Pith. sign in

REVIEW 6 major objections 7 minor 38 references

Benford's Law, applied to probability masses rather than digits, identifies one-sided manipulation of discrete scores that standard density-continuity tests miss.

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 ↗

A Benford's-Law-based framework for RDD manipulation testing that selects a bandwidth and tests score symmetry around the cutoff, but it tests global symmetry rather than local manipulation.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection Benford-based manipulation test is a clever diagnostic for asymmetry but doesn't test the RDD null; the central claim of detecting hidden manipulation is unsupported. the 6 major comments →

arxiv 2607.13564 v1 pith:IF4T7R6W submitted 2026-07-15 stat.ME econ.EMmath.STstat.TH

Manipulation testing based on Benford's Law for discrete scores

Roy Cerqueti , Marco Ventura This is my paper
classification stat.ME econ.EMmath.STstat.TH MSC 62G1062P20
keywords Regression discontinuity designBenford's LawManipulation testDiscrete running variableDensity asymmetryMean absolute deviationBandwidth selectionFraud detection
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 reading

Benford's Law—the digit-frequency pattern used to flag accounting fraud—can be retooled as a manipulation test for regression-discontinuity designs in which the running variable is discrete. The paper's central claim is that measuring how well the left and right sides of the score distribution, taken separately, conform to Benford's proportions reveals directional asymmetries that conventional density-continuity tests, which compare the two sides in one symmetric exercise, cannot see. The authors supply a data-driven bandwidth from nested Benford thresholds and two test statistics: a normalized distance between the two sides' threshold positions and a mean-absolute-deviation (MAD) conformity score per side, judged against standard critical values. If the claim holds, empirical researchers gain a precautionary, almost parameter-free diagnostic that can localize the side of the cutoff where sorting pressure exists. The trade-off the paper accepts is that its null hypothesis is Benford-symmetry over the selected window, not merely continuity at the cutoff—so the test answers a broader question than the classic density test.

Core claim

The central discovery is that Benford's Law can be applied to the running variable's probability distribution rather than to its digits. Threshold points are placed on each side of the cutoff so that the probability mass between successive thresholds equals the Benford proportions log10(1 + 1/j); the construction is iterated toward the cutoff, and the bandwidth is chosen at the last nested window in which both tails still conform to Benford within a mean-absolute-deviation bound. Symmetry is then tested in two ways: by comparing the normalized distances of left and right thresholds from the cutoff (statistic D), and by reflecting each side's thresholds across the cutoff and checking whether

What carries the argument

The load-bearing object is Benford's Law as a discrete nine-point probability distribution, B_j = log10(1 + 1/j), used as a yardstick for partitioning the score's support around the cutoff rather than as a property of digits. The iterative threshold construction turns that yardstick into a bandwidth selector—the last nested window in which both tails conform—and the reflected-threshold construction turns the same yardstick into the directional test statistics D, MAD+, and MAD-. This machinery replaces researcher-chosen bandwidth, kernel, and polynomial parameters with a parameter-free reference distribution and, crucially, reports left- and right-side deviations separately.

Load-bearing premise

The load-bearing premise is that 'no manipulation' is equivalent to both sides of the score reproducing Benford's proportions over a wide automatically chosen window; a distribution that is continuous at the cutoff but globally skewed will be classified as manipulated even though no sorting occurred.

What would settle it

Generate a discrete score that cannot be manipulated by design—for example a bounded test score with a known right-skewed distribution, cutoff placed at a point where the density is continuous—and apply the paper's MAD test. If either MAD+ or MAD- exceeds the close-conformity threshold (0.006), the test has rejected a manipulation-free design, showing that it responds to global shape rather than to sorting.

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

If this is right

  • Researchers studying discrete-score RDDs can run a manipulation check whose bandwidth is selected by the data's conformity to Benford's Law, eliminating discretionary bandwidth/kernel/polynomial choices.
  • A violation on only one side of the cutoff localizes the asymmetry, letting the researcher tie the deviation to the incentive to sort above or below the threshold.
  • The method complements conventional density-continuity tests: a non-rejection there does not clear a design if one directional MAD exceeds its threshold.
  • Because only Benford's proportions are used, not the actual digit frequencies, the method also applies to bounded or short-support scores whose digits do not follow Benford's Law.
  • The D-statistic and the directional MADs supply a finer diagnostic vocabulary—both which tail deviates and how far—rather than a single equality-of-densities p-value.

Where Pith is reading between the lines

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

  • Extension: because the null is Benford-symmetry over a wide automatic window, the test measures global skewness as well as sorting; a prudent user would pair it with a local continuity-at-cutoff check and inspect the chosen window's width before calling a rejection 'manipulation'.
  • Extension: the D statistic is presented without a null distribution, so a permutation or bootstrap version—randomly relabeling sides around the cutoff—could turn it into a formal test with calibrated size.
  • Paper's own caveat: the concluding section acknowledges that MAD depends on sample size and that calibrated critical values are still needed; the 0.006/0.012/0.015 boundaries are a convention from the Benford conformity literature, not a tested decision rule for this setting.
  • Extension: a natural validation exercise is to apply the directional MADs to datasets where sorting is externally verified (e.g., known eligibility-report fraud) and check that the flagged side matches the true direction of manipulation.
Share X Bluesky LinkedIn Reddit HN

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

6 major / 7 minor

Summary. This paper proposes a Benford's-Law-based manipulation test for regression discontinuity designs with discrete running variables. The procedure first constructs nested thresholds on each side of the cutoff so that interval probabilities match the Benford first-digit vector (Section 3.1), and selects a bandwidth as the largest iteration for which both directional MAD statistics fall below a threshold (Eq. 11). Two tests are then proposed: a tolerance-threshold comparison of normalized cutoff distances (Eq. 14) and a directional MAD comparison of reflected thresholds against Benford probabilities (Eq. 17). The paper reports simulations from a symmetric normal distribution with a shifted cutoff and applies the method to Meyersson (2014) and Londoño-Vélez et al. (2020), claiming that it detects asymmetries missed by McCrary-type tests while removing researcher-chosen parameters.

Significance. The question addressed is relevant: manipulation tests for discrete-score RDDs are less developed than their continuous counterparts, and a diagnostic that separates left- and right-side deviations could be useful. The paper is clearly written and the iterative bandwidth idea is creative. However, the statistical contribution is not established. The proposed tests do not target the RDD manipulation null, no critical values or size/power analyses are provided, and the threshold used is borrowed from a different Benford testing context. The empirical applications demonstrate detection of asymmetry, not manipulation. Unless these foundational issues are resolved, the stated claims are unsupported.

major comments (6)
  1. [§3.1, Eqs. (2)–(6)] The iterative bandwidth selection defines thresholds X^{(-)}_{n,j}, X^{(+)}_{n,j} by requiring that the probability mass in each interval equals γ_n B_j. Hence, by construction, the empirical histograms in Eq. (8) are aligned with Benford probabilities; the MAD sequence measures only finite-sample or discreteness deviations from quantiles chosen to match Benford, not the fit of the data to Benford's Law. This circularity means that 'BL-consistent bandwidth' does not have the interpretation claimed, and the subsequent tests inherit this problem.
  2. [§4.2, Eq. (17); §5] The test's null is global symmetry of the Benford-binned distributions on the two sides of the cutoff over the selected bandwidth. This is not the RDD manipulation null of continuity/local comparability at the cutoff. A density continuous at c but globally skewed, or with different tail shapes, will violate Eq. (17) and be flagged as manipulated. The simulations do not address this: they start from a symmetric distribution and shift the cutoff, so asymmetry is induced by comparing unequal intervals rather than by manipulation under the RDD null. No size or power analysis under a continuous-but-asymmetric density is provided, so the claim that the method 'spots hidden manipulation' (Section 6, abstract) is unsupported.
  3. [§4.1, Eq. (14); §5.1; §7] The tolerance-threshold approach reports D=0.0737 but provides no critical value, decision rule, or null distribution. Section 7 concedes that operational thresholds are lacking. Without a rejection rule, Eq. (14) cannot be used as a test.
  4. [§4.2, Eq. (17); §5.1] The λ=0.006 threshold is taken from Nigrini (2012), where it is calibrated for first-digit MAD tests on digit frequencies. Here it is applied to a data-dependent, transformed set of probabilities from Eqs. (15)–(16); its sampling distribution is not derived. The paper acknowledges the sample-size dependence of MAD, but the ad hoc rescaling by sqrt(M) and the ECP calculations do not produce calibrated tests. Consequently, the 'close conformity' statements in the applications have no stated type I or type II error rates.
  5. [§6] The two applications are presented as evidence that the method detects manipulation that McCrary-type tests miss. In both cases the original McCrary test found no discontinuity and the paper acknowledges the distributions are asymmetric. Detecting asymmetry is not detecting manipulation; these examples are consistent with natural skewness. This evidence therefore contradicts, rather than supports, the abstract's claim of 'hidden manipulation'.
  6. [§1; §3; §5; §6] The paper motivates the method for discrete running variables, but no discrete example or simulation is provided: §5 simulates a continuous Normal distribution and §6 applies the procedure to continuous margins/SISBEN scores. For discrete scores, the thresholds defined by Eqs. (2)–(6) need not exist when the support has gaps; the paper does not discuss this or provide a discrete-data variant. The claimed scope is therefore not demonstrated.
minor comments (7)
  1. [Eq. (8)] The notation r_k 1(...) appears to use values rather than counts; clarify whether the empirical distribution is based on frequencies or weighted values.
  2. [Eq. (14)] The formula mixes a mean absolute deviation with a square root; as written it is not a standard distance. If the intent is Euclidean distance, use sqrt((1/9) sum (...)^2); if MAD, use a mean of absolute differences.
  3. [Eq. (16)] The interval (Y^{(-)}_j, Y^{(+)}_{j-1}) mixes left- and right-side thresholds; check whether the second endpoint should be Y^{(-)}_{j-1}.
  4. [Abstract; §1] The claim that the procedure 'eliminates researcher-chosen parameters' is contradicted by the method's reliance on α (or λ), N, and the simulation's β; these are user-specified.
  5. [Table 2] The caption 'perfect symmetry' is misleading because the table reports results for MAD(+) under the symmetric-data simulation; rephrase to describe the simulation design.
  6. [§5.1] The choice of the 'close conformity' threshold (0.006) based on 'only 0.5% of cases' is not justified; report the full distribution or sensitivity to alternative thresholds.
  7. [References] 'Necomb' should be 'Newcomb' (1881).

Circularity Check

1 steps flagged

Bandwidth selection imposes Benford proportions by construction, so the claimed BL-optimal bandwidth is self-referential; the asymmetry tests themselves are not circular.

specific steps
  1. self definitional [Section 3.1, Eqs. (2), (8), (11), and Remark]
    "We then select two sets of endogenous thresholds ... such that ... satisfying the following property: ∑ f(x)1(x∈[a,b)) = γ_1B_1 ... γ_1B_9. ... We assume that this empirical dataset has the function f used in the steps above as its empirical probability density function. Using the theoretical thresholds suggested by BL ... we compute the empirical distribution, i.e., the histograms, by setting XB_{n,j} = ... The idea to optimally select the BW consists in finding n✱ such that ... leads to a compliance of the empirical probability distributions in (8) with the BL according to a distance criteri"

    The thresholds X_{n,j} are defined, in Eq. (2), as the points at which the probability masses of the running variable's distribution equal γ_n B_j. The paper then states that f is the empirical probability density function of the dataset, so the histograms XB_{n,j} computed on those same intervals equal B_j by construction (up to finite-sample discreteness). Hence MAD_n = (1/9)Σ|XB_{n,j} − B_j| is not an independent measure of conformity to Benford's Law; it is the approximation error of the quantile-construction itself. Selecting the largest n with MAD < α therefore selects the largest n for which the BL binning can be implemented, not an interval in which the data independently comply with BL. The 'bandwidth consistent with BL' is thus self-referential.

full rationale

The core asymmetry tests in Section 4 are not tautological: Eq. (13) compares reflected BL-quantile distances, and Eqs. (15)–(17) compare reflected probability masses against Benford proportions. These are genuine symmetry checks and would not reduce to their inputs by construction. The circularity is concentrated in the bandwidth selection, which is advertised as a main contribution ('an innovative method for selecting a bandwidth consistent with BL'). Because the thresholds are defined so that the empirical masses equal γ_nB_j, the MAD criterion of Eq. (11) measures only how well the imposed quantile partition can be realized in a discrete sample, not whether the data satisfy Benford's Law. The paper's own Remark acknowledges that it is 'partitioning the R range according to the probabilities consistent with the BL,' which is precisely the construction that makes the subsequent 'compliance' automatic. This is a partial circularity: one central component reduces by definition, while the symmetry tests retain independent content. The self-citations (Cerqueti and Maggi 2021; Arezzo and Cerqueti 2023) are peripheral and not load-bearing; the critical thresholds come from Nigrini (2012). The concern that the test's null is global symmetry rather than RDD continuity is a correctness/false-positive issue, not circularity, and is not scored here.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 0 invented entities

The central claim rests on several free parameters (α, N, β in simulation) and on the domain assumption that global symmetry is the right null for RDD manipulation. The Benford weights themselves are a modeling choice, not a derived requirement.

free parameters (4)
  • α / λ (MAD acceptance threshold) = 0.006 in Monte Carlo and applications (Nigrini's 'close conformity')
    Used both for the optimal bandwidth (eq. 11) and for the symmetry decision (Section 4.2). The value 0.006 is selected post hoc because only 0.5% of MAD(−) values fell between candidate thresholds.
  • N (maximum iterations) = 10 in applications; unspecified in simulations
    Terminates the iterative bandwidth search; results depend on N because MAD increases with n.
  • β (cutoff shift in simulation) = unspecified
    Creates the asymmetric bandwidth in the Monte Carlo; the reported rejection rates are a function of this shift.
  • Benford first-digit probability vector B = B_j = log10(1+1/j), j=1..9
    The specific bin weights used to define thresholds; any probability vector would generate a similar quantile-symmetry test, so the 'natural law' justification does not uniquely determine the procedure.
axioms (4)
  • domain assumption No manipulation in a discrete-score RDD is equivalent to symmetry of the density over the bandwidth selected by the iterative procedure.
    Introduced in the Introduction and operationalized in Section 4.2's symmetry rule. This is stronger than the local-continuity null of McCrary-type tests and is not implied by the absence of manipulation.
  • domain assumption The MAD thresholds of Nigrini (2012) are valid critical values for the test statistics constructed here.
    Section 5.1 adopts 0.006/0.012/0.015 without verifying their null distribution for data-dependent, quantile-based bins; the paper later acknowledges sample-size dependence.
  • standard math The empirical density f and the threshold equations (2)-(6) are well-defined and solvable for the observed discrete data.
    Needed to compute thresholds; discreteness can break exactness of the quantile equations.
  • domain assumption An n* satisfying (11) exists and is unique for the datasets of interest.
    The procedure excludes 'uninformative instances' when no n* exists, which is a selection that can bias the empirical analysis.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Manipulation testing based on Benford's Law for discrete scores." pith.science (2026). https://pith.science/paper/IF4T7R6W

@misc{pith2026260713564,
  author       = {Pith},
  title        = {Pith review of: Manipulation testing based on Benford's Law for discrete scores},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IF4T7R6W}},
  note         = {Machine review of arXiv:2607.13564}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

This paper addresses the problem of running variable manipulation in Regression Discontinuity Designs. Leveraging the observation that manipulation often alters the density balance around the cutoff, we detect these structural imbalances using Benford's Law -a natural statistical regularity widely applied in fraud detection. Our framework serves as a vital precautionary safeguard alongside traditional McCrary-type tests. It eliminates researcher-chosen parameters that can skew outcomes, while delivering a deeper diagnostic breakdown of the density's behavior. Crucially, whereas the classic McCrary test can overlook systemic imbalances due to its rigid symmetric setup, our method separates the data into directional components. This allows researchers to pinpoint the exact origin of a deviation and spot hidden manipulation that standard frameworks fail to capture. To achieve this, we introduce an innovative method for selecting a bandwidth consistent with BL, and construct two distinct, complementary tests using threshold values adapted from Nigrini (2012) that successfully transition the law's application from digits to probabilities. Empirical applications confirm the enhanced protective value of this diagnostic framework.

Figures

Figures reproduced from arXiv: 2607.13564 by Marco Ventura, Roy Cerqueti.

Figure 1
Figure 1. Figure 1: Iteration for deriving the optimal BW in the case above the cutoff [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Visual representation of the two testing procedures. The first method compares ()() [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Simulation exercise Since the cutoff has been shifted rightward, we expect |LB| < UB, implying that the interval on the LHS of the cutoff point is shorter than the corresponding interval on the RHS, thereby endogenously generating an asymmetric BW around zero. In principle, the exercise can be repeated for different values of β as long as LB < 0; namely, for sufficiently small shifts. Given this setup, we … view at source ↗
Figure 4
Figure 4. Figure 4: Density plot of the running variable for the Meyersson [2014] case. [PITH_FULL_IMAGE:figures/full_fig_p016_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Density plot of the SISBEN score running variable (Londoño Vélez et al., 2020) [PITH_FULL_IMAGE:figures/full_fig_p017_5.png] 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

38 extracted references · 1 linked inside Pith

  1. [1]

    Journal of econometrics , volume=

    Manipulation of the running variable in the regression discontinuity design: A density test , author=. Journal of econometrics , volume=. 2008 , publisher=

    2008

  2. [2]

    Econometrica , volume=

    Islamic Rule and the Empowerment of the Poor and Pious , author=. Econometrica , volume=. 2014 , publisher=

    2014

  3. [3]

    The Stata Journal , volume=

    rdrobust: Software for regression-discontinuity designs , author=. The Stata Journal , volume=. 2017 , publisher=

    2017

  4. [4]

    The Stata Journal , volume=

    Manipulation testing based on density discontinuity , author=. The Stata Journal , volume=. 2018 , publisher=

    2018

  5. [5]

    German Economic Review , volume=

    Fact and fiction in EU-governmental economic data , author=. German Economic Review , volume=. 2011 , publisher=

    2011

  6. [6]

    Economic Modelling , volume=

    The short-run effects of public incentives for innovation in Italy , author=. Economic Modelling , volume=. 2023 , publisher=

    2023

  7. [7]

    2019 , publisher=

    A practical introduction to regression discontinuity designs: Foundations , author=. 2019 , publisher=

    2019

  8. [8]

    The American Statistician , volume=

    Benford's Law (Letters to the Editor) , author=. The American Statistician , volume=

  9. [9]

    arXiv preprint arXiv:2506.09915 , year=

    How much is too much? Measuring divergence from Benford's Law with the Equivalent Contamination Proportion (ECP) , author=. arXiv preprint arXiv:2506.09915 , year=

    Pith/arXiv arXiv

  10. [10]

    Journal of Accounting, Auditing & Finance , volume=

    Persistent patterns in stock returns, stock volumes, and accounting data in the US capital markets , author=. Journal of Accounting, Auditing & Finance , volume=. 2015 , publisher=

    2015

  11. [11]

    International Journal of Accounting Information Systems , volume=

    Study on the effect of sample size on type I error, in the first, second and first-two digits excessmad tests , author=. International Journal of Accounting Information Systems , volume=. 2023 , publisher=

    2023

  12. [12]

    cry wolf

    Moderating "cry wolf" events with excess MAD in Benford's Law research and practice , author=. Journal of forensic accounting research , volume=. 2016 , publisher=

    2016

  13. [13]

    International Journal of Accounting Information Systems , volume=

    Divergence from Benford's law fails to measure financial statement accuracy , author=. International Journal of Accounting Information Systems , volume=. 2025 , publisher=

    2025

  14. [14]

    Journal of Informetrics , volume=

    Use of Benford's law on academic publishing networks , author=. Journal of Informetrics , volume=. 2021 , publisher=

    2021

  15. [15]

    International Journal of Research in Marketing , volume=

    Price developments after a nominal shock: Benford's Law and psychological pricing after the euro introduction , author=. International Journal of Research in Marketing , volume=. 2005 , publisher=

    2005

  16. [16]

    Significance , volume=

    The promises and pitfalls of Benford's law , author=. Significance , volume=. 2016 , publisher=

    2016

  17. [17]

    Journal of Applied Statistics , volume=

    Not the first digit! Using benford's law to detect fraudulent scientif ic data , author=. Journal of Applied Statistics , volume=. 2007 , publisher=

    2007

  18. [18]

    Physica a: Statistical Mechanics and Its Applications , volume=

    A Benford's Law view of inspections' reasonability , author=. Physica a: Statistical Mechanics and Its Applications , volume=. 2023 , publisher=

    2023

  19. [19]

    Journal of the American Statistical Association , volume=

    Simple local polynomial density estimators , author=. Journal of the American Statistical Association , volume=. 2020 , publisher=

    2020

  20. [20]

    Regression discontinuity designs: Theory and applications , pages=

    Party bias in union representation elections: Testing for manipulation in the regression discontinuity design when the running variable is discrete , author=. Regression discontinuity designs: Theory and applications , pages=. 2017 , publisher=

    2017

  21. [21]

    Journal of Business & Economic Statistics , volume=

    Estimation and inference of discontinuity in density , author=. Journal of Business & Economic Statistics , volume=. 2013 , publisher=

    2013

  22. [22]

    American Journal of Mathematics , volume=

    Note on the frequency of use of different digits in natural number , author=. American Journal of Mathematics , volume=

  23. [23]

    Proceedings of the American Philosophical Society , pages=

    The law of anomalous numbers , author=. Proceedings of the American Philosophical Society , pages=. 1938 , publisher=

    1938

  24. [24]

    2012 , publisher=

    Benford's Law: Applications for forensic accounting, auditing, and fraud detection , author=. 2012 , publisher=

    2012

  25. [25]

    German Economic Review , volume=

    Benford's Law as an Indicator of Fraud in Economics , author=. German Economic Review , volume=. 2009 , publisher=

    2009

  26. [26]

    An introduction to

    Berger, Arno and Hill, Theodore P , year=. An introduction to

  27. [27]

    2019 , publisher=

    Cerioli, Andrea and Barabesi, Lucio and Cerasa, Andrea and Menegatti, Mario and Perrotta, Domenico , journal=. 2019 , publisher=

    2019

  28. [28]

    Regularities and discrepancies of credit default swaps: a data science approach through

    Ausloos, Marcel and Castellano, Rosella and Cerqueti, Roy , journal=. Regularities and discrepancies of credit default swaps: a data science approach through. 2016 , publisher=

    2016

  29. [29]

    The Review of economic studies , volume=

    Optimal bandwidth choice for the regression discontinuity estimator , author=. The Review of economic studies , volume=. 2012 , publisher=

    2012

  30. [30]

    American Economic Journal: Economic Policy , volume=

    Upstream and downstream impacts of college merit-based financial aid for low-income students: Ser Pilo Paga in Colombia , author=. American Economic Journal: Economic Policy , volume=. 2020 , publisher=

    2020

  31. [31]

    The Stata Journal , volume=

    Robust data-driven inference in the regression-discontinuity design , author=. The Stata Journal , volume=. 2014 , publisher=

    2014

  32. [32]

    The American Mathematical Monthly , volume=

    The surprising accuracy of Benford's law in mathematics , author=. The American Mathematical Monthly , volume=. 2020 , publisher=

    2020

  33. [33]

    , author=

    Rdrobust: an R package for robust nonparametric inference in regression-discontinuity designs. , author=. R J. , volume=. 2015 , publisher=

    2015

  34. [34]

    The European Physical Journal B , volume=

    Benford's law predicted digit distribution of aggregated income taxes: the surprising conformity of Italian cities and regions , author=. The European Physical Journal B , volume=. 2014 , publisher=

    2014

  35. [35]

    Benford's laws tests on

    Ausloos, Marcel and Ficcadenti, Valerio and Dhesi, Gurjeet and Shakeel, Muhammad , journal=. Benford's laws tests on. 2021 , publisher=

    2021

  36. [36]

    Annual Review of Economics , volume=

    Regression discontinuity designs , author=. Annual Review of Economics , volume=. 2022 , publisher=

    2022

  37. [37]

    Chaos, Solitons & Fractals , volume=

    Data validity and statistical conformity with Benford's Law , author=. Chaos, Solitons & Fractals , volume=. 2021 , publisher=

    2021

  38. [38]

    2024 , publisher=

    A practical introduction to regression discontinuity designs: Extensions , author=. 2024 , publisher=

    2024

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.