A MODEL FOR PREDICTING TRANSIT RIDERSHIP-INTERIM TECHNICAL REPORT NO 7

THE PURPOSE WAS TO DEVELOP A MODEL FOR PREDICTING RIDERSHIP IN AN URBAN TRANSIT SYSTEM. TWO CATEGORIES OF INDEPENDENT VARIABLES WERE IDENTIFIED WHICH WERE ASSUMED TO GENERATE TRANSIT DEMAND: HOUSEHOLD CHARACTERISTICS AND CHARACTERISTICS OF THE TWO AVAILABLE TRANSPORTATION MODES, AUTOMOBILES AND TRANSIT BUSES. EIGHT SOCIO-ECONOMIC VARIABLES WERE USED IN THE ANALYSIS OF DEMAND FACTORS RELATED TO HOUSEHOLD CHARACTERISTICS: (1) FAMILY INCOME; (2) AUTOS OWNED PER HOUSEHOLD; (3) AGE DISTRIBUTION; (4) WORKERS PER HOUSEHOLD; (5) THE RATIO OF LICENSED DRIVERS TO AUTOS AVAILABLE; (6) MEAN HOUSEHOLD SIZE; (7) PROPORTION OF FEMALES; AND (8) AREA POPULATION DENSITY. TWO VARIABLES WERE USED IN THE ANALYSIS OF DEMAND FACTORS RELATED TO MODAL CHOICE: (1) THE RATIO OF AUTO TO TRANSIT TRAVEL TIME, AND (2) THE ACTUAL TIME DIFFERENCE BETWEEN TRANSIT AND AUTO BETWEEN SPECIFIC POINTS. MULTIPLE REGRESSION ANALYSIS WAS EMPLOYED TO TEST THE VALUES DETERMINED FOR EACH VARIABLE AS WELL AS FUNCTIONS OF OVERALL TRANSIT DEMAND. IT WAS FOUND THAT THE LINEAR REGRESSION MODEL WHICH RESULTED FROM THE STATISTICAL ANALYSIS WAS NOT A USEFUL PREDICTOR OF TRANSIT DEMAND. IT COULD NOT BE USED TO DETERMINE POTENTIAL TRANSIT RIDERSHIP WITHOUT A SIZEABLE ERROR IN THE PREDICTION. ONE PRIMARY CAUSE OF THIS PROBLEM WAS THAT ONLY THREE OF THE TEN INDEPENDENT VARIABLES WERE FOUND TO CONTRIBUTE DIRECTLY TO PASSENGER LEVELS: HOUSEHOLD INCOME, HOUSEHOLD SIZE, AND THE DIFFERENCE IN TRAVEL TIME BETWEEN AUTOS AND PUBLIC TRANSIT. TWO SPECIFIC RESTRICTIONS ARE IDENTIFIED WHICH LIMITED USE OF THE MODEL DEVELOPED. FIRST, OTHER RELEVANT VARIABLES WERE NOT ANALYZED, SUCH AS TRANSIT SCHEDULING AND ROUTING. SECOND, THE MODEL ASSUMED A LINEAR RELATIONSHIP BETWEEN INDEPENDENT AND DEPENDENT VARIABLES WHICH DID NOT OBTAIN FROM THE DATA. RATHER, VARIABLES SUCH AS AGE HAD U-SHAPED RELATIONSHIPS, IN WHICH CASE RIDERSHIP WAS GREATEST AMONG THE YOUNG AND OLD, WITH THE LEAST PATRONAGE FROM PERSONS IN-BETWEEN. OTHER VARIABLES PRODUCED OTHER NON-LINEAR RELATIONSHIPS WHICH DID NOT ADAPT TO THE EMPIRICAL EQUATION. ADDITIONAL RESEARCH EMPLOYING NON-LINEAR VARIABLES SHOULD BE ATTEMPTED. ONE UNEXPECTED BY-PRODUCT OF THE RESEARCH IS DISCUSSED IN DETAIL. THIS CONCERNS AN ANALYSIS OF PEAK-HOUR VS. OFF-PEAK RIDERSHIP AND FARES. THE REPORT DEMONSTRATES THAT REDUCED FARES FOR OFF-PEAK PASSENGERS ARE BOTH FEASIBLE AND DESIRABLE, AND CORRELATES THE PROPOSAL TO TRANSIT DEMAND. /UMTA/

  • Supplemental Notes:
    • PROJ NO CAL-MTD-10
  • Corporate Authors:

    Sacramento State College

    ,    
  • Authors:
    • Gutowsky, A R
  • Publication Date: 1971-4

Subject/Index Terms

Filing Info

  • Accession Number: 00228213
  • Record Type: Publication
  • Source Agency: Urban Mass Transportation Administration
  • Files: TRIS, USDOT
  • Created Date: Oct 9 1972 12:00AM