Analyzing Changes in Urban Form and Commuting Time*

*The authors thank the Québec Community Transit Authority for the use of travel survey data and Isabelle Anken, Anne-Marie Séguin, Corinne Thomas, Martin Lee-Gosselin, and Pierre Lemieux for their help at various stage of this research, as well as the Social Sciences and Humanities Research Council of Canada, the Geomatics for Informed Decisions (GEOIDE) Network, Québec's Formation des Chercheurs et Aide à la Recherche (Researcher's Formation and Help to Research—FCAR) Research Fund, the Desjardins Foundation, and the Québec Ministry of Transport for their financial support. The authors are grateful for the very helpful and encouraging comments of the anonymous reviewers.

Abstract

Urban work trips have changed in important ways during the last decades. In Québec City, a medium-sized Canadian metropolitan area, commuting distances increased for both male and female workers between 1977 and 1996, while durations increased for male workers and decreased for female workers. This article seeks to identify spatial and social factors responsible for these changes. We develop a disaggregate model of trip duration estimated on the basis of large samples derived from travel surveys comparable through time. Using categorical variables to specify change, we are able to separate change effects from level effects attributable to various dimensions of urban form. Our analysis clearly indicates that, once travel mode and key social factors are controlled for, the shift from a monocentric to a dispersed city form is responsible, in the Québec metropolitan area, for increasing commuting time. This is contrary to findings in larger metropolitan areas, where, it has been argued, the suburbanization of jobs maintains stability in commuting duration.

Key Words:

• accessibility
• commuting time
• structural change
• urban form

Notes

^aD_ia is the distance from home to the central urban axis; D_ja is the distance from place of work to the central urban axis; A is the accessibility index from residential location i or job location j to employment E or residences R, either by bus b or by car c. These indices are measured in units dependent on the values of E, R, and the function f in the accessibility equations. Only their relative values are of interest.

Note: The numbers shown in bold characters indicate highest loadings on factors.

^aThe rotation method is varimax with Kaiser normalization.

Note: The numbers in bold indicate the variables for which the coefficients are significantly different in 1996 compared to 1977.

¹ This is not to say that this research question is not tackled at all. A number of studies discuss this changing relationship. See, for instance, Hodge (1996).

² It is solely by experimentation that we have found that taking the third power of cos θ and subtracting the result from 1 gives a well-behaved index for further multivariate analysis. Figure 2 shows that a commute passing through the CBD has a larger θ than a commute toward the city center, but more generally, distances between home and CBD and between workplace and CBD have to be held constant for the effect of the directional index to be fully felt on commuting duration, as revealed by our regression models. Clearly, more work has to be done on directional biases before we fully understand this aspect of the relation between urban form and commuting time.

³ These two surveys have the advantage of having been conducted under the supervision of the same manager, who was especially attentive to maintaining the comparability between the data-collection methods and tabulation procedures used. We have also examined key variables, such as modal split, at five-year intervals—in 1977, 1981, 1986, 1991, and 1996—and have found that 1977 and 1996 are in line with the trends observed in between. These surveys, however, have the disadvantage of not asking any questions on income.

⁴ Distances and durations are through the shortest path on the street network from places of residence to places of work, without intermediate stops. As detailed in Thériault et al. (1999), impedance factors are introduced in the computations through speed and turns penalties, but not for congestion, since this factor is marginal in the QMA.

⁵ This procedure may appear unwarranted, but we had to apply it, since centroid coordinates of trip-ends enter the computation of both the time duration (the dependent variable) and the six spatial factors (independent variables). As a result, there is no random variation in the response for cases having the same trip-ends, while there is such a variation for the social and transportation factors on those cases. Clearly, this has the effect of unduly enhancing the statistical significance of the spatial-factors parameters compared to the nonspatial factors. Randomly selecting only one case among each set of observations having the same trip-ends drastically reduces the sample size but ensures better comparability among the parameters.

*The authors thank the Québec Community Transit Authority for the use of travel survey data and Isabelle Anken, Anne-Marie Séguin, Corinne Thomas, Martin Lee-Gosselin, and Pierre Lemieux for their help at various stage of this research, as well as the Social Sciences and Humanities Research Council of Canada, the Geomatics for Informed Decisions (GEOIDE) Network, Québec's Formation des Chercheurs et Aide à la Recherche (Researcher's Formation and Help to Research—FCAR) Research Fund, the Desjardins Foundation, and the Québec Ministry of Transport for their financial support. The authors are grateful for the very helpful and encouraging comments of the anonymous reviewers.