Introduction to Consensus

One of the core goals of Delphi studies is to find out whether experts agree on specific topics. This agreement — or lack of it — is called consensus.

But consensus is more than a simple majority vote. It reflects how stable and aligned opinions are across your expert panel.

Measuring consensus helps you distinguish:

• Well-supported, shared expectations
• Controversial or highly uncertain scenarios

What Does Consensus Mean in Delphi?

In Delphi, consensus means that experts have converged in their opinions.

• High consensus = Experts mostly agree.
• Low consensus = Opinions are widely spread or polarized.

Importantly, no Delphi guarantees consensus — sometimes, persistent disagreement is just as informative.

For example, strong divergence might indicate:

• Lack of evidence
• Emerging technologies
• Deeply held conflicting values

That's why consensus metrics are not just technical: They guide your interpretation of results.

How Do You Measure Consensus?

There are several ways to measure consensus rigorously:

1. Mean and Interquartile Range (IQR)

Mean:

Sum of all ratings divided by the number of ratings.

IQR (Interquartile Range):

Shows the spread of the middle 50% of all responses.

• Q1 = 25th percentile
• Q3 = 75th percentile
• IQR = Q3 - Q1

Why use IQR?

• Robust against outliers
• Ideal for Likert-scale data
• Easy to interpret

Typical thresholds:

• IQR ≤ 1 → High consensus
• IQR > 1 → Moderate or low consensus

Example:

Scenario: Widespread adoption of AI diagnostics by 2030.

Mean rating: 4 (Likert scale 1–5)

IQR: 0.75

Interpretation: High consensus — most experts agree this is likely.

2. Percentage Agreement

This is a simpler measure: What proportion of participants gave ratings within a defined range?

Common threshold:

• ≥75% agreement within 1 point

Example:

80% of experts rated between 4 and 5 on likelihood.

Interpretation: Strong consensus.

3. Convergence Rate Across Rounds

In multi-round Delphi studies, you want to see how opinions shift over time.

Convergence Rate:

• Measures how many experts change their ratings between rounds.

Low convergence: Opinions are stable.

High convergence: Panel is still adapting or uncertain.

Example:

Between Round 1 and 2:

• 90% of ratings remained unchanged or moved by only 1 point.

Interpretation: High convergence and stabilization.

Tracking convergence is essential for deciding when to stop the Delphi process.

Why Is Consensus Important?

• It helps you identify scenarios with clear expert support.
• It shows where further research or discussion is needed.
• It improves credibility — funders and stakeholders prefer recommendations backed by consensus.
• It informs priority setting, especially in policy and strategic planning.

Example Scenario

Imagine your Delphi study explores future technologies in healthcare.

Scenario 1: AI diagnostics widely adopted

• Median: 4.5
• IQR: 0.75
• 85% agreement
• High consensus

Scenario 2: Blockchain in patient records

• Median: 3
• IQR: 2
• 60% agreement
• Low consensus

Interpretation: Scenario 1 is widely accepted; Scenario 2 remains controversial.

How Durvey Helps

Durvey automatically:

• Calculates median, IQR, and percentage agreement in real time.
• Tracks convergence rates across rounds.
• Flags high- and low-consensus items so you can focus your analysis.
• Exports all consensus metrics and paper ready texts for publication or reporting.

No extra spreadsheets or software needed.

Example Interpretation for a Report

"Consensus was achieved for 12 of 20 scenarios (IQR ≤1 and ≥75% agreement). Items lacking consensus were primarily related to data privacy technologies, indicating divergent expert expectations."

Durvey calculates all of this automatically, helping you:

• Save time
• Interpret results with confidence
• Communicate your findings professionally