ITTI Observatory

Documentation & References

Data sources, methodology, and how to interpret the Observatory's ETTI and GTBI figures.

How to Use the Observatory

The Observatory has two tabs: the International Trauma Observatory, a query tool for assembling your own country/year data and charts across ETTI and GTBI, and the Nigeria Trauma Observatory (NTO), a fixed worked example built the same way. This guide walks through both.

1. Choose a tab and an indicator

At the top of the page, pick International Trauma Observatory. Below that, pick ETTI or GTBI - the query tool works one indicator at a time, though panels from both can be combined into a chart later.

  • Each indicator has its own set of countries, years, and variables.
  • Switching between ETTI and GTBI doesn't clear panels you've already added to the other one.

2. Add data panels

Click + Select Data, search for a country, check one or more years, then click Add to selected data to confirm. Each country/year becomes its own panel with a colored dot and border so it's easy to spot later.

  • You can check multiple years for the same country in one pass through the search window.
  • A panel shows every recorded variable for that country/year, not just one.
  • Panels persist as you switch between the ETTI and GTBI tabs.

3. Check panels to chart

Each panel has a checkbox in its top-left corner - check the ones you want to chart. You can mix panels from ETTI and GTBI. Use Select all / Deselect all to toggle every panel for the current indicator at once.

  • Checking a panel doesn't remove or hide it - it just marks it as chartable.
  • Panels you leave unchecked stay visible but are left out of the chart, table, and stats views.

4. Build a chart

In the chart builder bar below the panels, pick a chart type and variable from the dropdowns, then click Create chart. For bar, line, and radar charts with more than one country selected, a Split by country checkbox appears. Once a chart appears, use Download to save it, or Save to profile if logged in.

  • Chart type options include bar, line, pie, radar, scatter, and stacked domain/exposure-type breakdowns where available.
  • Stacked domain charts use EVS/TIE/PDL/ITS; stacked exposure charts use each GTBI exposure type's YLL - no variable dropdown needed for either.
  • Split by country breaks a single chart into small multiples, one per country.
  • Download saves an SVG; Save to profile requires being logged in.

5. Switch views

Above the chart builder sit five tabs - Chart, Map, Table, Stats, Timeline - all reading the same checked panels.

  • Switching views never changes or clears which panels are checked.
  • Map uses whichever indicator tab (ETTI or GTBI) is currently open, regardless of what's checked.

6. Try the Nigeria example

Click the Nigeria Trauma Observatory (NTO) tab at the top for a ready-made set of Nigeria charts and maps - no setup needed.

  • Includes every recorded ETTI and GTBI year for Nigeria as read-only panels.
  • Charts are pre-built and organized into Trends, Comparisons, and Breakdowns.
  • Meant as a working example of what the International tab's query tool can produce.

7. Check the references

Every section of this Docs page documents the data sources, formulas, and known gaps behind the figures you'll see in the Observatory, including the full APA-formatted reference list behind the Nigeria stressor map. If a value reads "Data Pending," it means the underlying source doesn't yet have a usable number for that field - not that it's zero.

Shared Data Source

Both ETTI and GTBI are built substantially from the same underlying event-level data source: the Armed Conflict Location & Event Data Project (ACLED), supplemented for GTBI by UCDP, GBD 2019, WHO GHE, UNHCR, Our World in Data, the Global Terrorism Index, PRIO, CFR's Global Conflict Tracker, and OCHA (see the full GTBI source table below).

ETTI — Election Trauma Temperature Index

ETTI (Election Trauma Temperature Index) is a comparative measure of election-related trauma intensity - the broader stress environment surrounding an election, not whether it was free and fair or how democratic the country is more broadly. It combines four normalized (0-100) domains: EVS, TIE, PDL, and ITS. Higher scores mean a heavier combined trauma burden across those domains in that election's coverage window; lower scores mean relatively less cumulative trauma. Read scores comparatively, not as fixed thresholds - a moderate ETTI score is not necessarily "safe," it just indicates a lower combined burden than the highest-scoring elections in this dataset. ETTI should not be used as a proxy for election integrity, democratic quality, or regime type.

The finalized panel covers 30 country-election observations across 15 countries, with each entry built from a roughly two-year election-coverage window (January 1 of the year before the election through December 31 of the election year), using ACLED-derived event categories converted into normalized 0-100 component scores.

Several fields that would strengthen the model are absent or only partially observed in the current ACLED-derived data: injuries in EVS, the search/media distress signal in PDL, and the court-challenge/legal-dispute fields in ITS. These show as "Data Pending" throughout the Observatory. This is a gap in the underlying evidence base, not the extraction pipeline - once real numbers are available upstream, re-running the extraction script picks them up automatically with no other changes needed.

EVS — Election Violence Severity

Captures the direct harmful burden associated with election-linked conflict events: fatalities, kidnappings/abductions, and attack or incident counts, normalized to 0-100. It's the clearest representation of overt coercive violence in the election cycle. Injuries are not operationalized - ACLED doesn't provide injury data, so this component is acknowledged as unavailable rather than estimated.

EVS_raw = 0.40·D + 0.20·K + 0.15·A
  • D — Deaths — total fatalities recorded across all events in the election's coverage window.
  • K — Kidnappings — count of abduction / forced-disappearance events.
  • A — Attacks/Incidents — total event count, excluding Protests and Riots.

Injuries (I) appear in the source workbook's weighting scheme but are not currently populated - ACLED has no injury field, so this term is omitted from the raw score rather than estimated. EVS_raw is then min-max normalized to 0-100 across all country-elections in the panel: EVS = (EVS_raw − min(EVS_raw)) / (max(EVS_raw) − min(EVS_raw)) × 100.

TIE — Threat & Intimidation Environment

Captures coercion that may not register as mass lethality but still shapes the electoral atmosphere: political intimidation, threats to candidates/press/voters, harassment, and politically-motivated arrests or detentions, mapped from ACLED-coded event categories to reflect wider patterns of fear, pressure, and disruption around the vote.

TIE_raw = 0.35·P + 0.30·T + 0.20·H + 0.15·R
  • P — Political Intimidation — violence against civilians (attacks, sexual violence) and armed clashes/battles.
  • T — Threats to Candidates/Press/Voters — violent demonstrations, mob violence, and explosions/remote violence.
  • H — Harassment — protests met with intervention or excessive force, strategic developments, and looting/property destruction.
  • R — Politically-Motivated Arrests/Detentions — arrests, disrupted weapons use, and changes to group/activity status.

TIE_raw is then min-max normalized to 0-100 across the panel: TIE = (TIE_raw − min(TIE_raw)) / (max(TIE_raw) − min(TIE_raw)) × 100.

PDL — Psychological Distress Load

Reflects the broader societal burden generated by repeated disruptive events, proxied through a societal-distress measure built from total ACLED events in the coverage window. A search/media distress signal that would strengthen this component remains unpopulated - it would require Google Trends or comparable media data that isn't available from ACLED.

PDL_raw = 0.40·S
  • S — Societal Distress Composite — total ACLED event count in the election's coverage window (estimated proxy).
  • G — Search/Media Distress Signal — left blank for every row; would require Google Trends or comparable media data not available from ACLED.

The workbook's design reserves a second weighted term for G (Search/Media Distress Signal), but since G is entirely unpopulated, PDL_raw currently reduces to the S term alone. PDL_raw is then min-max normalized to 0-100 across the panel: PDL = (PDL_raw − min(PDL_raw)) / (max(PDL_raw) − min(PDL_raw)) × 100.

ITS — Institutional Trauma Score

Reflects the extent to which an election cycle strains civic order and public institutions, relying chiefly on protests and security interventions. Court challenges and legal disputes are marked unavailable in the current data, so ITS should be read as a partial institutional-stress indicator, not a complete legal-institutional accounting.

ITS_raw = 0.20·P + 0.20·S
  • C — Court Challenges — not available in current data (no ACLED equivalent); term omitted.
  • L — Legal Disputes — not available in current data (no ACLED equivalent); term omitted.
  • P — Protests — peaceful-protest event count.
  • S — Security Interventions — protests met with intervention, arrests, attacks, and violent demonstrations.

The workbook's design reserves weighted terms for C (Court Challenges) and L (Legal Disputes), but both are unpopulated for every row, so ITS_raw currently reduces to the P and S terms alone. ITS_raw is then min-max normalized to 0-100 across the panel: ITS = (ITS_raw − min(ITS_raw)) / (max(ITS_raw) − min(ITS_raw)) × 100.

ETTI Composite Formula

ETTI = 0.30·EVS + 0.20·TIE + 0.30·PDL + 0.20·ITS
  • EVS — Election Violence Severity (0-100, see above) — weighted 30%.
  • TIE — Threat & Intimidation Environment (0-100, see above) — weighted 20%.
  • PDL — Psychological Distress Load (0-100, see above) — weighted 30%.
  • ITS — Institutional Trauma Score (0-100, see above) — weighted 20%.

All four domain scores are already normalized to a common 0-100 scale before this step, so ETTI is a direct weighted average of them rather than a further-normalized composite. EVS and PDL carry the heaviest weight (30% each), reflecting direct violence and societal distress as the two dominant contributors to election-related trauma load, with TIE and ITS each contributing 20%.

GTBI — Global Trauma Burden Index

GTBI (Global Trauma Burden Index) is a comparative, population-level index showing how heavily a country-year is burdened by trauma-related exposures and consequences - armed conflict, communal violence, terrorism, political repression, forced displacement, disaster-related trauma, and related public-health shocks - not a clinical diagnosis or an exact PTSD-prevalence estimate. Higher values signal a stronger accumulated burden, especially where several severe or persistent exposure channels coexist in the same country-year; lower values suggest a lighter relative burden, not an absence of trauma or risk. Treat it as a normalized comparative scale - focus on broad differences, trends over time, and clustering patterns rather than over-reading small gaps between nearby scores, and don't sum the underlying per-exposure rows yourself; the GTBI field is the intended, already-synthesized comparative signal.

The panel currently covers country-year observations from 2015 to 2025 across 14 countries, built from multiple exposure-specific rows per country-year (armed conflict, communal violence, terrorism, political repression, forced displacement, disaster-related trauma) rather than a single flat number. Coverage grows as new country data is added upstream, so exact counts here may lag behind the latest data.

As with most cross-national trauma datasets, data quality isn't uniform across countries, years, or exposure categories - some contexts are backed by stronger empirical reporting, others rely more on structured estimates and harmonized assumptions. The workbook also still contains missing or invalid entries in some exposure-level fields (blank cells and formula-error placeholders), so GTBI represents a best-available structured estimate, not a finalized, error-free census of all trauma burden everywhere.

Formula

  • YLD (Years Lived with Disability) = Incidents × Disability Weight × Duration, per exposure type.
  • YLL (Years of Life Lost) = Deaths × 73.8 (standard life expectancy), per exposure type.
  • TBU (Trauma Burden Unit), per country-year = ΣYLL + ΣYLD + Σ(Severity Weight × Exposure % / 100) across every exposure type recorded for that country-year.
  • Burden Rate = (TBU / Population) × 100,000.
  • GTBI = 100 × (1 − e^(−Burden Rate / 100)), so the composite score saturates toward 100 as burden rate grows.
  • Trauma Level bands (Low / Moderate / High / Severe) are applied to the 0–100 GTBI score; the source workbook grades individual exposure-type events on its own severity scale, but doesn't define country-year-level bands, so these thresholds (<25 / <50 / <75 / ≥75) are the Observatory's own convention pending confirmation from the data owner.

GTBI Sources

SourceCitationUsed forNotes
ACLED 2025Armed Conflict Location & Event Data Project (ACLED). (2025). ACLED Data Export Tool [Data set]. ACLED. https://acleddata.comIncidents (I); Deaths (S); Event Type; Primary SourcePrimary source for all conflict metrics. Filtered by country, year, and event type. Fatalities are summed for Deaths (S); events are counted for Incidents (I).
UCDP BRD v24Uppsala Conflict Data Program (UCDP). (2024). UCDP Battle-Related Deaths Dataset v24.1 [Data set]. Uppsala University. https://ucdp.uu.se/downloads/Deaths (S), when ACLED is incompleteUsed as the secondary source for battle deaths where ACLED coverage has gaps.
GBD 2019GBD 2019 Diseases and Injuries Collaborators. (2020). Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019. The Lancet, 396(10258), 1204-1222. https://doi.org/10.1016/S0140-6736(20)30925-9Disability Weight (DW); Duration (Ls); YLD parametersDisability weights and duration-of-disability figures used in the YLD calculation are drawn from this study.
WHO GHE 2024World Health Organization. (2024). Global Health Estimates 2024 [Data set]. WHO. https://www.who.int/data/global-health-estimatesMortality cross-check; disability burdenUsed to cross-validate YLL estimates. WHO GHE coverage runs through 2021.
UNHCR 2024United Nations High Commissioner for Refugees (UNHCR). (2024). UNHCR Refugee Population Statistics Database [Data set]. UNHCR. https://www.unhcr.org/refugee-statisticsEstimated Exposed Prevalence % for forced displacementAbsolute displacement figures are converted to a percentage of population for the exposure column.
OWID 2024Roser, M., Nagdy, M., & Ritchie, H. (2024). War and peace. Our World in Data. https://ourworldindata.org/war-and-peacePopulation; conflict death series cross-checkPopulation figures (drawn from the UN World Population Prospects via Our World in Data) are the denominator for the burden-rate calculation.
GTI 2025Institute for Economics and Peace. (2025). Global Terrorism Index 2025. IEP. https://www.economicsandpeace.org/research/gti/Terrorism incident counts; deathsUsed for terrorism-type rows where ACLED's terrorism classification is incomplete.
PRIO 2024Peace Research Institute Oslo (PRIO). (2024). PRIO Conflict Site Dataset [Data set]. PRIO. https://www.prio.org/data/Conflict recurrence classificationInforms the temporal-decay parameter and conflict onset vs. recurrence coding.
CFR GCT 2025Council on Foreign Relations. (2025). Global Conflict Tracker [Online database]. CFR. https://www.cfr.org/global-conflict-trackerConflict status verification; key event descriptionsCross-referenced for active-conflict classification. Not a primary numerical source.
OCHA 2024United Nations Office for the Coordination of Humanitarian Affairs (OCHA). (2024). Humanitarian Data Exchange [Data set]. OCHA. https://data.humdata.orgDisplacement, casualty, and exposure estimates for acute humanitarian crisesCross-national humanitarian data source for acute-crisis country-years.
WHO Global TB ReportWorld Health Organization. (2024). Global tuberculosis report 2024. WHO. https://www.who.int/publications/i/item/9789240101531TB incidence and mortality (Public Health Threats tab)Used for Indonesia and Thailand TB-burden country-years.
WHO Malaria ReportWorld Health Organization. (2024). World malaria report 2024. WHO. https://www.who.int/publications/i/item/9789240104440Malaria incidence and mortality (Public Health Threats tab)Used for Congo, Ghana, and Kenya malaria-burden country-years.
UNAIDSJoint United Nations Programme on HIV/AIDS. (2024). UNAIDS global AIDS update 2024. UNAIDS. https://www.unaids.org/en/resources/documents/2024/global-aids-update-2024HIV/AIDS prevalence and mortality (Public Health Threats tab)Used for Kenya HIV-burden country-years.
PAHO/WHOPan American Health Organization. (2024). PAHO/WHO Health Information Platform [Data set]. PAHO. https://www.paho.orgDengue, Zika, and regional infectious-disease data for the AmericasUsed for Brazil and Mexico dengue/Zika country-years.

NTO — Nigeria Geographic Stressor Severity Map

Severity is scored 0-10 per geopolitical zone for 2020-2026, alongside each zone's dominant stressor categories: banditry, mass kidnappings, and school raids in the North West; the Boko Haram/ISWAP insurgency and mass displacement in the North East; farmer-herder conflict and sectarian violence in the North Central; separatist violence and sit-at-home enforcement in the South East; oil pollution and environmental degradation in the South South; and cultism, gang violence, and urban economic strain in the South West.

Map by Blessing Alims. Published July 24, 2026.

References

SourceCitation
Nigeria Trauma Observatory 2026Nigeria Trauma Observatory. (2026). Geographic stressor severity report: Regional mapping of dominant stressors across Nigeria's six geopolitical zones (Technical Working Paper No. 3). Nigeria Trauma Observatory.
ACLED 2025 (Nigeria)Armed Conflict Location & Event Data Project. (2025). ACLED data export tool: Nigeria conflict and protest events (2020-2026) [Data set]. ACLED. https://acleddata.com
GBD 2019GBD 2019 Diseases and Injuries Collaborators. (2020). Global burden of 369 diseases and injuries in 204 countries and territories, 1990-2019: A systematic analysis for the Global Burden of Disease Study 2019. The Lancet, 396(10258), 1204-1222. https://doi.org/10.1016/S0140-6736(20)30925-9
UNHCR 2024 (Nigeria)United Nations High Commissioner for Refugees. (2024). UNHCR refugee and IDP population statistics database: Nigeria country report [Data set]. UNHCR. https://www.unhcr.org/refugee-statistics
References follow APA 7th edition: a 0.5-inch hanging indent, alphabetical ordering by author/organization, double line spacing in 12pt Times New Roman in the source document, and italics applied to standalone report/dataset and journal titles per APA guidelines.

Missing-Value Convention

Any variable without a usable number is recorded as "Data Pending" rather than a null or a numeric placeholder like -1, so it's never mistaken for a real value in a chart or export. A country with no recorded years at all for an indicator still has that indicator's section present, with a single "Data Pending" year.