This cross-sectional, observational study was conducted among school-going children in the state of Sikkim, India, with the aim of identifying the interplay between coping styles, behavioral regulation, prosocial traits, and anxiety. The investigation combined quantitative psychometric assessment with multivariate statistical modelling. Data were collected from March 2018 to November 2019 across all four administrative districts: East, West, North, and South to ensure representation of both urbanized and rural schooling environments.

A list of all private and public schools registered with the State Education Department was obtained, and 30 schools were randomly selected across districts (East = 12, West = 7, North = 5, South = 6). Of these, 11 schools did not participate due to administrative constraints, including scheduling difficulties and withdrawal citing time limitations, resulting in participation from 19 schools (East = 7, West = 6, North = 3, South = 3). Written informed consent was obtained for 65% of eligible students (N = 3250). Of these, approximately 40% were not available for assessment due to school-level attrition or absence on the day of data collection, yielding 1812 children who completed child-administered assessments. Complete data on all study variables were available for 1001 children. The primary source of missing data was non-return of parent-completed questionnaires, specifically the socio-demographic form and the Strengths and Difficulties Questionnaire–Parent version. As these measures were required for behavioral and socioeconomic analyses, cases with incomplete parent-reported data were excluded from the final analytic sample using listwise deletion. Comparisons of available demographic characteristics (age, gender, and district) between included and excluded participants did not indicate marked differences, although differential nonresponse related to socioeconomic factors cannot be ruled out. This potential for systematic bias is acknowledged and considered when interpreting the findings. The final sample consisted of 1001 children aged 7–10 years enrolled in classes II–V in English-medium schools recognized by the Department of School Education, Government of Sikkim (Selection process detailed in Figure 1). A multistage stratified random sampling approach was used, with schools selected through proportional allocation based on district-level enrollment. Eligibility criteria included enrolment as a full-time student within the specified age range, ability to understand instructions in English or Nepali, and provision of parental consent and child assent. Children with diagnosed neurodevelopmental disorders, sensory impairments, or chronic medical conditions likely to affect behavioral assessment were excluded.

The primary anxiety outcome was measured using the Revised Children’s Manifest Anxiety Scale–Second Edition (RCMAS-2; 11). The scale consists of 49 items scored dichotomously (“Yes/No”), producing subscale raw scores for Physiological Anxiety, Worry/Oversensitivity, Social Concerns/Concentration, and two validity indicators (Defensiveness, Inconsistent Responding), along with a Total Anxiety Raw Score. In the present study, analyses were performed using raw scores, and severity classification was based on U.S. normative T-score conversions provided in the official RCMAS-2 manual, as no Indian population norms are currently available. Cut-offs for interpretive categories (e.g., typical, mild, moderate, severe) were therefore derived using the manual’s age-based T-score ranges.

Behavioral characteristics were assessed using the Strengths and Difficulties Questionnaire–Parent Version (SDQ-P; 34, 35), generating scores for internalizing, externalizing, and prosocial behavior. Coping was measured using the Coping Strategies Inventory–Short Form (CSI-SF; 36), which yields adaptive and maladaptive coping indices based on eight domain scores. Socioeconomic status was assessed using the Kuppuswamy Scale (2020), and demographic variables were retained as covariates (28). For analysis, we computed Spearman correlations, hierarchical multiple regression models, mediation with bootstrap confidence intervals, and structural equation modelling (SEM). Prior to SEM, measurement models for latent constructs derived from SDQ-P and CSI-SF were evaluated to ensure acceptable factor structure and reliability.

Data were analyzed using Python, RStudio, MS Excel, and GraphPad Prism 7. Descriptive statistics summarized demographic variables, SES distribution, and mean scores of all scales. Categorical associations were examined using Chi-square tests. Spearman correlation coefficients assessed relationships among continuous variables. To identify predictors of anxiety, multiple linear regression models were estimated using the RCMAS-2 total raw score as the dependent variable. Independent variables included internalizing and externalizing behavior, adaptive and maladaptive coping, prosocial behavior, age, gender, class level, and SES. Before modelling, multicollinearity was evaluated using Variance Inflation Factors (VIFs), and all predictors had VIFs below 5, indicating acceptable levels of collinearity. Although the Kuppuswamy Scale is ordinal, its three components (education, occupation, income) generate a composite score with quasi-continuous properties; therefore, SES was treated as a continuous predictor to preserve variability and statistical power, consistent with prior Indian research using multidimensional SES indices in child and adolescent mental health studies. For higher-order interpretation of pathways, Structural Equation Modelling (SEM) was performed (37). For SEM analysis, confirmatory factor analysis (CFA) was conducted to evaluate the measurement structure of the latent constructs included in the model. CFA was performed at the subscale level. Adaptive and maladaptive coping were modelled as separate latent constructs using Coping Strategies Inventory–Short Form (CSI-SF) subscale scores. Adaptive coping was indicated by problem solving, cognitive restructuring, emotional expression, and social support, while maladaptive coping was indicated by problem avoidance, wishful thinking, self-criticism, and social withdrawal. Behavioral tendencies were modelled using Strengths and Difficulties Questionnaire–Parent version (SDQ-P) subscales, with internalizing behavior indicated by emotional and peer problems and externalizing behavior indicated by conduct problems and hyperactivity. Anxiety was modelled as a latent construct using RCMAS-2 domain scores. Latent constructs for Behavior, Coping and Anxiety were developed based on theoretical considerations and correlation patterns. The model was estimated using maximum likelihood with robust standard errors. Model fit was evaluated using standard indices, including the Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA) and Standardized Root Mean Square Residual (SRMR). Indirect effects were tested using 2,000-iteration bias-corrected bootstrap procedures, and mediation effects were interpreted using standardized coefficients with 95% confidence intervals.

The study involved human participants and was reviewed and approved by the Institutional Ethics Committee and the Institutional Review Committee of Sikkim Manipal Institute of Medical Sciences, Sikkim Manipal University. Written permission was obtained from the participating schools prior to data collection. Written informed consent was obtained from the parents or legal guardians of all participating children, and verbal assent was obtained from the children themselves prior to assessment. All study procedures were conducted in accordance with applicable institutional guidelines and ethical standards for research involving human participants.

The study included 1,001 school-going children aged 7–10 years (M = 8.6, SD = 1.1) from Sikkim’s four districts, with a near-even gender distribution (52.8% male, 47.2% female). Socioeconomic status (SES) per the 2020 Kuppuswamy Scale showed 17.5% upper class, 33.4% upper-middle, 26.7% lower-middle, 18.2% upper-lower, and 4.9% lower class, with significant district-wise variation (χ² = 23.64, df = 12, p = 0.022). East Sikkim had the highest upper-class representation (19.5%), while North Sikkim had the most lower-middle families (38.5%). Parental education and occupation correlated positively with income (r = 0.48–0.61, p < 0.01). (Tables 1, 2).

69.2% of children were classified as having mild anxiety, 21.0% moderate, and 1.5% severe as RCMAS-2 scale. Anxiety scores were significantly higher in East and South districts compared to North and West (p < 0.05). Anxiety increased with age (r = 0.10, p < 0.05), but gender differences were non-significant.

The Strengths and Difficulties Questionnaire (SDQ-P) indicated higher internalizing (emotional and peer problems) than externalizing (conduct and hyperactivity) scores, with 39.6% of children in the borderline/abnormal range for total difficulties. Prosocial behavior was inversely associated with total difficulties (r = -0.41, p < 0.001) and anxiety (r = -0.04, p < 0.05). District differences in SDQ-P scores were significant (p < 0.05), but gender and SES showed minimal impact. Internal consistency was acceptable (Cronbach’s α = 0.82) (Table 3).

The Coping Strategies Inventory (CSI-SF) showed strong internal consistency (α = 0.45–0.88). Maladaptive coping strategies (problem avoidance, wishful thinking, self-criticism, social withdrawal) were positively correlated with each other (r = 0.52–0.87) and with internalization (r = 0.48, p < 0.001). Adaptive coping strategies (problem-solving, cognitive restructuring, emotional expression, social support) were also intercorrelated (r = 0.46–0.75) but showed weaker associations with internalization (r = 0.33, p < 0.01). Maladaptive coping was the strongest correlate of RCMAS-2 scores (r = 0.15, p < 0.01), followed by internalization (r = 0.08, p < 0.05) and externalization (r = 0.07, p < 0.05) (Tables 4–6). Adaptive coping and prosocial behavior had negligible or inverse correlations with anxiety (r = -0.05 and -0.04, respectively). (Figures 2, 3).

The structural equation model assessing predictors of RCMAS-2 total anxiety score showed that maladaptive coping was the only significantly associated with anxiety (β = 0.18, p < 0.001). Adaptive coping demonstrated a small inverse trend (β = –0.08, p = 0.057), followed by prosocial behavior (β = -0.11, p = 0.035). Internalization (β = 0.05, p = 0.104) and externalization (β = 0.05, p = 0.206) did not exhibit significant direct effects on anxiety. Internalization predicted higher maladaptive coping (β = 0.04, p = 0.149), and externalization showed no meaningful influence on either coping domain (Engagement: β = 0.01, p = 0.674; Disengagement: β = 0.02, p = 0.489). As a result, the indirect pathways linking internalization and externalization to anxiety through maladaptive coping were small and not statistically significant (INT → Dis → RCMAS: indirect β = 0.0072, p ≈ 0.094; EXT → Dis → RCMAS: indirect β = 0.0034, p ≈ 0.105). Overall, the model indicates that maladaptive coping is the primary mechanism associated with elevated RCMAS-2 anxiety, while internalizing and externalizing behaviors contribute weak and largely indirect influences. The PCA-derived RCMAS factor explained 42.2% of variance (Figure 4, Supplementary Table 1).

Confirmatory factor analysis demonstrated acceptable measurement properties for all latent constructs included in the structural model. Standardized factor loadings for adaptive coping ranged from 0.61 to 0.67, and for maladaptive coping from 0.54 to 0.73, indicating moderate to strong representation of the respective constructs by their indicators. Anxiety domain scores showed strong loadings on the latent anxiety factor, ranging from 0.69 to 0.81. Internalizing and externalizing behavioral constructs demonstrated modest statistically significant, with standardized estimates ranging from 0.43 to 0.49 for internalizing behavior and 0.43 to 0.76 for externalizing behavior. These findings indicate that behavioral tendencies were measured with adequate but lower precision relative to coping and anxiety constructs. Overall model fit indices indicated acceptable fit of the measurement model, supporting the use of the specified latent constructs in subsequent structural equation modelling (Supplementary Table 2, 3). The relative strength of factor loadings across constructs was consistent with the magnitude of path coefficients observed in the structural model.