VARIATION IN WORK TRIP LENGTH IS EXAMINED AS RELATED TO MODE OF TRAVEL, SPATIAL DISTRIBUTION OF RESIDENCE AND EMPLOYMENT LOCATIONS, AND SOCIOECONOMIC CHARACTERISTICS OF WORKERS. USING MULTIPLE REGRESSION ANALYSIS ON DUMMY VARIABLES, 15 CHARACTERISTICS OF WORKERS IN A METROPOLITAN AREA WERE RELATED TO WORK TRIP LENGTH. STATISTICALLY SIGNIFICANT RELATIONSHIPS (P LESS THAN 0.001) WERE FOUND FOR 13 OF 15 VARIABLES, BUT THE EXPLAINED VARIANCE OF TEN OF THESE VARIABLES WAS LESS THAN ONE PERCENT. TRAVEL MODE AND THE TWO LOCATION VARIABLES EACH EXPLAINED MORE VARIATION THAN ANY OF THE SOCIOECONOMIC VARIABLES. BOTH ADDITIVE AND MULTIPLICATIVE MODELS WERE TESTED: THE MODELS PERFORMED EQUALLY WELL AND THE STATISTICAL SIGNIFICANCE OF THE VARIABLES DID NOT APPRECIABLY DIFFER. AN IMPROVED UNDERSTANDING OF WORK TRIP LENGTH VARIATION IS NEEDED TO PROVIDE A THEORETICALLY SOUND BASIS FOR CALIBRATION OF A GENERAL CLASS OF SPATIAL INTERACTION MODELS INCLUDING TRIP DISTRIBUTION, RESIDENTIAL LOCATION, AND WORK AND SHOPPING LOCATION MODELS. THIS RESEARCH INDICATES THAT DISAGGREGATION OF A TRIP DISTRIBUTION MODEL BY WORKERS' SOCIOECONOMIC CHARACTERISTICS WILL NOT IMPROVE ITS PERFORMANCE. DISAGGREGATION BY ZONE OF RESIDENCE OR EMPLOYMENT WILL RESULT IN ONLY SLIGHT IMPROVEMENTS IN MODEL PERFORMANCE.(Author/publisher).
Abstract