I've been thinking a lot about park-and rides and mode shift lately in the context of the Northern Branch restoration project, and I've come up with a new generalization. I even thought of a catchy summary for it, like Isaac Newton: people in cars tend to stay in cars.
Of course I'm standing on the shoulders of giants here. Donald Shoup found (PDF) that on average, when parking is free, drivers will prefer to spend three minutes circling for a close parking space than three minutes walking from a further space. Retailers fear any reduction in parking capacity or increase in parking cost out of fear that their driving customers will choose a competitor with easier or cheaper parking.
This extends to mode choices as well. In an illuminating report, "Guaranteed Parking, Guaranteed Driving" (PDF), Rachel Weinberger and her colleagues found that car-owning Jackson Heights residents who owned off-street parking spaces near their houses tended to drive more than car-owning Park Slope residents who had to circle for on-street spaces. I find this to be true when I walk, shop and eat in my own neighborhood: my car-free neighbors tend to be more visible and available than those with guaranteed parking, and those who have cars but no guaranteed spaces are in the middle.
In my view, park-and-rides have a similar effect. When I take the train or the bus, I walk past all the neighborhood shops and restaurants, and I know that if I need groceries, hardware, medicine, a haircut or a cup of coffee, I can stop on the way. If I need something I can't get near my apartment - a suitcase, a television, a kabob or a bagel - I can get off the train at a nearby station and walk a little further. If I need something that's not available at all in my neighborhood - a hair dryer, a package of frozen dosai, quality clothes, books - I can take the train or bus to Manhattan or to another part of Queens.
My neighbors who drive to work or for social trips know that they will probably not be able to find parking near any of the neighborhood stores or restaurants, so they stop at places with abundant parking on the way home. If they have guaranteed parking and they need something from outside the neighborhood, they are more likely to drive to a "cheaper" car-oriented store like Costco or Target than take the train to Manhattan.
None of my neighbors commute via park-and-rides, but the pattern seems pretty clear. If you need something, you can buy it right near your job downtown, or there might be somewhere to buy it near the train station or bus stop. Chances are, though, that you'll get off the train and immediately drive to a store with easy parking. What's least likely is that you'll drive home, park and then go to the store. So park-and-rides are bad for business at the transit stop, and near the homes of the parkers.
There is one more way that this works: in self-identification. You might think that transit riders would identify as transit riders and drivers as drivers, but it's not that simple. There are people who drive to the park-and-ride, people who drive when they're off work, and people who only occasionally drive up to the mountains for a weekend. Whenever they're not riding, they tend to be on foot or on transit. We might therefore expect them to identify at least partly as transit riders, but in my experience they all identify as drivers first, even if they hardly drive.
This leads me to my Law of Transportation Mode Inertia: on a given trip people in cars tend to stay in cars. But note the word "tend." Obviously there are plenty of people who transfer from car to foot or bus or train in their commutes, but there are particular conditions for these transfers, and most drivers will resist doing it more than once.
It is because of this principle that I get so frustrated with mainstream American transit planners' love of park-and-ride facilities, as seen most recently in the Northern Branch project. More on that soon.