Variations in racial and ethnic groups' trust in researchers associated with willingness to participate in research

The study addresses why non-White adults in the U.S. are underrepresented in biomedical research and whether different dimensions of trust in researchers vary across racial/ethnic groups and relate to willingness to participate. New Jersey’s diverse and rapidly growing Asian and Hispanic populations, coupled with persistent health disparities, present a critical context to examine attitudes toward research. Prior work often aggregates Asian and Hispanic subgroups and typically includes only two or three racial/ethnic groups, obscuring within-group differences. The authors hypothesized that using a standardized multi-item trust scale across five racial/ethnic groups would identify shared and unique trust-related factors associated with willingness to participate in various research activities. They further hypothesized that willingness would differ by research activity type and by subgroup, informing tailored recruitment strategies.

Existing literature links lower research participation among minoritized groups to medical mistrust and experiences of discrimination, but studies often have limited racial/ethnic diversity and aggregate subgroups, especially within Asian and Hispanic populations. Prior qualitative and quantitative studies report: (1) mixed or limited understanding of clinical trials among Chinese immigrants and the influence of trusted intermediaries (e.g., adult children, physicians) on participation; (2) variable associations between perceived discrimination, mistrust, and willingness, with some evidence that Black and Hispanic adults may still express willingness for certain low-risk studies; (3) lower willingness among Asian Americans in large multi-site U.S. cohorts compared with other groups, though site composition and historical context may affect results; and (4) the need to disaggregate Asian and Hispanic subgroups due to culture-specific beliefs (e.g., concerns about blood among some Chinese) and differing migration histories. The authors situate their work within calls to prospectively assess trust dimensions (safety/beneficence, equity, transparency) and their link to participation within each subgroup rather than assuming a uniform trust–participation relationship.

Design and ethics: Cross-sectional, prospective survey study conducted October 2020–February 2022. The study qualified for Common Rule Exemption 2(i) and was approved by the Rutgers University IRB. No personally identifying information was collected; consent was obtained via agreement to proceed with anonymous surveys. Participants and recruitment: 437 adults were approached at community presentations, health fairs, and cultural events across New Jersey, with intentional outreach to Black, Hispanic, Chinese, and non-Chinese Asian communities. Inclusion: age ≥18 years. Of those approached, 293 (67%) completed both surveys fully. Respondents self-identified as Hispanic (n=102), Black (n=49), Chinese (n=48), non-Chinese Asian (n=53), or White (n=41). Women comprised 72% of the sample; 38% were aged ≥55 years. Educational attainment varied by group (e.g., 51% of Hispanic respondents had high school or less; 62% of Chinese and 71% of White respondents had college or more). Survey instruments and administration: Two anonymous instruments were administered on paper or online in English, Spanish, simplified Chinese (Mandarin), Korean, Tagalog, Urdu, Gujarati, and Hindi, with forward–backward translation by native speakers. Staff fluent in participants’ preferred language could administer surveys verbally when available. - Trust in Research: 12 Likert items (1=Strongly disagree to 5=Strongly agree) assessing beliefs about researchers’ motivations and behaviors (safety/beneficence, equity, transparency). Seven negatively worded items were reverse coded so higher total scores indicated greater trust. Total possible score: 12–60. - Research Activities: Eight yes/no items on willingness to participate in hypothetical studies: anonymous health survey, medical record review, blood draw, genetic analysis, medication study, equipment-based study, overnight stay, and participation without remuneration. Statistical analysis: Conducted in SAS 9.2. For the 12-item trust scale, descriptive statistics, correlations, reliability, exploratory factor analysis (EFA; principal factor method, Varimax rotation) and confirmatory factor analysis (CFA) were performed. Factor retention considered eigenvalues >1, scree plot, and interpretability. Items with cross-loadings or low loadings were dropped iteratively (items 5, 8, 10 cross-loaded and were removed before final EFA). CFA fit assessed via CFI (>0.90 acceptable) and RMSEA (≤0.06 good; ≤0.08 acceptable). Factor scores (continuous) were generated for each participant. Group differences across race/ethnicity, gender, education, and age were tested via ANOVA with post hoc comparisons. Willingness patterns were examined using hierarchical cluster analysis (Ward’s method) on the eight dichotomous willingness responses; the pseudo F and T-squared statistics suggested two main clusters with a small third cluster. Chi-squared tests compared racial/ethnic distributions across clusters. A total willingness score was computed (range 0–6) summing affirmative responses excluding health survey and no-payment items. Mediation analysis (regression-based with Sobel test) evaluated whether trust factor scores mediated associations between race/ethnicity and total willingness.

Sample characteristics: 293 complete responses; 38% aged ≥55 years; 72% women. Education differed by race/ethnicity (p<0.001). Overall perceptions: 80.2% (235/293) agreed their attitude toward research is positive, yet 74.7% (219/293) endorsed at least one negative perception; the most common was that researchers act differently toward minority participants than toward White participants (30.0%, 87/293). No demographic differences between those with none vs at least one negative perception. Trust factor structure (EFA/CFA): Three factors emerged and were retained: - Researchers as fiduciaries (items 1,2,3,9): safety/beneficence. Example loadings: item 3 “It's safe to be in a medical research study” 0.68; item 2 “would not involve me in harmful research” 0.61. - Transparency in research (items 4,6,7): disclosure and respect. Example loadings: item 4 (reverse-scored) 0.70; item 6 0.68; item 7 (reverse-scored) 0.49. - Equity in research (items 11,12): fairness across racial/ethnic groups and preference for concordance. Example loadings: item 11 (reverse-scored) 0.52; item 12 (reverse-scored) 0.47. Items 5, 8, and 10 cross-loaded and were excluded from the factor solution. Group differences in trust factors: - Fiduciaries: Chinese respondents had significantly lower scores vs White (p<0.001), Hispanic (p<0.001), and non-Chinese Asian (p=0.004). - Equity: All non-White groups had lower equity scores vs Whites (p<0.005), with Black and Hispanic respondents lowest. - Transparency: Higher among those with a bachelor’s degree vs high school or less (p=0.007) and among women vs men (p=0.008). Additional item-level observations: Education (p<0.001) correlated with agreement that researchers are more interested in careers than learning about health/disease (reverse-coded), but this did not align with factor scores. Only Black respondents showed greater agreement that researchers act differently toward minority vs White participants (p<0.001). Willingness clusters and distributions: - Cluster 1 “Mostly surveys” (n=136): high willingness for health surveys (95%); lower for medical record review (47%), blood draw (26%), genetic analysis (4%), overnight stay (7%). Equipment 0%; medication 2%; no payment 64%. - Cluster 2 “Blood/genetic analysis” (n=128): high willingness for blood draw (99%), genetic analysis (93%), health surveys (98%), medical record review (83%), overnight stay (62%); equipment 80%; medication 45%; no payment 72%. - Cluster 3 “Equipment” (n=29): defining willingness for equipment-based study (100%); medication 31%; no payment 52%. Racial/ethnic distribution across clusters: Chinese were most likely in “Mostly surveys” (85%); Hispanic (61%) and White (58%) were most likely in “Blood/genetic analysis.” Non-Chinese Asian resembled Chinese; Black respondents were more evenly distributed across clusters. Mediation analysis (total willingness score 0–6): Chinese respondents had lower willingness than Whites (p=0.003); Hispanic respondents had higher willingness than Whites (p=0.026). Only Chinese race/ethnicity was associated with lower fiduciary trust (B = -0.54; 95% CI -0.85 to -0.23; p=0.0007). Fiduciary trust significantly mediated the association between Chinese ethnicity and willingness (Sobel Z = -3.22; p<0.001; direct effect c’ = -0.64; 95% CI -1.36 to 0.08; p=0.08; indirect effect = -0.69; 95% CI -1.06 to -0.32; p=0.001). No other significant mediation effects were found. Descriptive means (Table 1 highlights): Mean trust scores were lowest among Black respondents vs White (25.9 vs 30.9; p<0.05). Willingness categories were lower among Chinese (3.3), non-Chinese Asian (3.4), and Black (4.1) vs Hispanic (5.2; p<0.05).

The findings support that the relationship between trust and willingness to participate in research is heterogeneous across racial/ethnic groups and varies by the dimension of trust assessed. Chinese respondents’ lower trust in researchers as fiduciaries (safety/beneficence) appears pivotal, mediating their lower willingness to participate in procedures beyond anonymous surveys. In contrast, despite perceiving inequities in how participants are treated, Hispanic and Black respondents’ willingness to participate in minimally risky activities (blood draw, genetic analysis) remained relatively high (particularly among Hispanic respondents), indicating that mistrust in equity does not uniformly suppress willingness. These results address the research question by demonstrating that distinct trust domains (fiduciary duty, transparency, equity) operate differently across groups and differentially influence willingness for specific research modalities. The study underscores the need for subgroup-specific recruitment and consent strategies emphasizing safety for Chinese communities, while for Hispanic participants, historical inequity messaging alone may be less critical than addressing logistical and value propositions for participation. The work also situates current patterns relative to pre-pandemic multi-site studies, suggesting that local context and the COVID-19 era may further shape trust and willingness profiles.

This study identifies three trust domains relevant to biomedical researchers—fiduciary responsibility, equity, and transparency—and shows that their salience and impact on willingness to participate differ across racial/ethnic groups. Chinese respondents’ lower fiduciary trust mediates reduced willingness for more invasive/procedural studies, while Hispanic respondents maintain high willingness for blood and genetic studies despite lower perceived equity. Broad assumptions linking mistrust to low participation are therefore unreliable. Researchers should prospectively assess trust domains and tailor recruitment, consent framing, and study design to each subgroup (e.g., emphasizing safety and risk mitigation in Chinese communities). Future research should link theoretical willingness to actual enrollment, disaggregate Asian and Hispanic subgroups further (e.g., Asian Indian, Puerto Rican, Dominican), evaluate the roles of non-racial factors (e.g., income, immigration status, LGBTQIA+ status, disability), and examine pandemic-related shifts in trust and participation.

- Cross-sectional survey of theoretical willingness may not translate to actual participation; unmeasured motivators (e.g., personal illness, incentives) could alter behavior. - Convenience sampling at community events during the COVID-19 pandemic may bias toward health-conscious, more acculturated, or less risk-averse individuals and those more willing to engage researchers. - Small overall sample with uneven subgroup sizes; limited power for detailed subgroup analyses (e.g., Asian Indian, Puerto Rican, Dominican), and insufficient numbers for broader intersectional analyses (e.g., income, LGBTQIA+ status, disability, immigration status). - Potential social desirability biases and translation nuances across languages; possible Happy Migrant effect. - The trust scale required item removals due to cross-loadings; some items did not align with derived factors, and education effects were observed on certain items. - Pandemic-era context may limit generalizability relative to pre-pandemic studies.