CID scoring: SASAC Hindutva Harassment Field Manual

Document: Hindutva Harassment Field Manual Organization: South Asia Scholar Activist Collective (SASAC) Document type: TYPE 6 — Advocacy Document Rubric version: v0.3.2 Scoring date: 2026-03-22 Source file: hhfm-full.md (6,728 words)

---

Type classification rationale

The Hindutva Harassment Field Manual is a guidance document for individuals who self-identify as targets of Hindutva-associated harassment. It provides identification frameworks, response protocols, and resource directories organized by audience (targets, allies, students, employers). It does not collect original data, does not produce a dataset, and does not claim empirical findings. Its purpose is to advance a normative position: Hindutva-associated harassment is a definable, identifiable, and organized phenomenon, and the manual equips readers to respond. TYPE 6.

D2 (Classification Rigor) and D3 (Case Capture & Sampling) are N/A. Weights redistribute proportionally across the six applicable dimensions: D1 at 17.9%, D4 at 22.4%, D5 at 14.9%, D6 at 26.9%, D7 at 7.5%, D8 at 10.4%.

---

Dimension scores

D1 — Definitional precision | 3 / 10

The manual’s core term is “Hindutva.” It appears 80 times across 6,728 words. “Hindu” appears 74 times. The document includes a definitions/glossary section, one of only two structural elements detected by pipeline audit (2/10 sections present). That section exists, which separates this document from those that define nothing at all.

Operational specificity is the problem. For an advocacy document scored under the adapted pathway, the rubric asks: are characterizing terms defined with criteria, or used editorially? The manual uses “Hindutva” as both a political ideology descriptor and a harassment-origin label without drawing a line between political advocacy and conduct that constitutes harassment. “Hindu nationalist” (7 instances), “right-wing” (3), “far-right” (1), and “supremacist” (1) appear alongside “Hindutva” without differentiation criteria. Could five independent readers apply these terms to the same set of incidents and agree? The definitions section exists but does not provide decision rules for borderline cases. The gap between “Hindu political activism” and “Hindutva harassment” is left to reader judgment.

Score of 3: some definitions provided, but core characterizing terms lack the operational precision required for consistent independent application.

D2 — Classification rigor | N/A

Not applicable for TYPE 6. Weight redistributed.

D3 — Case capture & sampling | N/A

Not applicable for TYPE 6. Weight redistributed. Non-compensatory sampling integrity limit does not apply.

D4 — Coverage symmetry | 2 / 10

The pipeline detected the title as “Intersectional Hate,” a universalist framing. The content is particularist. The scope-claim alignment analysis found: anti-Hindu content accounts for 100% of directional terms. Zero anti-Muslim, anti-Christian, anti-Sikh, or anti-Dalit directional content.

Identity term directionality confirms this. “Hindu” appears as target 31 times and as agent 4 times (target/agent ratio: 7.8). “Dalit” appears as target 7 times, agent 0. “Muslim” appears as target 1 time, agent 0. The manual frames one community as the exclusive source of harassment and multiple communities as exclusive targets.

The Swap Test fails. Remove identity markers from the manual’s identification criteria for “attacks” and “organized harassment campaigns.” Apply them symmetrically. The criteria would not produce symmetric classifications. They are built around specific organizational names (RSS, HSS, BJP, Hindu YUVA, CoHNA, HAF) rather than behavioral criteria that could apply regardless of which group is acting.

Section headings confirm unidirectional framing: “What is Hindutva and How can It Impact You?”, “How do you recognize Hindutva influences on campus?”, “How You Can Fight Against Hindutva.” No section addresses the possibility of harassment flowing in any other direction, or that South Asian scholars might face organized campaigns from non-Hindutva actors.

The document could have scored higher here if it were transparent about its particularist scope. A field manual that says “this guide is specifically about X type of harassment” and sticks to that framing is not penalized for narrow scope under the rubric. But “Intersectional Hate” as a framing implies breadth the content does not deliver.

Score of 2: significant mismatch between framing and actual coverage. Unidirectional coverage with no acknowledgment of asymmetry. Public framing implies broader scope than content delivers.

D5 — Source independence | 2 / 10

The source type split tells the story: 1 academic source, 22 media, 1 government, 107 advocacy or other. That is 107 out of 131 total URLs — 81.7% advocacy sourcing.

Self-referential sourcing is heavy. The top domain is hindutvaharassmentfieldmanual.org with 16 links — the document citing itself. southasiacollective.org (SASAC’s own site) contributes 8 more. Combined self-citation: 24/131 URLs, 18.3% of all sources.

The SASAC citation ecosystem appears here in full. equalitylabs.org contributes 4 links. Rutgers (the CSRR report, already scored by CID at 3.7 Advocacy-Grade) contributes 5. These organizations share overlapping scholarly networks and cite each other’s work as independent corroboration. The Provenance Trace for this document would loop through the same cluster documented in CID-0007 and CID-0008: SASAC → Equality Labs → Rutgers CSRR → media coverage → SASAC citing that coverage.

One academic source out of 131. One government source. The Herfindahl Index of 0.0316 technically indicates low concentration across many different domains. But domain diversity masks organizational homogeneity. Seventy-two unique domains drawn overwhelmingly from the same advocacy network is not source independence.

Has the organization published a finding that contradicts or complicates its prior work? No evidence of this in the document.

Score of 2: circular institutional structure where affiliated organizations serve as each other’s primary sources. External validation not possible because the verification chain loops within the same network.

D6 — Verification standards | 3 / 10

For TYPE 6 under the adapted pathway, D6 evaluates citation accuracy: does each cited source actually say what the document claims it says? The manual has 131 URLs across 72 unique domains, which is substantial citation infrastructure relative to its 6,728-word length.

Link quantity is not the issue. Verification chain quality is. When 81.7% of sources are advocacy materials, “citation accuracy” means checking whether advocacy source A correctly quotes advocacy source B, not whether the underlying empirical claim is independently verifiable. The one academic source and one government source cannot anchor verification for the document’s claims.

The pipeline’s quantitative claims extraction found 6 claims, but most are URL-encoding artifacts (values like 2580 and 2016 parsed from encoded URLs). The one substantive claim, “two-third of American adults have witnessed online harassment,” appears to reference a Pew Research Center finding, which is verifiable. But this is a general statistic about online harassment, not specific to the manual’s claims about Hindutva-associated harassment.

Structure audit: no methodology section, no data availability section, no corrections policy. There is no mechanism for an independent reader to verify the manual’s core claim that the described harassment patterns are organized, systematic, and attributable to a specific ideological movement.

Data access tier: Tier 3. No documented access pathway to underlying evidence. The document asks readers to trust the organizational framing.

Score of 3: some citation infrastructure exists but the verification chain leads predominantly to affiliated advocacy sources. No independent verification pathway for core claims. Tier 3 data access.

D7 — Transparency & governance | 4 / 10

The structure audit detected funding disclosure, one of only two structural elements present. This is notable; many advocacy documents in the CID corpus disclose nothing.

What is missing: no conflict of interest statement, no published data ethics policy, no clear governance structure. SASAC operates as a “collective,” and the organizational boundary between the collective, its individual members’ academic affiliations, and allied organizations (Equality Labs, Savera) is not documented. No board with governance authority is evident. No external financial audit referenced.

The funding disclosure meets the minimum for the 4–6 band: “990 filings available but minimal beyond legal requirements.” Without a COI statement and with opaque governance structure, the score stays in the lower half of this band.

Score of 4: funding disclosure present but governance otherwise opaque. No COI statement. Organizational structure unclear.

D8 — Counter-evidence | 1 / 10

For TYPE 6 under the adapted pathway, D8 weights limitations acknowledgment and corrections policy more than counter-evidence engagement.

The structure audit found: no limitations section, no counter-evidence or opposing views, no corrections/errata policy. Zero out of the three elements the adapted pathway prioritizes.

No limitation of the framework is acknowledged anywhere. False positive risk from the identification criteria goes unaddressed. Scholars who contest the Hindutva harassment framing are absent from the citations. No correction or revision to prior characterizations is referenced. The HAF (Hindu American Foundation) appears 7 times in citations — as a subject organization, not as a source whose perspective is engaged.

No evidence of methodological evolution, corrections, or engagement with criticism in any form.

Score of 1: no limitations section, no corrections policy, no engagement with counter-evidence. Below the floor for advocacy documents, which at minimum should acknowledge framework limitations.

---

Weighted score computation

Standard weights (v0.3.2, TYPE 6 effective)

Dimension	Score	Effective weight	Weighted
D1	3	17.9%	0.537
D4	2	22.4%	0.448
D5	2	14.9%	0.298
D6	3	26.9%	0.807
D7	4	7.5%	0.300
D8	1	10.4%	0.104
Total		100.0%	2.49

Grade: Advocacy-Grade (2.0–3.9)

Non-compensatory checks

• Sampling integrity limit (D3 < 3 → cap at 5.9): D3 is N/A for TYPE 6. Does not apply.
• Data access limit (D6 < 7 → blocks Research-Grade): D6 = 3. Research-Grade blocked. Moot at this score level.

---

Sensitivity analysis

Equal weights (all applicable dimensions at 16.67%)

Dimension	Score	Weight	Weighted
D1	3	16.67%	0.500
D4	2	16.67%	0.333
D5	2	16.67%	0.333
D6	3	16.67%	0.500
D7	4	16.67%	0.667
D8	1	16.67%	0.167
Total		100.0%	2.50

Grade: Advocacy-Grade.

Verification-heavy (D6 at 25%, others reduced proportionally)

Dimension	Score	Weight	Weighted
D1	3	18.37%	0.551
D4	2	22.96%	0.459
D5	2	15.31%	0.306
D6	3	25.00%	0.750
D7	4	7.65%	0.306
D8	1	10.71%	0.107
Total		100.0%	2.48

Grade: Advocacy-Grade.

Grade stability

All three weighting schemes produce scores between 2.48 and 2.50. The grade is stable. No weighting scheme moves the document out of the Advocacy-Grade band. The score is not near any band boundary.

---

Calibration context

The SASAC Hindutva Harassment Field Manual at 2.49 sits below the SASAC Hindu Nationalism Reporting Guide (CID-scored at 4.3, Deficient) and below the Rutgers CSRR report (CID-scored at 3.7, Advocacy-Grade). This is consistent: the Reporting Guide at least attempts a research framing with some structural infrastructure, while the Field Manual is a pure guidance document with minimal methodological apparatus.

The HHFM closes a documented loop in the CID corpus. CID-0007 and CID-0008 identified the SASAC citation ecosystem as a case study in circular amplification. This scoring confirms the sourcing pattern at the document level: 81.7% advocacy sources, 18.3% self-citation, and a verification chain that loops through the same cluster of affiliated organizations.

---

Pipeline-sourced evidence summary

All scoring is grounded in the eight MAI Analyzer output files dated 2026-03-22. Key pipeline findings referenced:

• Source split: 1 academic, 22 media, 1 government, 107 advocacy/other (file 03)
• Self-citation rate: 24/131 URLs to own domains (file 03)
• Herfindahl Index: 0.0316 (file 03, file 08)
• Structure audit: 2/10 sections detected — definitions and funding disclosure only (file 04)
• Identity directionality: Hindu as target 31 times, agent 4 times; 100% anti-Hindu directional content (files 05, 06)
• Quantitative claims: 6 detected, most are URL-encoding artifacts; 1 substantive (“two-third of American adults”) (file 02)
• Denominator flags: 5 flagged, all “percentage without visible denominator” — all appear to be URL-encoding false positives (file 07)
• Detected title: “Intersectional Hate” — classified AMBIGUOUS (file 06)