Multi-round worked example

This example walks through three items across two rounds with real numbers — showing how statistics become zones, dispositions, and feedback.

Setup

Parameter	Value
Panel	10 experts (all respond both rounds)
Scale	5-point Likert
Rules	≥80% in zone, median agree ≥ 4, IQR ≤ 1

Delphi Studio computes quartiles by linear interpolation (R type 7, the NumPy and pandas default) and takes the mean of the two middle values for even-numbered medians. The figures below match what the platform displays.

Item A — clear consensus in Round 1

Statement: "Annual competency review should be mandatory for telehealth nurses."

Round 1 ratings

Score	1	2	3	4	5
Count	0	0	1	3	6

Statistic	Value
Median	5
IQR	1
% agree (4–5)	90%
Zone	Agree
Disposition	retain

Item A is excluded from Round 2. No feedback shown to panelists for this item.

Item B — indeterminate, revised for Round 2

Round 1 statement: "Nurses should read ECGs during telehealth visits."

Round 1 ratings

Score	1	2	3	4	5
Count	1	2	4	2	1

Statistic	Value
Median	3
IQR	1.5
% agree	30%
Zone	Indeterminate

Revision

Round 2 statement: "Nurses should recognize ECG red flags during telehealth visits and escalate per protocol."

Round 2 ratings

Score	1	2	3	4	5
Count	0	0	2	5	3

Statistic	Value
Median	4
IQR	0.75
% agree	80%
Zone	Agree
Disposition	retain

Item C — bimodal pattern

Statement: "Patients may receive prescriptions without synchronous video if audio quality is sufficient."

Round 1 ratings

Score	1	2	3	4	5
Count	3	1	0	1	5

Statistic	Value
Median	4.5
IQR	3.75
% agree	60%
Bimodal flag	Yes
Zone	Indeterminate

The item fails on two rules: agreement is 60% against a ≥80% threshold, and the IQR far exceeds the ≤1 ceiling. The bimodality flag does not itself block consensus — it marks the item for qualitative review of the two opposed clusters. Panelists see a histogram in feedback showing the polarized distribution.

Researcher action

Override not applied — item revised for Round 2 to clarify regulatory context. If still indeterminate after Round 2, document as unresolved in discussion.

Feedback packet (Round 2 preview)

For Item B, panelists see:

Element	Value shown
Prior own rating	3
Group median (R1)	3
IQR (R1)	2
Approved comments	2 de-identified excerpts
Peer identities	Hidden
Peer individual ratings	Hidden

Stopping check after Round 2

Item	Final disposition
A	retain (R1)
B	retain (R2)
C	revise → Round 3 or unresolved

Stopping signal	Value
Unstable items	1 (Item C)
Recommendation	still_moving if Round 3 planned
Max rounds	3 — if C still indeterminate, report caveat

Kendall's W (illustrative)

Round	Kendall's W
1	0.42
2	0.58

Increasing concordance supports convergence narrative in results section.

Takeaways

Concept	Demonstrated
Median + IQR + % threshold	Item A consensus
Revision lineage	Item B parent → child item
Bimodality guard	Item C false consensus prevention
Feedback lock	Item B Round 1 stats in R2 packet
Stopping logic	Mixed retain + contested

Next steps

• Interpreting results
• Running rounds
• Clinical guideline study