Table 5.

Estimated results for regressions using diabetes as dependent variable

Variable	OLS	Robust Estimation
Obesity rate	0.1329***	0.0981***
Hypertension rate	0.0279	0.0221*
Ln(GDP)	0.0036*	0.0023*
Ln(wage)	0.0066	0.0051*
Health center density	0.0037***	0.0023***
Hospital density	−0.0063*	−0.0055**
Household size	−0.0064*	0.0004
Gender	−0.001	−0.0005
Age	0.0022***	0.0013***
East	0.0064*	0.0031*
Middle	0.0008	0.0007
North	0.0024	0.0014
Constant	−0.0152	−0.0709***
F (Wald) test for all variables	24.07***	37.60***
R square	0.5124
Adjusted R square	0.4911
Hausman Test for two regressions	χ2-statistic—46.22***
Skewness/Kurtosis test for Normality	χ2-statistic—18.81***
Shapiro-Wilk W test for Normality	z-statistic—8.615***
Shapiro-Francia W test for Normality	z-statistic—7.829***

Note: The null hypothesis for F or Wald test is that the concerned coefficients are jointly equal to zero. The null hypothesis for normality test is the normal distribution of the model

*, **, and *** give the coefficients’ significance indicated by estimated standard errors or bootstrap standard errors at 10%, 5% and 1% level individually