| Dim | Dimension | Score | Weight | Flag |
|---|---|---|---|---|
| D1 | Definitional Precision | 8 | 12% | — |
| D2 | Classification Rigor | 8 | 18% | — |
| D3 | Case Capture & Sampling | 8 | 15% | csohate.org appears once in citations — verify recruitment methodology does not include advocacy distribution |
| D4 | Coverage Symmetry | 9 | 15% | — |
| D5 | Source Independence | 8 | 10% | csohate.org appears once — provenance verification pending |
| D6 | Verification Standards | 5 | 18% | ⚑ Scoring rule limits grade — no documented data access pathway |
| D7 | Transparency & Governance | 6 | 5% | Survey funding not disclosed in report |
| D8 | Counter-Evidence | 7 | 7% | — |
| Composite Score | 7.4 | Adequate | ||
Scope: The survey data is not available for download and Carnegie does not document a formal research data request process. This is the second Carnegie survey with this documented gap. If Carnegie documents an equivalent access pathway to Pew's archive process, both surveys (CID-0004 and CID-0010) would be reassessed for Research-Grade.
Scope: csohate.org appears once in citations. Relationship to CSOH organizational structure requires verification before this is cleared. Does not suggest sampling bias from available evidence.
Scope: Who funded the YouGov data collection for a survey covering Trump approval and immigration enforcement is material information for readers assessing potential influence on question design or reporting.
Consistent with CID-0004 (2020 survey). Political and religious identity constructs operationalized through validated question batteries. 'Discrimination' measured through specific behavioral experience questions, not general self-classification. 2026 additions (Trump approval, immigration enforcement, DOGE attitudes) operationalized with clear Likert scales. Score 8 rather than 9 because discrimination construct relies on self-reported perception rather than behavioral definition — appropriate for survey research.
Adapted for Survey type. Identical methodology to 2020 survey — YouGov panel with ACS matching. The longitudinal consistency (same team, same platform, comparable methodology across six years) is itself a methodological strength: changes in findings reflect population changes rather than measurement changes.
csohate.org appears once in citations — verify recruitment methodology does not include advocacy distribution
Adapted for Survey type. 170/191 denominator flag rate (89%) confirms near-universal denominator reporting. Zero directional content confirmed by scope-claim alignment analysis. csohate.org appearing once is noted for provenance verification but does not suggest sampling bias from available evidence.
Same as 2020 survey. Symmetric question design produces symmetric coverage. 2026 focus on Trump 2.0, immigration policy, and discrimination experiences is appropriate given political context and does not introduce asymmetry. All religious and political subgroups receive equivalent treatment.
csohate.org appears once — provenance verification pending
Clean source profile consistent with CID-0004. Carnegie is a nonpartisan foreign policy institution. cloudfront.net CDN citations are the report's own data tables. csohate.org single citation requires a provenance note but does not establish independence failure from available evidence.
⚑ Scoring rule limits grade — no documented data access pathway
Identical gap to 2020 survey. Data not available for open download. Carnegie does not document a formal data request process in the report. Under the revised rule (v0.3.1), a documented formal request process would satisfy the standard — as it does for Pew. The pattern across two Carnegie surveys (2020 and 2026) confirms this is systemic rather than an oversight.
Survey funding not disclosed in report
Carnegie institutional transparency is adequate. Authors named with institutional affiliations. Funding for the YouGov panel not disclosed in the report — who paid for data collection on a politically sensitive survey covering Trump approval and immigration enforcement is material information. Same gap as 2020 survey.
Limitations section present. Methodology section documents constraints on generalizability. No recommendations section — methodologically appropriate for this report type, correctly suppresses advocacy orientation flag. Slightly stronger than 2020 survey on this dimension.
How this report's findings have been cited or applied after publication. Severity reflects the gap between what the report establishes and how it was represented.
Claimed scope: The Economic Times framed the survey primarily as a bilateral relations story, headlining '55% of Indian Americans disapprove of Trump's India policies' and foregrounding crises in 'trade, technology transfers, and strategic cooperation that have strained the Quad alliance.' The article also asserted the survey 'warns that ongoing U.S. policies risk eroding decades of progress in the bilateral relationship.'
Established scope: The report established that 55% of Indian Americans disapprove of Trump's handling of U.S.-India relations, but explicitly noted that one-quarter of respondents held no opinion—'suggesting that the issue has low salience with many Indian Americans.' The authors concluded that foreign policy, including U.S.-India relations, 'remains of secondary concern for most respondents' and that dissatisfaction with Trump's India handling has 'not meaningfully reshaped vote intentions.' The report contains no warning about eroding bilateral progress as a top-line finding.
The Economic Times article (February 20, 2026) repackaged the survey primarily as a U.S.-India diplomatic story, which serves the interests of an Indian financial readership more than a diaspora politics readership. The actual report's conclusion explicitly cautions that 'foreign policy barely registers' for Indian Americans when evaluating politics. The ET's framing of the bilateral relationship as the central finding inverts the report's hierarchy of findings, where domestic economics and discrimination led. No margin of error was cited in the ET article. Source: https://economictimes.com/news/india/55-of-indian-americans-disapprove-of-trumps-india-policies-survey/articleshow/128588061.cms
Claimed scope: A New India Abroad op-ed (October 2025, responding to the 2024 wave but applying the argument to the 2026 wave's similar caste finding) and CoHNA's public commentary cited the survey's finding that only 5-7% of Indian Americans reported caste-based discrimination to conclude that 'caste discrimination is not a widespread reality in the U.S., and existing anti-discrimination laws are more than sufficient to address any incidents.' CoHNA further characterized the survey's question asking whether respondents support anti-caste laws as 'a loaded question, similar to asking, Do you support genocide?', framing the 77% support finding as an artifact of question design rather than genuine opinion.
Established scope: The 2026 report found that caste-based discrimination was reported by 5% of respondents who personally experienced any discrimination—accurately a minority form compared to skin color (36%), country of origin (21%), and religion (17%). The report does not claim caste discrimination is the dominant form. The 2024 wave (July 2025) additionally found 77% of Indian Americans support anti-caste legislation. The report's authors presented both findings neutrally, noting only that 'pro-Hindu groups typically perceived as close to the BJP and its ecosystem have been the most vocally opposed to caste-based regulation in the United States.' The report does not characterize the anti-caste question as 'loaded.'
CoHNA and the aligned New India Abroad op-ed engaged in selective citation of the caste prevalence finding (5-7%) while dismissing the 77% support for anti-caste legislation as methodologically flawed without evidence. This is a documented pro-Hindu advocacy posture on the Carnegie surveys: CoHNA's own press release (2023) explicitly cited the Carnegie 2021 survey's low caste-discrimination numbers as 'demolishing' Equality Labs' claims, and the same framing recurs in the 2025/2026 cycle. The actual survey reports neither confirm nor deny the advocacy position; they present both low prevalence and high legislative support as data. The selective deployment of the low-prevalence number—while dismissing the 77% finding as 'predictable'—constitutes a meaningful escalation of the caste debate beyond the report's scope. Sources: https://www.newindiaabroad.com/english/opinion/understanding-the-carnegie-indian-american-survey-the-hindu-view; https://hinduamerican.org/blog/carnegie-indian-american-attitudes-survey-2025/; https://cohna.org/new-survey-analyzes-awareness-of-caste-in-the-us/
Scope: The report's finding that 71% disapprove of Trump is accurate and accurately cited. However, the report equally emphasized that (a) Democratic support has not rebounded commensurately; (b) independents have grown to 29%; (c) discrimination prevalence has remained statistically unchanged since 2020 across three survey waves; and (d) the community is ideologically centrist, not polarized leftward. The topline approval figure is correct but presented without the nuancing caveat that disapproval of Trump has not translated to Democratic gains.
The '7 in 10 disapprove' frame is accurate within the ±3.6 percent margin of error and does not misstate data. However, isolating it as the report's 'clear message' (Roundglass wording) omits the report's equally prominent finding of Democratic softening—what Carnegie itself called a 'rising dissatisfaction with both major parties.' This is a simplification rather than a distortion, and severity is accordingly low. The margin of error (±3.6 pp) was cited in NRI Pulse's coverage but was omitted from Roundglass and WION's social posts. Sources: https://www.facebook.com/roundglassindiacenter/posts/122272908170072067/; https://www.facebook.com/WIONews/videos/807189014975326/
Scope: The report found that 1,000 Indian American adults were surveyed with a ±3.6 percent overall margin of error, using YouGov's sample-matching procedure drawing from a panel of 500,000 U.S.-based residents. The 2026 survey also incorporated a revised sample frame to capture mixed-race Indian Americans, and the authors themselves cautioned that 'one must exercise some degree of caution when examining trends over time' as a result of this methodological change.
NRI Pulse's February 19, 2026 article ('Indian Americans Face Surge in Online Hate, Strongly Disapprove of Trump') is a model of responsible secondary citation within this dataset. It correctly cited the ±3.6 percent margin of error, accurately named the YouGov partnership, described the sample frame (online panel, 1,000 adults), and noted the expanded multiracial sample in the 2026 wave. Key statistics—71% disapproval, 48% online racism exposure, 46% Democratic identification—are all accurately transcribed from the source. Caste discrimination prevalence (5%) was noted accurately in context alongside skin color (36%) as the dominant form. No methodological limitations were omitted or distorted. This represents a responsible citation pattern. Source: https://nripulse.com/indian-americans-face-surge-in-online-hate-strongly-disapprove-of-trump-carnegie-survey/
Scope: The 2026 report established that Indian Americans' Democratic identification has declined from 52% in 2020 to 46% in 2026, Republican identification has risen from 15% to 19%, and independents have grown to 29%. The report explicitly states that 'claims of a wholesale realignment are overstated' and that the community 'remains more Democratic-leaning than the American electorate as a whole.' The survey's ideological findings show moderates as the largest bloc (32%), with conservatives at 22% and liberals at 21%.
The American Bazaar's February 20, 2026 article accurately reported the topline approval figures, Democratic/Republican/independent breakdown with percentages, and margin of error. It correctly conveyed the survey's central tension—widespread Trump disapproval coexisting with no commensurate Democratic rebound—and cited Mamdani enthusiasm findings accurately. The article did not overstate the partisan shift toward Republicans and included the report's own caution language about 'dissatisfaction with both major parties.' This is a responsible, accurate representation consistent with the 7.1 score given to the source report. Source: https://americanbazaaronline.com/2026/02/20/7-in-10-indian-americans-disapprove-of-trump-carnegie-survey-475502/
Scope: The 2026 Carnegie IAAS found that 5% of Indian American respondents reported personal caste-based discrimination in the relevant period—the lowest-reported category, behind skin color (36%), country of origin (21%), religion (17%), and gender (16%). The 2024 wave found 7%. The report's authors themselves note the question of who discriminates on the basis of caste remains unresolved (the 2021 wave found one-third of caste-discrimination reporters identified the discriminator as non-Indian, a finding the authors called 'perplexing'). The Carnegie surveys are cross-sectional and rely on self-reported discrimination; they do not confirm or deny the Equality Labs 25% physical/verbal assault statistic.
Equality Labs did not directly cite or respond to the 2026 IAAS as of March 21, 2026. However, the organization's foundational claim—that 1 in 4 caste-oppressed South Asian Americans face physical/verbal assault—is structurally in tension with three successive Carnegie IAAS waves finding 5-7% self-reported caste discrimination. The Carnegie 2021 report's footnote explicitly called the Equality Labs 2018 survey 'unscientific' (as cited by CoHNA). Pro-Hindu groups (CoHNA, HAF) have used the Carnegie low-prevalence findings as a counter-citation against Equality Labs. The result is a circular dynamic in which Equality Labs' data and Carnegie's data are mobilized by opposing advocacy coalitions without the underlying methodological differences being explained to general audiences. The escalation field is false because Equality Labs has not yet cited or distorted the 2026 IAAS specifically. Source: https://www.equalitylabs.org; https://cohna.org/new-survey-analyzes-awareness-of-caste-in-the-us/
Scope: The report is based on a nationally representative online survey of 1,000 Indian American adults conducted November 25, 2025–January 6, 2026, with ±3.6% overall margin of error. The report covers partisan identity, Trump approval, policy priorities, discrimination experiences, and political affect. A methodological revision in 2026 incorporated multiracial Indian Americans for the first time, and the authors explicitly flagged this as requiring 'appropriate caution' in trend comparisons. Caste discrimination was measured as self-reported personal experience; no causal attribution to perpetrator group was collected in this wave.
The op-ed published in New India Abroad on February 21, 2026, was written by the report's own co-authors (Badrinathan, Kapur, Robaina, Vaishnav) and functions as the most responsible secondary citation in the dataset—accurately summarizing survey percentages, explicitly noting behavioral self-censorship findings, and placing discrimination prevalence in comparative context. The disclosure that the op-ed 'does not necessarily reflect the official policy or position of India Abroad' was included, and the authors' institutional affiliations (American University, Johns Hopkins-SAIS, Carnegie) were stated. No inflation of findings, no omission of the stable-across-three-waves discrimination trend, and no omission of the methodological caveat regarding multiracial respondents. This is a gold-standard secondary citation by the primary investigators. Source: https://www.newindiaabroad.com/english/opinion/indian-americans-discrimination-and-the-question-of-belonging