Somewhat surprisingly, although the original model specified that the road distance would be a nonlinear function of the straight line distance, only the coefficient on the linear distance term was statistically significantly different from zero. The intercept was weakly significant (i.e., at 10%). A reestimation of the model, forcing a zero intercept, revealed a linear circuity factor of 1.45, applicable to the entire study area and any straight line distance.
