Statistical realism of synthetic responses is not a reliable proxy for treatment-effect accuracy.

SocioLens project,

Claim LD07-HQ-C07 (v1.0)

Statistical realism of synthetic responses is not a reliable proxy for treatment-effect accuracy.

Research opportunities this claim motivates

What this claim leaves open for research. Each is a hypothesis about a gap, not a finding.

Can synthetic respondents recover strong treatment-effect heterogeneity?Response-level realism does not predict treatment-effect accuracy, so CATE recovery must be tested directly.

Provenance and status

Claim type: Fact
Publication status: Eligible for publication, with qualifiers: preprint
Verification state: Checked against the primary source; the source is a preprint. 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: Claim that effect forecasting always fails; contradiction of Ashokkumar’s archive-forecasting estimand
Limitations: Preprint; specific intervention families/model set
Source (as cited): Zonghan Li & Feng Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458
Source type: Preprint (recorded as “Preprint”)
Institutional position: Independent academic
Maturity of the evidence: Emerging evidence
Recorded date: 25 September 2026
Currentness / correction: This is the current version. No correction or concern is recorded on its sources.

Sources this version cites

Li & Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458 (Preprint)
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): FACT (as printed: “FACT”)
Publication status (coded): ELIGIBLE (PREPRINT) (as printed: “ELIGIBLE WITH PREPRINT QUALIFIER”)
Verification state (coded): CONTENT_VERIFIED_PRIMARY (PREPRINT)
Source type (coded): PREPRINT
Maturity (coded): EMERGING

#	Canonical field	Canonical value (verbatim)	Normalized (Claim Value Vocabulary v1.0.2)
1	CLAIM ID	LD07-HQ-C07	—
2	PUBLIC-FACING CLAIM	Statistical realism of synthetic responses is not a reliable proxy for treatment-effect accuracy.	—
3	CLAIM TYPE	FACT	base_types: FACT
4	RESEARCH STRAND	C	—
5	MATURITY	EMERGING	states: EMERGING
6	SOURCE	Zonghan Li & Feng Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458	source_ids: SRC-LD07-LI-JI-REALISM-ATE-2026
7	SOURCE TYPE	Preprint	classes: PREPRINT
8	INSTITUTIONAL POSITION	Independent academic	—
9	SCOPE	Cross-national climate-psychology experiments	—
10	DESIGN	Real-participant-profile-conditioned synthetic responses compared with human outcomes and supervised baselines	—
11	SAMPLE / DATASET	59,508 participants, 62 countries, 11 interventions; two replication datasets totaling 20,785	—
12	MODEL / TOOL / VERSION	GPT-4o-mini, Gemini 2.5 Flash Lite, Claude 3 Haiku	—
13	COMPARATOR	Human response/treatment effects; OLS/LASSO	—
14	MAIN RESULT	Realism–ATE rank relation near zero (rho≈.10); similar response MAE but ~2.1x treatment-effect error; optimizing realism can worsen ATE accuracy	—
15	INFERENTIAL TARGET	Average treatment effects	—
16	DOES NOT LICENSE	Claim that effect forecasting always fails; contradiction of Ashokkumar’s archive-forecasting estimand	—
17	VERIFICATION STATE	CLEARED WITH QUALIFIER	state: CONTENT_VERIFIED_PRIMARY · qualifiers: PREPRINT · verified_by: UNKNOWN
18	LIMITATIONS	Preprint; specific intervention families/model set	—
19	CONTRADICTORY / SUPPORTING SOURCES	Ashokkumar positive on different estimand	—
20	GEOGRAPHY / LANGUAGE	62 countries	—
21	DISTRIBUTIONAL LEDGER	Country/outcome heterogeneity central	—
22	R01 / FIELD-MAP LINKS	D9, R-1, L-B	—
23	WHAT WOULD CHANGE THIS CLAIM?	Independent study showing realism metric reliably predicts causal-effect validity out of sample	—
24	PUBLICATION STATUS	ELIGIBLE WITH PREPRINT QUALIFIER	state: ELIGIBLE · qualifiers: PREPRINT
25	RECORDED DATE	2026-09-25	—

Field 22 references resolved against Field Map v0.2: D9 (DOMAIN); R-1 (LENS); L-B (LENS)

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

Pinned Sources (field 6, version-specific)

“Zonghan Li & Feng Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458” → Li & Ji, “Statistical realism is not evidence that LLMs…” (SRC-LD07-LI-JI-REALISM-ATE-2026--v1.0) — Li & Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458; class PREPRINT (explicit public provenance pin)

Cited by Living Debate versions

AI in the research apparatus (v0.1) LD-07
AI in the research apparatus (v0.2) LD-07

Cite this page

SocioLens project. (2026, September 25). Statistical realism of synthetic responses is not a reliable proxy for treatment-effect accuracy [Claim LD07-HQ-C07, version v1.0]. SocioLens. https://sociolens-web.pages.dev/claims/ld07-hq-c07/v1.0/

@misc{sociolens_ld07_hq_c07_v1_0,
  author = {{SocioLens project}},
  title = {Statistical realism of synthetic responses is not a reliable proxy for treatment-effect accuracy.},
  year = {2026},
  date = {2026-09-25},
  howpublished = {\url{https://sociolens-web.pages.dev/claims/ld07-hq-c07/v1.0/}},
  note = {Claim LD07-HQ-C07, 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 5e4b8f9eff53d1ac2faacaa293e943786e6525a0ad41ced5c82c42ad17831568.

Frozen Drive-derived inputs

R03 — LD-07 Canonical Claim Ledger v1.0 — LOCKED (v1.0) — Drive ID 1xVz_LrQAA8EA1tPhVWIrDgOUU3Zidlg9; revision: NOT_EXPOSED_BY_DRIVE_TEXT_EXPORT; Drive modified 2026-09-29T04:31:14.120Z; block “ENTRY 07 — LD07-HQ-C07” SHA-256 b1d20a4d6dc9231ebb3a46ca27dc9b9e880c97be00960c93e86714802fe967f2; SocioLens — Programme Status & Authority Register v0.6 status “CURRENT.”; frozen fixture fixtures/frozen/r03-ld07-claim-ledger-v1.0-canonical.

Builder decisions (not Drive data)

VERSION = v1.0 — basis: Claim version taken from the canonical ledger artifact version (R03 — LD-07 Canonical Claim Ledger v1.0); Implementation Architecture Selection §5
CURRENT_POINTER = true — basis: SocioLens — Programme Status & Authority Register v0.6 status for "R03 — LD-07 Canonical Claim Ledger v1.0 — LOCKED": "CURRENT."
FIELD_17_CONTEXT_QUALIFIERS = PREPRINT — basis: R03 — HQ Cross-Batch Synthesis & Adjudication (v0.1, Drive 1V-HOVaPFwF8BTf60-LGPsJXf59V3kRlQ)
FIELD_6_SOURCE_REFERENCE = SRC-LD07-LI-JI-REALISM-ATE-2026--v1.0 — basis: field 6 of ENTRY 07 — LD07-HQ-C07 cites "Zonghan Li & Feng Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458"; the frozen field-6 segment of this ClaimVersion's ledger block; Drive-derived identity authority (Issue #39 Source inventory): Batch 02 §6; ledger C07; KEY EVIDENCE.
SOURCE_PROVENANCE_PIN = SRC-LD07-LI-JI-REALISM-ATE-2026--v1.0 — basis: Issue #5 proof rendering (Architecture Proof Protocol v0.1 AP-08 / Amendment v1.2 §6): explicit version-pinned public provenance of this ClaimVersion — the pinned field-6 Source record "Zonghan Li & Feng Ji, “Statistical realism is not evidence that LLMs can estimate treatment effects in social science experiments,” arXiv:2604.02458"; Drive-derived identity authority (Issue #39 Source inventory): Batch 02 §6; ledger C07; KEY EVIDENCE.