Research Platform Extension

Generated: 12 May 2026

The Atlas app was extended to support multi-study research workflows that randomise participants to packeted case collections while reusing existing collection-taking, timing, marking, and attempt models.

Randomisation on site Pseudo-ID workflow Per-packet marking config CSV/JSON export

Implemented Data Model

ResearchStudy: study container with slug, active flag, demographic collection toggle, and randomisation mode.
ResearchStudyArm: packet definition linked to existing CaseCollection, including weighting and marking scheme metadata.
ResearchParticipant: pseudo-ID keyed participant with demographic fields, consent flag, assigned arm, and generated CID credential link.

Migration generated: atlas/migrations/0103_researchstudy_researchstudyarm_researchparticipant.py.

Participant Flow

Participant visits: /atlas/research/run/<study_slug>/<pseudo_id>/.
Site captures consent and simple demographics (configurable per study).
Participant is randomised to a packet (study arm) by configured strategy.
A dedicated CidUser credential is created and attached to the participant if needed.
Assigned packet CaseCollection is launched via existing candidate collection routes.

This reuses current Atlas taking/timing/marking behavior and avoids changes to candidate-facing scoring internals.

Randomisation Logic

Mode	Behavior
Completely random	Weighted random choice over active arms using `allocation_weight`.
Balanced (equal-fill)	Assigns to least-filled active arm; tie broken randomly.
Balanced within group	If enabled, balancing is applied within `candidate_group` strata.

Marking and Scoring

Existing answer models remain in use: CidReportAnswer / UserReportAnswer.
Existing collection mark views remain functional for non-technical markers.
Per-packet configuration fields added on study arm:
- marking_scheme_name
- marking_guidance
- scoring_mode and max_case_score for normalised exports

Current marker UI itself is unchanged; packet-specific configuration is reflected in study management and exports.

Management and Export

Management Views

/atlas/research/ list studies
/atlas/research/create/ create study
/atlas/research/<pk>/ study detail + packets + participants
/atlas/research/<pk>/update/ update study
/atlas/research/<pk>/arm/<arm_id>/update/ update packet

Exports

/atlas/research/<pk>/export.csv
/atlas/research/<pk>/export.json

Export rows include demographics, assignment metadata, credential ID, attempt status, and raw/normalised scores.

Files Added/Updated

Updated: atlas/models.py, atlas/forms.py, atlas/views.py, atlas/urls.py, atlas/admin.py
Added templates:
- atlas/templates/atlas/research_study_list.html
- atlas/templates/atlas/research_study_form.html
- atlas/templates/atlas/research_study_detail.html
- atlas/templates/atlas/research_study_arm_form.html
- atlas/templates/atlas/research_participant_entry.html
Added migration: atlas/migrations/0103_researchstudy_researchstudyarm_researchparticipant.py

Compatibility Notes

No existing collection, case, or answer model behavior was removed.
Randomisation and participant routing are additive and isolated to new research URLs.
Author/editor permission pattern is respected for study management views.