# GAIN — SDG Indicators for Forcibly Displaced Populations: Methodology Record *How each SDG indicator was derived from the different survey microdata, the disaggregation used, and the caveats. Companion to `gain_sdg_values_calculated.csv` (computed) and `gain_sdg_values_promoted.csv` (report-extracted).* Last updated: 2026-08-24. --- ## 1. Disaggregation — the population-group variable, per survey family Every value is broken down by displacement status. The variable that carries that status **differs by survey family**: | Survey family | Datasets in hand | Group variable | Groups it distinguishes | |---|---|---|---| | **UNHCR FDS** (Forced Displacement Survey) | South Sudan 2023, Pakistan 2024, **Zambia 2025** | `Intro_07` / `Intro_07_1` | Refugees · Host community · (Former refugees · Returnees) | | **UNHCR Profiling** | Cameroon 2024, DR Congo Grand Kasaï 2023 | `HOUSEHOLD_STATUS` (CMR) · `s00q0_HHtype_HHH` (COD) | IDP · Host · Returned IDP · Repatriated refugee | | **WB–UNHCR JDC HFPS** | Ethiopia, Chad, Burkina, Djibouti, Sudan, Jordan | (survey-specific stratum) | Refugee / IDP / host | | **CBPS** (Cox's Bazar Panel Survey — **Rohingya**) | Bangladesh 2019–2020 | `stratum` | **Camp (Rohingya)** · High-spillover host · Low-spillover host | | **SRHCS** (Syrian Refugee HH Survey) | Jordan, Lebanon, Iraq/Kurdistan 2016 | Syrian-refugee sample frame | Syrian refugees | | **National MICS / DHS** | Bangladesh MICS7, Afghanistan MICS6 | **none** | General population **only** — camps excluded from frame | > **Critical caveat:** the **national MICS/DHS surveys carry no refugee/IDP flag** — their > sampling frames exclude the camps. They are used as the **general-population baseline** for > comparison, **not** as a source of refugee-disaggregated values. --- ## 2. How each SDG indicator was computed For every indicator: the internationally-agreed definition (fixed), then the **survey-specific variable mapping** (what changes per survey), and the survey weight. ### SDG 2.1.2 — Food insecurity - **Definition:** WFP **Food Consumption Score** = Σ (7-day frequency of 8 food groups × weight: cereals ×2, pulses ×3, vegetables ×1, fruit ×1, meat/fish ×4, dairy ×4, sugar ×0.5, oil ×0.5). Classified **Poor ≤21 / Borderline ≤35 / Acceptable >35**; *food-insecure = FCS ≤35*. Own-use farming is **excluded** (per 19th ICLS). Where a survey ships a pre-computed reduced Coping Strategy Index (**rCSI**) level, that is used directly (level 3 = food-insecure). - **Variable mapping:** - South Sudan FDS → `Food_div1..8`; weight `wgh_samp_resc_pop` - Pakistan & Zambia FDS → `Food11aa, Food12a, Food13a, Food14a, Food15a, Food16a, Food17a, Food18a`; weight `wgh_strata_spec` - Cameroon Profiling → `rCSI_LEVEL` (pre-computed); by `HOUSEHOLD_STATUS` ### SDG 6.1.1 / 6.2.1 — Improved drinking water / sanitation - **Definition:** WHO/UNICEF **JMP ladder** — % using an *improved* source/facility. Water improved = piped/faucet, borehole/tubewell, protected well/spring, rainwater, packaged/ delivered. Sanitation improved = flush/pour-flush, VIP, pit latrine with slab, composting. - **Variable mapping:** - FDS (SSD/PAK/ZMB) → `BD01` (water), `SAN01` (sanitation); classified by JMP code list - DRC Profiling → `s06q15_WaterSource`, `s06q25_Toilet` (codes 1–4 = improved sanitation) - National MICS → `WS1`, `WS11` ### SDG 8.5.2 — Unemployment rate - **Definition (ILO):** unemployed ÷ (employed + unemployed). **Employed** = worked ≥1 hr for pay/profit or temporarily absent. **Unemployed** = not employed **and** actively sought work **and** available to start. - **Variable mapping:** - Pakistan FDS → employed `EMP01–EMP05`; sought `EMP25a`; available `EMP29` - South Sudan FDS → **not computed** (its non-employed job-search question routes differently; `EMP20/EMP22` measure *underemployment* of the already-employed, so were not used) ### SDG 16.9.1 — Birth registration, under-5 - **Definition:** children under 5 whose birth is registered ÷ all under-5. - **Variable mapping:** roster level, filtered to age <5. - SSD & PAK FDS → `HH_07b` (birth registered); age `ageYears` (SSD) / `agetouse` (PAK) - DRC Profiling → `s01q04_BirthCertificate` (member file, joined to household group via `ID_Household`) ### SDG 4.1.2 — School attendance, ages 6–17 - **Definition:** children 6–17 currently attending/enrolled ÷ all 6–17. - **Variable mapping:** roster level. - SSD FDS → `HH_Educ02a` (currently attending) - DRC Profiling → `s03q03_InSchoolNow` - *Pakistan:* the auto-detected variable was pre-school (`HH_Educ00`) — **not used** (wrong band). --- ## 3. Computed values (as of 2026-08-24) 16 calculated values across **5 countries** (refugee/IDP value; host value in the CSV): | Country | 2.1.2 Food | 6.1.1 Water | 6.2.1 Sanit. | 8.5.2 Unemp. | 16.9.1 Birth reg | 4.1.2 School | |---|---|---|---|---|---|---| | South Sudan (FDS) | 90% | 97% | 58% | — | 62% | 81% | | Pakistan (FDS) | 47% | 79% | 74% | 11% | 42% | — | | Zambia (FDS) | 76% | 98% | 75% | — | — | — | | DR Congo (Profiling) | — | 30% | 8% | — | — | — | | Cameroon (Profiling) | 51% | — | — | — | — | — | **Recurring finding:** refugees in *managed settlements* often out-perform surrounding hosts on services (Zambia water 98% vs 75%, sanitation 75% vs 29%; South Sudan sanitation 58% vs 18%; birth registration 62% vs 27%), while **food insecurity is severe for both** (South Sudan 90%, Zambia 76%). Full host comparisons and per-cell method notes are in `gain_sdg_values_calculated.csv`. --- ## 4. Caveats recorded (per survey / indicator) - **National MICS (Bangladesh MICS7, Afghanistan MICS6):** general population only; the frame excludes camps → used as **baseline**, never as a refugee value. (Bangladesh national water came out ~47% via `WS1` codes — a JMP-tier/arsenic mapping quirk; the national figure is really ~98% and needs the improved/limited tiering before use.) - **South Sudan 8.5.2 (unemployment):** skipped — a wrong 0–1% resulted from using the underemployment variables; the correct non-employed job-search item must be located first. - **DR Congo (Grand Kasaï Profiling):** **unweighted** (no design weight ships with the file); member-level birth-certificate and enrolment came out 0% for all groups (the yes/no code is not `1` as assumed) — **WASH only** was retained until the value labels are inspected. - **Cameroon Profiling:** the distributed **CSV strips value labels** (WASH variables are bare codes 1–4) — only the self-documenting `rCSI_LEVEL` was computable from the CSV; the `.dta`/`.sav` would carry the labels for the rest. - **CBPS (Rohingya):** rich panel, but **modules differ by round and by group** — e.g. the skip-meals item (`s04c`) was asked to the **host** sample only in Round 2, so a clean camp-vs-host cell requires selecting a variable asked to both; and the COVID-subsample weight is missing for the camp records. - **Zambia FDS:** ships as **`.rds`** (R-native) not readable Stata `.dta` — read with `readRDS`. --- ## 5. Method summary (repeatable recipe) > Read the value labels (codebook) → identify the population-group variable + survey weight → > map the survey's variables to the indicator definition → apply the fixed formula (WFP FCS / JMP > ladder / ILO / registration ratio) → weight → disaggregate by group → tag `calc`. The **formulas are fixed and international**; only the **variable names change per survey** (~5–15 min of variable-location per survey). Mapping once per *family* (FDS, Profiling, HFPS, MICS/DHS, CBPS, SRHCS) covers all countries within it. --- ## 6. Population group carried on every value (map tooltip) Each value on the map is tagged with the **specific displaced population** it describes — never generic "displaced." The tooltip chip is colour-coded: **Refugees** (incl. Syrian, Rohingya, Palestine, resettled) = UNHCR blue; **IDPs** (incl. victims/displaced) = amber; **Venezuelans** = teal; anything else (e.g. a national LFS that only *includes* non-nationals) = grey, so a non-clean population is visible at a glance. **Refugees ≠ migrants.** Census/foreign-born stock is *not* relabelled as refugees. Group by source: - Refugees: South Sudan, Pakistan, Zambia, Kenya (FDS/RHHS); Egypt, Uganda, Italy, UK (reports) - IDPs: DR Congo, Cameroon (Profiling); Somalia, Nigeria, Honduras, Colombia, Burkina (reports) - Syrian refugees: Jordan, Iraq (SRHCS, est.); Rohingya refugees: Bangladesh (CBPS); Venezuelans: Ecuador, Peru ## 7. SDG 10.7.4 (share of population who are refugees) — not harvestable from reports Scanned all downloaded reports for a clean *"X% of the population are refugees/IDPs"* figure. **Not available as a percentage:** refugee-focused reports state population *characteristics*, not the population *share*; census releases report only foreign-born/migrant **stock** — which must **not** be relabelled as refugees. A clean 10.7.4 would have to come from UNHCR Refugee Data Finder (refugee count ÷ total population), not from these report PDFs. ## 8. Integrity note — a figure pulled after review **Nigeria 4.1.2 removed:** the NBS IDP figure (63.4%) is *share who **never** attended school* — the inverse of attendance — so it was dropped rather than shown under "School attendance." Replaced coverage with a clean **Uganda 4.1.2 = 71%** (primary net enrolment, refugee sample, vs 78% national — UBOS UNHS). ## 9. Full-report sweep (all 83 PDFs) — adds, corrections, new indicator Scanned every downloaded PDF for indicator + displacement-population + percentage in-context. Census releases (Armenia, Belize, Cambodia, Kazakhstan, Kosovo, Thailand, Turkmenistan, S. Africa, Zambia, Djibouti, Indonesia, Mexico, Morocco RGPH, Côte d'Ivoire, Mali, Malawi, Spain, Kyrgyzstan, Uganda census) returned **no refugee/IDP-disaggregated indicator** — they carry migrant/foreign-born stock only, which is **not** relabelled as refugees. - **New indicator SDG 7.1.1 (access to electricity)** added to the schema (user-flagged). Values: Morocco 99.8%, Nigeria (IDP) 61%. - **Added:** Egypt 2.1.2 = 58% (refugees food-insecure); Morocco 6.1.1/6.2.1/7.1.1 (labelled *"forced migrants (incl. refugees)"* — HCP survey population is mixed, **not** refugees-only). - **Corrections (earlier mis-extractions):** Burkina 1.2.1 56%→**93%** (multidim. poverty incidence, IDP vs 75% host) and 2.1.2 35.6%→**76%** (food insecurity) per INSD ESEP-PDI p33/p37; Palestine 8.5.2 29.2%→**42.2%** (29.2% was West Bank only; 42.2% is the State-of-Palestine total). - **Left out (mislabeled):** France/Norway "64%" (activity/employment rate, not unemployment); Somalia "70% not attending" and Nigeria "never attended" (inverse-signed vs attendance). ## 10. Temporal scope rule (2026-08-25) — 2018+ only, prefer 2021+, government-produced Per EGRISS timeline: **IRRS adopted 2018, GAIN implementation monitoring from 2021.** The map excludes **pre-2018 data** (won't be in GAIN to most extent) and prefers **government-produced** official statistics in the 2021+ window. **Removed for being pre-2018:** - **Lebanon** — MICS 2011 (Palestinian refugees), all 6 computed indicators — dropped entirely. - **Iraq** — SRHCS 2015 (Syrian, estimated) — dropped entirely. - **Jordan 2.1.2** — SRHCS 2015 (estimated) — dropped; Jordan keeps only 8.5.2 (DOS LFS 2024). - **Somalia 6.1.1 / 6.2.1** — SHFS-W1 2016 — dropped; Somalia keeps its 2022 JDC Displacement Phone Survey values (1.2.1, 6.2.1, 11.1.1). Nigeria 2018 IDP survey (school 70%) retained (2018 = IRRS adoption year). All other map values are 2021+ except a few 2018-2020 (Bangladesh CBPS 2020, Nigeria 2018). ## 11. Alignment to EGRISS priority SDG indicators (Paper 3, Oct 2024) Indicators now framed against **EGRISS Methodological Paper 3 — "Capturing Priority SDG Indicators in Refugee, IDP and Statelessness Contexts"**: **14 priority indicators** (12 IAEG-prioritised + 2 statelessness): 1.2.1, 1.4.2, 2.2.1, 3.1.2, 4.1.1, 5.1.1, 6.1.1, 7.1.1, 8.3.1, 8.5.2, 10.3.1, 11.1.1, 16.1.4, 16.9.1. - **On the map & priority (6):** 1.2.1, 6.1.1, 7.1.1, 8.5.2, 11.1.1, 16.9.1 → marked with ◆. - **On the map but NOT the priority set (3):** 2.1.2 (priority uses 2.2.1 stunting), 4.1.2 (priority uses 4.1.1 proficiency), 6.2.1 (only 6.1.1 water is priority). Kept as supplementary. - **Priority indicators still MISSING (8):** 1.4.2 land tenure, 2.2.1 stunting, 3.1.2 skilled birth attendance, 4.1.1 learning proficiency, 5.1.1 gender frameworks, 8.3.1 informal employment, 10.3.1 discrimination, 16.1.4 feeling safe — the target set for future computation. Indicator names updated to official/EGRISS titles; official UN SDG icons embedded (data-URI) as the card badges. ## §12 — SDG 6.1.1 drinking water: two definitions on the map SDG 6.1.1 is **safely managed** drinking water: an improved source, accessible on premises, available when needed, and **free from contamination (E. coli test)**. Where a survey publishes that full definition — MICS with water-quality testing, e.g. Lebanon MICS 2023 (Syrian settlements 7%, Palestinian camps 8%) — the map shows the safely-managed value. Everywhere else the value is **computed from microdata that cannot test water quality** (UNHCR FDS, DHS, World Bank household surveys). There we use the **basic-water proxy** — an improved source within a ≤30-minute round trip — which is systematically higher than safely-managed. On the map these values carry an **asterisk (\*)** and the card's source line reads "basic drinking water." A basic-water 6.1.1 (e.g. Morocco 99%) is therefore **not directly comparable** to a safely-managed 6.1.1 (Lebanon 7%).