{"id":"7653367b-a573-4373-857c-a020979568b9","arxiv_id":"1908.07390","paper_version":1,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":0.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A review-style compilation of standard statistical concepts (descriptive statistics, inference, regression, factor analysis, clustering, classification) with no new results.","lead":"This paper is a textbook-style overview of standard statistical methods, from descriptive statistics to classifiers. It contains no new data, analysis, or theory, and is aimed at readers who need a rapid reference for common concepts.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The overview's central promise of reliable 'fast access' is undermined by corrupted formulas, e.g., the chi-square PDF in §3.1, invalid F-distribution constraints, and a mislabeled Q3.","rationale":"The reader's verdict is well-founded. The paper claims only to be an accessible summary, so its correctness rests entirely on faithful transcription of standard material. The three flagged items are load-bearing because they appear in the inference and descriptive sections that a user would consult first. I re-read the full text and did not find any compensating support: there is no code release, no worked verification, and no formal proof; the paper is a self-contained exposition. The soft spot is therefore not disagreement with statistical consensus but plain internal accuracy. A single decisive check — evaluating the chi-square formula as printed — will settle whether the concern lands. If it does, the paper cannot serve as 'fast access' without requiring the reader to consult textbooks, which defeats its stated purpose. Since the paper makes no novel empirical or theoretical claim, the appropriate disposition remains UNVERDICTED (unchanged) rather than accept or reject.","tokens_in":21992,"tokens_out":4811,"duration_ms":45599,"concrete_test":"Recompute the chi-square PDF exactly as printed in §3.1 at n=5, x=1 and compare with the standard density value 1/(2^{n/2}Γ(n/2)) x^{n/2−1} e^{−x/2}; if the printed expression contains a spurious integral and an 'e^{−X} dX' factor and either cannot be evaluated or fails to integrate to 1, the overview cannot be used without external checking. This is the weakest link because the chi-square function is central to the inference section.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's stated aim is to provide 'fast access' to statistical concepts; for a quick-reference overview, the formulas and conditions must be accurate transcriptions of the cited sources. The most load-bearing assumption is therefore transcription accuracy, and the text fails it in several non-cosmetic places. In §3.1 the chi-square PDF is printed as '1/(2^{n/2}) ∫_0^∞ x^{n/2-1} e^{-X} dX • x^{n/2-1} e^{-x/2}', which is not a density and cannot be used to compute probabilities. The same section states the F-distribution is defined only for 'm > 2 and n > 4'; these are not the support conditions for the F pdf (which holds for m,n > 0) but rather restrictions tied to moments (n>2 for the mean, n>4 for the variance), so a reader applying them to the density would reject valid parameter pairs. In §2 the 75th percentile is computed correctly as 23 but labeled 'P25 = Q3', which corrupts the quartile notation and the IQR calculation. These are not stylistic slips; they directly undermine the paper's usefulness as an unmediated reference.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents a broad overview of statistical data analysis concepts for practitioners who use statistical software. It covers variable types, descriptive statistics, common probability distributions, sampling distributions, hypothesis testing, linear regression, exploratory factor analysis, cluster analysis, and popular classifiers, with the stated goal of providing 'fast access' to theoretical concepts. The paper restates standard textbook material, includes worked numerical examples (e.g., frequency tables, mean, median, percentiles) and illustrative figures, but does not contain original derivations, simulations, or data analyses.","tokens_in":22124,"tokens_out":5599,"duration_ms":49670,"significance":"If the formulas and definitions were accurately transcribed, the overview could serve a useful purpose as a quick reference for non-specialist users of statistical software, and the breadth of topics is appropriate. The paper explicitly credits its sources and includes concrete examples that help readers map definitions to computations. However, the manuscript's central promise of reliable 'fast access' depends entirely on whether the reproduced formulas and conditions are correct; the current text contains several transcription errors in load-bearing formulas, so the overview cannot be used without consulting the original textbooks.","major_comments":[{"comment":"The chi-square PDF as printed, fX(x)= 1/(2^{n/2}) ∫_0^∞ x^{n/2−1} e^{−X} dX • x^{n/2−1} e^{−x/2}, is not a probability density function: the integral factor is a spurious constant involving the dummy variable X, and the normalization is missing. The correct density is 1/(2^{n/2} Γ(n/2)) x^{n/2−1} e^{−x/2} for x>0. Because the paper's purpose is an unmediated quick reference, this error directly undermines the stated goal.","section":"3.1"},{"comment":"For the Snedecor F-distribution, the text states 'where τ(u)=... and m >2 and n >4'. These inequalities are not support conditions for the density; the F density is defined for all m,n>0. The restrictions n>2 and n>4 are required only for the existence of the mean and variance, respectively. As written, a reader following the paper would incorrectly reject valid parameter pairs such as (m,n)=(1,1), which are routinely used in F tests.","section":"3.1"},{"comment":"In the percentile example, the 75th percentile is computed as (X15+X16)/2 = 23, which is correct, but the line is labeled 'P25 = Q3 = ...'. This directly contradicts the paper's own definition that P25 is the first quartile Q1 and P75 is the third quartile Q3, and it corrupts the notation used in the following IQR definition (IQR = Q3 − Q1). The label should be 'P75 = Q3'.","section":"2"},{"comment":"The sampling-distribution table contains a garbled chi-square row: 'S′2 (N−1)S′2 / σ² ∼ X²(n−1)' should read (n−1)S′²/σ² ∼ χ²(n−1). In Section 6.1, the Euclidean distance formula is printed as 'd(p,q) = v√ ∑...' with an unexplained 'v' before the radical. These additional transcription errors, alongside those in Section 3.1, indicate that the manuscript cannot currently serve as a reliable reference without external checking.","section":"3.2 and 6.1"}],"minor_comments":[{"comment":"The sentence about the chi-square distribution says the central limit theorem ensures that the X² distribution approaches the normal distribution 'for high values of p'; this should be 'for high values of n', since p is not a parameter of this distribution.","section":"3.1"},{"comment":"The gamma function is denoted τ(u) in the Student-t and F PDFs, but the definition writes 'dX' instead of 'dx' inside the integral; the dummy variable should match the integration variable.","section":"3.1"},{"comment":"In the ANOVA paragraph, 'n is the number of variables' should be 'n is the sample size' when giving the F degrees of freedom (p, n−p−1).","section":"4"},{"comment":"The binomial distribution text says 'exactly n successes out of N Bernoulli trials' while the PMF uses n as the number of trials; the wording should be 'x successes out of n trials'.","section":"3.1"},{"comment":"The instance-based learner is referred to as 'IBl'; the standard Weka name is 'IBk' (k-nearest neighbors).","section":"7"},{"comment":"The author name Marôco is spelled inconsistently: 'Marˆoco' in most citations but 'Morˆoco' in Section 4 and Section 5.1; please unify the spelling.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"This manuscript is a compilation of standard textbook material with no new research contribution, simulations, or data analysis. If the journal's scope requires original contributions, the fit is questionable; my recommendation of major_revision assumes an overview article is within scope. The companion publication (Sarmento and Costa, 2017) is cited but the code is not included in this manuscript, so the 'fast access' claim rests solely on the accuracy of the textual formulas."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, this one is a rehash of standard textbook statistics, explicitly described by the authors as a summary to support their own R/Python code. There is no new result, no data analysis, no derivation, so as a research submission it is out of place. But it is not a bad outline: the sections cover descriptive stats, inference, regression, factor analysis, clustering, and classifiers, and the worked examples (age, height, color) are simple and mostly correct.\n\nThe problem is that the text does not deliver what it promises. For a quick reference, the formulas are the product. And in Section 3.1 the chi-square PDF is printed as a garbled expression with an integral over x that makes no sense as a density; the F-distribution is said to be defined only for m > 2 and n > 4, which are moment restrictions, not support conditions (the pdf actually requires m, n > 0); and in Section 2 the 75th percentile is computed correctly as 23 but then labeled 'P25 = Q3'. These are not cosmetic typos. A reader who uses the overview without consulting the original textbooks will be actively misled on the density functions.\n\nThere are also smaller signs of carelessness: duplicated words like 'OutliersOutliers', garbled equations in the appendix-like parts, and a few references that are incomplete. None of that changes the fact that the verbal explanations are conventional and the structure is sensible. The paper would be a passable teaching note for absolute beginners if the equations were corrected and the scope stated as a tutorial rather than a reference.\n\nMy bottom line: do not send this to a research journal. It has no research content, and the transcription errors are load-bearing for its stated purpose. If the authors fix the formulas, it could be a legitimate technical report or a chapter in an educational collection. For this version, desk reject.","headline":"A sincere but careless overview of standard statistics whose transcription errors defeat its stated purpose as a fast-access reference.","tokens_in":22661,"tokens_out":2158,"would_cite":false,"duration_ms":20635,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"One compact document gathers the theory behind everyday statistical analysis software, from descriptive statistics to classifiers.","keywords":["statistical data analysis","descriptive statistics","probability distributions","hypothesis testing","linear regression","factor analysis","cluster analysis","classification"],"falsifier":"Compare the printed chi-square density in Section 3.1 with the standard expression $\\frac{1}{2^{n/2}\\Gamma(n/2)}x^{n/2-1}e^{-x/2}$: the printed version contains an integral sign, an undefined $X$, and a stray factor. The same section prints the t and F densities with $\\tau(u)$ where $\\Gamma(u)$ is intended and attributes them to a \"Kerns, 2010\" that is missing from the reference list; a reader checking any of these against the sources would find that the overview cannot be relied on as a stand-alone quick reference.","tokens_in":21722,"feed_emoji":"📊","tokens_out":15910,"duration_ms":133725,"temperature":0.7,"pith_summary":"This paper sets out to be a compact, fast-access reference for the statistical concepts a software user needs when analyzing data: variable types, descriptive measures, probability distributions, sampling distributions, hypothesis tests, regression, factor analysis, clustering, and classifiers. It is written to accompany the authors' R and Python code, so the goal is not to teach derivations but to let a reader look up what a method assumes, what formula it uses, and how to interpret its output. A reader who follows it should be able to get from raw data to a reported result—mean and quartiles, a fitted regression, a factor rotation, a cluster assignment, or a classifier—without opening several textbooks. The document's usefulness stands on the accuracy of the formulas and decision rules it reproduces from its cited sources.","feed_headline":"One document gathers statistics theory from means to classifiers","feed_subtitle":"It collects descriptive statistics, inference, regression, factor analysis, clustering, and classifiers in one look-up source.","key_machinery":"The carrying structure is a catalog of formulas and rules, grouped by the stage of an analysis. Descriptive statistics are carried by frequency distributions and the mean, median, mode, percentile, range, and variance formulas; inference is carried by the probability densities for the normal, chi-square, Student t, F, and binomial distributions, together with the sampling-distribution table and the central limit theorem; hypothesis testing is carried by p-value and confidence-interval decision rules; regression by the sums-of-squares decomposition and F/t tests; factor analysis by the factor model, communalities, adequacy measures, and rotations; clustering by distance measures and linkage rules; classification by the named algorithms. The paper's claim to usefulness is that these entries are self-contained, so a reader can look up one formula and its interpretation without reading the whole document.","core_discovery":"The central claim of the overview is that the theory supporting everyday statistical software can be organized into a single coherent chain of concepts, and that a reader can move through that chain at reference speed: classify variables, summarize them with measures of central tendency and variability, choose a probability model, use sampling distributions and the central limit theorem for inference, apply hypothesis-testing decision rules with p-values and confidence intervals, then proceed to multivariate tools—linear regression, exploratory factor analysis, cluster analysis, and common classifiers. Each concept is presented as a formula or a decision rule, often with a small worked example, and estimation details are deliberately left to the software. The intended outcome is that a user who understands where a given test or estimator sits in this chain will be able to configure software settings and interpret results correctly, rather than treating the software as a black box.","pith_inferences":["Editorial inference: the same catalog format could be extended into a troubleshooting chart that maps each software setting (t-test vs. z-test, PCA vs. factor extraction, linkage method) to the distributional assumptions the overview lists.","Editorial inference: a natural stress test is to ask whether a reader could reproduce every worked example in the paper (age mean 21.75, median 21.5, third quartile 23, factor-loading illustration) in any standard statistical package; agreement would confirm the reference is self-contained.","Editorial inference: because the overview defers estimation to software, its most useful next step would be a side-by-side table of R, Python, and SPSS commands that correspond to each formula, making the theory-to-tool link explicit."],"forward_implications":["If the overview works as intended, a user of the companion R/Python material can look up the equation behind each output and the interpretive rule for it in one pass.","The organization by analysis stage gives a beginner a map from raw data to reported results: summarize, choose a distribution, test, then model.","The explicit warnings about p-value misinterpretation and the distinction between \"reject\" and \"fail to reject\" carry directly into the companion software examples.","A reader who needs only one concept—say, the F-distribution or the KMO measure—can extract it without reading the surrounding chapters."],"supporting_citations":[{"why":"Supplies the variable classification, percentile rule, sampling-distribution table, and hypothesis-test framework that structure the inference sections.","marker":"(Marôco, 2011)"},{"why":"Provides the definitions of mean, median, mode, range, variance, and standard deviation used in the descriptive section.","marker":"(McCune, 2009)"},{"why":"Gives the percentile and quartile reporting, boxplot five-number summary, time-chart, and histogram conventions.","marker":"(Rumsey, 2010)"},{"why":"Cited for the chi-square density used in Section 3.1.","marker":"(Kerns, 2018)"},{"why":"Cited for the Student t and Snedecor F density formulas in Section 3.1.","marker":"(Kerns, 2010)"},{"why":"Supplies the linear regression equation and the factor-analysis model assumptions with factor loadings.","marker":"(Morôco, 2014)"},{"why":"Motivates the factor-retention decision in exploratory factor analysis.","marker":"(Hayton et al., 2004)"},{"why":"Underlies the eigenvalue-greater-than-one retention criterion and the varimax rotation method.","marker":"(Kaiser, 1958)"},{"why":"Provides the OneR rule-based classifier and the entropy-based decision-tree rationale in the classifier section.","marker":"(Holte, 1993)"},{"why":"Describes bagging, the ensemble method summarized in the classifier section.","marker":"(Breiman, 1996)"}],"fun_headline_variants":["Statistics theory from A to Z in one reference","One-stop guide: from means to classifiers","Concise stats overview: variables to classifiers","Statistics distilled: from descriptive to predictive"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that every formula and definition in the overview is transcribed accurately enough from its source that a reader can rely on it without checking the original textbooks; if even one formula is corrupted, a reader following the overview will get a wrong result with no sign that anything is amiss.","fun_headline_variants_meta":{"raw":{"variants":["Statistics theory from A to Z in one reference","One-stop guide: from means to classifiers","Concise stats overview: variables to classifiers","Statistics distilled: from descriptive to predictive"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000508,"raw_usage":{"total_tokens":2409,"prompt_tokens":813,"completion_tokens":1596,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":429,"completion_tokens_details":{"reasoning_tokens":1541}},"tokens_in":429,"tokens_out":1596,"duration_ms":12171,"temperature":1.0,"reasoning_tokens":1541,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:36:50.957369+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compare the printed chi-square density in Section 3.1 with the standard expression $\\frac{1}{2^{n/2}\\Gamma(n/2)}x^{n/2-1}e^{-x/2}$: the printed version contains an integral sign, an undefined $X$, and a stray factor. The same section prints the t and F densities with $\\tau(u)$ where $\\Gamma(u)$ is intended and attributes them to a \"Kerns, 2010\" that is missing from the reference list; a reader checking any of these against the sources would find that the overview cannot be relied on as a stand-alone quick reference.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the definitions of mean, median, mode, range, variance, and standard deviation used in the descriptive section."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the percentile and quartile reporting, boxplot five-number summary, time-chart, and histogram conventions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Cited for the chi-square density used in Section 3.1."},{"cited_title":"C., Allen, D","cited_arxiv_id":null,"evidence_quote":"Motivates the factor-retention decision in exploratory factor analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the OneR rule-based classifier and the entropy-based decision-tree rationale in the classifier section."}],"review_version":1}