What the Data Tells Us
Over 200 cross-variable correlations reveal which factors most strongly predict life expectancy, which findings defy expectation, and why ecological correlations can be deeply misleading. This page synthesizes the key insights from the entire dataset.
Learning-adjusted years of schooling (r = 0.84) is the strongest single predictor of national life expectancy — stronger than physician density or health spending.
Sanitation access, clean water, and electricity are among the top 10 predictors of life expectancy. Basic infrastructure outperforms most medical metrics at the country level.
Country-level correlations can be deeply misleading. Obesity and alcohol appear beneficial at the national level — but both are clearly harmful at the individual level. Confounding by wealth drives most of these paradoxes.
The Gini index (income inequality) is negatively correlated with life expectancy. More equal societies tend to have longer lifespans, independent of absolute wealth levels.
- CVD is the leading cause of death globally, dominating in Europe & Central Asia
- Eastern European countries (Georgia, Ukraine, Bulgaria) have among the highest CVD rates
- CVD death rates negatively correlate with GDP — wealthier nations have far lower rates
- Cancer death rates are highest in high-income nations with ageing populations
- Mongolia, Greenland, and Hungary are outliers with very high cancer mortality
- Cancer rates correlate positively with life expectancy — people live long enough to develop cancer
- HIV/AIDS, TB, and malaria deaths are heavily concentrated in Sub-Saharan Africa
- Infectious disease rate is the strongest negative correlate of GDP per capita
- Countries with high infectious rates often have young populations and low median age
- Life satisfaction correlates strongly with LE but weakly with health spending
- Suicide rates are paradoxically higher in many wealthy countries
- Social factors (trust, freedom, support networks) predict happiness better than income
- Daily calorie and protein supply correlate strongly with LE — reflecting food security
- Child stunting prevalence is one of the most powerful negative predictors of LE
- The diet page reveals national dietary patterns linked to specific cause-of-death profiles
- Air pollution (PM2.5) is a significant predictor of respiratory and CVD mortality
- Indoor air pollution kills more people than outdoor pollution in developing nations
- Renewable energy adoption correlates with lower pollution-related death rates
Many variables in this dataset are driven by the same underlying factor: national economic development. Wealthier countries simultaneously have higher obesity rates, more internet access, better sanitation, more physicians, and longer life expectancy. When obesity appears to correlate positively with life expectancy, it is not because obesity is beneficial. Rather, both variables are independently associated with GDP. This is known as confounding.
All data in this explorer is at the country level. Relationships observed between countries do not necessarily hold for individuals within those countries. For example, countries where people sleep more tend to have lower life expectancy (because developing nations sleep more). But at the individual level, adequate sleep is well-established as beneficial for longevity. Drawing individual conclusions from aggregate data is the ecological fallacy.
Not all variables have the same coverage. Clean water access is available for 258 countries, but daily step data covers only 46 countries. Correlations based on small samples are less reliable. Always check the sample size (n) when interpreting any correlation. A strong-looking r value with n=33 is far less convincing than a moderate r with n=250.
| Variable | Category | Unit | Countries (n) | Coverage |
|---|
Cross-Domain Health Insights
What the data reveals about global patterns in health, wealth, and longevity
The Human Mortality Explorer aggregates data from 30+ variables across lifestyle, economic, environmental, dietary, and wellbeing domains for over 260 countries. By computing Pearson correlations across all variable pairs, we can identify which factors most strongly associate with life expectancy and which correlations defy intuition.
The strongest predictors of national life expectancy are infrastructure and education variables: learning-adjusted years of schooling (r = 0.84), sanitation access (r = 0.82), and internet penetration (r = 0.80). These outperform direct health metrics like physician density (r = 0.75), suggesting that the social determinants of health operate primarily through structural development rather than healthcare alone.
Counter-intuitive findings include the positive correlation between obesity and life expectancy, the negative correlation between sleep duration and life expectancy, and the near-zero correlation between diabetes prevalence and daily steps at the country level. Each of these reflects confounding by national wealth and the ecological fallacy rather than genuine causal relationships. The insights page is designed to highlight these traps and educate users about the limitations of ecological data.
Beware the Ecological Fallacy
Why country-level correlations can mislead about individual health
The ecological fallacy is the most important methodological caveat in cross-country health analysis. When we observe that countries with higher meat consumption live longer, it does not mean eating more meat makes individuals live longer — it means wealthier countries eat more meat AND have better healthcare. Every insight on this page comes with a “confounding alert” that identifies the likely confounders and explains why the correlation exists. Understanding these statistical traps is essential for anyone interpreting global health data, whether in journalism, policy-making, or academic research.
What is the strongest predictor of life expectancy?
Learning-adjusted years of schooling is the single strongest correlate (r = 0.84), followed by sanitation access (r = 0.82) and internet access (r = 0.80). Infrastructure and education variables consistently outperform direct health metrics as predictors of national life expectancy.
Why does obesity positively correlate with life expectancy?
This is a classic confounding effect. Wealthier countries have both higher obesity rates (due to food abundance and sedentary lifestyles) and higher life expectancy (due to better healthcare and infrastructure). Obesity does not cause longer life. Both variables are independently driven by economic development.
What is the ecological fallacy?
The ecological fallacy occurs when conclusions about individuals are drawn from group-level data. For example, countries with higher average sleep duration have lower life expectancy, but this does not mean sleeping more shortens life. The country-level pattern reflects that developing nations (with lower life expectancy) also have longer sleep durations. Individual-level research shows adequate sleep is beneficial for health.
How many correlations are computed?
The correlation matrix includes 200 variable pairs, computed from over 30 variables spanning lifestyle, economic, environmental, dietary, and wellbeing domains. Each correlation is a Pearson r with significance testing (p-value) and sample size reporting.