{"id":"d8f9ca3e-e1f0-4acc-9de0-715c2f62e5ee","arxiv_id":"2607.11778","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Laws and policies define “statistical purposes” by aggregate public-benefit production plus confidentiality/non-enforcement; a broader definition adding integrity, transparency, and harm avoidance is proposed for NSO use.","lead":"The paper traces how U.S. and international statistical agencies have used “for statistical purposes only,” extracts two core legal criteria (aggregate public-benefit statistics plus confidentiality and non-enforcement use), and proposes a broader operational definition that also requires scientific integrity, transparency, and avoidance of group or partisan harm. Official-statistics agencies and data-governance bodies can use the definition as an explicit decision standard w","discovery_kind":"review","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The manuscript is a careful policy/ethics synthesis, not an empirical or formal derivation. Its strongest claim—the extraction of the two predominant criteria from U.S. law and policy—is well-supported by the cited statutes (1929 Census Act, Title 13, CIPSEA, Evidence Act) and historical episodes. The broader definition is offered as a proposal informed by scientific-integrity and ethical frameworks, not as a logical entailment that those frameworks redefine the statutory term. The reader’s weakest_assumption therefore correctly identifies a normative step, but that step is not load-bearing for the paper’s actual claim; the authors already flag the need for further discussion and empirical research on stakeholder interpretation. Consequently the CONDITIONAL verdict (definition offered for consideration; further work needed) remains appropriate and requires no adjustment.","tokens_in":12977,"tokens_out":458,"duration_ms":4221,"concrete_test":"Independently re-read CIPSEA §502(5)–(7) and Evidence Act 44 U.S.C. §3563(a),(d) against the two criteria stated in the abstract and Section 7; confirm that both criteria are textually supported and that the broader definition in Section 6 is introduced only with “we propose … for … consideration.” If that framing holds, the concern does not land.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper’s central claim is a historical-legal synthesis of two predominant statutory criteria for “statistical purposes,” followed by an explicitly labeled proposal for a broader applied definition. The reader’s weakest_assumption correctly flags the normative leap from Belmont/Menlo/FIPPs and Evidence Act responsibilities (Sections 4 and 6) into that definition, but the authors themselves present the definition as “for consideration” rather than as a deduced necessity, and they close by calling for further empirical work on public understanding. Because the paper does not claim that the ethical frameworks are themselves statutory definitions of the term, the leap is not load-bearing for the descriptive claim that is actually advanced. No internal inconsistency or unsupported factual assertion undermines the two-criteria extraction or the proposal’s framing.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper traces the historical and legal evolution of the phrase “for statistical purposes only” in U.S. federal statistics, from Taft’s 1910 census proclamation through the 1929 Census Act, wartime breaches, Title 13, the Privacy Act, CIPSEA (2002), and the Evidence Act (2018). It extracts two predominant statutory criteria: (1) production of aggregate statistical information intended for general public benefit, and (2) confidentiality protections that prohibit use of identifiable data for legal or regulatory action against data subjects. It then situates the term within scientific-integrity principles, ASA/ISI ethical guidelines, and the Belmont/Menlo/FIPPs frameworks, and proposes a broader applied definition (Section 6) that adds transparency, scientific integrity, and avoidance of group or partisan harm. The paper closes with calls for empirical research on public interpretation of the phrase.","tokens_in":13146,"tokens_out":1028,"duration_ms":8003,"significance":"If the two-criteria extraction and the proposed broader definition are adopted as working guidance, NSOs would gain a clearer operational and ethical anchor for communications with respondents, internal data-use decisions, and stewardship of administrative data. The historical synthesis is carefully documented with primary statutory citations and is of direct practical value to statistical agencies facing declining trust and response rates. The explicit framing of the broader definition as “for consideration,” together with the forward-looking research agenda in Section 7, makes the contribution usable without overclaiming normative necessity.","major_comments":[{"comment":"Section 6 (and the abstract’s criterion (1)): the claim that “intention of creating a general public benefit” is one of the two predominant criteria extracted from underlying laws is only weakly supported by the statutory language actually quoted. CIPSEA §§502(5),(7) and 13 U.S.C. §9 emphasize aggregate description without identification and prohibition of non-statistical uses that affect rights or benefits; they do not themselves codify “public benefit.” The public-benefit language appears mainly in Taft’s 1910 proclamation and in selected international examples (UK ONS, Statistics NZ). The manuscript should either (a) qualify the first criterion as an implicit or historical rather than statutory element, or (b) supply additional U.S. statutory or OMB policy language that explicitly embeds public benefit inside the definition of statistical purpose.","section":null},{"comment":"Sections 4 and 6: the leap from the Evidence Act’s four fundamental responsibilities (44 U.S.C. §3563) and the Belmont/Menlo/FIPPs frameworks into the multi-sentence “broader applied definition” is presented as a synthesis “for consideration,” yet the abstract and introduction still frame the paper as showing that reviews of laws and policies “identify” the two criteria and then “provide a broader definition.” To keep the descriptive claim load-bearing and the normative proposal clearly optional, the authors should add an explicit demarcation sentence at the start of Section 6 stating that the expanded definition is a policy recommendation, not a restatement of existing statutory text.","section":null}],"minor_comments":[{"comment":"Section 2.1: the discussion of wartime uses of census data (Japanese-American internment, Second War Powers Act) is historically important but could be tightened; a short table or timeline of key statutory changes would help readers track the evolution of the confidentiality guarantee.","section":null},{"comment":"Section 5: the international comparison is limited to brief definitional quotations. A single comparative table (EUROSTAT, Statistics Canada, UNSD, UK ONS, Statistics NZ) would make the claimed “broad similarity” more transparent.","section":null},{"comment":"References: Eltinge (2025, 2026) are cited for related discussion; if these are forthcoming or under review, a note on availability would assist readers.","section":null},{"comment":"Typographical: “RELEV ANT” in the Evidence Act quotation (Section 4) contains a stray space; “official” appears with a ligature that may render inconsistently.","section":null},{"comment":"Section 7: the proposed empirical research agenda is valuable; specifying one or two concrete cognitive-interview or survey-experiment designs would strengthen the call for future work.","section":null}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is a solid historical-legal synthesis with a clearly labeled policy proposal. The two major comments are essentially framing and demarcation issues rather than factual errors; once the authors clarify that “public benefit” is an implicit/historical rather than strictly statutory criterion and that the Section 6 definition is optional guidance, the paper is ready for publication. Fit for a statistics/official-statistics journal is good; novelty is interpretive rather than empirical, which is appropriate for the venue."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The paper’s real contribution is a clean extraction of two criteria that actually sit in the statutes (CIPSEA 502(5)/(7), Title 13 §9, Evidence Act 3563) and a multi-sentence broader definition offered explicitly “for consideration.” That definition is the only new piece; everything else is careful restatement of Taft 1910, the 1929 Act, wartime breaches, Privacy Act/FIPPs, Belmont/Menlo, ASA ethics, and parallel international language.\n\nWhat it does well: the historical arc is accurate and readable, the two-criteria claim is directly supported by the quoted text, and the authors never pretend the ethical frameworks are themselves statutory definitions of the term. They flag the need for empirical work on public understanding and keep the proposal provisional. Citations are on-point; self-cites to Eltinge are peripheral. No circularity, no invented entities beyond the proposed definition itself.\n\nSoft spots are real but proportionate. The leap from Belmont/Menlo/FIPPs and the four Evidence Act responsibilities into an operational NSO definition is normative, not deduced; the paper itself labels it a proposal, so the leap is not load-bearing for the descriptive claim. The international section is thin (one-sentence glosses). No data, no math, no formal verification—none claimed. Impact stays inside official-statistics governance and respondent messaging.\n\nThis is for people who write NSO policy, disclosure standards, or respondent materials, and for anyone teaching the legal-ethical boundary of official statistics. It is not a methods paper. A serious editor should send it to peer review; the synthesis is careful enough and the definition is concrete enough to deserve referee time, even if the final version stays modest. I would cite the two-criteria extraction and the proposed definition if I were writing on statistical confidentiality or agency trust. Bring it to reading group if the group cares about data-governance language; otherwise skip.","headline":"Solid historical-legal synthesis of two statutory criteria for “statistical purposes,” plus a candid proposal for a broader NSO definition; modest novelty, clean framing, worth a referee.","tokens_in":13741,"tokens_out":481,"would_cite":true,"duration_ms":4543,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Laws already define 'statistical purposes' by two criteria—aggregate public-benefit statistics and a hard ban on using data to harm respondents—and a broader operational definition is needed.","keywords":["statistical purposes","confidentiality","public benefit","disclosure avoidance","privacy protection","public trust","data quality","official statistics"],"falsifier":"Empirical cognitive testing or legal review showing that either (a) the two-criteria reading is not in fact the predominant interpretation across current U.S. statutes and agency practice, or (b) adding the broader integrity and group-harm language produces no measurable change in respondent understanding, trust, or agency decision quality.","tokens_in":13888,"feed_emoji":"📊","tokens_out":701,"duration_ms":5778,"temperature":0.7,"pith_summary":"National statistical offices routinely tell people that data will be used only for 'statistical purposes,' yet the phrase is rarely defined clearly for the public and is poorly understood by respondents. A review of U.S. statutes, regulations, and agency practice shows that the term has long rested on two concrete criteria: statistics must describe relatively large population groups and serve a general public benefit, and identifiable information must never be used for legal, regulatory, or other actions that harm the people or organizations who supplied it. The same phrase also sits inside a larger web of scientific-integrity rules, professional ethics codes, and the four statutory responsibilities of federal statistical agencies. The authors therefore offer a broader working definition that makes those implicit obligations explicit—accuracy, relevance, objectivity, transparency about limitations, stewardship of the public’s trust, and avoidance of group or purely partisan harm—so that the phrase can serve as a more reliable guide for both public messaging and day-to-day decisions.","feed_headline":"What 'statistical purposes only' actually means in law","feed_subtitle":"Two criteria already govern the phrase; a broader definition would make the guardrails usable day to day.","key_machinery":"The two-criterion statutory core (aggregate public-benefit statistics plus functional separation that bars use of identifiable data for administrative, regulatory, or enforcement action against the data subject), expanded into an explicit multi-part operational definition that also requires scientific integrity, transparency, and avoidance of group or purely private/partisan harm.","core_discovery":"Reviews of underlying laws and policies identify two predominant criteria for 'statistical purposes': (1) production of statistical information about relatively large population aggregates, with the intention of creating a general public benefit; and (2) protection of the confidentiality of data collected about data subjects and a related prohibition against the use of that information for legal or regulatory action against those individuals or organizations. A broader applied definition that also incorporates scientific integrity, transparency about limitations, and avoidance of group or partisan harm is therefore warranted for NSO decision-making.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Two legal criteria already define 'statistical purposes only'","Laws pin 'statistical purposes' to aggregates plus confidentiality","What laws require for data used 'for statistical purposes only'","NSOs' 'statistical purposes only' rests on two clear criteria","A usable definition of statistical purposes beyond bare legality"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The claim assumes that ethical frameworks and the four fundamental responsibilities of statistical agencies can and should be treated as load-bearing parts of the operational meaning of 'statistical purposes,' even though those frameworks are not themselves the statutory definition of the term.","fun_headline_variants_meta":{"raw":{"variants":["Two legal criteria already define 'statistical purposes only'","Laws pin 'statistical purposes' to aggregates plus confidentiality","What laws require for data used 'for statistical purposes only'","NSOs' 'statistical purposes only' rests on two clear criteria","A usable definition of statistical purposes beyond bare legality"]},"model":"grok-4.5","effort":"low","cost_usd":0.002454,"raw_usage":{"total_tokens":977,"prompt_tokens":756,"num_sources_used":0,"completion_tokens":80,"cost_in_usd_ticks":24540000,"prompt_tokens_details":{"text_tokens":756,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":141,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":756,"tokens_out":80,"duration_ms":1871,"temperature":1.0,"reasoning_tokens":141,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T03:14:23.839329+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Empirical cognitive testing or legal review showing that either (a) the two-criteria reading is not in fact the predominant interpretation across current U.S. statutes and agency practice, or (b) adding the broader integrity and group-harm language produces no measurable change in respondent understanding, trust, or agency decision quality.","supporting_citations":[],"review_version":1}