Reference · What sits under each option

The record behind each response option

The database category each source actually recorded for this option, and a verdict on how good that fit really is. This is the verifiable record — every card below traces to a real logged category in a real source database.


“In plain words” below is supporting context, not part of that record — a separately written description of how this reason plays out in practice, filed under the same option because it belongs to the same idea, not because it's been matched to a specific category.

In plain words

Not matched row-for-row to the categories below.

Categories looked at and recoded to nothing

Not every raw category earns a response option — some were considered and deliberately excluded, which is itself part of the record. ACLED's “Peaceful protest” is the biggest example: 386,822 events, correctly mapped to no option, because peaceful protest doesn't displace anyone. These don't appear under any option above for that reason.

How much recorded volume each source carries

Total volume across every crosswalk row for that source — no options here, just size. IOM DTM and UNHCR are flagged rather than drawn at zero: their crosswalk rows aren't volume-comparable displacement counts, so a zero-length bar would misreport “found nothing” when the truth is “doesn't measure it the same way.”

Which options each source's categories map to

Structure, not size: ribbon width here is the count of distinct source categories mapped to each option, not recorded volume — IDMC GIDD dominating the chart above would swamp this picture too, so this one deliberately answers a different question: how many ways does each source connect to each option. Hover a ribbon for the count.