When Robert Cervero and Kara Kockelman published their highly-cited article “Travel demand and the 3Ds: Density, diversity, and design” in 1997, they seemingly changed the transportation planning discourse forever. The idea of characterizing the built environment using three measures that happen to start with the letter D was catchy, and catch on it did.
By 2010, the 3Ds had grown to seven with the addition of “destination accessibility” (what I would call regional accessibility), “distance to transit,” “demand management,” and “demographics” (a category of control variables, not a characteristic of the built environment). I have heard rumors of an additional D or two since then.
The Ds now prevail in the academic literature on the effect of the built environment on travel behavior. I do not consider this a good thing.
At a minimum, the terminology is confusing. “Diversity” means the mix of land uses in a given area, but one could easily mistake it to mean the socio-demographic mix of residents of the area, which falls instead under “demographics.” “Design” is usually measured as the connectivity of the street network, not as the aesthetic qualities of the street environment that the word generally implies. The Ds are a catchy shorthand, but not an especially clear one.
Another problem is that researchers treat the D characteristics as independent, when in fact they are interdependent. Researchers are usually careful to test for “multicollinearity,” the situation in which variables are highly correlated with each other, making it difficult to estimate the effect of any one of them on another. Most studies show that the D characteristics are not as correlated as one might think. It is possible, for example, to have a good mix of land uses in an area with a low level of street connectivity.
But certain values of the D characteristics do often go together, reflecting the era when a particular neighborhood came into being. For example, pre-World War II neighborhoods are more likely to have both grid street networks (“design”) and neighborhood-scale shopping within walking distance (“diversity”).
The Ds also tend to go together because they influence each other. Density in particular has a strong effect on other D characteristics: Higher densities support better transit service (“distance to transit”) and a closer proximity of retail and services (“diversity”). Greater land use “diversity,” coupled with better street connectivity (“design”), produces shorter distances to destinations and thus better “destination accessibility.”
Treating the Ds as independent of each other creates the risk of overestimating their influence. It also creates the risk of underestimating their influence. If two or more characteristics together have synergistic effects, they produce a total effect greater than the sum of each independent effect.
Rarely do researchers discuss these relationships, let alone account for them in their analyses. Particularly troubling is the use of density as a predictor of travel behavior without an explanation of how density influences behavior. Density influences the other D characteristics that more directly determine the choices available to travelers and thus more directly shape their travel choices. Researchers too often justify their reliance on density and the other Ds by citing other researchers’ reliance on the same measures. Even a minimal attempt to justify the Ds based on behavioral theory rather than accepted practice would be an improvement.