Demonstrate dangers of normalizing

Generate random CO2, area & energy from Barros data

## Loading objects:
##   US.GHG.dt

Compare raw and area-normalized random data

## Randomly generated CO2 (kg) & Area (km2) are uncorrelated: 0.0163
## `geom_smooth()` using formula 'y ~ x'

## Correlation Area-normalized CO2 (kg km-2) & area (km2): -0.2626
## `geom_smooth()` using formula 'y ~ x'

## 
## Call:
## lm(formula = CO2.km2 ~ area.ran)
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
##  -9734973  -1409980   -540263    654075 120309598 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  4468517     558315   8.004 8.57e-15 ***
## area.ran      -19533       3216  -6.074 2.47e-09 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 6272000 on 498 degrees of freedom
## Multiple R-squared:  0.06898,    Adjusted R-squared:  0.06711 
## F-statistic:  36.9 on 1 and 498 DF,  p-value: 2.473e-09
## ..as are area-normalized CO2 (kg/km2) & ln(Area): -0.4855
## `geom_smooth()` using formula 'y ~ x'

## 
## Call:
## lm(formula = CO2.km2 ~ log(area.ran))
## 
## Residuals:
##       Min        1Q    Median        3Q       Max 
## -19962710  -1398123     72259   1241339 108063412 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)   17381189    1303399   13.34   <2e-16 ***
## log(area.ran) -3355781     270745  -12.39   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 5682000 on 498 degrees of freedom
## Multiple R-squared:  0.2358, Adjusted R-squared:  0.2342 
## F-statistic: 153.6 on 1 and 498 DF,  p-value: < 2.2e-16

Compare raw and energy-normalized random data

## Correlation of CO2 (kg) & electricity generation (kWh): -0.0502
## `geom_smooth()` using formula 'y ~ x'

## Energy-normalized CO2 (kg kWh-1) & generation (kWh) are correlated: -0.4212
## `geom_smooth()` using formula 'y ~ x'

## 
## Call:
## lm(formula = CO2.GWh ~ elec.ran)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -325.82  -78.63  -27.15   37.43 2411.73 
## 
## Coefficients:
##               Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  2.527e+02  1.879e+01   13.45   <2e-16 ***
## elec.ran    -6.056e-05  5.844e-06  -10.36   <2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 206.1 on 498 degrees of freedom
## Multiple R-squared:  0.1774, Adjusted R-squared:  0.1757 
## F-statistic: 107.4 on 1 and 498 DF,  p-value: < 2.2e-16
## Scherer & Pfister (2016) model involves area on the right hand side
## Energy-normalized CO2 (kg kWh-1) & area/generation are correlated: 0.0718
## `geom_smooth()` using formula 'y ~ x'

## 
## Call:
## lm(formula = CO2.GWh ~ ATE.km2.GWh)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -516.51  -71.32  -52.77  -25.52 2563.89 
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)    81.37      10.19   7.984 9.89e-15 ***
## ATE.km2.GWh  7321.29    4555.62   1.607    0.109    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 226.7 on 498 degrees of freedom
## Multiple R-squared:  0.005159,   Adjusted R-squared:  0.003162 
## F-statistic: 2.583 on 1 and 498 DF,  p-value: 0.1087