Construct Browser

Crop output per unit of harvested area (tonnes per hectare).

Disposition toward deliberative, reflective cognitive processing (System 2) versus intuitive processing (System 1). Measured via Cognitive Reflection Test (CRT). Consistently predicts better discernment between true and false news headlines, independent of political ideology.

Binary indicator: 1 if Polity IV score is between -5 and +5, indicating a mixed or incoherent regime type. Anocracies are predicted to have higher conflict risk than either full democracies or full autocracies (inverted-U).

Mixed regime type (Polity score approximately -5 to +5) that combines elements of democracy and autocracy. Anocracies face elevated coup risk compared to both full democracies and consolidated autocracies — the 'inverted U' hypothesis. Partial liberalization creates political competition without establishing strong civilian control institutions.

Age of the loan applicant in years

The hypothesized phenomenon where corrections strengthen rather than weaken false beliefs. Originally reported by Nyhan & Reifler (2010) but largely failed to replicate in subsequent studies. Now considered rare or nonexistent under standard conditions — corrections generally work, they just do not work completely.

Property that predicted probabilities match observed frequencies of the outcome, meaning that among all individuals assigned a predicted risk of X%, approximately X% actually experience the outcome.

Ratio of the potential supporter's military capabilities to the NAG's target's capabilities. Computed as average of shares of military expenditures and military personnel (CAPRAT). Values < 1 indicate weaker supporter; > 1 indicate stronger. Weak dissatisfied states ('frustration games') and capable constrained states ('opportunity games') both increase NAG support probability.

Ratio of the stronger state's CINC score to the sum of both states' CINC scores, capturing power asymmetry.

Binary indicator: 1 if a civil war began in this country-year, 0 otherwise. Fearon & Laitin define civil wars as internal conflicts with organized violence, at least 1,000 battle deaths, and effective resistance by both sides.

Count of civilian fatalities from one-sided violence perpetrated by rebel organizations (UCDP One-Sided Violence dataset). Operationalized as 'best estimate' of annual civilian deaths (rebbest). Overdispersed count variable modeled with negative binomial regression. Key dependent variable for testing how external sponsorship affects rebel restraint of violence against noncombatants.

Share of GDP lost due to climate change impacts, from integrated assessment damage functions.

Extreme weather events (drought, flood, cyclone, heat wave) and climate variability affecting food production and access. Operationalized via NDVI anomalies, rainfall deficits (CHIRPS data), temperature extremes, or crop damage assessments. Primary driver of food crises in pastoral and rain-fed agricultural regions. Climate shocks interact with conflict and poverty to compound food insecurity.

Variance in problem-solving approaches, mental models, and heuristics within a group. Operationalized via diverse cognitive toolkit measures or functional background diversity. Distinct from demographic diversity — cognitive diversity predicts group performance; demographic diversity does not after controlling for cognitive diversity.

Composite of reaction time (PVT), working memory (n-back), executive function (Stroop), and decision quality measured after controlled sleep restriction. Performance declines are cumulative: 6h/night for 14 days produces impairment equivalent to 48 hours total sleep deprivation. Crucially, subjective sleepiness plateaus after ~3 days while objective impairment continues to worsen.

National average score on internationally comparable standardized tests (PISA, TIMSS). Captures educational quality.

Binary indicator for the Cold War period (pre-1991). Coup rates were substantially higher during the Cold War when superpower competition created permissive environments for military intervention. Post-Cold War democratic norms, international organizations, and conditional aid reduced the international tolerance for coups. Structural break in coup patterns around 1991.

A single statistical factor explaining 30-50% of variance in group performance across diverse tasks — analogous to 'g' for individuals. Measured via group performance battery (Brainstorming, Typing, Matrix Reasoning, Moral Judgments). NOT correlated with average or maximum member IQ.

Direct military intervention by a sponsor state in a civil conflict, involving deployment of troops to fight alongside rebel forces against the target government. Represents the highest level of sponsor involvement and provides direct control over the conflict process. Coded 2 in the NAGs support typology (vs. 1 for material/logistical support, 0 for no support). Predicted probability: ~2% for non-personalist targets, ~6.6% for personalist targets (230% increase).

Mean one-way commute duration in minutes for metro area workers. The primary time-cost of agglomeration. US average ~27 minutes; NYC ~40 minutes. Increases approximately 8 minutes per doubling of metro population. ACS and OECD provide annual estimates.

Armed conflict as a primary driver of acute food insecurity. Conflict disrupts food production, destroys agricultural assets, displaces populations, blocks humanitarian access, and collapses markets. Conflict-affected countries account for the majority of people in IPC Phase 3+ (Crisis or worse). The conflict-hunger nexus is bidirectional: food insecurity can also fuel conflict.

The tendency for previously encoded misinformation to continue shaping reasoning and inference even after an effective correction has been encoded and accepted. Measured via inference questions about a scenario where causal misinformation was retracted.

The extent to which public power is NOT exercised for private gain, including petty and grand corruption.

The speed at which poorer economies catch up to richer ones in terms of per-capita income, conditional on structural characteristics.

Evenness of speaking time distribution within a group. Measured as 1 - Gini coefficient of turn counts. Groups where one member dominates discussion show lower collective intelligence, even if that member is the smartest.

Reduced Coping Strategy Index: a rapid household-level measure of food insecurity severity based on frequency of five universal coping behaviors (eating less preferred foods, borrowing food, reducing portion sizes, restricting adult consumption, reducing number of meals). Higher scores indicate worse food insecurity. Used by WFP alongside FCS for real-time monitoring.

Degree to which a correction reduces reliance on misinformation in subsequent reasoning. Measured as proportion of misinformation-consistent inferences eliminated. Corrections typically reduce but do not eliminate CIE — residual influence remains even after strong corrections.

Strategies adopted by authoritarian leaders to neutralize military threats to their rule: creating parallel security forces with conflicting jurisdictions, frequent purges of senior officers, loyalty-based promotions over competence, restricting inter-unit coordination, and appointing ethnic/family loyalists to command positions. Effective at reducing coup risk but systematically degrades military cohesion and battlefield effectiveness.

Binary indicator: an illegal and overt attempt by the military or other elites within the state apparatus to unseat the sitting executive. Following Powell & Thyne (2011), includes both successful and failed attempts. Excludes popular revolutions, civil wars, and foreign invasions unless led by domestic military/elite actors.

Regional diffusion or demonstration effects whereby coups in neighboring or culturally similar states increase coup risk domestically. Mechanisms include: (1) demonstration that coups are feasible, (2) diffusion of coup-facilitating networks among military officers trained together, (3) regional instability creating permissive international environments. Operationalized as count of coups in neighboring states or regional coup rate.

Model-generated probability of a coup attempt occurring in a given country-year. Produced by statistical or machine learning models using structural, institutional, and contextual predictors. Examples include PITF models, ViEWS system (Hegre et al. 2019), and various logistic regression/random forest approaches in the literature.

Count or recency of prior coup attempts in a country. The single strongest predictor of future coups — the 'coup trap' dynamic where initial coups lower the normative and practical barriers to subsequent attempts. Countries with recent coup histories face 3-5x higher coup risk than countries without. Operationalized as binary (any prior coup), count, or years since last coup.

Binary indicator: a coup attempt that successfully results in the displacement of the incumbent chief executive and seizure of executive power for at least 7 days. Subset of coup_attempt. Success rate globally ~50% across 1950-2010 period.

Total loan amount in Deutsche Marks or equivalent

Binary indicator of whether a loan applicant defaulted on their credit obligation

Inter-annual variation in crop yields driven by climate variability, extreme weather, pest/disease pressure, and input availability. Climate variation explains roughly one-third of global crop yield variability (Ray et al. 2015). Key transmission mechanism from climate shocks to food insecurity, especially in rain-fed agriculture systems.

Points or runs allowed per possession or opportunity, measuring a team's ability to prevent opponent scoring

A mode of rebel sponsorship where the sponsor provides troops, training, sanctuaries, or access to military infrastructure, enabling hierarchical monitoring and sanctioning of rebel compliance. Sufficient support types: troops (secondary warring party), training/expertise, access to territory, access to military infrastructure. Higher visibility increases domestic and international accountability costs.

Degree to which political leaders are chosen through free elections with civil liberties protection. Polity2, V-Dem, Freedom House.

Fairness criterion requiring equal positive prediction rates across demographic groups, ensuring that the proportion of individuals receiving a positive classification is independent of group membership.

Per capita daily caloric availability (kcal/capita/day) derived from FAO food balance sheets. Measures national food supply adequacy — the total food available for consumption after accounting for production, imports, exports, stock changes, and non-food uses. Does not capture distribution within country or household-level access.

Number of diplomatic representations (embassies, consulates, and other missions) a state maintains abroad in a given year. Captures the logistical infrastructure enabling a state to project repression transnationally. Source: Diplometrics Diplomatic Representation dataset (Moyer et al. 2021), 1960-2020. Mean in sample: 50.8, range: 1-170.

External support involving deployment of foreign troops for combat operations on behalf of a recipient party. Coded separately from indirect support (weapons, training, funding, intelligence, logistics). Direct intervention first exceeded indirect support in 2015 and continues to grow. Conceptually distinct from delegation/orchestration framing by requiring actual combat role.

Inverted V-Dem Civil Liberties Index (CLI), measuring the intensity of domestic state repression. The CLI aggregates three component indices: physical violence index (torture, killings), political civil liberties index (censorship, parties, civil society), and private civil liberties index (forced labor, property rights, religion). Higher values indicate more repression. Lagged one year in analysis. Source: V-Dem v12, Coppedge et al. 2022.

The Polity IV score of the less democratic state in a dyad. Captures the binding constraint: a dyad is only jointly democratic if both members clear the threshold.

Short-term negative deviation in GDP growth rate, typically measured as annual real GDP growth or growth relative to trend. Sudden economic downturns increase coup risk by creating popular discontent, weakening regime legitimacy, and reducing the regime's ability to buy military loyalty through patronage.

Highest level of formal schooling completed, operationalized as years of schooling or credential level.

Skill rating of a team based on results against rated opponents, updated after each game using a transfer function

Binary dummy indicating whether the ethnic group a rebel group claims to fight for is also politically relevant in the sponsor state. Matched using Ethnic Power Relations (EPR) data, ACD2EPR, and TEK data. Coded 1 when co-ethnicity exists. Strongly increases probability of orchestration by substituting goal alignment for hierarchical control instruments.

The degree to which the ethnic composition of the military (especially officer corps) diverges from the general population. Ethnic homogeneity in the military (via ethnic stacking) may increase loyalty to the regime but creates grievances among excluded groups that can fuel civil war (Roessler 2011, Harkness 2014).

The deliberate appointment of co-ethnics or loyalists to key command positions in the military and security forces. A form of coup-proofing that increases short-term regime security by ensuring military leaders share identity-based loyalty with the ruler, but creates long-term instability by generating grievances among excluded ethnic groups and degrading meritocratic military performance (Harkness 2014, Roessler 2011).

Binary indicator: 1 if event occurred, 0 if censored

The mechanism by which external state funding reduces rebel incentives to maintain civilian cooperation. Rebels relying on external resources no longer need to extract resources from the population through taxation or civilian support networks, reducing the cost-benefit calculus that otherwise deters mass atrocities. Creates agency slack in the principal-agent relationship between sponsor and rebel.

The share of armed conflict-dyads or intrastate conflicts that receive at least one form of external support from state or non-state actors in a given time period. ESD documents: 80% of all intrastate conflicts (1975-2017) received external support; 72% of interstate conflicts. External support is the rule, not the exception, in modern armed conflict.

Rate of incorrect positive predictions, measuring how often a classifier incorrectly labels negative instances as positive, potentially causing harm through false accusations or unnecessary interventions.

Percentage of working-age women (15-64) who are employed or actively seeking employment. ILO modeled estimates. Global average ~47% vs male ~72%. Shows strong non-linear relationship with GDP — U-shaped curve (high in low-income agriculture, dips in middle-income industrialization, rises again in high-income service economies).

WFP's standard proxy indicator for household food access and dietary quality. Composite score based on dietary diversity, food frequency, and relative nutritional importance of food groups consumed over the past 7 days. Thresholds: poor (<21), borderline (21.5-35), acceptable (>35). Collected in WFP vulnerability assessments and used in HungerMap LIVE nowcasting.

Share of domestic food supply derived from imports, typically measured as cereal import dependency ratio. Countries highly dependent on food imports are vulnerable to global price shocks, supply chain disruptions, and currency depreciation. Small island developing states and conflict-affected countries often have extreme import dependency.

Experience-based metric (SDG 2.1.2) measuring severity of food insecurity through 8 yes/no questions about behaviors and experiences (worry about food, skipping meals, going hungry, going a whole day without eating). Collected via Gallup World Poll in 140+ countries since 2014. Yields prevalence of moderate and severe food insecurity. Advantage over PoU: captures lived experience, not just caloric supply.

Rate of increase in food prices at market or national level. Rapid food price spikes erode purchasing power of poor households, directly increasing food insecurity. Monitored by WFP (ALPS indicator), FAO Food Price Index, and FEWS NET market data. Particularly impactful in countries with high food import dependency and low social protection coverage.

External military assistance (arms transfers, training, financial support) provided to a state's armed forces. Theoretically ambiguous effect on coup risk: may professionalize the military and increase civilian control, but may also empower the military and create dependencies that make aid withdrawal destabilizing. Cold War-era aid was often coup-permissive.

Share of respondents answering 'yes' to 'Are you satisfied or dissatisfied with your freedom to choose what you do with your life?' Ranges 0-1.

A rivalry game type in which the focal state is dissatisfied with the status quo (SATISFLEV = -1) but lacks the capabilities to challenge it directly (CAPRAT < 1). Supporting NAGs targeting the rival serves as a low-cost substitution strategy to harass the rival and potentially weaken its capabilities without risking direct retaliation.

Estimated annual GDP loss from workforce sleep deprivation through three channels: mortality (shorter lifespan), absenteeism (more sick days), and presenteeism (reduced on-the-job productivity). Hafner et al. (2017) estimate: US loses $411B/year (2.28% GDP), Japan $138B (2.92% GDP), UK $50B (1.86% GDP). Calculated via human capital approach.

Total economic output divided by population, measured in constant purchasing power parity dollars. The standard measure of average material living standards across countries.

Logged real GDP per capita as a structural predictor of coup risk. Lower income levels are robustly associated with higher coup probability — the 'coup trap' mechanism where poverty creates grievances and lowers opportunity costs of plotting. One of the most robust predictors across studies (Londregan & Poole 1990, Gassebner et al. 2016).

Annual percentage change in real GDP per capita. Growth-rate effects compound; level effects do not.

The counterintuitive finding that countries with higher gender equality show LARGER gender differences in STEM field choice and personality traits. Proposed mechanism: economic security in egalitarian countries allows intrinsic preferences to drive career choice, while economic pressure in unequal countries drives women toward high-paying STEM regardless of preference. Contested — may partly reflect measurement artifacts.

WEF composite index (0-1) averaging four sub-indices: Economic Participation & Opportunity, Educational Attainment, Health & Survival, Political Empowerment. Each sub-index normalizes to the female-to-male ratio. 1.0 = full parity. Published annually since 2006 for 146 countries.

Male-female wage differential, measured as raw gap or residual gap after controlling for observable characteristics.

Residual of regressing national average response to 'Have you donated money to a charity in the past month?' on log GDP per capita. Captures pro-social behavior beyond what income explains.

Composite index (0-100 scale) combining four indicators: prevalence of undernourishment (FAO), child wasting (WHO), child stunting (WHO), and child mortality (UN IGME). Published annually by Welthungerhilfe and Concern Worldwide for ~130 countries. Provides a single summary metric of hunger severity at country level.

Quality of public services, civil service, and credibility of government commitment to policies. WGI Government Effectiveness.

The trade-off faced by indirect governance principals choosing between hierarchical control and agent independence benefits. Delegation ensures compliance but forfeits rebels' local legitimacy and plausible deniability; orchestration captures independence benefits but sacrifices compliance guarantees. Borrowed from indirect governance theory (Abbott et al. 2016, 2020a, 2020b) and applied to state-rebel sponsor relationships.

Total health spending per person in purchasing power parity dollars, capturing aggregate resource allocation to health systems.

Healthy life expectancy at birth (years), from WHO/GHO data. Captures both longevity and health-adjusted quality of life.

Annual pre-tax income received by all members of a household, inflation-adjusted. Primary measure of material resources for health-relevant consumption.

Ratio of median housing costs (rent or mortgage) to median household income in a metro area. The primary congestion cost of agglomeration. US metros range from 20% (Sun Belt) to 50%+ (San Francisco, New York). When housing costs are subtracted, the real urban wage premium shrinks or vanishes for many workers.

Count of human rights organization secretariats in the sponsor state (hrosecretariat_sum). States with stronger human rights advocacy sectors face greater domestic pressure to condition foreign support on rebels' conduct toward civilians. Operationalizes the 'principled' dimension of sponsor oversight beyond formal democratic institutions.

The stock of knowledge, skills, and health embodied in the labor force. Accumulated through education, training, and health investment.

Repeated exposure to a statement increases its perceived truth, independent of actual veracity. Processing fluency from repetition is misattributed to truth. Effect size typically d=0.3-0.5 per additional exposure.

Degree of dispersion in income distribution, most commonly measured by the Gini coefficient (0-1 scale).

Deaths per 1000 live births in the first year of life, a key indicator of child health and healthcare system performance.

Reduction in susceptibility to misinformation after exposure to weakened doses of manipulative arguments plus refutational preemption. Measured as the difference in belief change between inoculated and control groups when subsequently exposed to misinformation. Meta-analytic effect: d=0.29 (Banas & Rains, 2010).

The degree to which a leader faces meaningful checks from domestic political elites, military officers, or party structures that limit unilateral foreign policy decisions. High under institutionalized regimes (party-based, military junta, monarchy); minimal under personalist rule where other elites lack independent power bases to challenge the leader.

The degree to which the target state is excluded from international economic, diplomatic, and military networks. Personalist regimes tend toward greater isolation due to limited trade openness, fewer stable alliances, and repressive practices that draw international condemnation. Reduces the reputational cost for sponsors of directly intervening in the target's civil conflict.

A persistent, militarized dispute relationship between two states characterized by repeated conflicts, mutual threat perception, and ongoing competition. Operationalized as binary indicator from Goertz, Diehl & Balas (2016) peace and rivalry dataset. Strong positive predictor of both general rebel sponsorship emergence and combat support provision. Sponsoring rebels is an indirect, lower-cost strategy for rivals to weaken each other.

Integrated Food Security Phase Classification: a 5-phase scale (Minimal, Stressed, Crisis, Emergency, Famine) classifying the severity of acute food insecurity at subnational level. Used by IPC Global Partnership across 30+ countries. Based on convergence of evidence from food consumption, livelihood change, nutritional status, and mortality. The standard framework for humanitarian food crisis assessment and response planning.

The informal transfer of knowledge between co-located workers and firms through face-to-face interaction, labor mobility, and observation. Measured via patent citation localization, co-invention networks, or productivity growth after arrival of star scientists. Jaffe (1993): patent citations are 5-10x more likely to cite geographically proximate patents, controlling for technology class.

Log of the incumbent ruler's cumulative time in office (years). Controls for the possibility that crackdowns occur around regime consolidation periods and that repression spikes during transitions. Source: Bell, Besaw, and Frank (2021). Mean: 4.60 log-years.

Number of years the current chief executive has been in power. Non-linear relationship with coup risk: new leaders face high risk during consolidation period, risk declines as they establish control and patronage networks, but may rise again in very long tenures as succession anxieties emerge and loyalty networks calcify.

Average number of years a newborn is expected to live given current age-specific mortality rates.

Cantril Self-Anchoring Striving Scale: respondents rate their life on a ladder from 0 (worst possible life) to 10 (best possible life). National average. Primary outcome in WHR analyses.

Above grade living area in square feet

Deaths per 100,000 live births from pregnancy-related causes. WHO/UNICEF/UNFPA estimates. Ranges from <5 (Nordic, Japan) to >800 (Sub-Saharan Africa). The single largest gender-specific health inequality. Strongly correlated with skilled birth attendance and health expenditure.

Annual average surface temperature (degrees Celsius) for a country, from gridded meteorological datasets.

Average years of formal education completed by adults aged 25+. Primary macro proxy for human capital stock.

Binary indicator: 1 if a dyad experienced MID onset in a given year, per the COW MID dataset.

The battlefield capacity of state armed forces, including command coordination, tactical competence, and ability to generate intelligence and civilian cooperation. Systematically reduced under personalist regimes due to coup-proofing strategies. Acts as partial mediator between personalism and combat support: weaker target militaries lower the cost and risk of sponsor troop intervention.

The degree to which the armed forces operate independently from civilian political control. High military autonomy — separate intelligence apparatus, independent budget authority, professional promotion criteria, corporate identity — increases both the capability and motivation for military intervention in politics. Reduced by coup-proofing but at the cost of military effectiveness.

Military expenditure as a percentage of GDP or total government expenditure. Has theoretically ambiguous effects on coup risk: higher spending may buy military loyalty and reduce grievances, but also empowers the military as an institution and increases its capacity to act. Empirical evidence is mixed (Gassebner et al. 2016).

Total spending in observation period

Percentage of a country's territory that is mountainous, per the Geomorphic Units of the World dataset. Logged in regression models. Higher values increase insurgency viability by providing rebel sanctuary.

Three or more states coordinating their provision of external support to achieve a common goal. Coalition-based support was rare before 2001 (Gulf War Coalition 1990-91 is the first recorded instance) but grew dramatically after 9/11, reaching nearly one-third of all support instances by 2017. Implications for effective principal discipline over rebel recipients.

Count of distinct state actors providing external support to a rebel group (num_supp). Multiple principals create a collective action problem: no single sponsor can effectively monitor and discipline the rebel agent, each free-riding on others' oversight efforts. Higher values associated with more civilian abuse due to reduced effective monitoring.

The effect of state-NAG cooperation on the probability of direct conflict escalation between the supporter and the target. State-NAG support significantly increases target's MID initiation (coeff 1.441**, Table 4) and level of hostility against the supporter, making NAG sponsorship a strategy that ultimately intensifies rather than substitutes for direct confrontation.

Degree to which a daytime nap (10-30 minutes) restores cognitive performance after sleep restriction. A 26-minute nap improves pilot performance by 34% and alertness by 54% (NASA nap study). However, naps do not fully compensate for chronic sleep debt — they provide temporary relief, not recovery. Longer naps (>30 min) risk sleep inertia.

Binary indicator: 1 if the state became independent within the prior two years. New states face elevated conflict risk due to weak institutions, contested borders, and unresolved ethnic or regional claims.

An armed opposition organization or rebel group that provides external support to another warring party in a different conflict. ESD is the first global dataset to systematically track non-state actors as supporters (not just recipients). Non-state supporters peaked at 38 active groups in 2012 (including al-Shabaab training Boko Haram). More than a quarter of all support to rebel groups comes from non-state actors.

Binary indicator: 1 if the state has territory that is physically separated from the main territory (e.g., islands, exclaves). Noncontiguity creates zones where state capacity is lower and rebel control is easier to establish.

Points or runs scored per possession or opportunity, measuring a team's ability to convert opportunities into scoring

Binary indicator: 1 if fuel exports exceeded one-third of export revenues. Oil states may face conflict due to prize capture incentives, weaker tax-based state institutions, or Dutch disease effects on state capacity.

A rivalry game type in which the focal state is dissatisfied with the status quo (SATISFLEV = -1) and has the capability to change it (CAPRAT > 1) but is deterred from direct confrontation by political constraints. Supporting NAGs allows the capable state to pursue strategic goals while managing domestic or international accountability costs.

A mode of rebel sponsorship where the sponsor provides purely material, financial, intelligence, or logistical support without hierarchical control instruments. The sponsor cannot directly monitor or sanction rebel compliance but benefits from plausible deniability and rebels' local legitimacy. Coded 1 in the binary DV; contrasted with delegation (0). Necessary support types: weapons, materiel/logistics, funding, intelligence.

Overall material and finish quality of house on 1-10 scale

The creation of multiple, competing armed bodies (presidential guards, paramilitaries, intelligence agencies with arrest powers) with overlapping jurisdictions. A key coup-proofing mechanism that prevents any single military unit from accumulating enough power to execute a coup. Reduces coup risk but fragments the security apparatus and degrades conventional military effectiveness.

Patents per capita or per worker in a metro area. Proxy for localized innovation output. Correlation with population density: r~0.35. Highly concentrated: top 20 MSAs produce ~60% of US patents. OECD REGPAT database provides geocoded patent data back to 1977.

GDP per capita, typically logged and lagged one year. One of the strongest predictors of civil war onset: wealthier countries have lower risk, reflecting higher opportunity costs of rebellion and greater state capacity to deter it.

Average of two Gallup binary questions: is corruption widespread throughout government/business in this country? Ranges 0-1 (higher = more corruption perceived).

A regime type in which power is concentrated in a single leader who dominates both the military and party (if any), with weak formal institutions and no independent elite power bases. Operationalized as: (1) binary indicator from GWF typology (Geddes, Wright & Frantz 2014) distinguishing personalist from all other regime types including democracies; (2) continuous latent personalism index 0-1 (Geddes, Wright & Frantz 2018) covering 118 authoritarian regimes 1946-2010.

The stock of produced means of production including machinery, equipment, structures, and infrastructure. Accumulated through investment (savings) and depreciated over time.

Gross fixed capital formation as share of GDP. Key control in augmented Solow growth regressions.

The ability of a state sponsor to credibly deny involvement in supporting a rebel group. A key benefit of orchestration over delegation: lower visibility of material and financial support reduces responsibility attribution for rebel atrocities, international sanctions, and direct military retaliation by the target state. Eroded by visible troop deployments or training missions (delegation).

Binary indicator: 1 if there was a three-or-more-point change in Polity score in the previous three years. Captures regime transitions and political turbulence that may lower the cost of civil war initiation.

Revised combined Polity score measuring level of authoritarianism/democracy on a scale from -10 (full autocracy) to +10 (full democracy). Used as control for regime type. Source: Marshall and Gurr (2020). Mean in sample: -4.46.

The number of people per unit area (typically per km²), which affects disease transmission dynamics and outbreak potential.

Annual rate of change in total population, including natural increase and net migration. In growth models, higher population growth dilutes per-capita capital.

Total population, logged and lagged one year. Larger populations increase civil war risk — more people means more potential recruits and more heterogeneous grievances, though the mechanism is contested.

Total annual precipitation (mm) from gridded meteorological data.

Correctness of model predictions, measuring how well a classifier or regression model maps inputs to true outcomes across the full population or subgroups.

Calibration and discrimination of market-aggregated probability estimates compared to polls, models, and expert panels. Prediction markets consistently outperform polls for election forecasting and match or beat expert panels for geopolitical questions. Brier scores typically 0.15-0.25 for well-functioning markets.

FAO's flagship indicator (SDG 2.1.1): the estimated share of a population whose habitual food consumption is insufficient to meet dietary energy requirements. Derived from food balance sheets, food consumption surveys, and population distribution parameters. Country-level, annual. Widely used but criticized for masking within-country variation and relying on caloric adequacy alone.

External support provided to government forces (rather than rebel groups) in intrastate conflicts. After 9/11, the share of conflict-dyads with exclusively rebel-sided support fell to near zero by 2016, while government-sided support reached 77% of all active conflict-dyads by 2017. Reflects the international anti-terrorism norm shift that reframed rebel groups as terrorist organizations.

Degree to which private property rights are legally protected against state expropriation and private predation.

Number of purchases in observation period

Count of rebel groups (other than the supported group) fighting the target government in the previous year (lagged). Source: UCDP Dyadic Conflict Data. Creates a market mechanism: competition between groups disciplines rebel behavior, making orchestration viable without direct sponsor monitoring or sanctioning.

Any form of external state support provided to rebel groups in civil conflicts, including material resources (weapons, financing), logistical support (safe haven, training camps, transport), and direct combat support. Emergence of sponsorship is not predicted by personalism (H2); only the type of sponsorship varies.

Days since last purchase

Number of years since the last regime transition (3+ point Polity score change). Longer-lived regimes face lower coup risk due to institutional consolidation, established patronage networks, and routinized civil-military relations. New regimes are vulnerable during transition periods.

Probability that two randomly selected individuals belong to different religious groups (Herfindahl index). Ranges 0-1. Included as a control alongside ethnic fractionalization; not found to be a robust predictor of civil war onset.

Number of years the rivalry has been in existence up to time t. Longer rivalries are associated with higher probability of NAG support (+204% over full range in strategic rivalries, Table 3). Entrenched rivalries feature more polarized preferences and greater frustration with the status quo.

Aggregate maximum hostility level of MIDs in the rivalry divided by rivalry duration. Captures the relative severity of the rivalry over time. Higher intensity rivalries are associated with more NAG support (+22% in strategic rivalry sample, Table 3).

Confidence in and abidance by the rules of society, including contract enforcement, property rights, and the courts. WGI Rule of Law indicator.

Sale price of residential property in dollars

Gross or net enrollment ratio at primary, secondary, or tertiary level. Flow proxy for human capital investment.

Count of intergovernmental organizations to which both states belong simultaneously.

Total hours of sleep per 24-hour period, measured via polysomnography (gold standard), actigraphy (wearables), or self-report (ATUS, NHANES). Recommended 7-9 hours for adults (AASM). US average: 6.8 hours on workdays. Population-level data available from ATUS (N>200K), NHANES sleep module, and Fitbit aggregate studies (N>6M nights).

Composite measure of sleep efficiency (% time in bed asleep), number of awakenings, time in deep/REM sleep, and sleep onset latency. Pittsburgh Sleep Quality Index (PSQI) is the standard self-report instrument (0-21 scale, >5 = poor quality). Wearable devices now provide objective proxies at population scale.

Marginal economic damage of one additional tonne of CO2, in USD. Ranges from ~$50 to ~$185/tonne.

The distortion of independent judgments when group members observe each others' estimates before finalizing their own. Reduces effective sample size of the crowd, destroying the error-cancellation mechanism that produces wisdom. Even small amounts of social information can cut crowd accuracy by 25-50%.

Objective lack of social contacts or relationships, measured by network size or frequency of social interaction.

Share of population covered by social safety net programs (cash transfers, food assistance, school feeding, public works). Effective social protection buffers households against food insecurity during shocks. Measured by World Bank ASPIRE database. Coverage ranges from <10% in fragile states to >80% in OECD countries.

Average group member score on the Reading the Mind in the Eyes Test (RME). The strongest individual-level predictor of collective intelligence — groups with higher average social perceptiveness coordinate and integrate information more effectively.

Share of respondents answering 'yes' to 'If you were in trouble, do you have relatives or friends you can count on?' Gallup World Poll. Ranges 0-1.

Perceived expertise and trustworthiness of the information source. Moderates correction effectiveness — corrections from high-credibility sources are more effective, but the advantage decays over time (sleeper effect). Typically measured via expertise + trustworthiness scales.

Binary indicator for whether a rebel group's external sponsor is a democracy (democ_supdum), or continuous proportion of democratic sponsors (perc_dem). Democratic sponsors are more sensitive to human rights concerns and face stronger domestic accountability pressures, making them more likely to condition support on rebel restraint of civilian targeting. Moderates the negative effect of external funding on civilian safety.

Sponsor's military expenditures (logged), from Correlates of War National Material Capabilities Dataset v5.0. Counterintuitively associated with MORE orchestration: powerful states can sustain credible shadow-of-hierarchy threats, can absorb efficiency losses from rebel non-compliance, and can exploit rebels' local legitimacy while still deterring defection.

Latent measure of state capacity aggregating 21 variables across three conceptual pillars: extractive capacity, coercive capacity, and administrative capacity. Estimated via item response theory model. Source: Hanson and Sigman (2021), State Capacity Dataset v1. Mean in sample: -0.059, range: -1.541 to 1.28.

A state's level of satisfaction with the prevailing outcome in a rivalry, operationalized via prior MID outcomes: +1 (satisfied, won previous dispute), -1 (dissatisfied, lost), 0 (neutral/drawn). Derived from Maoz & Mor (2002) rivalry supergame framework. Dissatisfied states with low capabilities are the primary candidates for NAG support as a substitution strategy.

Percentage of children under 5 with height-for-age below -2 standard deviations of the WHO Child Growth Standards median. Primary indicator of chronic undernutrition and long-term food insecurity. Reflects cumulative effects of inadequate nutrition, repeated infections, and poor care practices. Links to existing Praxis construct stunting_prevalence in nutrition_stunting domain.

Binary indicator of whether two states share a land border or narrow sea corridor. Standard opportunity control.

Duration from index event to outcome or censoring

The portion of output not explained by the quantity of inputs used in production. Reflects technological progress, efficiency, and institutional quality.

The lower of the two states' ratios of bilateral trade to GDP, capturing the dyadic member with least to gain from trade continuity.

Binary indicator equal to 1 if an authoritarian state carried out one or more transnational repression events (threats, arrests, extraditions, abductions, assassinations) against its own citizens abroad in a given country-year, 0 otherwise. Source: AAAD (Dukalskis 2021).

Count of transnational repression events carried out by an authoritarian state against its own citizens abroad in a given country-year. Ranges from 0 to 61 (Uzbekistan 2005). Source: AAAD (Dukalskis 2021).

Average causal effect of treatment on outcome across all units

The percentage by which wages in dense urban areas exceed wages in less dense areas, controlling for worker characteristics. Raw premium ~30% for doubling city size; after controlling for sorting (education, ability), residual agglomeration premium is ~4-8% per doubling of density. Measured via Mincerian wage equations with city-size or density controls.

Global Acute Malnutrition: percentage of children under 5 with weight-for-height below -2 SD (moderate + severe wasting). Key indicator of acute food insecurity at population level, used in IPC nutrition analyses and SMART surveys. Threshold of 15% GAM triggers emergency classification. Links to existing Praxis construct wasting_prevalence.

Estimated probability of winning a game given current state, capturing pre-game expectations and in-game dynamics

Percentage of seats in national parliament (lower or single house) held by women. IPU data. Global average ~26.5% (2024). Strongest predictor: electoral gender quotas. Nordic countries lead (~45%) without quotas; Rwanda leads globally (61%) with constitutional mandate.

Probability of occupational injury or error per shift as a function of prior sleep. Workers sleeping <6 hours have 1.7x the accident risk of those sleeping 7-8 hours. After DST spring-forward (population loses 1 hour), US workplace injuries increase 5.7% and severity increases 67.6% on the following Monday.