Table 5.
Study | Benefit/cohort (billions $) | 95% CIa |
---|---|---|
Grosse et al. (2002) linear lead | 213.83 | 147.27–280.39 |
Pooled analysis, linear lead | 148.58 | 82.18–215.82 |
Pooled analysis, natural-log lead | 318.98 | 196.30–441.67 |
CIs cannot be used to compare linear and log lead specifications because the linear specification is incorrect and the 95% CI calculated from it suffers from uncorrected residual heteroskedasticity.