The identification questions, built for one country
The questions and response options never change. The parts the instrument lets vary — the examples after each “e.g.”, the office and document named when asking about international protection — are drafted from what was recorded in that country and shown in blue. Choose the country and the populations to identify, check and edit the blue text, then download the questionnaire with its instructions, derivation rules and translation template.
Options with local examples:5–7 of 73–40–2▮ document image▮ document: links only
2 Who to identify
3 Forced-to-flee questionVersionExamples
How askedLanguage
Shortest and Mid-length are the
other two official variants from the question-testing document — each merges some of
the Long version's options into broader categories, and Mid-length uses an alternate
definition of "forced to flee". Only drafted in English so far; pick English to use them.
Version, example length and language apply to the forced-to-flee question;
the rest of the module follows the language and the populations chosen in step 2.
Subnational example setWhere a region’s recorded events differ from the national picture, its
own examples replace the national ones in the forced-to-flee question.
Preview with another population’s examples
A refugee from another country is not well served by this country’s own
examples. Pick a population hosted here to see the forced-to-flee question with the examples
recorded for their country of origin instead.
4 Review and edit
Anything in blue can be edited: click it.
It was drafted for this country from the databases — an office renamed, a document everyone
calls something else, an example that does not fit. Purple
marks your edits: kept in this browser, carried into every file you download and into the link you copy.
Forced-to-flee question — with this country’s examples
Each example and the record it was drafted from. The map (step 0) shows the same evidence as populations and events: see this country on the map ↗
Option
Example
Type
Source
Evidence
International protection question — the office and document named here, two versions
Full questionnaire — every item, exactly as it downloads
The two questions above, followed by the rest of the module for the populations
chosen in step 2. Items always asked: forced to flee, country of the home fled, moved to another
country, applied for protection, outcome. Everything else appears because a sub-category you
selected needs it.
No office or document is on record for this
country yet. Draft Version A (the office) and Version B (the document) with whoever registers
claims there, then report them so the record is updated.
5 Try a respondent
Pick a respondent profile, or set the answers yourself. The questions that would be
asked are marked in the questionnaire above and the resulting category follows the rules in the
derivation sheet, for the populations chosen in step 2. Use it for interviewer training and to
check a programmed form.
6 Your files
All five are for the country, populations, version and language chosen above, and
carry your edits.
Questionnaire
The two customised questions, the full module, and a table of every customisation.
Instructions
For the coordinator: placement, checks before fielding, pretest protocol. For interviewers: question by question.
Derivation sheet
How answers become categories: the rules, with Stata, R and Python code.
Translation template
Every text with the six UN-language drafts, a column for the survey language, and terminology notes.
KoBo / ODK form
The module as an XLSForm: skip logic, the valid-reason calculations, the localised lists, six languages. Upload it to KoBo, ODK or SurveyCTO.
These are drafts for review, not enumerator text: the examples come from event
databases, the office and document names from UNHCR help pages and its registration survey, and
the translations are unreviewed.
The office is where the claim is lodged, never who decides it. Eligibility
panels, appeals boards and hotlines are excluded even where they are well known,
because a respondent never went near them.
The office doesn’t travel everywhere. In many countries the claim is
lodged online, by post, at a police station, or happens automatically with no
office a respondent would visit — those are flagged in the card with the actual
channel, so the wording can be adapted rather than asked as written.
Source confidence is high, medium or low per country. Check medium and low
against a country source before fielding; the checks list in step 6 says which.
Cross-checked against UNHCR’s internal Registration Baseline Survey (2024/25). Where the operation itself answered, the note under the module records what it said — joint, parallel or split registration, the year registration was handed over to the Government, which document types UNHCR issues — and flags the handful of countries where its answer doesn’t match the office named here. It also adds 29 countries the public scrape never covered, with the registrar known but no office or document named yet (LOW confidence).
No IDP document is named. Of the major contexts checked, only a handful have a
verifiable IDP status document; most have none at all, which is a finding about the
instruments, not a gap in the search.
The translations are unreviewed. Translating an instrument is a specialist
job — TRAPD, or forward-and-back translation with reconciliation —
because a question has to be understood the same way, not merely mean the
same thing. Cognitive testing on an unreviewed translation tests the translation,
not the question.
“Flee” is rendered as leaving under duress, not fleeing in
panic. The item defines it as leaving “due to events that posed a
threat”, so the sense is compelled departure. Five of these languages have
a default verb that instead means running away in fear — fuir,
huir, الفرار, спасаясь бегством,
逃离 — and each of those invites a respondent to picture a panicked
escape and answer “no” if their own departure was deliberate: packing
over days, or leaving after a threat rather than during an attack. That is exactly
the population these questions exist to count. All five now use a duress
construction.
Still open:persecution, a legal term of art in the refugee
definition that in everyday registers reads as ordinary harassment — a
materially different threshold, and not something an automatic process can
settle.
The names are UCDP’s, not a respondent’s. UCDP writes
“JAS” where a Nigerian respondent says Boko Haram. Actor names are
never translated — they are proper nouns.
Identity groups are never named. UCDP codes communal violence as, for
example, “Christians (Nigeria) – Muslims (Nigeria)”, and at
subnational level as named ethnic groups. Correct as conflict coding, unusable in
an instrument a government enumerator reads aloud. Those become “communal or
intercommunal violence”.
Naming actors risks anchoring. Someone displaced by a group not named may
conclude the option does not cover them. Category phrasing is preferred where the
data allows, and every example is tagged with which it is.
Subnational sets appear only where they say something different. A region
whose examples merely repeat the national list is not shown.
