. 2009 Sep 28;106(41):17290–17295. doi: 10.1073/pnas.0904491106

Table 2.

Annual change in life expectancy associated with an increase of one percentage point in annual GDP growth in regression models without lagged effects or with lag effects up to three years

Dependent variable	Lag 0	Lag 1	Lag 2	Lag 3	d	R²
Annual change in life expectancy for the whole population	−0.20^*				2.79	0.29
	−0.24^**	0.09			2.32	0.37
	−0.21^**	0.10	0.01		3.06	0.42
	−0.22^*	0.09	0.03	−0.04	2.94	0.26
Annual change in life expectancy, white population	−0.19^*				2.78	0.35
	−0.24^**	−0.11			2.29	0.42
	−0.21^**	−0.12	0.01		3.00	0.42
	−0.22^*	−0.11	0.03	−0.05	2.88	0.49
Annual change in life expectancy, nonwhite population	−0.24^**				2.36	0.40
	−0.24^**	−0.02			2.02	0.43
	−0.21^**	−0.02	0.03		2.70	0.46
	−0.21^*	−0.01	0.01	0.04	2.70	0.49

*, P < 0.05;

**, P < 0.01. d is the Durbin-Watson statistic. Because d ≈ 2 · (1 − r), where r is the sample autocorrelation of the residuals, d > 2 implies negative autocorrelation of the residuals resulting in possible overestimation of standard errors and the underestimation of statistical significance.