# Ambassador Evaluation Dashboard Project Guide

## Project Purpose

This dashboard evaluates **Ambassadors to Promote Sexual Health in Diverse Communities in Baltimore**, a Baltimore City Health Department funded project implemented by **RnD Associates**.

Tagline: **Breaking Barriers, Building Healthier Communities TOGETHER**

The dashboard is meant to be public, funder-ready, and useful for program improvement. It should show the story of the Ambassadorial Model, document reach and performance, summarize evaluation responses, surface qualitative participant voice, and make practical recommendations for improving programming.

## Program Model

The Ambassadorial Model is a community-driven public health approach that brings sexual health education, HIV/STI prevention, PrEP/PEP awareness, and trusted service connections into spaces where people already live, work, learn, worship, heal, and gather.

Core idea: **trust is public health infrastructure.**

Four pillars:

1. **Trusted Ambassadors**: peer-led and community-rooted organizations reduce distance between residents and health systems.
2. **Relevant and Culturally Grounded Education**: education is integrated into youth programs, mentoring, women’s circles, faith conversations, shops, wellness retreats, storytelling, and community events.
3. **Stigma Reduction Through Conversation and Storytelling**: peer dialogue, podcasts, workshops, and creative community engagement create safer places to ask questions.
4. **Service Connection and Navigation**: education should connect people to HIV/STI testing, PrEP/PEP services, counseling, wellness resources, and trusted referral systems.

Priority zip codes: `21215`, `21213`, `21218`, `21224`, `21217`.

RAP goal:

- Raise awareness about HIV & STI, PrEP, PEP, and sexual health.
- Address barriers to sexual health information and practice.
- Provide referral to prevention and treatment services.

## Source Files

The dashboard is generated from files in the parent workspace:

- `Project Overview.docx`: project story, funder/recipient, model, pillars, ambassadors, scope, and populations of reach.
- `Performance Measures Sheet (2).xlsx`: performance scorecard and event-detail attendance log.
- `Training & Events Evaluation Responses 1 - July to December 2025.xlsx`: evaluation responses from July through December 2025.
- `Training & Events Evaluation Responses 2 - Activities from Jan to May 2026.xlsx`: evaluation responses from January through May 2026.
- `Evaluation - PSH in Black Churches.xlsx`: expected newest evaluation source for the PSH in Black Churches program segment.

Generated public outputs:

- `data/evaluation-data.json`: canonical dashboard data.
- `data/evaluation-data.js`: same data wrapped for direct file-open previews.
- `assets/ambassador-evaluation-report-updated.pdf`: generated static PDF report.
- `index.html`: interactive dashboard.
- `report.html`: browser-printable report view.

## Data Rules

- Treat the summed event-detail attendance in the current performance workbook as the authoritative attendance total.
- Append new evaluation workbooks as new program segments instead of replacing prior survey waves unless the user explicitly requests replacement.
- Preserve the `sourceLabel` / `programSegment` fields in public rows, quotes, segment summaries, and workbook exports.
- If an expected required source is missing, keep `sourceStatus.allExpectedSourcesLoaded` false and do not describe the refresh as complete.
- Preserve public detail rows as requested, but redact direct email, phone, and street-address patterns if they appear in public quote or survey text.
- Keep participant quotes as close to verbatim as possible.
- Exclude empty/non-informative quote values like `NA`, `N/A`, `None`, and `No` from the quote wall.
- A quote can be coded to more than one theme.
- Current report timing is interim/current-to-date through available May 2026 data.

## Qualitative Analysis

The quote wall and theme charts are generated from the survey field:

`10. Share a few things that you learned today`

Current theme families:

- PrEP/PEP awareness
- HIV/STI knowledge
- Healthy relationships
- Prevention behaviors
- Testing motivation
- Data and disparities
- Ambassador model
- General learning

Theme coding is keyword-assisted and should be reviewed before final/public submission if the report will be used as a formal evaluation product.

## Recommendations Logic

Recommendations should remain actionable and improvement-oriented. Each recommendation should include:

- Evidence from performance, event, survey, or quote data.
- A practical next step.
- A suggested metric to track.

Current recommendation areas:

- Increase onsite testing conversion.
- Turn PrEP/PEP curiosity into navigation follow-through.
- Close remaining topic-level knowledge gaps.
- Strengthen youth-program reach and segmentation.
- Institutionalize ambassador learning loops.
- Tighten source-data consistency before final reporting.

## Updating the Dashboard

After source workbooks or the project overview change, regenerate data from the dashboard folder:

```powershell
& 'C:\Users\Dr Sam\.cache\codex-runtimes\codex-primary-runtime\dependencies\python\python.exe' scripts\build_data.py
```

Then preview by opening `index.html` directly, or serve locally:

```powershell
& 'C:\Users\Dr Sam\.cache\codex-runtimes\codex-primary-runtime\dependencies\python\python.exe' -m http.server 4173
```

Open:

`http://localhost:4173`

## GitHub and Vercel Notes

This is a static dashboard. It can be deployed from the `dashboard` directory to Vercel or served by GitHub Pages.

Vercel:

- Use `dashboard` as the project root.
- Build command can be left as `npm run build`.
- Output directory can be left empty/default because files are static at project root.

GitHub:

- Commit the `dashboard` directory and source files.
- If using GitHub Pages, publish from the `dashboard` folder or copy the dashboard contents to the Pages root.

## Checks Future Agents Should Preserve

- Attendance headline is recalculated from the current performance workbook event-detail rows.
- Survey response count should be **115** with the current source files.
- Quote count may be lower than 115 because blank/NA quote responses are excluded.
- PDF and dashboard should use the same `evaluation-data.json`.
- Recommendations should never become generic; keep evidence and metrics attached.
- Public raw/detail rows should be reviewed for direct identifiers before deployment.