Heat Map Methodology

The heat map is built from reports published in the Hub's document library. Each report names its author or issuing body and records how many independent sources corroborate it. Sources include institutional and official statements, news reporting, policy briefs, scholarly work and civil-society and faith-based reporting.

Reports are read in 14-day monitoring cycles. Scores are refreshed every cycle, and earlier cycles are kept, never overwritten: any past cycle can be reopened from the dashboard and compared side by side with another.

Each report is assigned to one or more of 5 categories, and scored on the component elements of each category it addresses:

Cross-border items. A report concerning more than one country, such as a cross-border armed group, cross-border displacement or a regional peace initiative, is attributed to every country it concerns rather than forced into one, so transnational developments are not understated.

Each element is scored on a ten-point scale, where 10 is severe or critical. Elements a report does not address are left unscored and excluded, never counted as zero. A report's score for a category is the average of its scored elements.

Reports are then weighted before they are combined:

• Corroboration. A report backed by more independent sources counts more: ×1.0 for one source, rising by 0.25 for each further source, up to ×2.0 at five or more.
• Recency. Within a cycle, a report 14 days older than the cycle's end counts half as much as one from its last day, and never less than a quarter, so the reading reflects the most current situation. The “all time” view is an overall record and is not weighted by recency.

Each cell of the heat map, one country and one category in one cycle, combines its reports in two parts:

The share given to the highest report keeps a single critical event from being averaged away by several routine reports. In addition, a cell is flagged critical whenever any report in it scores 9 or more, whatever its combined score. A country's Overall column is the average of its category scores.

Worked example

A plain average would give 5.3. The weighted mean is 6.3; blended with the highest report (9), the cell scores 7.1 (High), and carries a critical flag because report C scores 9.

Regional overview. Regional rows combine every report concerning any member country, counting each report once per region even when it names several members. Regions may overlap.

No score reaches the public heat map without human approval. Reports are entered and scored by CRTP analysts, then reviewed and approved by a senior administrator before publication. An approved report that is later edited is withdrawn and re-enters review.

The platform records how each report's scores were produced (analyst-scored, AI-proposed, or AI-proposed and then adjusted by an analyst), who last set them and when. That label appears with each report on the map.

Limitations

• Scores are structured expert judgements, not measurements. They are only as complete as the reporting behind them.
• A country or category with no score has no published report in that cycle; it does not mean the situation is calm.
• Coverage favours places and issues that are widely reported. Local and faith-based sources are sought to offset this.
• Changes to the weights apply to every cycle, past ones included, so comparisons stay like for like.

Questions about this methodology are welcome through the contact page.