The Community Trust Index assesses the level of trust in the Nepal Red Cross-supported flood early warning system (EWS), focusing on perceptions of its actions, competencies, and the role of volunteers and staff in strengthening community preparedness. The analysis draws on survey rounds conducted in March 2025 in Sudur Paschim, Lumbini Province, and Kathmandu Valley, and in March 2026 in Koshi Province. These surveys were implemented to measure trust in flood EWS initiatives that support communities and local authorities in improving anticipatory action and mitigating disaster risks.

Summary

  • Moderate overall performance across all Early Warning System (EWS) pillars, with an average score of 5.9.

  • The highest score was observed for Pillar 4 (Response and Capacities), while Pillar 2 (Detection, Monitoring and Forecasting) received the lowest score.

  • Effectiveness within Pillar 3 (Warnings, Dissemination and Communication) recorded the highest sub-dimension score (6.61), while Feedback showed comparatively stronger performance in Pillars 3 and 4.

  • Across all pillars, Transparency and Participation emerged as the main areas requiring improvement.

  • Lower scores were associated with lower education levels, limited receipt of early warning information, lower levels of engagement with the Red Cross, and older age groups, with respondents aged 60 years and above reporting the lowest scores.

Sampling

The survey was conducted by the Nepal Red Cross Society (NRCS) as part of the Community Trust Index – Early Warning System module. Data collection was carried out by NRCS volunteers using a standardized questionnaire, with a focus on issues of trust within the framework of the Building Trust initiative.

NRCS implemented three survey waves. The first two waves were conducted between April and May 2025 across selected districts in the western provinces and the Kathmandu Valley. A third wave was later carried out in March 2026 in the Eastern Province (Koshi).

In total, the survey reached 5,309 respondents, of whom 4,958 met the eligibility criteria.

See metrics: Metrics

Geographic

The sample distribution shows some differences compared to the population structure across the surveyed districts. Kathmandu is notably under-represented in the sample (21.2% vs. 29.9%), while Morang (19.2% vs. 10.9%) and Sunsari (14.7% vs. 8.6%) are over-represented relative to official population figures. Other districts show smaller deviations: Kailali is slightly under-represented (12.2% vs. 13.2%), while Banke, Bardiya, Bhaktapur, Kanchanpur, and Lalitpur are all marginally under-represented compared to their population shares.

These differences will be taken into account during the analysis, and weighting or other adjustments may be applied to ensure that the results remain representative.

Coverage

Gender and Age

The sample broadly reflects the population distribution by age and gender, with only minor differences. Among women, age groups closely match the population, although women aged 60+ are slightly under-represented. Among men, those aged 40–59 are somewhat over-represented, while men aged 18–39 are slightly under-represented. Overall, these differences are small and will be considered in the analysis, with adjustments applied if needed to improve representativeness.

Education

The sample broadly reflects the population distribution by education level, with some differences across categories. Secondary school education remains the largest group, though it is slightly lower in the sample (39.5% vs. 42.2%). Respondents with primary education (27.5% vs. 22.5%) and no formal education (16.0% vs. 9.7%) are somewhat over-represented, while university-educated respondents are under-represented (17.1% vs. 25.6%). These differences will be considered in the analysis, with adjustments applied if needed to improve representativeness.

Employment

The sample differs more noticeably from the population distribution in economic activity status. Employed respondents are substantially over-represented in the sample (67.4% vs. 42.9%), while those who are not economically active (17.8% vs. 34.3%) and not usually active (7.5% vs. 19.4%) are under-represented. Respondents who are active but unemployed are slightly over-represented (7.2% vs. 3.2%). These differences will be considered in the analysis, with adjustments applied where necessary to improve representativeness.

Source: https://censusnepal.cbs.gov.np/results/files/result-folder/Labour%20Force%20and%20Economic%20Activities%20in%20Nepal.pdf

Limitations

The survey findings should be interpreted with caution due to some differences between the sample and population distribution:

  • Age and gender: Men aged 40–59 are over-represented, while younger men and women aged 60+ are slightly under-represented.

  • Education: No formal and primary education groups are over-represented, while university-educated respondents are under-represented.

  • Employment: Employed respondents are over-represented, while economically inactive groups are under-represented.

  • Geography: Some regional and district-level differences remain.

Post-stratification adjustments were applied to improve representativeness; however, some sampling bias may remain, and findings should be considered broadly indicative rather than fully representative. ————————————————————————

Survey Results

Drivers of trust

Pillar 1

Pillar 2

Pillar 3

Pillar 4

Perception of Risk

Contextual questions

This section presents findings on community members’ experiences with and behaviors toward the Red Cross. These questions explore interactions, perceptions, and engagement patterns, offering insights into how the Red Cross is viewed and utilized within the community.

Experiences

The survey shows that 39.7% of respondents reported having received aid or support, while 60.3% had not. Regarding early warning information, 39.8% reported receiving warnings from the Nepal Red Cross, compared with 56.4% who reported receiving other warnings. This suggests that respondents were more likely to receive warning information from sources other than the Nepal Red Cross.

Trust and EWS

Others Questions

Score

This score is derived from responses to questions that assess perceptions of competencies and values, providing a comprehensive measure of trust. A higher score indicates stronger trust, suggesting that community members believe their needs are being addressed and their values are respected. Learn more about scoring method: Methods

Overall Score

The results show moderate to relatively strong performance across the four dimensions of the early warning system. 

  • Preparedness & Response Capabilities (Pillar 4) received the highest overall score (5.92), followed by Warning Dissemination & Communication (Pillar 3) (5.86).

  • Disaster Risk Knowledge (Pillar 1) scored 5.48, while Detection, Monitoring and Forecasting (Pillar 2) had the lowest overall score (5.35).

  • Feedback and effectiveness generally received stronger ratings across dimensions.

  • Participation and transparency tended to score comparatively lower, indicating areas for improvement.

Learn more about weighting process: Weighting

Scores by Pillar

Combined Pillar Scores

This chart presents the scores of key subdimensions across the four pillars:

  • Feedback received the highest combined score (6.09), followed by Effectiveness (5.90) and Inclusiveness (5.71).

  • Participation (5.35) and Transparency (5.26) received the lowest scores, indicating potential areas for improvement.

Dimension definition

Score by factors

This chart compares scores across different demographic and contextual groups.

  • Geographic variation: Considerable differences are observed across locations. Sunsari scored higher than Morang for Disaster Risk Knowledge (5.86 vs. 5.16) and Detection/Forecasting (5.90 vs. 4.79), while Banke reported the highest scores for Warning Dissemination (7.29) and Preparedness & Response (7.30).

  • Language differences: Respondents speaking Awadi (7.09; 7.15) and Newari (6.25 for detection/forecasting) generally reported stronger perceptions, while Bhojpuri (4.46–4.85) and Maithili (4.76–5.07) speakers tended to report lower scores across pillars.

  • Age and gender: Younger respondents (18–39) consistently reported slightly higher scores than those aged 60+ (e.g., 5.64 vs. 5.12 for risk knowledge; 5.87 vs. 5.12 for preparedness). Differences between men and women were small across all dimensions.

  • Education: Scores generally increased with educational attainment. University-educated respondents reported stronger results across pillars (e.g., 6.17, 5.95, 6.14, 6.13) than respondents with no formal education (5.05, 4.63, 4.83, 5.05).

  • Experience and exposure: Respondents who received early warnings from Nepal Red Cross consistently reported much stronger scores across all pillars (6.29–6.85) compared with those who had not (4.80–5.05). Similarly, previous beneficiaries also reported higher scores (6.19–6.29) than non-beneficiaries (4.90–5.29).

  • Disability and employment: Differences by disability status were relatively small, while economically inactive groups generally reported lower scores than employed or active groups.

Distribution of mean scores for values and competencies per demographic questions

Pillars 1 & 2

Pillars 3 & 4

Score by respondent profile

This chart compares scores across key dimensions by respondents’ level of engagement with Nepal Red Cross activities, including receipt of early warnings and previous beneficiary status.

  • Respondents who received early warnings from Nepal Red Cross consistently reported the strongest scores across all four pillars, with overall scores ranging from 6.44–6.91, compared with 4.80–5.06 among other respondents.

  • People who received aid or services also reported stronger perceptions, with overall scores ranging from 5.98–6.41, higher than respondents without prior engagement.

  • The largest differences are seen in awareness, participation, and inclusiveness. For example, Warning Dissemination awareness was 6.91 among respondents receiving early warnings compared with 4.45 among others, while Preparedness inclusiveness reached 7.09 compared with 5.20 among others.

  • Across all subdimensions, respondents with direct exposure through early warnings or previous support consistently reported more positive experiences and stronger perceptions of system performance.

Pillars 1 & 2

Pillars 3 & 4

Methods and Metrics

Methods

Scoring methodology

To determine the score, we employ the following method:

  1. Survey Structure The CTI – Early Warning and Anticipation module organizes all questions according to the four pillars of the Early Warning for All (EW4All) initiative:
  • Disaster Risk Knowledge
  • Detection, Monitoring and Forecasting
  • Warning, Dissemination and Communication
  • Preparedness and Response Capacity

Each pillar includes several sub-dimensions designed to assess different aspects of early warning system implementation.

  1. Sub-Dimension Scoring

    Each sub-dimension comprises several survey items (questions).Respondents answer on a Likert-type scale (1 to 4 - Don’t not is excluded). For each sub-dimension:

    Sub-dimension Score = ∑ (Weighted Response Scores) / Number of Items

If weights are not empirically derived, equal weighting is typically applied to each item.

  1. Pillar Scoring

Subsequent to the calculation of all sub-dimension scores, the score for each pillar is derived as the arithmetic mean of their respective sub-dimension scores.

For pillar (1 to i)

  • Pillar i Score = ∑(Sub-dimension Scores for Pillar i) /𝑛

where 𝑛 is the number of sub-dimensions in each category.

  1. Overall Scoring

    The final Community Trust Index score is the arithmetic mean of the Pillar scores available:

  • CTI Score = ∑(Pillar i Score)/N

where N is the number of pillars.

Metrics

Gender

Respondents by Gender
Gender Total Respondents Percentage (%)
Female 2434 49.1
Male 2519 50.8
Other or did not answer 5 0.1
Total 4958 100.0

Age

Respondents by Age Group
Age Group Total Respondents Percentage (%)
18-39 2537 51.2
40-59 1765 35.6
60+ 656 13.2
Total 4958 100.0

Geographic

Respondents by District and Province
Province District Total Respondents Percentage (%)
Bagmati Kathmandu 1051 65.9
Bagmati Lalitpur 307 19.2
Bagmati Bhaktapur 238 14.9
Bagmati TOTAL 1596 100.0
Koshi Province Morang 952 56.6
Koshi Province Sunsari 731 43.4
Koshi Province TOTAL 1683 100.0
Lumbini Banke 412 56.4
Lumbini Bardiya 319 43.6
Lumbini TOTAL 731 100.0
Sudur Paschim Kailali 607 64.0
Sudur Paschim Kanchanpur 341 36.0
Sudur Paschim TOTAL 948 100.0

Relationship with RC

Respondents by relationship with RC
Profile Total Respondents
Received early warnings from NS: Yes 1971
Once beneficiary: Yes 1969

Weighting

Weighting vs. unweighting

To correct demographic deviations from the overall population, we applied a technique called raking. This method adjusts results based on variables such as age, gender, province, education level, and geographic (urban/rural) to align the sample with the population distribution. Data sources included the 2021 Nepal Population and Housing Census (Central Bureau of Statistics, https://censusnepal.cbs.gov.np/Home/Index/EN).

The weighted results are broadly consistent with the unweighted findings, with only small variations across dimensions. Across the four pillars, the overall patterns remain largely unchanged, with stronger performance generally observed in effectiveness, feedback, and inclusiveness, while participatio tends to score relatively lower. Differences between weighted and unweighted results are modest, indicating that weighting slightly adjusts scores but does not substantially alter the overall performance patterns or key findings.

Drivers Correlation

Correlation matrix

This chart shows the relationships between sub-dimensions across pillars, with the following key findings:

  • Pillar 1: Moderate relationship between Awareness and Effectiveness

  • Pillar 2: Moderate relationship between Responsiveness and Effectiveness

  • Pillar 3: Moderate relationship between Participation and Transparency

  • Pillar 4: Moderate relationship between Participation and Effectiveness, and Effectiveness and Inclusiveness

How to read the correlation matrix:

1. Find the row and column

  • Example: Effectiveness × Inclusiveness = 0.65

  • That means these two dimensions have a moderately strong positive relationship.

2. Read the number inside the circle

  • Values range from –1 to +1

  • Interpretation:

Value Meaning
0.00 No relationship
0.10–0.30 Weak
0.30–0.50 Mild
0.50–0.70 Moderate
0.70–1.00 Strong
Negative values Variables move in opposite directions

Significance testing

Significance testing

When testing for significant differences between groups, a t-test was used to compare mean scores across dissemination and response questions. The table summarizes whether differences between People received support, people receved warnings, and others are statistically significant.

the results suggest strong alignment across most dimensions, particularly for Dissemination and Response, while greater variation is observed in some drivers within Disaster Risk Knowledge and Detection, Monitoring & Forecasting, especially for participation, inclusiveness, and responsiveness.

Pillar Drivers Received early warnings from NS: Yes-Once beneficiary: Yes
Disaster Awareness Yes
Disaster Effectiveness Yes
Disaster Participation No
Disaster Feedback Yes
Disaster Inclusiness No
Disaster Transparency Yes
Detection Awareness Yes
Detection Responsiveness No
Detection Effectiveness Yes
Detection Inclusiness No
Detection Participation No
Detection Feedback Yes
Detection Transparency Yes
Dissemination Awareness Yes
Dissemination Responsiveness Yes
Dissemination Effectiveness Yes
Dissemination Inclusiness Yes
Dissemination Participation Yes
Dissemination Feedback Yes
Dissemination Transparency Yes
Response Awareness Yes
Response Responsiveness Yes
Response Participation Yes
Response Effectiveness Yes
Response Feedback Yes
Response Inclusiness Yes
Response Transparency Yes