no new derivation; existing CGF objects can be reused through adaptors.
  • Custom CGFs can be added through a public constructor, so the framework can extend beyond its built-in menu of operations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The framework's broader leverage may be to make saddlepoint inference a default workflow for latent-count models, much as general-purpose sampling engines abstracted away the mechanics of simulation; the same CGF objects could also be reused for approximate Bayes factors or profile likelihoods.
  • A natural stress test is to compare the constrained optimization's convergence and speed against the classical nested two-step method on the same models, since the paper argues for but does not benchmark this advantage.
  • If the discrepancy diagnostic proves reliable, it offers a decision rule: compute the diagnostic alongside the estimate and escalate to exact methods only when it is large relative to the standard error.
  • For the capture-recapture application, the promised simulation study is the key external validation; confirming that the estimated discrepancy for population size is an order of magnitude below sampling error would solidify the practical case in that domain.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. This paper proposes a unified computational framework for saddlepoint likelihood estimation. The user specifies a model through high-level operations (sums, thinning/splitting, random stopped sums, linear transformations, partial summaries, etc.); the software automatically assembles the CGF, its derivatives, and the saddlepoint log-likelihood, and maximizes it via constrained optimization over (t,θ). A second contribution is a discrepancy diagnostic, taken from the authors' Oketch et al. (2025) preprint, that approximates the difference between saddlepoint and exact MLEs even when the exact likelihood is unavailable. The framework is implemented in the R package saddlepoint. Applications include a multivariate Poisson model validated against exact likelihood, a random stopped sum, a composite model, and a two-source capture-recapture model with latent identities.

Significance. If the framework performs as claimed, it would make saddlepoint likelihood estimation and its error assessment accessible to practitioners for a wide class of models with intractable likelihoods. The paper ships running R code, and the §5.1 comparison against exact likelihood provides a useful sanity check; the exponential-family check in §5.2 (zero discrepancy) is consistent with Goodman (2022). The constrained optimization formulation is a sensible alternative to the usual nested optimization. However, the paper's load-bearing claim about the discrepancy diagnostic in the capture-recapture setting is not yet supported by in-paper evidence, and there is a technical error in the thinning CGF formula. These issues are fixable but require revision.

major comments (3)
  1. [§5.4 / §3.5, Eq. (8)] The central claim—that the discrepancy diagnostic can be trusted when the exact likelihood is unavailable—is not yet supported for the flagship setting. Eq. (8) is taken from Oketch et al. (2025) without derivation; the only in-paper validation is the multivariate Poisson example (§5.1, Fig. 3), where exact likelihoods are tractable and the regime lacks the features of the capture-recapture model (inequality constraints, integer N, subunitary CGF). The text's assertion that δ̂_N is an order of magnitude below sampling error is deferred to simulations 'to be reported in an upcoming paper' (§5.4). This is a validation gap, not a demonstrated contradiction; it can be closed by a proof of (8)'s validity for the subunitary/constrained setting or by a tractable validation example in that regime.
  2. [§2.2, thinning formula] The formula for the CGF of a p-thinned variable, K_Y(t) = K_X{p e^t + 1 − p}, is incorrect. Since K_X is the cumulant generating function, the correct relation is K_Y(t) = K_X(log(p e^t + 1 − p)). The displayed expression would be correct only if K_X denoted the moment generating function. The composition used later in §5.2 via K_N(K_X(t)) is correct, so this appears to be a typo, but as written in a core building block it must be corrected and the affected examples checked.
  3. [§5.4, Code Block 4 / Eq. (8)] The capture-recapture model's N is a positive integer, but the constrained optimization in (10) and the derivative-based diagnostic (8) treat θ as continuous. The paper does not state how N̂_spa is obtained (real-valued maximizer, rounded value?) nor how δ̂_N is defined relative to the integer-valued exact MLE. The sentence that δ̂_N 'can be read directly as the number of animals we are likely to over- or under-estimate' presupposes a definition that is not given. Please clarify the optimization over N and the interpretation of δ̂_N, or restrict the discrepancy claim to continuous parameters.
minor comments (4)
  1. [§5.1, Code Block 1 vs Fig. 3] Code Block 1 specifies m=5 and d=3, but the Fig. 3 caption reports m=20 and d=10. This mismatch weakens the code-to-paper correspondence and should be aligned.
  2. [§2.10 / §5.1] The phrase 'typically a small fraction of the standard error' is used, but the largest discrepancy in Fig. 3 is about 20% of a standard error. Consider reporting a typical (e.g., median) fraction rather than the maximum, or softening the wording.
  3. [Supplement S3] The term 'subunitary CGF' is nonstandard. It would help to state explicitly that the probability vector sums to less than one because the remaining probability is assigned to a zero-count cell, and to explain why the saddlepoint approximation with this CGF remains a valid likelihood approximation.
  4. [§3.5, Eq. (8)] Equation (8) uses sums over j1,...,j6 without specifying the summation convention. Please state that Einstein summation or explicit multi-index summation is intended, and define the dimensions of the tensors.

Circularity Check

1 steps flagged

The capture-recapture validation of the discrepancy diagnostic rests on the authors' own Oketch et al. (2025) preprint; the framework itself is not circular.

specific steps
  1. self citation load bearing [Section 5.4, final paragraph; Eq. (8) in Section 3.5; Section 2.10]
    "Results in (Oketch et al., 2025) show that δ̂N is typically an order of magnitude smaller than the sampling error of ˆN. In the two-source capture-recapture context, the authors have confirmed this through extensive simulations to be reported in an upcoming paper."

    The advertised diagnostic is implemented by Eq. (8), introduced as 'Oketch et al. (2025, Definition 2)'. The flagship demonstration is the capture-recapture model in §5.4, where exact likelihood is unavailable; there, the claim that δ̂N is an order of magnitude below sampling error is attributed to the same authors' preprint and to simulations 'to be reported in an upcoming paper'. The only in-paper check of Eq. (8) against true discrepancies is the multivariate Poisson example in §5.1 (Fig. 3, row 3), so the §5.4 assurance is not independently established in the manuscript and rests on a self-citation chain.

full rationale

The central framework—CGF building blocks, automatic assembly of K, K', K'', and constrained optimization of the saddlepoint log-likelihood—is self-contained and does not reduce to a fit or to a self-citation. The examples in §5.1–5.3 compare saddlepoint estimates with exact likelihoods where available, and §5.2's zero discrepancy in the exponential-family case is consistent with a published, peer-reviewed result (Goodman 2022), albeit by a co-author. The only load-bearing self-citation is the discrepancy approximation in Eq. (8), taken from Oketch et al. (2025), by the same authors. That formula is checked against true discrepancies in one tractable example (§5.1, Fig. 3 row 3) and returns zero in the exponential-family case, but in the capture-recapture application (§5.4), where the exact likelihood is unavailable and the diagnostic is central, the quantitative claim is deferred to a same-author 'upcoming paper'. This is a validation gap rather than a definitional circularity: the paper does not define the discrepancy in terms of the approximation, and if Eq. (8) is independently verified for discrete constrained settings, the concern disappears. Accordingly, the score is 4 rather than 6 or higher: some self-citation is load-bearing, but the central modeling framework has substantial independent content.

Axiom & Free-Parameter Ledger

0 free parameters · 5 axioms · 1 invented entities

The framework introduces no fitted free parameters: the saddlepoint approximation and the discrepancy formula are determined by the CGF and its derivatives, and the example-specific numbers (α=14, β=7; m, d) are simulation settings, not fitted constants. Load-bearing assumptions are the regularity of the saddlepoint expansion, the accuracy rates from Goodman (2022), the discrepancy formula from the authors' own Oketch et al. (2025) preprint, and the Poisson-encounter construct in Supplement S2. The only invented construct is the 'subunitary CGF' device for zero-count cells, which is a definition, not an empirical entity.

axioms (5)
  • standard math Saddlepoint expansion validity: CGF exists in an open neighborhood of zero and the saddlepoint equation K'(t̂;θ) = y has a unique solution (Daniels 1954).
    Invoked in §1 Eqs. (1)-(3) to define the saddlepoint likelihood; regularity conditions are not checked per example.
  • domain assumption Goodman (2022) discrepancy rates (n⁻², n⁻³ᐟ², n⁻¹) apply to the framework's model class.
    Cited in §1 to justify using (3); proved under specific limiting frameworks, with hypotheses unverified for each §5 example.
  • domain assumption Oketch et al. (2025) discrepancy formula (8) is accurate for the framework's models.
    §3.5 takes formula (8) from the authors' own arXiv preprint 2504.07324; only validated in §5.1 within this paper.
  • standard math Constrained optimization (10) converges to the saddlepoint MLE.
    §4 relies on Boyd & Vandenberghe (2004) and Johnson (2022); assumes standard smoothness and constraint qualifications.
  • domain assumption Poisson encounter process in Supplement S2 yields the latent-identity probabilities (S3) for the capture-recapture model.
    Supplement S2 gives only a sketch ('After some algebra...'); the derivation is not shown and the assumption drives the adaptor h(θ).
invented entities (1)
  • Subunitary CGF (K_subunitary) no independent evidence
    purpose: Handles zero-count latent identity cells in multinomial CGFs so the saddlepoint equation has a finite solution and K'' has nonzero determinant (Supplement S3).
    A definitional device, not an empirical entity: derived as KY(t,−∞) restricted to Y_{d+1}=0. Internally coherent; independent evidence not applicable.

pith-pipeline@v1.3.0-alltime-deepseek · 18193 in / 18470 out tokens · 145344 ms · 2026-08-01T17:51:33.814768+00:00 · methodology

0 comments
read the original abstract

The saddlepoint approximation provides highly accurate approximations to probability density and mass functions using only the corresponding moment generating functions (MGFs). Recent work has increasingly seen the saddlepoint approximation applied to likelihood functions, enabling likelihood-based inference in models where exact likelihoods are intractable. However, existing implementations have largely been developed on a model-by-model basis, and the methodology remains underutilized because of the conceptual and computational challenges of working with MGFs. We introduce a unified framework for model construction and computation using the saddlepoint approximation. The framework is based on a collection of model-building operations that preserve access to MGFs while allowing complex distributions to be constructed from simpler components. With these components, users need only provide a high-level specification of the model structure, from which the software automatically assembles the necessary generating functions, saddlepoints, and gradients, and performs the optimization of the saddlepoint likelihood. We also introduce a diagnostic that quantifies the difference between saddlepoint and exact likelihood estimates, even when the exact likelihood is unavailable. The framework is implemented in the R package saddlepoint and provides fast, convenient computation of parameter estimates, standard errors, and the discrepancy diagnostic. Numerous examples illustrate the scope and flexibility of the approach.

Figures

Figures reproduced from arXiv: 2607.17464 by Godrick Oketch, Jesse Goodman, Rachel M. Fewster.

Figure 1
Figure 1. Figure 1: Schematic showing how we assemble the CGF of [PITH_FULL_IMAGE:figures/full_fig_p016_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Flowchart depicting the integration of constrained optimization into the framework [PITH_FULL_IMAGE:figures/full_fig_p021_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The saddlepoint-based estimates closely match those from the exact likelihood, [PITH_FULL_IMAGE:figures/full_fig_p024_3.png] view at source ↗
Figure 3
Figure 3. Figure 3: Comparison of α and β estimates (row 1), their standard errors (row 2), and the approximated versus true discrepancies (row 3), obtained from the saddlepoint likelihood and exact likelihood for the multivariate Poisson model in Section 5.1. Each of the six panels shows estimated values based on 100 datasets simulated using α = 14 and β = 7, with each dataset consisting of m = 20 i.i.d. copies of the vector… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

51 extracted references · 14 canonical work pages

  1. [1]

    Reid , journal =

    N. Reid , journal =. Saddlepoint Methods and Statistical Inference , volume =

  2. [2]

    H. E. Daniels , journal =. Saddlepoint Approximations in Statistics , volume =

  3. [3]

    Small Sample Asymptotics , volume =

    Christopher Field and Elvezio Ronchetti , journal =. Small Sample Asymptotics , volume =

  4. [4]

    Annals of Statistics , FJOURNAL =

    Jesse Goodman , TITLE =. Annals of Statistics , FJOURNAL =. 2022 , VOLUME =. doi:10.1214/22-AOS2169 , SICI =

    doi:10.1214/22-aos2169 2022

  5. [5]

    Butler, Ronald W. , year=. Saddlepoint Approximations with Applications , DOI=

  6. [6]

    Hermite Polynomials , booktitle =

    Sel. Hermite Polynomials , booktitle =. doi:https://doi.org/10.1002/9781119425465.ch3 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1002/9781119425465.ch3 , year =

    work page doi:10.1002/9781119425465.ch3

  7. [7]

    The American Statistician , volume =

    Constantino Goutis and George Casella , title =. The American Statistician , volume =. 1999 , publisher =. doi:10.1080/00031305.1999.10474463 , URL =

    Pith/arXiv arXiv 1999

  8. [8]

    1987 , publisher=

    Tensor Methods in Statistics , author=. 1987 , publisher=

    1987

  9. [9]

    Barndorff-Nielsen and D

    O. Barndorff-Nielsen and D. R. Cox , journal =. Edgeworth and Saddle-Point Approximations with Statistical Applications , volume =

  10. [10]

    Davison and Konstantinos Fokianos , journal =

    Xanthi Pedeli and Anthony C. Davison and Konstantinos Fokianos , journal =. Likelihood Estimation for the

  11. [11]

    and Bravington, M

    Zhang, W. and Bravington, M. V. and Fewster, R. M. , title =. Biometrics , volume =. doi:https://doi.org/10.1111/biom.13030 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/biom.13030 , year =

    work page doi:10.1111/biom.13030

  12. [12]

    and Price, S

    Zhang, W. and Price, S. J. and Bonner, S. J. , title =. Environmental and Ecological Statistics , volume =. doi:https://doi.org/10.1007/s10651-021-00492-6 , url =

    work page doi:10.1007/s10651-021-00492-6

  13. [13]

    and Yoshizaki, Jun and Bailey, Larissa L

    Link, William A. and Yoshizaki, Jun and Bailey, Larissa L. and Pollock, Kenneth H. , title =. Biometrics , volume =. doi:https://doi.org/10.1111/j.1541-0420.2009.01244.x , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1541-0420.2009.01244.x , year =

    arXiv 2009

  14. [14]

    Bonner and Matthew R

    Simon J. Bonner and Matthew R. Schofield and Patrik Noren and Steven J. Price , title =. The Annals of Applied Statistics , number =. 2016 , doi =

    2016

  15. [15]

    Davison, A. C. and Hautphenne, S. and Kraus, A. , title =. Biometrics , volume =. doi:https://doi.org/10.1111/biom.13282 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/biom.13282 , year =

    work page doi:10.1111/biom.13282

  16. [16]

    Vale, R. T. R. and Fewster, R. M. and Carroll, E. L. and Patenaude, N. J. , title =. Biometrics , volume =. doi:https://doi.org/10.1111/biom.12195 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/biom.12195 , year =

    work page doi:10.1111/biom.12195

  17. [17]

    Hazelton , title =

    Martin L. Hazelton , title =. The Annals of Applied Statistics , number =. 2015 , doi =

    2015

  18. [18]

    Biometrics , volume =

    Bonner, Simon J and Holmberg, Jason , title =. Biometrics , volume =. doi:https://doi.org/10.1111/biom.12045 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1111/biom.12045 , year =

    work page doi:10.1111/biom.12045

  19. [19]

    and Conn, Paul B

    McClintock, Brett T. and Conn, Paul B. and Alonso, Robert S. and Crooks, Kevin R. , title =. Ecology , volume =. doi:https://doi.org/10.1890/12-1613.1 , url =. https://esajournals.onlinelibrary.wiley.com/doi/pdf/10.1890/12-1613.1 , year =

    work page doi:10.1890/12-1613.1

  20. [20]

    2016 , pages=

    Journal of Statistical Software , author=. 2016 , pages=. doi:10.18637/jss.v070.i05 , number=

    doi:10.18637/jss.v070.i05 2016

  21. [21]

    Johnson , howpublished =

    Steven G. Johnson , howpublished =. 2022 , title =

    2022

  22. [22]

    Bell , title =

    Bradley M. Bell , title =

  23. [23]

    2004 , publisher=

    Convex Optimization , author=. 2004 , publisher=

    2004

  24. [24]

    1989 , publisher=

    Asymptotic techniques for use in statistics , author=. 1989 , publisher=

    1989

  25. [25]

    2024 , url =

    R: A Language and Environment for Statistical Computing , author =. 2024 , url =

    2024

  26. [26]

    Journal of Statistical Software , year =

    Dirk Eddelbuettel and Romain Fran. Journal of Statistical Software , year =

  27. [27]

    Journal of Applied Statistics , volume =

    Dimitris Karlis , title =. Journal of Applied Statistics , volume =. 2003 , publisher =. doi:10.1080/0266476022000018510 , URL =

    work page doi:10.1080/0266476022000018510 2003

  28. [28]

    1992 , publisher=

    Bivariate Discrete Distributions (1st ed.) , author=. 1992 , publisher=

    1992

  29. [29]

    J. O. Irwin , journal =. The Place of Mathematics in Medical and Biological Statistics , volume =

  30. [30]

    On the compound Poisson-gamma distribution , url =

    Withers, Christopher and Nadarajah, Saralees , journal =. On the compound Poisson-gamma distribution , url =

  31. [31]

    Meraou and Noriah M

    Mohammed A. Meraou and Noriah M. Al-Kandari and Mohammad Z. Raqab and Debasis Kundu , title =. Journal of Statistical Computation and Simulation , volume =. 2022 , publisher =. doi:10.1080/00949655.2021.1981324 , URL =

    arXiv 2022

  32. [32]

    2018 , journal =

    Boxiang Liu and Thomas Quertermous , title =. 2018 , journal =. doi:10.32614/RJ-2018-011 , url =

    work page doi:10.32614/rj-2018-011 2018

  33. [33]

    Data augmentation in multi-way contingency tables with fixed marginal totals , journal =

    Adrian Dobra and Claudia Tebaldi and Mike West , keywords =. Data augmentation in multi-way contingency tables with fixed marginal totals , journal =. 2006 , issn =. doi:https://doi.org/10.1016/j.jspi.2004.07.002 , url =

    work page doi:10.1016/j.jspi.2004.07.002 2006

  34. [34]

    and McCrea, Rachel S

    Zhang, Wei and Bonner, Simon J. and McCrea, Rachel S. , title = ". Biometrics , volume =. 2022 , month =. doi:10.1111/biom.13789 , url =

    work page doi:10.1111/biom.13789 2022

  35. [35]

    Biometrika , volume =

    Hazelton, M L and McVeagh, M R and van Brunt, B , title = ". Biometrika , volume =. 2021 , month =. doi:10.1093/biomet/asaa083 , url =

    work page doi:10.1093/biomet/asaa083 2021

  36. [36]

    1976 , PAGES =

    Rudin, Walter , TITLE =. 1976 , PAGES =

    1976

  37. [37]

    and Ristić, Miroslav M

    Altun, Emrah and Cordeiro, Gauss M. and Ristić, Miroslav M. , title =. Journal of Applied Statistics , volume =. 2022 , publisher =. doi:10.1080/02664763.2021.1884846 , URL =

    arXiv 2022

  38. [38]

    Andrew T. A. Wood and James G. Booth and Ronald W. Butler , journal =. Saddlepoint Approximations to the

  39. [39]

    Pollock and Cavell Brownie and Raymond A

    Jun Yoshizaki and Kenneth H. Pollock and Cavell Brownie and Raymond A. Webster , journal =. Modeling Misidentification Errors in Capture-Recapture Studies Using Photographic Identification of Evolving Marks , urldate =

  40. [40]

    Pollock and William A

    Jun Yoshizaki and Cavell Brownie and Kenneth H. Pollock and William A. Link , journal =. Modeling misidentification errors that result from use of genetic tags in capture–recapture studies , volume =

  41. [41]

    Schofield and Simon J

    Matthew R. Schofield and Simon J. Bonner , journal =. Connecting the Latent Multinomial , urldate =

  42. [42]

    Saddle Point Approximation for the Distribution of the Sum of Independent Random Variables , urldate =

    Robert Lugannani and Stephen Rice , journal =. Saddle Point Approximation for the Distribution of the Sum of Independent Random Variables , urldate =

  43. [43]

    Neil Arnason , journal =

    Carl James Schwarz and A. Neil Arnason , journal =. A General Methodology for the Analysis of Capture-Recapture Experiments in Open Populations , urldate =

  44. [44]

    A Conversation with

    Peter Whittle , journal =. A Conversation with

  45. [45]

    Saddlepoint Approximations

    DasGupta, Anirban. Saddlepoint Approximations. Asymptotic Theory of Statistics and Probability. 2008. doi:10.1007/978-0-387-75971-5_14

    work page doi:10.1007/978-0-387-75971-5_14 2008

  46. [46]

    Flegg , title =

    Yong See Foo and Jennifer A. Flegg , title =. arXiv:2308.16465 , year =

    Pith/arXiv arXiv

  47. [47]

    Whittaker, E. T. and Watson, G. N. , year=. The fundamental properties of analytic functions; Taylor’s, Laurent’s and Liouville’s Theorems , booktitle=

  48. [48]

    , title =

    McClintock, Brett T. , title =. Ecology and Evolution , volume =. doi:https://doi.org/10.1002/ece3.1676 , url =. https://onlinelibrary.wiley.com/doi/pdf/10.1002/ece3.1676 , year =

    work page doi:10.1002/ece3.1676

  49. [49]

    Skaug , journal =

    Tore Selland Kleppe and Hans J. Skaug , journal =. Building and Fitting Non-

  50. [50]

    2024 , url =

    Author , title =. 2024 , url =

    2024

  51. [51]

    and Goodman, Jesse , title =

    Oketch, Godrick and Fewster, Rachel M. and Goodman, Jesse , title =. 2025 , howpublished =

    2025