Regression of fuel consumption with respect to car weight in the xauto data

In the dataset uploaded using the SSC package xauto, we fitted a linear regression model of fuel consumption (in nipperkins per mile) with respect to car weight (in US tons), separately for the 2 car model origin groups (non-US models and US models). In each regression model, the regression slope was a weight effect on fuel consumption (expressed in nipperkins per additional ton-mile), and the intercept was a mean fuel consumption (expressed in nipperkins per mile) for a fantasy car with zero weight.

Table 1. Regression of fuel consumption with respect to weight (long version using insingap)

ParameterEstimate(95%CI)P
Non-US (N=22):
Weight (US tons)11.059(8.130,13.987)1.5x10-7
Constant-1.764(-5.382,1.853).32
US (N=52):
Weight (US tons)7.885(6.484,9.286)2.2x10-15
Constant0.537(-1.549,2.622).61

Table 2. Regression of fuel consumption with respect to weight (wide version using xrewide)

Non-US (N=22): US (N=52):
ParameterEstimate(95%CI)PEstimate(95%CI)P
Weight (US tons)11.059(8.130,13.987)1.5x10-77.885(6.484,9.286)2.2x10-15
Constant-1.764(-5.382,1.853).320.537(-1.549,2.622).61