Core concepts

These definitions describe the fundamental building blocks of a Delphi study as implemented in Delphi Studio. Together they form a coherent, auditable workflow that supports rigorous modified e-Delphi and related consensus methodologies.

Round

A round is one complete iterative rating cycle within a Delphi study. It is the primary mechanism by which expert judgment is collected, aggregated, and returned to the panel through controlled, anonymized feedback.

Each round advances the study toward consensus by allowing panelists to revise their ratings after reviewing structured group information. In Delphi Studio, rounds are configured in the study wizard and governed by the immutable rules captured in the launch snapshot at study activation.

Round states

Delphi Studio manages every round through a strict, auditable state machine:

State	Description	What happens in Delphi Studio
draft	Round has been configured but is not yet visible to panelists	Researcher defines timing, item set version, and revision windows. Still editable.
scheduled	Round will open automatically at the defined opensAt timestamp	Hidden from panelists. Dashboard shows countdown.
open	Round is active; panelists may submit ratings and comments	Real-time response monitoring and preliminary analytics available to the research team.
closed	No new ratings or comments are accepted	Round is finalized. Analysis pipelines become available.
analyzed	Statistics, zone classifications, and item dispositions have been computed	Researcher reviews results, approves feedback snapshot, and may initiate item revisions.

Round state-transition diagram

Legend

• Blue nodes — Primarily researcher-controlled actions (via the study dashboard).
• Green nodes — Automated transitions managed by Delphi Studio.
• Pink nodes — Terminal or decision states.

See Running rounds for operational detail.

Rating

A rating is the atomic unit of data in a Delphi study: one panelist's response to one item in one round.

Each rating can contain:

• One or more scored dimensions (e.g., Agreement as the primary dimension, plus optional secondary dimensions such as Importance, Clarity, Feasibility, or Relevance).
• An optional free-text comment (required on low scores or unable-to-rate responses when the governance rule is enabled).
• An Unable to rate flag, which excludes the response from median, IQR, and zone-percentage calculations but is reported separately for transparency.

Ratings are private to the individual panelist until the round closes. Delphi Studio stores every rating with full provenance (panelist pseudonym, round, item version, timestamp, dimension values, and comment) to ensure all downstream analytics and exports remain fully traceable.

Consensus zones and status

After a round closes and is analyzed, each item is classified into one of three zones using the consensus rules frozen in the launch snapshot:

Zone	Meaning
Agree	The panel meets or exceeds the pre-specified agreement threshold (e.g., ≥80% in agree zone and median ≥ threshold and IQR ≤ threshold).
Disagree	The panel meets or exceeds the pre-specified disagreement threshold (symmetric to Agree).
Indeterminate	Neither the Agree nor Disagree threshold is met. The item carries forward or is flagged for researcher review and possible revision.

Consensus status is derived from the zone and recorded as:

• consensus_in
• consensus_out
• no_consensus (indeterminate)
• consensus_with_subgroup_divergence — panel-wide agreement downgraded because a prespecified stakeholder group of sufficient size showed substantial opposition. This status is non-terminal: the item carries forward for re-rating rather than being retained.

When bimodality detection is enabled, items showing strong polarization are additionally flagged even if they meet zone thresholds.

Controlled feedback

Controlled feedback is the structured information returned to panelists between rounds. It includes:

• Quantitative summaries (group median, IQR, response histograms, zone percentages)
• Approved, de-identified qualitative comments
• The panelist's own prior rating for self-comparison

In Delphi Studio, feedback can be locked by the research team before the next round opens. Once locked, an immutable per-panelist snapshot is generated and stored. Panelists receive the same approved group-level feedback, alongside their own prior response for reflection. Feedback content, lock timestamp, and delivered snapshots support auditable ACCORD- and CREDES-informed reporting.

Collaborator roles

Delphi Studio uses role-based access control:

Role	Typical permissions
owner	Full control over study configuration, activation, round management, item revisions, exports, and user management.
co_investigator	Can design methodology, manage the panel, configure and run rounds, and review analytics (subject to owner oversight).
analyst	Access to results, exports, qualitative comment review, and consensus analytics. Cannot alter protocol or rounds.
read_only	View-only access to study configuration, progress, and results (useful for oversight or external collaborators).

All role assignments and permission changes are logged in the audit trail.

How the concepts fit together

The diagram below shows the core entity relationships and workflow in Delphi Studio.

This structure ensures that every decision (methodology, item wording, feedback content) is versioned or snapshotted, every rating is traceable, and the entire study remains reproducible and defensible for publication or regulatory review.

Next steps

• Glossary — full term reference
• The Delphi lifecycle — design through report
• Structure — methodological architecture of a Delphi study
• Modified e-Delphi — default preset and thresholds