Skip to main content
. 2023 Aug 11;39(7):e00145922. doi: 10.1590/0102-311XEN145922

Table 2. Pooled and random-effect regressions.

Disease/Pollutant Pooled OLS Random effects LM test
Coefficients Robust SE N R2 Coefficients Robust SE N R2
Respiratory diseases
PM10 0.1123 * 0.0374 615 0.683 0.092 * 0.024 615 Within = 0.132 Probability > χ2 = 0.000
Between = 0.774
Overall = 0.663
PM2.5 0.1524 0.093 191 0.678 0.066 0.038 191 Within = 0.329 Probability > χ2 = 0.000
Between = 0.734
Overall = 0.648
Stroke
PM10 0.208 * 0.0422 615 0.471 0.145 * 0.042 615 Within = 0.413 Probability > χ2 = 0.000
Between = 0.003
Overall = 0.044
PM2.5 0.1731 * 0.0572 191 0.689 0.024 0.049 191 Within = 0.016 Probability > χ2 = 0.000
Between = 0.389
Overall = 0.473
Diabetes
PM10 0.0332 * 0.0122 615 0.269 -0.01 0.013 615 Within = 0.025 Probability > χ2 = 0.000
Between = 0.27
Overall = 0.216
PM2.5 -0.081 0.0595 191 0.16 -0.002 0.036 191 Within = 0.013 Probability > χ2 = 0.000
Between = 0.161
Overall = 0.095
Cancer
PM10 -0.0021 0.0197 615 0.481 -0.01 0.007 615 Within = 0.237 Probability > χ2 = 0.000
Between = 0.516
Overall = 0.47
PM2.5 0.0303 0.0218 191 0.688 -0.001 0.016 191 Within = 0.261 Probability > χ2 = 0.000
Between = 0.722
Overall = 0.667

GDP: gross domestic product; LM: Breusch-Pagan Lagrange multiplier test; OLS: ordinary least squares; PM: particulate matter; SE: standard error.

Notes: models include as control variables the logarithm of annual GDP per capita, the average age of population, the number of physicians per 1,000 inhabitants, and a time trend as a dummy variable. PM concentrations are expressed as logarithms.

* Indicates significance at a 1% level.