A new statistical criterion, on examining the independent variables used in the trip generation process, objectively chooses the best single variable or combination of variables to forecast person trips per household and autos per household. These predictive equations relating travel characteristics to the behavior of the household is a functional part of the systematic planning for the network of highways and mass transit facilities to meet this demand. The statistical criterion minimizes the total error of the predictive equation considering errors in the simulation of the dependent variables for present day conditions, errors in estimating the independent variables, and errors in the forecasting equations using the independent variables. The error in the simulation of the dependent variable was measured by the standard error of estimate in the regression equations used for reproducing the tri-state transportations home interview survey data. The error in estimating the independent variables was determined by forcing the independent variables to take on systematic errors and then measuring the effect of these errors on the estimation of the dependent variable. The error in the forecasting equation was measured by the testing of the trip generation equation for its accuracy of estimation (stability) over 2 periods of time. The statistical criterion is illustrated by sample calculations determining the best variable and equation for forecasting the dependent variable of autos per household. (Author/publisher)
Abstract