Who watches the watchmen?

Quis custodiet ipsos custodes? — Juvenal

Rubric v0.3.2 44 reports 15 organizations 1999–2026 Avg 5.0/10

Why This Exists

Research reports about India and Indian Americans shape laws, news coverage, and university courses. But who checks whether those reports used good methods? We do. The Citation Integrity Dashboard scores 44 published reports on how well their research was conducted — not on what they concluded.

A report can be factually correct and still have poor methodology. A widely-cited report can fail basic research standards.

These reports have been cited in congressional hearings, State Department briefings, and major newsrooms. We publish our scoring criteria, weights, and every piece of evidence openly — before any report is evaluated.

How CID Scores

A three-step, pre-registered review
  1. 1
    We read the full report. Every scored report is read cover to cover. We examine the methodology, data collection, sourcing, and analytical framework — not the conclusions.
  2. 2
    We score eight dimensions. Each report is evaluated on eight aspects of research quality: how terms are defined, whether sources are independent, whether data can be verified by outsiders.
  3. 3
    Everything is published. The rubric, weights, and scoring criteria were published before any report was evaluated. If an organization disagrees, they can submit a response that we publish unedited.

Corpus at a Glance

44 reports · 15 organizations · 1999–2026
  • 1 Research-Grade
  • 8 Adequate
  • 21 Deficient
  • 14 Advocacy-Grade
Average score 5.0/10 across the corpus. Rubric weights were frozen in git before any report was scored — every score links back to evidence. See the rubric for band definitions.

Evaluated Reports

44 reports · sort, filter, and collapse
Sort:
Filter:

Organizations

7 with two or more scored reports

Citation Loops

4 documented patterns

Rubric Preview

Eight dimensions · non-compensatory caps · five grade bands

Every report is evaluated on eight dimensions of methodological rigor. Weights are published before any report is scored — not adjusted after the fact.

Dimension Weight What it measures
D1 Definitional Precision 12% Are the key terms defined clearly enough that someone else could apply them the same way?
D2 Classification Rigor 18% Would different analysts looking at the same data sort it into the same categories?
D3 Case Capture & Sampling 15% Does the data actually represent what the report claims it represents?
D4 Coverage Symmetry 15% Does the report cover its topic evenly, or does it only look in one direction?
D5 Source Independence 10% Do the sources check out independently, or do they all trace back to the same place?
D6 Verification Standards 18% Could an outsider verify the claims by checking the underlying evidence?
D7 Transparency & Governance 5% Is it clear who funded the work, who wrote it, and whether they have conflicts of interest?
D8 Counter-Evidence 7% Does the report address criticism and acknowledge what it can't prove?
D3 < 3 Overall score capped at 5.9 — sampling cannot support the report's claims.
D6 < 7 Cannot reach Research-Grade — underlying data is not independently verifiable.
8.0–10.0 Research-Grade
6.0–7.9 Adequate
4.0–5.9 Deficient
2.0–3.9 Advocacy-Grade
0.0–1.9 Unreliable