Panel management

The Panel area in Delphi Studio is the central place to recruit experts, manage participation, track consent, and monitor engagement throughout the study. It is designed to protect panelist privacy while giving researchers the tools they need to run a rigorous and well-documented Delphi process.

Recruitment workflow

Recruiting and onboarding panelists follows a clear, auditable sequence:

1. Add panelist — Researchers add individuals either one at a time or in bulk via CSV.
2. Send invitation — An email containing a secure magic link is sent to the panelist.
3. Panelist clicks magic link — They access the study without needing to create an account.
4. Consent — Panelists review and accept the consent form before they can participate.
5. Rate in open rounds — Once consented, panelists can rate items in any currently open round.

This workflow ensures that consent is obtained before any data collection and that all steps are recorded for audit purposes.

Adding panelists

Delphi Studio supports two methods for adding panelists:

Method	When to use	Notes
Single add	One-off or small invitations	Quick entry for individual experts
Bulk CSV	Panels of 10 or more	Validation plus dry-run preview before commit

The CSV importer accepts exactly four columns; any other column is rejected:

Field	Required	Purpose
email	Yes	Used to deliver the magic link
name	Yes	Stored in encrypted form for identity management
country	Optional	Two-letter ISO code, used when reporting panel composition
stakeholder_role	Optional	Primary analytic stratum for balance and cross-stakeholder rules

Bulk import runs a dry-run preview first so you can fix validation errors (bad emails, duplicates, unknown columns) before any rows are written. Import is all-or-nothing: a file containing any invalid row imports no rows. Emails are hashed for deduplication. Raw email addresses and names are stored in an encrypted panelist_identities table, separate from the ratings data.

The Panel view lists every panelist by pseudonym, tracks invitation state, and surfaces response rates and follow-up actions.

Pseudonymization

To protect panelist privacy, Delphi Studio uses strong pseudonymization:

What researchers see	What is actually stored
Pseudonyms (Panelist 1, Panelist 2, …)	Encrypted name and email address
Expertise fields and credentials	As provided during enrollment
All ratings and comments	Linked only to the pseudonym

Researchers work with pseudonyms in standard analytic and directory views. An authorized panel manager may reveal identity only through a deliberate action that requires an operational reason and commits an audit event before returning the identity. This design keeps identity access separate from routine review without preventing necessary panel operations.

Reusing eligible experts across studies

The organization-wide Expert directory is separate from the study panel. It indexes pseudonymous participation and shows a recruitment pool only for identity-linked experts who explicitly permit future contact.

Adding an eligible expert to a target study records the criteria snapshot, rationale, stakeholder role, and COI decision and creates a pending panelist. It does not send an email; sending and resending are separate commands. Historical participation alone never creates future-contact permission.

See Expert directory and recruitment for eligibility gates, directory decisions, candidate recruitment, and identity boundaries.

Invitation states

Each panelist moves through defined states during the study:

State	Description	Available actions
pending	Panelist has been added but not yet invited	Send invitation
invited	Invitation has been sent	Resend invitation
accepted	Panelist has consented and is active	Include in upcoming rounds
completed	Reserved for panelists who finish all required rounds; not currently set automatically	—
withdrawn	Panelist has withdrawn — including by declining the consent form	Excluded from future rounds; withdrawal is audited and not reversible as “active” participation
unavailable	Researcher-marked incapacity or unreachability (not disengagement)	Reminders stop; excluded from prospective round denominators; prior-round data stays valid; state is reversible back to accepted, with audit
declined	Reserved; a panelist who declines consent is recorded as withdrawn, not declined	—
non_responder	Reserved for attrition flagging; not currently set automatically	—

completed, declined, and non_responder exist in the data model but are not currently set by the platform. In practice, a panelist who declines the consent form (or a re-acknowledgment) moves to withdrawn.

Unavailable vs withdrawn

Use unavailable when a panelist becomes unreachable or incapacitated but their earlier ratings should remain valid for analysis. Use withdrawal when the panelist (or you) permanently ends participation. Both suppress reminders; only unavailable is designed to be reversed if the panelist returns. Marking unavailable requires a reason and is recorded in the audit trail.

These states help researchers track engagement, report attrition causes honestly, and manage follow-up efficiently.

Stakeholder roles and panel balance

When your study defines stakeholder groups (for example patients, clinicians, and methodologists), each panelist carries:

Field	Role
Primary stakeholder role	Analytic stratum used for subgroup breakdowns and optional cross-stakeholder consensus rules. Dual-role experts are never double-counted.
Secondary affiliations	Descriptive only — useful for reporting panel composition

The Panel view compares enrolled counts against recruitment targets set in the wizard and surfaces balance warnings when groups are missing, a single group dominates, or panelists lack an assigned primary role.

Consent & re-consent

Consent is tightly linked to the study protocol. When a panelist consents, their consent is bound to a specific protocol hash.

If the protocol is later amended:

1. Delphi Studio generates a new protocol hash and records the amendment.
2. Affected panelists receive a re-acknowledgment request tied to the new hash.
3. What happens next depends on the study's re-consent mode (configured on Consent & comms):
   • re_consent_required — opening a new round is blocked study-wide until every outstanding re-acknowledgment is resolved.
   • notify_only — panelists are notified of the amendment, but nothing is blocked.

Declining a re-acknowledgment withdraws the panelist from the study.

All consent events are recorded in the audit log:

Event	Audit record
Initial consent	panelist.consent
Protocol amendment applied	protocol.amendment_applied
Re-acknowledgment accepted	panelist.reack_acknowledged
Re-acknowledgment declined	panelist.reack_declined (followed by panelist.withdraw)

This mechanism ensures that panelists are always consenting to the current version of the study.

Withdrawals, unavailability & suppression

Panelists may withdraw from the study at any time: by choosing Decline on the consent screen, via the withdrawal link in every study email footer, or by replying to a study email (handled by the inbound-email webhook). The withdrawal is immediately recorded, and the panelist is excluded from all future rounds. Researchers can also mark a panelist unavailable (with a reason) when incapacity—not voluntary withdrawal—is the issue.

Email suppression is handled automatically and operates at two scopes: hard bounces and spam complaints suppress the address platform-wide, while unsubscribes suppress it for that study only. Withdrawing in-app does not create a suppression entry — the withdrawn state itself stops study emails. Suppressed addresses will not receive further study emails, protecting both panelists and sending reputation.

If a late withdrawal or data correction changes who should count in a closed round, use derived re-analysis rather than rewriting frozen dispositions.

Monitoring participation

The panel dashboard provides real-time visibility into engagement:

Surface	What it shows
Round progress	Per-round response tracking, so you can spot panelists at risk of dropping out
Panel balance	Enrolled counts per stakeholder role against recruitment targets, with imbalance warnings
Re-acknowledgment banner	Outstanding consent re-acknowledgments after a protocol amendment

Invitation reminders are scheduled automatically at 3, 7, and 10 days after the invitation (configurable per study), so unaccepted invitations are followed up without manual work.

A dedicated Tasks page aggregates action items for the research team. Six issue types can appear: missing IRB number on an advanced draft, pending consent re-acknowledgments, a closed round awaiting a stopping decision, a round held past its close date, low response rate on a closing round, and a round closing soon.

Panel size guidance

Recommended panel sizes vary by study type:

Study type	Typical panel size	Notes
Modified e-Delphi	10–30 experts	Most common range for health services research
RAND/UCLA (by specialty)	9–12 per group	Smaller, highly specialized panels
Pilot studies	5–8	Should be documented as a limitation

Delphi Studio blocks activation when the enrolled panel is below the minimum panel size set during study design. Stakeholder-role imbalance produces a warning but does not block.

Email communications

Delphi Studio seeds one editable template per email kind for each study:

Email kind	Trigger
Invitation	Initial recruitment
Invitation reminder	Follow-up when the invite has not been accepted
Reminder	Non-responder follow-up during an open round
Round open	A new round becomes available
Round closing	Reminder before a round closes
Round complete	Notification that a round has closed
Study complete	Study has concluded
Withdrawal confirmation	Confirmation when a panelist withdraws
Re-consent	Required when the protocol changes

All email templates are editable on the Consent & comms screen in the study wizard. Quiet hours can be configured to delay sends during a nightly window. The window is applied in the panelist's timezone when one is on record and in UTC otherwise; the platform does not currently collect panelist timezones, so in practice quiet hours are evaluated in UTC.

Additional panel tools

• Per-panelist link copying — copy a panelist's secure access link directly from the panel list.
• Candidate recruitment from the expert directory — pull candidates from your organization's expert directory, with conflict-of-interest status captured at selection.
• Panelist detail page — per-panelist round progress, outlier flags, audit history, and stakeholder-role reassignment.
• Ethics & publication intake — acknowledgment preference (named, anonymous, or none) and related publication details captured when adding a panelist.
• Idempotent email sends — scheduled sends carry idempotency keys, so retries never double-send.
• Audited identity unmasking — revealing a panelist's real identity records a panelist.unmask audit event.

Next steps

• Panelist experience — what panelists see after they click the magic link
• Expert directory and recruitment — consent-aware reuse across studies
• Running rounds — open, close, and iterate
• Collaboration & roles — who on the research team can manage the panel
• Audit trail — consent and invitation events in the log