Skip to main content
Schizophrenia Bulletin logoLink to Schizophrenia Bulletin
. 2025 Oct 30;52(5):sbaf193. doi: 10.1093/schbul/sbaf193

Associations Between Area Deprivation and Paranoia in Serious Mental Illness

Sarah A Berretta 1, Robert A Ackerman 2, Raeanne C Moore 3, Colin A Depp 4, Philip D Harvey 5,6, Arundati Nagendra 7, Amy E Pinkham 8,✉
PMCID: PMC13645824  PMID: 41165228

Abstract

Background and Hypothesis

Though linked to paranoia in the general population and high-risk groups, neighborhood deprivation remains underexplored in serious mental illness. Paranoia also varies by demographic subgroup, potentially reflecting exposure to place-based disadvantage. Advances in the measurement of neighborhood deprivation and momentary experiences of suspiciousness could shed light on these relationships. This study uses neighborhood-level data to examine associations between area deprivation and paranoia in schizophrenia spectrum disorders (SSDs) and bipolar disorder (BD).

Study Design

Data from 293 participants with SSD (n = 148) and BD (n = 145) from three US sites were analyzed. Neighborhood disadvantage was quantified using the National Area Deprivation Index (ADI). Clinician-rated paranoia was assessed using the Positive and Negative Syndrome Scale; momentary feelings of threat/suspiciousness (MFTS) was measured across 30 days of Ecological Momentary Assessments (EMAs). Analyses examined the relationships between ADI and paranoia and tested if ADI explains observed demographic differences in paranoia. To assess specificity, hallucinations were examined in parallel.

Study Results

Greater neighborhood disadvantage was associated with higher clinician-rated paranoia and MFTS; no association was observed for hallucinations. Participants with SSD and those with racial backgrounds other than White had higher ADI scores and higher paranoia. The ADI-paranoia relationship did not differ in strength across clinical or demographic groups, nor across study sites. Area Deprivation Index significantly mediated the association between racial background and paranoia.

Conclusions

Neighborhood deprivation is associated with paranoia but not hallucinations, supporting a specific link between environment and paranoia. Improvement of neighborhood-level conditions may offer a pathway to reducing paranoia and improving functioning.

Keywords: psychosis, deprivation, environment, symptoms, suspiciousness, paranoia

Introduction

Paranoia, defined as unfounded suspiciousness and mistrust of others, is a prominent symptom of serious mental illness (SMI).1 Paranoid thoughts range from mild suspicion to severe delusions, affecting about half of those with psychotic disorders2-4 and occurring regularly in individuals with bipolar disorder.5 Paranoia is associated with negative functional and psychological outcomes, such as social isolation, poorer interpersonal functioning, mis-estimation of competencies, and heightened depression and anxiety.1,4,5 Thus, paranoia contributes to poorer quality of life, and understanding factors that contribute to paranoia is critical.6,7

Greater paranoia has been consistently observed among participants from racial backgrounds other than White.8-12 In one sample of individuals with depressive disorders, prevalence of paranoia was 15.5% among White individuals, compared to 31.5% among Black individuals, and 21.1% among Latino individuals.13 These differences may reflect exposure to broader contextual pressures, including chronic social threat and increased environmental and economic challenges, each linked to elevated threat perception.8,10,11 As such, assessing paranoia in individuals with SMI requires cultural sensitivity, as clinician-rated symptoms may be influenced by racial bias or misinterpretation of contextually grounded mistrust, particularly among racially minoritized group.14,15 Despite evidence that racial background may influence diagnostic patterns,16,17 few studies have examined whether similar patterns exist within SMI, such as in schizophrenia spectrum disorders (SSDs) or bipolar disorder (BD).

Neighborhood deprivation, characterized by unemployment, poverty, housing instability, and limited utilities, has been associated with elevated suspiciousness in the general population, with stress mediating the relationship.18 Similar patterns have been observed in high-risk groups—Wilson and colleagues found that young adults at clinical high risk for psychosis reported greater suspiciousness in high-crime, low-SES neighborhoods, even after adjusting for other positive symptoms.19 Deprivation has also been linked to increased incidence of psychosis, particularly in urban and economically unequal areas.20 Still, it remains unclear how neighborhood conditions shape paranoia in SMI in US contexts.

Neighborhood deprivation varies geographically and is correlated with racial background; individuals identifying with racial backgrounds other than White are more likely to reside in areas of higher disadvantage.21 These same populations are diagnosed with psychotic disorders at disproportionately higher rates.13,22 Emerging frameworks suggest that chronic exposure to adverse conditions, including community disinvestment and limited access to supportive resources, may heighten risk for paranoia.23,24 Within this framework, neighborhood disadvantage may not only contribute directly to paranoia but could explain observed differences in symptom expression across groups.

Although neighborhood disadvantage is increasingly recognized as a contributor to psychopathology, prior research has not systematically examined its role in paranoia. The Area Deprivation Index (ADI), a validated measure of the social exposome, enables fine-grained, cross-regional comparisons at the neighborhood level.25-27 Designed specifically for US populations, this spatial resolution makes it well-suited for investigating contextual factors that shape paranoid ideation in SMI.

In parallel, Ecological Momentary Assessment (EMA) provides an ecologically valid method for capturing paranoid thoughts as they occur in daily life within participants’ natural environments, reducing reliance on retrospective recall and clinician judgment. Few studies have integrated this approach with high-resolution environmental data. Integrating ADI with EMA could offer novel insights into how neighborhood disadvantage interacts with in vivo experiences of paranoia. Although the ADI has been utilized to examine links between neighborhood deprivation with medical rehospitalization28 and cognitive decline29; its potential to elucidate paranoia-specific pathways remains largely unexplored.

We aim to address these gaps by investigating how neighborhood disadvantage relates to paranoia across various groups. Specifically, we hypothesize that higher ADI scores, indicating greater deprivation, will be associated with increased paranoia, as measured by both clinician-rated and momentary aggregated assessments. We hypothesize that racial backgrounds other than White and a diagnosis of SSD (relative to BD) will be associated with greater severity of paranoia. Finally, we predict that ADI will mediate the relationship between participant racial background and paranoia.

An exploratory aim investigates whether the strength of the relationship between neighborhood disadvantage and paranoia differs by demographic or diagnostic group, or recruitment site. To assess the specificity of our findings for paranoia, we also examine whether ADI is similarly associated with severity of hallucinations.

Methods

Participants

This study utilizes data collected from 01 June 2018 through 28 February 2022 for a study focused on technology-based assessment of everyday functioning in SMI.30-32 Participants completed an initial laboratory visit, followed by 30 days of EMA surveys, and returned approximately 1 month later for a final laboratory visit. During each visit, participants completed interviews assessing current and past psychopathological symptoms, along with brief questionnaires and computerized tasks.

Participants with SSD or BD were recruited from The University of Miami Miller School of Medicine (UM), the University of California, San Diego (UCSD), and The University of Texas at Dallas (UTD). Participants, aged 18-65, were required to meet DSM-5 criteria for schizophrenia, schizoaffective disorder, or bipolar I or II disorder; current or past psychotic symptoms were not required. For those with BD, a staging model of severity was implemented to require at least one mood episode recurrence or incomplete remission following a first episode.31 All participants were clinically stable, with no psychiatric hospitalizations for at least 6 weeks and on a stable medication regimen with no dose changes >20% in the 2 weeks prior to enrollment.

Participants were excluded if they had a history of neurological or neurodegenerative disorders (e.g., stroke, epilepsy, dementia, and Parkinson’s) or a history of a traumatic brain injury with unconsciousness lasting >15 min. Individuals were also excluded for sensory impairments that could interfere with assessment (e.g., uncorrectable vision or significant hearing loss), pervasive developmental disorders or intellectual disability (IQ <70), a current substance use disorder, or limited English proficiency. All data were collected in English. All participants provided written informed consent, and the study was IRB approved at each site (UTD, #18-93; UM, #20180352; and UCSD, #180716).

The initial sample included 310 participants. Participants with poor EMA compliance (<30 completed surveys out of a possible 90 surveys across the 30-day study period) were excluded from analyses. Participants excluded for poor EMA adherence (n = 17) did not differ significantly from included participants on demographics (race, diagnosis, and age), PANSS P6 or P3 score, or National ADI (all P > .05). After filtering, the final analytic sample included 293 participants (93.1% of the original sample), who completed an average of 75 EMA surveys (SD = 15.24) out of a possible 90 (83% adherence). This final analytic sample contained 145 participants with BD and 148 participants with SSD. For our demographic analyses, self-reported racial background was dichotomized into White (n = 136) and Other racial background (n = 157) categories to ensure adequately sized and relatively balanced samples across groups. Racial backgrounds other than White, in this sample, included participants with Black, Asian, Native American, or unspecified racial identities. This grouping is used for statistical purposes.

Consistent with current guidance on selecting reference populations in health disparities research,33,34 White participants were used as the reference group, as they have historically showed the lowest levels of paranoia. Racial background was not defined using biological information such as genomics, but rather as a self-identified, socially defined characteristic reflecting lived experience. This analytic decision aligns with NIMHD recommendations to make explicit the assumptions underlying reference group selection and to interpret disparities in relation to structural positioning and exposure.34 Full participant demographics are detailed in Table 1.

Table 1.

Demographics

Demographics SSD (n = 148) BD (n = 145) Overall (n = 293)
n (%) n (%) n (%)
Sex
 Male 69 (46.62%) 51 (35.17%) 120 (40.96 %)
 Female 79 (52.38%) 94 (64.83%) 173 (59.04%)
Race
 White 47 (31.80%) 89 (61.42%) 136 (46.42%)
 Other (Non-White) 101 (68.2%) 56 (38.58%) 157 (53.58%)
  Black 83 (56.1%) 29 (22.76%) 112 (38.23%)
  Native American 1 (0.68%) 1 (0.69%) 2 (0.68%)
  Asian 3 (2.03%) 9 (6.21%) 12 (4.10%)
  Hawaiian 2 (1.35%) 1 (0.69%) 3 (1.02%)
  Not listed 12 (8.11%) 16 (11.03%) 28 (9.56%)
Ethnicity
 Hispanic 39 (26.35%) 38 (26.21%) 77 (26.28%)
 Non-Hispanic 109 (73.65%) 107 (73.79%) 216 (73.72%)
Site
 University of Texas at Dallas 63 (42.57%) 63 (43.45%) 131 (44.71%)
 University of Miami 42 (28.38%) 31 (21,38%) 75 (25.60%)
 University of California, San Diego 43 (29.05%) 51 (35.17%) 87 (29.69%)
(Mean ± SD) (Mean ± SD) (Mean ± SD)
Age 41.42 ± 10.69 39.28 ± 10.67 40.00 ± 10.94
Education (in years) 12.81 ± 2.37 14.61 ± 2.61 14.00 ± 2.64
Key variables (range)
 National ADI (0-100) 52.15 ± 29.71 39.31 ± 28.09 45.80 ± 29.58
 PANSS P6 (1-6) 3.33 ± 1.25 2.39 ± 1.20 2.86 ± 1.31
 MFTS (1-7) 2.41 ± 1.57 1.50 ± 1.06 1.96 ± 1.42
 PANSS P3 (1-6) 3.19 ± 1.51 1.36 ± 0.81 2.28 ± 1.52
 MAH (1-7) 2.19 ± 1.46 1.19 ± 0.53 1.69 ± 1.21

Abbreviations: ADI = Area Deprivation Index, PANSS = Positive and Negative Syndrome Scale, MAH = Momentary auditory hallucinations, MFTS = Momentary feelings of threat/suspiciousness, P6 = Suspiciousness/Persecution, P3 = Hallucinations.

Measures

Neighborhood Disadvantage

Neighborhood disadvantage was operationalized using the ADI, a composite measure capturing socioeconomic disadvantage across 17 dimensions of social determinants of health. These dimensions include educational attainment (e.g., percentage of adults aged ≥ 25 years with <9 years of education, percentage with at least a high school diploma), employment characteristics (e.g., percentage in white-collar occupations, unemployment rate), income metrics (e.g., median family income, income disparity), and housing-related factors (e.g., homeownership rate, median home value, crowding, and access to utilities). Additional dimensions encompass poverty levels, single-parent households, and access to transportation and communication resources.26,27,29 The ADI is derived using principal components analysis from the 2018-2022 five-year American Community Survey and the US Census, aligning with this study’s data collection period. Scored on a scale from 0 to 100, higher ADI scores indicate greater neighborhood disadvantage. The ADI is calculated at the census block group level, offering high geographic granularity. In this study, ADI values were assigned using 12-digit Federal Information Processing Standards (FIPS) codes corresponding to each participant’s census block group of current residence at the time of enrollment. While both state- and national-level ADI scores are available through the Neighborhood Atlas, our study used national rankings to ensure comparability across participants recruited from multiple geographic regions.27,29,30 Notably, the ADI does not account for racial background or ethnicity, allowing for independent control of these variables in analyses.

Paranoia Assessment

Paranoia was assessed via both field-standard clinician-rated interviews and EMA, targeting momentary feelings of threat/suspiciousness (MFTS). Between the baseline and final laboratory visits, participants completed 30 days of EMA. Participants received three surveys per day, each containing the question, “Since the past alarm, how much have you had thoughts that others might want to harm you, or that people are untrustworthy?” Responses were recorded using a 7-point Likert scale ranging from 1 (Not at all) to 7 (Very much). A single MFTS score was calculated as the average of all responses to this question across the 30-day period, comparable to the single clinician-rated paranoia score collected via PANSS.

At both baseline and follow-up lab visits, clinician-rated paranoia was measured using item P6 (Suspiciousness/persecution) of the Positive and Negative Syndrome Scale (PANSS), a 30-item clinician-administered rating scale in which each item is scored from 1 (absent) to 7 (extreme).35 PANSS administrators were trained to a high reliability (ICC > 0.80) with a gold standard assessor. A paired t-test revealed no significant difference between baseline and study-end P6 scores (t(291) = −0.121, P = .904, mean difference = −0.08, 95% CI [−0.253, 0.233]), and the correlation between baseline and study-end scores was strong (r = 0.536, P < .001, 95% CI [0.449, 0.613]). These two scores were averaged to create a single rating of clinical paranoia.

The correlation between average MFTS and average clinician-rated paranoia was moderate-to-strong (r = 0.459, P < .001, 95% CI [0.364, 0.545]), suggesting some degree of consistency across measures and also indicating potential utility for separate analyses.

Hallucination Assessment

To parallel paranoia assessments, hallucinations were assessed using the same 30-day EMA protocol and clinician-rated interview described above. For clinician-rated symptoms, hallucinations were measured using the PANSS P3 item, which captures hallucinations across sensory modalities (e.g., auditory, visual, olfactory, gustatory, and tactile hallucinations).35 Scores (from 1 [absent] to 7 [extreme]) were similarly collected at both baseline and final lab visits and averaged to create a single rating of clinician-rated hallucinations. For momentary experiences of auditory hallucinations, EMA surveys contained the question: “Since the past alarm, how much have you experienced hearing voices?” Responses again ranged from 1 (Not at all) to 7 (Very much). A single Momentary Auditory Hallucinations (MAH) score was computed by averaging responses across the study period.

Statistical Plan

All analyses were conducted using R version 3.5.2.36 Data preparation included data cleaning and verification of model assumptions. After the exclusion of participants with poor EMA adherence, all included participants had complete data.

To test our first hypothesis, correlations between main variables were assessed, and initial simple regressions explored the relationship between ADI and paranoia measures. Because we expected the heterogeneity of variance assumption to be violated, Welch’s t-tests examined group differences in ADI, MFTS, and clinician-rated paranoia (PANSS P6) across diagnostic (BD vs. SSD) and demographic background (White vs. Other) categories to test hypothesis two.

To test our primary hypothesis that neighborhood disadvantage would mediate the relationship between participant racial background and paranoia, mediation analyses were conducted using the lavaan package in R.36,37 These analyses examined both direct and indirect effects. The a-path in our primary mediation model represents the effect of participant racial background on neighborhood deprivation. Next, the b-path represents the effect of neighborhood deprivation on paranoia, controlling for racial background. Finally, the c′-path represents the direct effect of participant racial background on paranoia after accounting for area deprivation. Separate models were specified for paranoia as measured by EMA and PANSS across all regression and mediation models. Demographic and diagnostic group variables were effect-coded (−0.5 = White, 0.5 = Other; −0.5 = BD, and 0.5 = SSD), and ADI scores were grand-mean centered. Unstandardized path coefficients (bs) are reported throughout, except for the a and c′ paths, for which Cohen’s d is also reported due to the binary coding of the demographic variable. For interpretability, the mediation model figure displays standardized estimates (βs) generated from the Std.all output in lavaan.37

We also conducted analyses to examine the specificity of these effects to paranoia by testing parallel models with hallucinations (MAH and PANSS P3) as outcome variables in place of paranoia. Simple regressions and mediation models were tested, along with moderation analyses by race and diagnosis. To statistically compare the strength of the correlations between ADI and paranoia versus hallucinations, we used Fisher r-to-z transformations for dependent correlations.

Finally, exploratory moderated regression analyses incorporating demographic group, diagnosis, and their interaction terms were conducted to test tentative hypotheses regarding whether the relationship between neighborhood deprivation and paranoia differed by group. Because participants were clustered by county-level codes, we used a robust estimator for our standard errors to account for geospatial dependencies.38,39 Sex and ethnicity were initially included as covariates but were removed from final models due to limited explanatory power and lack of influence on the primary effects. In addition, to evaluate whether the relationship between neighborhood deprivation and paranoia varied across geographic location, we conducted a one-way ANOVA to test site differences in ADI and specified robust moderation models with site dummy-coded (Dallas as the reference group) to examine ADI × Site interaction effects on paranoia. To formally test whether the inclusion of site interaction terms significantly improved model fit, we compared nested models with and without the interaction term using standard linear models and conventional F-tests. ANOVA-based R2 change tests are not currently compatible with clustered models, which adjust standard errors to account for heteroskedasticity and non-independence (e.g., due to shared FIPS codes). As such, these comparisons do not account for clustering.

Results

Descriptive statistics for all key variables (i.e., ADI, clinician-rated paranoia, and MFTS, and hallucination measures) appear in Supplementary Table S1, including values for the full sample and stratifications by sex, ethnicity, and site (see Supplementary Data Content 1). Table 3 presents descriptive statistics by diagnostic group and racial background, alongside inferential tests of group differences.

Table 3.

Welch’s t-Test for ADI, Paranoia, and Hallucinations Across Diagnostic and Demographic Groups

Measure SSD (n = 157)
(M ± SD)
BD (n = 155)
(M ± SD)
t-value P-value Cohen’s d [95% CI]
National ADI 52.15 ± 29.71 39.31 ± 28.09 −3.80 .0002** 0.44 [0.21, 0.68]
MFTS 2.41 ± 1.57 1.50 ± 1.06 −5.88 <.0001*** 0.68 [0.45, 0.92]
PANSS P6 3.33 ± 1.25 2.39 ± 1.20 −6.57 <.0001*** 0.77 [0.53, 1.00]
MAH 2.19 ± 1.46 1.19 ± 0.53 −7.88 <.0001*** 0.91 [0.67, 1.15]
PANSS P3 3.19 ± 1.51 1.36 ± 0.81 −12.95 <.0001*** 1.5 [1.24, 1.76]
Measure White ( n  = 136)
( M  ± SD)
Other ( n  = 157)
( M  ± SD)
t -value P -value Cohen’s  d  [95% CI]
National ADI 38.60 ± 26.76 52.03 ± 30.55 −4.01 <.0001*** 0.47 [0.23, 0.70]
MFTS 1.73 ± 1.28 2.16 ± 1.50 −2.62 .009** 0.30 [0.07, 0.53]
PANSS P6 2.60 ± 1.32 3.09 ± 1.27 −3.27 .001** 0.38 [0.15, 0.62]
MAH 1.51 ± 0.99 1.85 ± 1.35 −2.45 .015* 0.28 [0.05, 0.51]
PANSS P3 1.99 ± 1.41 2.54 ± 1.58 −3.13 .002** 0.36 [0.13, 0.59]

Abbreviations: ADI = Area Deprivation Index, MAH: momentary auditory hallucinations, MFTS: momentary feelings of threat/suspiciousness, PANSS = Positive and Negative Syndrome Scale, P6 = Suspiciousness/Persecution, P3 = Hallucinations. Cohen’s d calculated with pooled SD. *  P< .05, **P < .01, ***P < .001.

Relationship between Area Deprivation Index and Paranoia

In the overall sample, ADI showed significant positive correlations with MFTS (r = 0.232, P < .001, 95% CI [0.120, 0.337]) and with clinician-rated paranoia (r = 0.252, P < .001, 95% CI [0.141, 0.356]), indicating small-to-moderate relationships between neighborhood disadvantage and paranoia. Regression models indicated that ADI significantly predicted both MFTS and clinician-rated paranoia (P’s < .001), accounting for approximately 5% of the variance in each (R2 = 0.05 and 0.06, respectively; Table 2).

Table 2.

Regression Results

Momentary feelings of threat/suspiciousness (MFTS) b (SE) β/d t P 95% CI
Simple regression (R2 = .050)
 ADI 0.011 (0.002) 0.23 5.90 .003** [0.006, 0.016]
Moderated regression (R2 = .142)
 ADI 0.008 (0.001) 0.17 5.62 .002** [0.004, 0.012]
 Demographic group 0.051 (0.102) 0.04 0.49 .639 [−0.203, 0.304]
 Diagnostic group 0.814 (0.103) 0.57 7.92 <.001*** [0.561, 1.067]
 ADI × Demographic group −0.005 (0.006) −0.11 −0.90 .412 [−0.021, 0.010]
 ADI × Diagnostic group 0.009 (0.008) 0.19 1.19 .293 [−0.011, 0.030]
Clinician-rated paranoia (PANSS P6) b  (SE) β/ d t P 95% CI
Simple regression (R2 = .060)
 ADI 0.011 (0.004) 0.25 3.07 .032* [0.001, 0.021]
Moderated regression (R2 = .163)
 ADI 0.008 (0.003) 0.18 3.20 .023* [0.001, 0.015]
 Demographic group 0.116 (0.234) 0.09 0.50 .638 [−0.464, 0.696]
 Diagnostic group 0.824 (0.149) 0.63 5.52 .002** [0.475, 1.192]
 ADI × Demographic group −0.008 (0.005) −0.18 −1.65 .163 [−0.021, 0.005]
 ADI × Diagnostic group 0.010 (0.007) 0.21 1.28 .261 [−0.010, 0.029]

Abbreviations: SE = standard error, adjusted to account for non-independence in geospatial location across our sample. β reported for continuous predictors, Cohen’s d reported for dichotomous predictors (Demographic and Diagnostic group). For demographic group, White participants were coded as −0.5, and Other as 0.5. Similarly, for diagnosis, participants diagnosed with BD participants were coded as −0.5 and SSD as 0.5. ADI was grand-mean centered.

* P < .05, **P < .01, ***P < .001

Group Differences in Paranoia

Table 3 summarizes the results of Welch’s t-tests comparing demographic and diagnostic groups on ADI and both PANSS P6 and MFTS. Participants with racial backgrounds other than White demonstrated consistently higher scores on both PANSS P6 and MFTS and on neighborhood deprivation, with the largest difference observed in ADI scores. For diagnostic groups, participants with SSD showed significantly higher scores across all measures compared to those with BD, with large effect sizes for paranoia/suspiciousness measures (Cohen’s d = 0.77 and 0.68). Welch’s t-tests showed no significant group differences based on sex assigned at birth or Hispanic/Latino ethnicity (all P’s > .21, see Supplementary Material).

Mediation Analyses

Mediation analyses revealed that ADI fully mediated the association between the demographic grouping variable (i.e., self-reported racial background, White or Other) and momentary paranoia, and partially mediated the association with clinician-rated aggregate scores for paranoia. Figure 1 shows the mediational models with path statistics.

Figure 1.

Figure 1

Mediational models for momentary paranoia (A) and clinician-rated paranoia (B). Standardized path coefficients for mediation models examining the relationship between racial background and paranoia through area deprivation. Panel (A) shows MFTS, and panel (B) shows clinician-rated paranoia (PANSS P6). All path coefficients shown in the figure are standardized (β); unstandardized values (b, or Cohen’s d used for binary predictors) are reported in the main text. P < .05, **P < .01, ***P < .001

For MFTS, the demographic grouping variable was significantly associated with neighborhood deprivation, such that individuals identifying with racial backgrounds other than White tended to live in areas with higher ADI scores (a path: b = 13.44, 95% CI [7.06, 19.96], d = 0.47, P < .001). In turn, greater area deprivation was associated with increased MFTS (b path: b = 0.01, 95% CI [0.004, 0.016], P = .001). The indirect effect of participant racial background on momentary paranoia through ADI was significant (a × b path: b = 0.13, 95% CI [0.04, 0.26], P = .02). After accounting for ADI, the direct effect of the demographic grouping variable on MFTS was not statistically significant (c′ path: b = 0.29, 95% CI [−0.04, 0.62], d = 0.20, P = .08), indicating a mediated relationship.

For clinician-rated paranoia, a similar pattern was observed. The demographic variable significantly predicted neighborhood deprivation (a path: b = 13.44, 95% CI [6.94, 19.68], d = 0.47, P < .001), and area deprivation significantly predicted higher PANSS paranoia (b path: b = 0.01, 95% CI [0.004, 0.015], P < .001). The indirect effect of the demographic grouping variable on clinician-rated paranoia through ADI was also significant (a × b path: b = 0.13, 95% CI [0.05, 0.24], P = .008). Notably, after accounting for area deprivation, the direct effect was reduced but remained significant (c′ path: b = 0.37, 95% CI [0.06, 0.68], d = 0.27, P = .02).

Exploratory Moderated Regression Analyses

Racial Background and Diagnosis

Moderated regression models with effect coding were conducted to test whether diagnostic or demographic status moderated the ADI-paranoia relationship, with observations clustered by census tract codes. In these robust models, ADI remained a significant predictor of both MFTS (P = .002) and clinician-rated paranoia (P = .023), and diagnosis was also a significant predictor for both outcomes (MFTS: P < .001; PANSS P6: P = .002). However, interaction terms between ADI × Diagnostic group and ADI × Demographic group were nonsignificant (all P’s > .16), suggesting that while neighborhood disadvantage predicts paranoia, these associations did not statistically differ across diagnostic or demographic groups in our sample (Table 2).

Recruitment Site

Descriptive statistics for ADI and paranoia by site are presented in Supplementary Table S1. A one-way ANOVA revealed a significant effect of site on ADI scores, F(2, 290) = 61.46, P < .001. Post-hoc Tukey comparisons indicated that participants in Dallas (M = 59.63, SD = 27.90) had significantly higher ADI scores than those in Miami (M = 49.58, SD = 24.70; P = .016) and San Diego (M = 21.79, SD = 19.66; P < .001), and that Miami had significantly higher ADI scores than San Diego (P < .001).

Despite regional differences in neighborhood deprivation, there was no evidence that site moderated the association between ADI and paranoia. The addition of the Site interaction term did not significantly improve model fit for either EMA-rated paranoia (ΔR2 = 0.010, F(2, 287) = 1.40, P = .247) or clinician-rated paranoia (ΔR2 = 0.014, F(2, 287) = 1.24, P = .292). This aligns with the full robust models clustered by FIPS, in which none of the interaction terms were statistically significant (e.g., EMA: Miami × ADI, b = 0.006, 95% CI [–0.014, 0.026], β = 0.11, P = .385; San Diego × ADI, b = –0.011, 95% CI [–0.023, 0.002], β = –0.24, P = .078).

Testing Symptom Specificity: ADI Effects on Hallucinations

Area Deprivation Index showed an association with MFTS (r = 0.23, P < .001), but was not significantly associated with MAH (r = .03, P = .597). Fisher r-to-z transformation confirmed that the ADI-MFTS correlation was significantly stronger than the ADI-MAH correlation (z = –3.41, P < .001). Area Deprivation Index was significantly associated with clinician-rated hallucinations (PANSS P3; r = 0.15, P = .010), though this association was weaker than the one observed for clinician-rated paranoia (PANSS P6; r = 0.25, P < .001). However, this difference did not reach statistical significance (Fisher z = –1.59, P = .113).

Simple regression models showed that ADI did not significantly predict hallucinations, either through MAH (b = 0.0013, β = 0.03, P = .647) or clinician-rated PANSS P3 scores (b = 0.0078, β = 0.15, P = .071). Moderation analyses further indicated that the relationship between ADI and hallucinations was not significantly moderated by race or diagnosis (all P’s > .37). Although individuals with racial backgrounds other than White were more likely to report hallucinations across both EMA and clinician-based measures (t(283.36) = –2.45, P = .015, 95% CI [–0.61, –0.07]; t(290.75) = –3.13, P = .002, 95% CI [–0.89, –0.20], respectively), analyses revealed no significant mediation effect of ADI on this relationship. Specifically, indirect effects of race on hallucinations via ADI were not significant for EMA-rated voices (b = −0.000, P = .992) or for PANSS P3 (b = 0.080, P = .101).

Discussion

The current study underscores the role of environmental factors, and specifically neighborhood disadvantage, in the expression of paranoia across diagnostic and demographic groups in SMI. While prior research has explored the associations between socioeconomic conditions, personal background, and psychosis, our findings build on this work by demonstrating that social and environmental disadvantage, as quantified through component measures of neighborhood characteristics, is associated with clinician-rated and MFTS. Moreover, this study is among the first to show that neighborhood disadvantage mediates the relationship between racial background and paranoia, offering insights into how place-based disadvantage and contextual stressors may explain demographic differences in symptom expression in clinical samples.

Overall, our hypotheses were supported. Higher ADI scores were significantly associated with increased clinician-rated paranoia and increased MFTS, with effect sizes ranging from small to medium. In addition, participants from racial backgrounds other than White were more likely to live in higher-ADI environments and reported greater paranoia and MFTS. Area Deprivation Index, diagnosis, and demographic group each independently predicted paranoia/suspiciousness in initial models. Mediation analyses suggested that neighborhood disadvantage may help explain observed differences in clinical symptom expression across demographic groups. This association likely reflects the combined influence of structural factors such as economic hardship, adverse social experiences, and other contextual stressors, underscoring the relevance of environmental context in understanding paranoia. Alternatively, paranoia itself may contribute to functional disability that results in income loss and subsequent residence in deprived areas, in a variant of the social drift hypothesis which posits that the symptoms of SMI, including paranoia, can lead to downward socioeconomic mobility over time.40

The differential associations observed between neighborhood disadvantage and our two paranoia measures suggest that clinician-rated paranoia and MFTS may reflect related but somewhat distinct constructs. Whereas PANSS clinician ratings evaluate a continuum from distrustful attitudes to persecutory delusions and emphasize the degree of impact on functioning, the EMA MFTS item may instead capture more transient, contextually grounded experiences of social threat. This distinction may help explain why neighborhood disadvantage fully mediated the relationship between race and MFTS, but only partially mediated the relationship with clinician-rated paranoia. By distinguishing between these constructs, our findings highlight the need for measurement approaches that are sensitive to both clinical severity and ecologically valid, lived experiences of mistrust.

The stronger association between race and clinician-rated paranoia, compared to MFTS, may also reflect known bias in clinical assessments. Although this difference was modest in our sample, prior research has shown that racially minoritized individuals with SSD, particularly Black individuals, are more likely to have their symptoms misinterpreted or overpathologized by clinicians who lack awareness of patients’ sociocultural context.14-16 These findings align with evidence that paranoia in racially minoritized groups may be shaped by lived experiences of discrimination and may not conform to the same cognitive-affective pathways observed in predominantly White samples.14,41,42 Future research should further examine how assessment modality and clinician perspective influence observed group differences in paranoid ideation.

In exploratory analyses, neither demographic or diagnostic group status significantly moderated the association between ADI and paranoia, although ADI remained a significant predictor of overall symptom severity. This pattern is consistent with prior research broadly linking neighborhood deprivation to increased paranoia18,19 and extends such work by explicitly demonstrating that the strength of these associations is statistically comparable across groups. In addition, while there were significant site-level differences in area deprivation, recruitment site did not moderate the relationship between ADI and paranoia or MFTS, suggesting that the association is consistent across geographic locations.

The findings of this study are supported by theoretical frameworks situating paranoia within a biopsychosocial context, emphasizing the interplay between environmental, psychological, and social factors.42 Previous international studies, such as those using the Index of Multiple Deprivation (IMD) in the United Kingdom, similarly demonstrate that neighborhood deprivation predicts paranoia.19 Our findings extend this work by highlighting the global relevance of environmental conditions in psychological vulnerability while emphasizing that place-based disadvantage may help account for observed variability in paranoia and suspiciousness symptoms across groups. Importantly, while the IMD emphasizes income, employment, and crime indicators, the ADI includes variables tied to US-specific structures, such as housing quality, education, and access to transportation and health care but does not include crime indicators. In addition, the use of EMA to assess MFTS offers a granular, ecologically valid perspective not utilized in the previous European study that also speaks to the consistency of the findings across measurement methods.

Previous work indicates that neighborhood disadvantage influences mental health outcomes through heightened cognitive vulnerabilities, such as low social rank, negative schemas, and extrinsic stressors that diminish social cohesion and exacerbate paranoia.43,44 Furthermore, socioenvironmental factors, such as income inequality and neighborhood composition, are closely tied to symptom dimensions in psychosis.45 Our findings build on this work by identifying paranoia as a key dimension linked to localized deprivation. Uneven resource distribution can erode trust in institutions, further amplifying the psychological impact of environmental disadvantage and contributing to heightened paranoia.12,24

These findings highlight the importance of addressing neighborhood-level conditions that contribute to elevated paranoia in SMI. Interventions that support housing stability, expand access to education and employment, and improve local infrastructure may help reduce external stressors linked to paranoia. Beyond broad contextual improvements, community-based programs that enhance engagement with local systems and foster positive social interaction may also be protective. For example, increasing access to green spaces and community centers has been shown to improve social connectedness and reduce stress-related symptoms.28,45 The ADI provides a useful tool for identifying neighborhoods that may benefit most from such support. Moreover, the current findings emphasize the heightened vulnerability of individuals with SSD, underscoring the importance of expanding access to mental health services and outreach strategies tailored to high-deprivation contexts (e.g., telehealth expansion, mobile outreach, and clinician training focused on trust-building in the context of paranoia).

A strength of the present analysis is the use of a transdiagnostic multi-site sample, which enhances the representativeness of the samples and promotes generalizability of the findings. Furthermore, the application of the ADI allows for comparisons with studies conducted in other contexts, including international settings. These strengths provide a robust foundation for exploring the relationship between localized deprivation, paranoia, and demographic factors.

However, the cross-sectional design of this study limits causal inferences, particularly regarding the temporal dynamics of ADI, paranoia, and race. Although demographic characteristics are typically stable over time, key questions remain about how the relationship between neighborhood conditions and paranoia may change with prolonged exposure, relocation, or changes in environmental context. It is also unclear how these relationships may manifest in individuals who are less clinically stable. It is possible that ADI may show an even stronger relationship to paranoia in those with more severe forms of illness, but this requires empirical testing.

Our decision to group participants into broad racial categories (e.g., White vs. non-White) represents a significant limitation of the current study. While this approach was necessary due to sample size constraints, it risks obscuring the heterogeneity of lived experiences within racialized groups and may perpetuate the same reductive assumptions our study seeks to interrogate. For example, the non-White group included a range of individuals whose social positions, histories, and exposures to threat likely differ in both form and magnitude. Treating these participants as a homogeneous group may mask important within-group variation and fails to account for the specific structural and interpersonal experiences of distinct communities, such as Black Americans.14,42 We, therefore, encourage future work that more rigorously examines how intersecting racial, ethnic, and contextual factors shape mental health risk, symptom expression, and clinical assessment.

Relatedly, although PANSS P6 and P3 are widely used and well-validated single-item indicators of paranoia and hallucinations, they may not capture the full complexity of these constructs. Future studies may benefit from including more detailed scales to assess symptom dimensions more precisely. And, while the correlation between P6 ratings at baseline and study end was moderate-to-strong (r = 0.54), it was not perfect. We attempted to mitigate the impact of this by averaging ratings across time and utilizing multiple assessment methods; however, the potential for rater-related measurement error in clinician-rated paranoia should be noted.

The reliance on neighborhood-level measures such as the ADI, without incorporating individual-level socioeconomic factors or subjective life experiences, limits comprehensive understanding of determinants of symptoms and functioning. The study also lacked data on participants’ perceptions of neighborhood disorder, such as feeling unsafe or mistrustful, which may interact with the objective conditions captured by the ADI. These perceptions may create a feedback loop, wherein perceived disorganization or danger increases stress and erodes trust, which in turn heightens vulnerability to paranoia, further sensitizing individuals to environmental threat cues and reinforcing negative appraisals of their surroundings.46 Future research should consider mixed methods designs to investigate how subjective perceptions and objective neighborhood context shape paranoid ideation.47 Finally, as momentary data was averaged across the 30-day period for each participant rather than examination of within-day fluctuations or diurnal patterns in paranoia, future work should examine how symptoms vary across the day and in response to environmental context.

Despite these limitations, this study underscores the role of neighborhood disadvantage in paranoia symptoms, with findings demonstrating that higher levels of area deprivation are associated with increased paranoia across both clinician-rated and EMA measures of paranoid thinking. These results highlight the need for policies that address place-based disadvantage to improve clinical outcomes for individuals with SMI. By identifying neighborhood disadvantage as a modifiable risk factor, this investigation offers guidance for structuring context-sensitive supports aimed at reducing the psychological burden of paranoia through improvements in local environmental conditions.

Supplementary Material

Supplement_ADI_0828_sbaf193

Acknowledgments

We would like to thank the individuals who participated in this study.

Contributor Information

Sarah A Berretta, School of Behavioral and Brain Sciences, The University of Texas at Dallas, Richardson, TX 75080, United States.

Robert A Ackerman, School of Behavioral and Brain Sciences, The University of Texas at Dallas, Richardson, TX 75080, United States.

Raeanne C Moore, Department of Psychiatry, University of California San Diego, La Jolla, CA 92093, United States.

Colin A Depp, Department of Psychiatry, University of California San Diego, La Jolla, CA 92093, United States.

Philip D Harvey, Department of Psychiatry, Miller School of Medicine, University of Miami, Miami, FL 33136, United States; Research Service, Bruce W. Carter Miami VA Medical Center, Miami, FL 33125, United States.

Arundati Nagendra, Schizophrenia & Psychosis Action Alliance (S&PAA), Alexandria, VA, United States.

Amy E Pinkham, School of Behavioral and Brain Sciences, The University of Texas at Dallas, Richardson, TX 75080, United States.

Funding

This work was supported by the National Institute of Mental Health at the National Institutes of Health (grant number R01MH112620 to A.E.P.). The data in this study are deposited in the NIMH RDoC repository (ID: C2941).

Conflicts of Interest

Dr. Moore is a co-founder of KeyWise AI, Inc. and a consultant for NeuroUX. Dr. Harvey has received consulting fees or travel reimbursements from Alkermes, Boehringer Ingelheim, Bristol Myers Squibb, Kynexis, Minerva Neurosciences, and Sumitomo Pharma. He receives royalties from WCG Endpoint Solutions, Inc., for the Brief Assessment of Cognition in Schizophrenia, which is contained in the MATRICS Consensus Cognitive Battery. He is chief scientific officer of i-Function, Inc. Dr. Pinkham is a consultant for Boehringer Ingelheim. Dr. Depp received consulting fees from WCG Clinic and Boehringer Ingelheim. The other authors have no potential biomedical conflicts of interest to declare.

References

  • 1. Freeman  D. Suspicious minds: the psychology of persecutory delusions. Clin Psychol Rev. 2007;27:425-457. 10.1016/j.cpr.2006.10.004 [DOI] [PubMed] [Google Scholar]
  • 2. Sartorius  N, Jablensky  A, Korten  A, et al.  Early manifestations and first-contact incidence of schizophrenia in different cultures: a preliminary report on the initial evaluation phase of the WHO collaborative study on determinants of outcome of severe mental disorders. Psychol Med. 1986;16:909-928. 10.1017/S0033291700011910 [DOI] [PubMed] [Google Scholar]
  • 3. Bentall  RP, Rowse  G, Shryane  N, et al.  The cognitive and affective structure of paranoid delusions: a transdiagnostic investigation of patients with schizophrenia spectrum disorders and depression. Arch Gen Psychiatry. 2009;66:236. 10.1001/archgenpsychiatry.2009.1 [DOI] [PubMed] [Google Scholar]
  • 4. Hajdúk  M, Klein  HS, Harvey  PD, Penn  DL, Pinkham  AE. Paranoia and interpersonal functioning across the continuum from healthy to pathological–network analysis. Br J Clin Psychol. 2019;58:19-34. 10.1111/bjc.12199 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Morgan  OA, Strassnig  MT, Moore  RC, et al.  Accuracy of immediate self-assessment of neurocognitive test performance: associations with psychiatric diagnosis and longitudinal psychotic symptoms. J Psychiatr Res. 2022;152:152-159. 10.1016/j.jpsychires.2022.06.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6. Orth  RD, Hur  J, Jacome  AM, et al.  Understanding the consequences of moment-by-moment fluctuations in mood and social experience for paranoid ideation in psychotic disorders. Schizophr Bull Open. 2022;3:sgac064. 10.1093/schizbullopen/sgac064 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7. Pinkham  A, Bass  E, Klein  H, Springfield  C, Kent  J, Aslan  S. Inducing paranoia is linked to increased resting amygdala cerebral blood flow in healthy individuals. J Exp Psychopathol. 2022;13:20438087221130693. 10.1177/20438087221130693 [DOI] [Google Scholar]
  • 8. Combs  DR, Penn  DL, Cassisi  J, et al.  Perceived racism as a predictor of paranoia among African Americans. J Black Psychol. 2006;32:87-104. 10.1177/0095798405283175 [DOI] [Google Scholar]
  • 9. Combs  DR, Penn  DL, Fenigstein  A. Ethnic differences in subclinical paranoia: an expansion of norms of the paranoia scale. Cultur Divers Ethnic Minor Psychol. 2002;8:248-256. 10.1037/1099-9809.8.3.248 [DOI] [PubMed] [Google Scholar]
  • 10. Morgan  C, Fisher  H, Hutchinson  G, et al.  Ethnicity, social disadvantage and psychotic-like experiences in a healthy population based sample. Acta Psychiatr Scand. 2009;119:226-235. 10.1111/j.1600-0447.2008.01301.x [DOI] [PubMed] [Google Scholar]
  • 11. Wolny  J, Moussa-Tooks  AB, Bailey  AJ, O'Donnell  BF, Hetrick  WP. Race and self-reported paranoia: increased item endorsement on subscales of the SPQ. Schizophr Res. 2023;253:30-39. 10.1016/j.schres.2021.11.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Gordon  S, Mote  J, Fulford  D. Qualitative analysis of paranoia reported in clinical interviews with Black and White adults with schizophrenia. Psychiatr Serv. 2023;74:1263-1269. 10.1176/appi.ps.20220089 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Cassano  P, Chang  T, Trinh  NH, Baer  L, Fava  M, Mischoulon  D. Differential impact of isolated psychotic symptoms on treatment outcome of major depressive disorder in the STAR*D cohort of Whites, Blacks and Latinos. J Affect Disord. 2013;150:578-584. 10.1016/j.jad.2013.02.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. Kingston  JL, Schlier  B, Lincoln  T, et al.  Paranoid thinking as a function of minority group status and intersectionality: an international examination of the role of negative beliefs. Schizophr Bull. 2023;49:1078-1087. 10.1093/schbul/sbad027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Olbert  CM, Nagendra  A, Buck  B. Meta-analysis of Black vs White racial disparity in schizophrenia diagnosis in the United States: do structured assessments attenuate racial disparities?  J Abnorm Psychol. 2018;127:104-115. 10.1037/abn0000309 [DOI] [PubMed] [Google Scholar]
  • 16. Schwartz  RC, Blankenship  DM. Racial disparities in psychotic disorder diagnosis: a review of empirical literature. World J Psychiatry. 2014;4:133-140. 10.5498/wjp.v4.i4.133 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17. Whaley  AL. Ethnicity/race, paranoia and hospitalization for mental health problems among men. Am J Public Health. 2004;94:78-81. 10.2105/AJPH.94.1.78 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Wickham  S, Taylor  P, Shevlin  M, Bentall  RP. The impact of social deprivation on paranoia, hallucinations, mania and depression: the role of discrimination, social support, stress and trust. PLoS One. 2014;9:e105140. 10.1371/journal.pone.0105140 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19. Wilson  C, Smith  ME, Thompson  E, et al.  Context matters: the impact of neighborhood crime and paranoid symptoms on psychosis risk assessment. Schizophr Res. 2016;171:56-61. 10.1016/j.schres.2016.01.007 [DOI] [PubMed] [Google Scholar]
  • 20. Kirkbride  JB, Jones  PB, Ullrich  S, Coid  JW. Social deprivation, inequality, and the neighborhood-level incidence of psychotic syndromes in East London. Schizophr Bull. 2014;40:169-180. 10.1093/schbul/sbs151 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21. Williams  DR, Collins  C. Racial residential segregation: a fundamental cause of racial disparities in health. Public Health Rep. 2001;116:404-416. 10.1016/S0033-3549(04)50068-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22. Anglin  DM. Racism and social determinants of psychosis. Annu Rev Clin Psychol. 2023;19:277-302. 10.1146/annurev-clinpsy-080921-074730 [DOI] [PubMed] [Google Scholar]
  • 23. Anglin  DM, Lui  F, DeLuca  V, Ellman  LM. From womb to neighborhood: a racial analysis of social determinants of psychosis in the United States. Curr Opin Psychol. 2021;37:89-93. 10.1016/j.copsyc.2020.08.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24. Anglin  DM, Lui  F, Thompson  JL. Racial microaggressions and major discriminatory events explain ethnoracial differences in psychotic experiences. Schizophr Res. 2023;254:1-8. https://pubmed.ncbi.nlm.nih.gov/36888999/ [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25. Kind  AJH, Buckingham  WR. Making neighborhood-disadvantage metrics accessible—the Neighborhood Atlas. N Engl J Med. 2018;378:2456-2458. 10.1056/NEJMp1802313 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26. Kind  AJH, Jencks  S, Brock  J, et al.  Neighborhood socioeconomic disadvantage and 30-day rehospitalization: a retrospective cohort study. Ann Intern Med. 2014;161:765-774. 10.7326/M13-2946 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27. University of Wisconsin School of Medicine and Public Health . Area Deprivation Index (Version v2.0) [dataset]. 2018. https://www.neighborhoodatlas.medicine.wisc.edu/
  • 28. Hu  J, Kind  AJH, Nerenz  D. Area deprivation index predicts readmission risk at an urban teaching hospital. Am J Med Qual. 2018;33:493-501. 10.1177/1062860617753063 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29. Vassilaki  M, Aakre  JA, Castillo  A, et al.  Association of neighborhood socioeconomic disadvantage and cognitive impairment. Alzheimers Dement. 2023;19:761-770. 10.1002/alz.12702 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30. Nagendra  A, Liu  CWJ, Mueser  KT, et al.  Do symptom severity, individual socioeconomic status, and neighborhood socioeconomic status explain differences in daily functioning in non-Latinx Black, non-Latinx White, and Latinx people with serious mental illnesses?  Clin Psychol Sci. 2024;12:220-236. 10.1177/21677026231169659 [DOI] [Google Scholar]
  • 31. Harvey  PD, Miller  ML, Moore  RC, Depp  CA, Parrish  EM, Pinkham  AE. Capturing clinical symptoms with ecological momentary assessment: convergence of momentary reports of psychotic and mood symptoms with diagnoses and standard clinical assessments. Innov Clin Neurosci. 2021;18:24-30. [PMC free article] [PubMed] [Google Scholar]
  • 32. Paquin  V, Ackerman  RA, Depp  CA, Moore  RC, Harvey  PD, Pinkham  AE. Media use and its associations with paranoia in schizophrenia and bipolar disorder: ecological momentary assessment. JMIR Ment Health. 2024;11:e59198. 10.2196/59198 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33. Fang  H, Hui  Q, Lynch  J, et al.  Harmonizing genetic ancestry and self-identified race/ethnicity in genome-wide association studies. Am J Hum Genet. 2019;105:763-772. 10.1016/j.ajhg.2019.08.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34. National Institute on Minority Health and Health Disparities . Strategies to Promote the Advancement of Health Disparities Science. Bethesda, MD: U.S. Department of Health and Human Services, National Institutes of Health, 2024. https://www.nimhd.nih.gov/resources/science-visioning-research-strategies. [Google Scholar]
  • 35. Kay  SR, Fiszbein  A, Opler  LA. The Positive and Negative Syndrome Scale (PANSS) for schizophrenia. Schizophr Bull. 1987;13:261-276. 10.1093/schbul/13.2.261 [DOI] [PubMed] [Google Scholar]
  • 36. R Core Team . R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, 2016. https://cran.r-project.org/doc/manuals/r-release/fullrefman.pdf. [Google Scholar]
  • 37. Rosseel  Y. lavaan: an R package for structural equation modeling. J Stat Softw. 2012;48. 10.18637/jss.v048.i02 [DOI] [Google Scholar]
  • 38. Blair  G, Cooper  J, Coppock  A, Humphreys  M, Sonnet  L. estimatr: fast estimators for design-based inference. R package version 1.0.4. 2025. https://declaredesign.org/r/estimatr/
  • 39. Pustejovsky  JE, Tipton  E. Small-sample methods for cluster-robust variance estimation and hypothesis testing in fixed effects models. J Bus Econ Stat. 2018;36:672-683. 10.1080/07350015.2016.1247004 [DOI] [Google Scholar]
  • 40. Dohrenwend  BP, Levav  I, Shrout  PE, et al.  Socioeconomic status and psychiatric disorders: the causation-selection issue. Science. 1992;255:946-952. 10.1126/science.1546291 [DOI] [PubMed] [Google Scholar]
  • 41. Liyanage  RD, Barlow  F, Saini  P, Nathan  T, McIntyre  J. Racial discrimination and paranoia: exploring the potential moderating role of ethnic identity. BJPsych Open. 2025;11:e10118. 10.1192/bjo.2025.10118 [DOI] [Google Scholar]
  • 42. Freeman  D, Loe  BS. Explaining paranoia: cognitive and social processes in the occurrence of extreme mistrust. BMJ Ment Health. 2023;26:e300880. 10.1136/bmjment-2023-300880 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43. McElroy  E, McIntyre  JC, Bentall  RP, et al.  Mental health, deprivation, and the neighborhood social environment: a network analysis. Clin Psychol Sci. 2019;7:719-734. 10.1177/2167702619830640 [DOI] [Google Scholar]
  • 44. Fone  D, Dunstan  F, Lloyd  K, Williams  G, Watkins  J, Palmer  S. Does social cohesion modify the association between area income deprivation and mental health? A multilevel analysis. Int J Epidemiol. 2007;36:338-345. 10.1093/ije/dym004 [DOI] [PubMed] [Google Scholar]
  • 45. Tibber  MS, Kirkbride  JB, Mutsatsa  S, et al.  Are socioenvironmental factors associated with psychotic symptoms in people with first-episode psychosis? A cross-sectional study of a West London clinical sample. BMJ Open. 2019;9:e030448. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46. Newbury  JB, Arseneault  L, Caspi  A, et al.  In the eye of the beholder: perceptions of neighborhood adversity and psychotic experiences in adolescence. Dev Psychopathol. 2017;29:1823-1837. 10.1017/S0954579417001420 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47. Cooper  C, Morgan  C, Byrne  M, et al.  Perceptions of disadvantage, ethnicity and psychosis. Br J Psychiatry. 2008;192:185-190. 10.1192/bjp.bp.107.042291 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplement_ADI_0828_sbaf193

Articles from Schizophrenia Bulletin are provided here courtesy of Oxford University Press

RESOURCES