Back to Dashboard
Researcher Flow dashboard/src/views
This is the researcher/admin path for preparing a study, monitoring live sessions, protecting data quality, and exporting evidence for analysis.
Study setup flow
1. Researcher signs in
The researcher uses
/dashboard/. Login issues an 8-hour researcher JWT. Admin-only views are hidden from non-admin researchers.2. Confirm study configuration
Review thresholds, cooldowns, readiness weights, active condition, phase, and config locks. Threshold changes are allowed only in calibration phase and require audit reasons.
3. Prepare task templates and content
Admins maintain templates, supported languages, locators, intervention content, and help keyword clusters. Every content change writes version history and audit entries.
4. Prepare team access
Admins create researcher accounts, activate/deactivate users, and control roles. Researcher actions that affect study data are audit logged.
5. Run pilot/calibration sessions
Watch baseline quality, signal completeness, phone/PWA behavior, practice task routes (e.g. WhatsApp family-message flow, SHA registration, KRA iTax login), proxy/extension capture behavior, and intervention timing before locking the study into main phase.
Live monitoring flow
| Dashboard area | Researcher question | Action if something looks wrong |
|---|---|---|
| Overview | Is enrollment healthy, are conditions balanced, and is attrition acceptable? | Pause recruitment or inspect assignment/config if condition balance drifts. |
| Sessions | Who is live, what state are they in, and did calibration clear? | Inspect a session, terminate only for safety/operational reasons, or exclude from analysis with reason. |
| Session signals | Are events complete and confidence high enough to trust readiness? | Check device type, capture mode, baseline quality, debug failures, tested route behavior, and raw event details. |
| Live Pipeline | Is event processing moving through normalise/readiness/governance/adaptation? | Investigate Redis, API errors, extension handoff, or slow pipeline responses. |
| Packaging choice | Does the selected path match the participant population? | Use phone/PWA as the default for older mobile users. The Chrome extension path is available for services that cannot be iframed. There is no Tauri desktop app in the current deployment. |
| Alerts | Are there study-integrity, data-quality, ethical, or technical issues? | Resolve only after review; keep the alert trail as operational evidence. |
| Analytics | Are trends emerging by condition, task, readiness state, or step difficulty? | Use for interpretation; avoid changing thresholds after main phase lock. |
Content and intervention review flow
1. Review governance hit rates
Identify which rules fire often and whether cooldowns are producing recovery rather than repeated interruption.
2. Inspect adaptation outcomes
Look for ignored, worsened, conflict, or repeated-ignored failures by step, language, level, and condition.
3. Edit content only when appropriate
During calibration, rewrite unclear guidance or add missing language/locator coverage. During main phase, avoid changes unless the protocol permits them.
4. Audit every change
Content, config, condition, exclusion, withdrawal, export, and destructive actions require traceable audit entries.
Data closeout flow
1. Resolve operational alerts
Check unresolved alerts before export so analysis notes can separate technical issues from participant behavior.
2. Review exclusions and withdrawals
Confirm every exclusion has a reason. Withdrawn participant records remain historically available unless the study protocol requires deletion.
3. Export analysis views
Use events, readiness, governance, adaptations, sessions, participants, and audit views. Exports are logged.
4. Preserve reproducibility
Keep config snapshots, audit logs, docs version, source commit, and export timestamps together with the analysis dataset.
Researcher role boundaries
- Researcher: monitor sessions, inspect signals, review analytics/alerts, export data, withdraw/exclude with reason where permitted.
- Admin: manage content, templates, language support, researcher accounts, config, study condition, and destructive operations.
- Everyone: preserve auditability; never edit operational state casually during main data collection.