Factor Analysis for the Study of Determinants of Public Transit Ridership

Abstract

Behavioral studies based on attitude survey questionnaires with numerous variables may be tainted with repetitions and correlations. To overcome these deficiencies, a factor analysis approach is demonstrated that produces clusters of uncorrelated factors. From 47 observable variables contained in the Ottawa-Carleton Transportation Commission (OC Transpo) attitude survey, only 8 underlying factors have emerged. Bus information service is the most important factor. In addition to factor analysis, this article reports on a logistic regression model, based on key factors, for estimating the odds of ridership.