In a review by 29 experts of 445 language-model benchmarks from leading machine-learning and NLP venues, only 16.0% used any statistical testing and about half (53.4%) presented evidence that they measure what they claim to measure; 12.3% relied wholly and another 27.0% partly on convenience sampling.

SocioLens project,

Claim LD06-SL-C05 (v1.0)

Provenance and status

Claim type: Estimate
Publication status: Eligible for publication, with a qualifier
Verification state: Checked against the primary source. Who checked this claim is not recorded for it; claim checks so far were AI-led, and no person has checked claims against their sources.
Does not license: A judgement on any single benchmark, or on benchmarks outside the reviewed venues
Limitations: Reflects what papers report, not what developers did; venue-based sample
Source (as cited): Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703
Source type: Peer-reviewed conference paper (recorded as “Peer-reviewed conference paper”)
Institutional position: Independent academic (University of Oxford-led)
Maturity of the evidence: Rapidly maturing evidence
Recorded date: 15 October 2026
Currentness / correction: This is the current version. No correction or concern is recorded on its sources.

Sources this version cites

Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703 (Peer-reviewed conference paper)
Read the original

Version

This is the current version, v1.0. Earlier versions stay available, and their records never change.

Versions: v1.0 · current

Raw record: every recorded field, with its coded values

All 25 canonical Claim fields, verbatim, with the coded values derived from them. The plain-language labels above are read from these.

Claim type (coded): ESTIMATE (as printed: “ESTIMATE”)
Publication status (coded): ELIGIBLE (QUALIFIED) (as printed: “ELIGIBLE WITH QUALIFIER”)
Verification state (coded): CONTENT_VERIFIED_PRIMARY
Source type (coded): PEER_REVIEWED_CONFERENCE_PAPER
Maturity (coded): RAPIDLY_MATURING

#	Canonical field	Canonical value (verbatim)	Normalized (Claim Value Vocabulary v1.0.2)
1	CLAIM ID	LD06-SL-C05	—
2	PUBLIC-FACING CLAIM	In a review by 29 experts of 445 language-model benchmarks from leading machine-learning and NLP venues, only 16.0% used any statistical testing and about half (53.4%) presented evidence that they measure what they claim to measure; 12.3% relied wholly and another 27.0% partly on convenience sampling.	—
3	CLAIM TYPE	ESTIMATE	base_types: ESTIMATE
4	RESEARCH STRAND	A — Benchmark validity and measurement	—
5	MATURITY	RAPIDLY MATURING	states: RAPIDLY_MATURING
6	SOURCE	Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703	source_ids: SRC-SL-BEAN-ET-AL-NEURIPS-2025-ARXIV-2511-04703
7	SOURCE TYPE	Peer-reviewed conference paper	classes: PEER_REVIEWED_CONFERENCE_PAPER
8	INSTITUTIONAL POSITION	Independent academic (University of Oxford-led)	—
9	SCOPE	Benchmarks published at leading NLP and machine-learning conferences	—
10	DESIGN	Systematic review with a structured coding scheme	—
11	SAMPLE / DATASET	445 benchmarks	—
12	MODEL / TOOL / VERSION	Not applicable	—
13	COMPARATOR	None	—
14	MAIN RESULT	16.0% conduct any statistical testing; 53.4% present construct-validity evidence; 78.2% define the phenomenon measured, of which 47.8% use contested definitions; convenience sampling used exclusively in 12.3% and in part in another 27.0%; 38.2% reuse earlier benchmark or exam data; eight recommendations	—
15	INFERENTIAL TARGET	How far benchmark practice meets basic measurement standards	—
16	DOES NOT LICENSE	A judgement on any single benchmark, or on benchmarks outside the reviewed venues	—
17	VERIFICATION STATE	CONTENT VERIFIED	state: CONTENT_VERIFIED_PRIMARY · verified_by: UNKNOWN
18	LIMITATIONS	Reflects what papers report, not what developers did; venue-based sample	—
19	CONTRADICTORY / SUPPORTING SOURCES	Consistent with LD06-SL-C06; the authors treat the problems as fixable	—
20	GEOGRAPHY / LANGUAGE	Not applicable / English	—
21	DISTRIBUTIONAL LEDGER	Not analysed in this claim	—
22	R01 / FIELD-MAP LINKS	LD-06, D4b	—
23	WHAT WOULD CHANGE THIS CLAIM?	A repeat review showing whether validity reporting spreads	—
24	PUBLICATION STATUS	ELIGIBLE WITH QUALIFIER	state: ELIGIBLE · qualifiers: QUALIFIED
25	RECORDED DATE	2026-10-15	—

Field 22 references resolved against Field Map v0.2: LD-06 (LIVING_DEBATE); D4b (SUB_AREA)

The sources as recorded, with their record IDs and coded source types:

Pinned Sources (field 6, version-specific)

“Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703” → Bean et al. (2025), “Measuring what Matters: Construct…” (SRC-SL-BEAN-ET-AL-NEURIPS-2025-ARXIV-2511-04703--v1.0) — Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703; class PEER_REVIEWED_CONFERENCE_PAPER (explicit public provenance pin)

Cited by Living Debate versions

The evaluation apparatus: Can benchmarks, leaderboards and audits carry the weight placed on them? (v0.1) LD-06

Cite this page

SocioLens project. (2026, October 15). In a review by 29 experts of 445 language-model benchmarks from leading machine-learning and NLP venues, only 16.0% used any statistical testing and about half (53.4%) presented evidence that they measure what they claim to measure; 12.3% relied wholly and another 27.0% partly on convenience sampling [Claim LD06-SL-C05, version v1.0]. SocioLens. https://sociolens-web.pages.dev/claims/ld06-sl-c05/v1.0/

@misc{sociolens_ld06_sl_c05_v1_0,
  author = {{SocioLens project}},
  title = {In a review by 29 experts of 445 language-model benchmarks from leading machine-learning and NLP venues, only 16.0\% used any statistical testing and about half (53.4\%) presented evidence that they measure what they claim to measure; 12.3\% relied wholly and another 27.0\% partly on convenience sampling.},
  year = {2026},
  date = {2026-10-15},
  howpublished = {\url{https://sociolens-web.pages.dev/claims/ld06-sl-c05/v1.0/}},
  note = {Claim LD06-SL-C05, version v1.0}
}

The link points to this version, whose record never changes. For a finding, cite the original study as well: SocioLens summarizes research and does not replace it.

Research-of-record provenance

Materialized by sociolens-materializer 0.6.0 at 2026-09-28T20:00:00.000Z; normalization contract SocioLens — Claim Value Vocabulary v1.0.2 — Pre-Spike Materialization Contract (v1.0.2). Content seal SHA-256 83e405898f537636e4107045cce9c1067cc6a1a2ea17c7b3852c811e93c4b9dc.

Repository-authored inputs (not Drive exports)

SocioLens claim — LD06-SL-C05 v1.0 (v1.0) — written in the SocioLens repository by Project Lead (Claude, AI) on 2026-10-15; no Drive identity exists. Authority: AGENTS.md "Authority" (the Project Lead decides research content within the research-integrity rules) and the Owner's choice in the SocioLens project chat on 15 October 2026 to build the LD-06 debate; a new Claim, first version, written from the project's LD-06 source notes, checked against full texts or official abstracts. Block “ENTRY 05 — LD06-SL-C05” SHA-256 744226d6eedc602f1609e4ffa520590b0e343d5f5ad452e30d6300a82a7ba7c1; fixture fixtures/authored/claims/LD06-SL-C05--v1.0.

Builder decisions (not Drive data)

VERSION = v1.0 — basis: Claim version taken from the repository-authored claim fixture version (SocioLens claim — LD06-SL-C05 v1.0); the first version of a new Claim of the LD-06 debate
CURRENT_POINTER = true — basis: repository-authored claim "SocioLens claim — LD06-SL-C05 v1.0" (Project Lead (Claude, AI), 2026-10-15); basis: AGENTS.md "Authority" (the Project Lead decides research content within the research-integrity rules) and the Owner's choice in the SocioLens project chat on 15 October 2026 to build the LD-06 debate; a new Claim, first version, written from the project's LD-06 source notes, checked against full texts or official abstracts
FIELD_6_SOURCE_REFERENCE = SRC-SL-BEAN-ET-AL-NEURIPS-2025-ARXIV-2511-04703--v1.0 — basis: field 6 of ENTRY 05 — LD06-SL-C05 cites "Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703"; the only field-6 segment of this ClaimVersion's repository-authored ledger block.
SOURCE_PROVENANCE_PIN = SRC-SL-BEAN-ET-AL-NEURIPS-2025-ARXIV-2511-04703--v1.0 — basis: explicit version-pinned public provenance of this ClaimVersion (Amendment v1.2 §6): the pinned field-6 Source record "Bean, Kearns, Romanou, Hafner, Mayne, Batzner et al., “Measuring what Matters: Construct Validity in Large Language Model Benchmarks,” NeurIPS 2025 Datasets and Benchmarks Track, arXiv:2511.04703".