Physical Measures Reduce Trips by How Much?

In an often-quoted study by Nelson\Nygaard it is said that physical measures such as net residential density can be used to reduce trip generation by up to 90%. I have often wondered about this figure, and I went to check the article. Unfortunately, the study says just that but the data on which it is based can not be used to support this hypothesis.

Here is the data from the table that makes the claim. Net Residential Density can reduce trip generation by "Up to 55%", Mix of Uses by "Up to 9%", Local-Serving Retail by "2%", Transit Service by "Up to 15%", Pedestrian/Bicycle Friendliness by "Up to 9%", which all adds up to 90%. Add in "Affordable Housing" (Up to 4%) and Free Transit Passes (25%), and you've reduced trips by 119%!

Catch the flaw in logic? The maximum values of these reduction ranges are not additive, unless you can prove that they are independent. I think it's pretty clear that density and transit service, for instance, are not independent in the figures on which these reduction factors are based. If they're not independent, the reductions aren't additive. In fact the individual reductions would be lower.

What exactly are these figures based on? Well, for density this is the difference between the highest and the lowest density area (assuming that everyone currently lives in the lowest density area and all of them will move to the highest-density area) calculated using the single-parameter formula from the paper Holtzclaw, Clear, Dittmar, Goldstein, and Haas, "Location Efficiency: Neighborhood and Socio-Economic Characteristics Determine Auto Ownership and Use - Studies in Chicago, Los Angeles and San Francisco", Transportation Planning and Technology (2002). This study was quite clear that the effects of socio-economic characteristics, transit, nearby shopping, and so forth could not be disentangled. It also discusses studies showing that variations in household size and income are more important than transit or density. They say that area-wide studies show that doubling the density of an entire metropolitan area would reduce VMT by about 20-25%. Would further doubling further reduce it by the same amount? No, it is quite clear in the paper that the relationship is not linear but logarithmic. Virtually all the reduction in trips occurs at very low density - you get four times as much decrease in going from very low density to about 20 households per net residential acre than you do by going from 20 households per acre to 200.

I won't even talk about self-selection bias nor about my own papers that measure the fact that when you look at apartment residents and house dwellers separately, the density effect is almost entirely in the house dwellers. Good thing because that means we can do something about that given that they are the majority and are unlikely to want to live in apartments. If apartments are all we build, driven by a simple-minded belief in provably incorrect linearity, we will only drive the majority further away. Of course if we want to concentrate on the single housing form with the lowest VMT, we should be building nothing but homeless shelters.

Enough of that. These effects are neither linear nor additive, but they exist. Just for fun, I looked at the Nelson\Nygaard Table 3, which shows the trip rate as a function of land use, but also shows, for each of the land uses from Single-Family Detached Housing to High-Rise Apartment, the values for residential density, housing units, jobs, presence of retail, index of transit service, intersection density, sidewalks, etc., and did a multiple linear regression on their predicted average trip generation figures. The result is that transit, jobs, and intersection density explain virtually all the trip generation data, density is not really a factor at all, and the transit index by itself is an excellent predictor.

Tags: Urban Planning Urban Design Walkable LEED-ND Transportation

[] permanent link Comments: 0