Who is displaced in each country, how many, and what displaced them — the evidence that decides which populations a survey there needs to identify and which forced-to-flee options need local examples. Open a country for its drafting brief, then build the questionnaire from it.
Nobody measures "why did you leave home" directly at scale. Every figure
here is assembled from organisations that count something adjacent, and the join between
what they count and the eight response options is the thing this project had to build.
That join is a judgement, not a fact, and it is written down in
config/crosswalk.yaml so you can disagree with it.
| Who | What they actually count | Which options that can speak to |
|---|---|---|
| IDMC Internal Displacement Monitoring Centre |
People. How many were displaced inside their own country, and what the agency recorded as the trigger — a named storm, a named armed conflict. | 1, 2, 6, 7 — the entire population half of this dashboard |
| UNHCR | People. Refugees and asylum seekers, by country of origin and country of asylum, plus how often each nationality's claims are recognised. | everything shown for refugees; recognition rates stand in for 3 |
| UCDP Uppsala Conflict Data Program |
Incidents. One record per event in which at least one person was killed in organised violence, placed on a map, from 1989, worldwide. No deaths, no record. | 1, 2, 4, 5 — the events half |
| ACLED | Incidents. Political violence and protest, whether or not anyone died — so several times more events than UCDP for the same conflicts. 68 countries. | same options as UCDP, as an alternative reading, never added to it |
| IOM DTM | What displaced people say. The only source here where the reason comes from the person rather than from an analyst. | a reality check on all of the above |
| V-Dem | Conditions, not displacement. Expert ratings of torture, political killing, religious freedom, discrimination by social group. | 3 and 4 — proof the cause exists here, not proof it moved anyone |
The three things most likely to trip you up. One: people and incidents are different units and are never added together — the two views exist precisely so they stay apart. Two: sources are not summed. IDMC derives its figures from IOM DTM in twelve countries, so treating them as independent agreement is circular; UCDP and ACLED code the same events. The only legitimate sum on this page is IDPs plus hosted refugees. Three: none of this is causation. "61% drought-attributed" describes co-occurrence in administrative statistics, not why any individual left.
What is missing, and it matters. No database anywhere has a category for persecution or for state repression short of killing. A Rohingya family displaced by military persecution is recorded by IDMC as armed conflict. Across IDMC's own methodology notes, the words "persecution", "ethnic", "discrimination" and "torture" appear zero times. Options 3 and 7 look empty in the data because nobody counts them, not because they are rare — which is an argument for keeping them on the questionnaire, not for dropping them.
| View | What it counts | Where it is drawn | Period |
|---|---|---|---|
| People currently displaced (IDPs) | A stock — people still displaced right now. A snapshot, so nobody is counted twice. | Country where the displacement happened | end-2025 |
| Total displacements recorded | A flow — movements recorded during the period. Somebody displaced three times counts three times. | Country where the displacement happened | — |
| Worst year on record | Each country's peak displaced population in any single year, with the whole trajectory in the tooltip. Conflict only — no disaster split exists for this series. | Country where the displacement happened | 1990–2025 |
| Refugees hosted, by cause in origin | People who crossed an international border, attributed to the cause mix of the country they came from. | Country hosting them — not where events happened | latest year |
| Refugee movements between countries | The same refugees, drawn as movements rather than totals. Each arrow runs from origin to host country, coloured by what caused the displacement. | An arrow from origin to host | latest year |
Watch the third column. Three views put the circle where displacement was caused. The last two are about where displaced people went. Germany shows 2.8m not because anything happened in Germany, but because it hosts that many people displaced elsewhere. That is the view that matters for designing a show card in a host country, precisely because the causing events happened somewhere else.
An IDP was displaced inside their own country and never crossed an international border — still their own government's responsibility. A refugee crossed a border and is under international protection. The same event produces both; the only difference is whether the person crossed a line on a map. The first three views are about IDPs. The last two are about refugees. This distinction is the reason the identification questions ask about border crossing and time away at all — the reason for fleeing can be identical.
The response options in the identification question are numbered 1 to 8. Three of them — 3. discrimination or persecution, 4. human rights violations by authorities, 7. man-made events — are counted by no displacement database anywhere. IDMC has no category for persecution: a Rohingya family displaced by military persecution is recorded as armed conflict. The layer adds documented evidence from human rights investigations for those three, across the twenty largest displacement contexts. Ringed countries have it — hover for the evidence, click to pin the sources.
The twenty largest displaced populations, and the caveats that apply to every figure on this page. Full source notes are behind “Where these numbers come from” above.
Caveats. IDMC figures here cover 2025 only, so the stock reflects people displaced and still displaced at end-2025, and long-settled protracted caseloads from earlier decades are understated. The refugee view attributes each hosted refugee to the cause mix of their origin country as a whole — it is a population-weighted estimate, not a count of individually-attributed people, and it assumes refugees leave for the same reasons that displace people internally in their country. Man-made events (code 7) is absent everywhere because no agency counts development-induced displacement.