Comments on When is bikeshare faster than transit? - Greater Greater Washington

Comment by David C

Tue, 19 Jun 2012 15:28:25 EDT

Yes I'm surprised.

This data shows that people are making a transportation decision to use bikeshare - because it will get them there faster.

Right. But it also shows they aren't valuing the other advantages of CaBi as much as I would expect them to.

Imagine that people were travelling from point A to point B and taking transit was 1% faster than CaBi. I wouldn't expect everyone to choose transit, because some people would value the other advantages. Now imagine that everyone is told that those who uses CaBi will be paid $1 when they get to point B. I expect the number of CaBi users to go up. Up that to $10, and it would go up even more. Up that to $1000 and the time difference wouldn't even matter any more.

So...I'm surprised because travel time isn't the only relevant consideration in transportation decisions AND I feel like CaBi wins on some of those other considerations.

Comment by MLD

Tue, 19 Jun 2012 15:16:19 EDT

@David C

You're surprised? You think that more people would choose to bike just for random "values" reasons or a small price difference? As a bike advocate you know that bikes are part of transportation. This data shows that people are making a transportation decision to use bikeshare - because it will get them there faster.

Also I think the data shows that lots of trips are under 3 miles (around 20 minutes on CaBi) and you'd be hard pressed to find a trip of less than 3 miles that's much faster on transit than on a bike once you factor in transit wait time (which this analysis does).

Comment by David C

Tue, 19 Jun 2012 15:11:01 EDT

As I think about it, I'm actually surprised that the bulk of trips aren't on the other side of the line, namely because speed isn't CaBi's only advantage.

1. Cost - The marginal cost of CaBi is $0. Whereas riding transit costs more for each trip. So users should be willing to take a little bit longer to get where they're going because it will save them money.

2. Exercise - On average, people value exercise. That's why they invest resources like time and money into it. People should be willing to take a bit longer to get somewhere via CaBi because it frees up time for exercise.

But then again...

3. Presentation - working against this is that many people don't want windblown hair and sweat-soaked clothing based on their destination, so this might dissuade members from biking, event when the above factors would make it desirable.

4. The Blockbuster effect - In a group, taking CaBi might only work if everyone is a member and everyone wants to take the trip. It's possible that even considering all of the above factors, members choose transit to stay with friends.

Still I would expect 2 and especially 1 to far outweigh 3 and 4 and so for the bulk of trips to be above the line, not below it.

Comment by Bob r.

Tue, 19 Jun 2012 11:28:22 EDT

It'd be interesting to see plots of average trip duration, both time and distance, for each hour throughout the week. Start Monday 1am, end Sunday at 12am. Can you see commute arrivals? What happens at lunch? Are school times important? And, after using trips from throughout the year, is it different in January than in July?

Comment by Jasper

Fri, 15 Jun 2012 11:11:04 EDT

@ karl:Cars get points that affect their insurance and they pay fines. Bikes keep on going. Riders are out there without helmets or lights but bitching when they get hit because no one can see them. If you want to play, you have to pay. No free rides!

Bikers pay with their bodies and lives when crashing. Much worse than getting dinged with a point on your driver's license, or paying a few bucks more for your insurance.

Also, you missed the sarcasm in my response.

Comment by David C

Thu, 14 Jun 2012 23:48:01 EDT

karl,

No cyclist without lights or reflectors is bitching when they get hit because no one can see them. I'm pretty sure of that, but perhaps you could prove me wrong.

Comment by karl

Thu, 14 Jun 2012 20:20:30 EDT

@Jasper

Cars get points that affect their insurance and they pay fines. Bikes keep on going. Riders are out there without helmets or lights but bitching when they get hit because no one can see them. If you want to play, you have to pay. No free rides!

Comment by MLD

Thu, 14 Jun 2012 16:48:41 EDT

I figured at least a few of the trips would be to a station where there is basically no transit service, and three of those are the same trip (4,7,8).

Also a few of those show the dearth of transit on capitol hill. 8th street has good bus service for the most part but the rest of it is convoluted and less useful than it could be.

And the fact that it's a pain to get to Georgetown on transit from any place that isn't directly North or East of it.

Comment by James Wong

Thu, 14 Jun 2012 16:17:08 EDT

For inquiring minds... here are the Lat/Lon pairs (formatted to plug into google maps) that fall in the bottom-right-most box of 10-15 min CaBi trips in 40-45 min transit...

38.889955,-77.000349 to 38.919488 , -77.000624
38.8803,-76.9862 to 38.900017 , -77.001967
38.90774,-77.071652 to 38.9176 , -77.0321
38.9022212,-77.059219 to 38.88412 , -77.04657
38.916442,-77.0682 to 38.930282 , -77.055599
38.88412,-77.04657 to 38.9022212 , -77.059219
38.88412,-77.04657 to 38.9022212 , -77.059219

(its too much to post here, but FYI all of these trips were between 8am and 9pm on weekdays, so there's decent transit service for all them)

Comment by David C

Thu, 14 Jun 2012 14:33:31 EDT

MLD, I'm equally interested in the 45 minute transit trip that can be done in 13 minutes by CaBi.

Comment by David C

Thu, 14 Jun 2012 14:31:53 EDT

Biking is always quicker the transit when no traffic signals are obeyed.

I know. It's awesome.

Comment by Richard Layman

Thu, 14 Jun 2012 13:38:55 EDT

Mike B -- interesting point. It would be very interesting to study people's "mode-amory" or willingness to substitute modes depending on the trip.

For me, I mostly bike (up to say 12 miles), at least when I am by myself. I am too cheap to take a cab. I do what you do though some of the time, in terms of taking Metro back home because it's uphill. Although that means schlepping the bike on transit, which you don't have to worry about (normally, cf. the MLD photo) with bike share.

Comment by UrbanEngineer

Thu, 14 Jun 2012 13:15:26 EDT

I'm curious to see how this data would be different if done only with Bikeshare riders who have agreed to abide by traffic laws.

Comment by Jasper

Thu, 14 Jun 2012 13:03:28 EDT

@Karl:Biking is always quicker the transit when no traffic signals are obeyed

Same goes for driving.

Comment by MLD

Thu, 14 Jun 2012 10:28:26 EDT

The scatterplot distribution tells me that people use CaBi for trips SPECIFICALLY WHEN it will be faster than transit. That is, CaBi is a tool in a person's transportation tool belt, and users choose between various modes the one that will work best for their trip. Trips on the other side of the dividing line are much less common: why would I take CaBi if transit is going to get me there faster?

It also shows me that most CaBi trips are under 20 minutes - which is about that 3 mile barrier that Richard is talking about.

Also, I would be REALLY interested to know what trip that dot is in the upper left - a CaBi trip of 44 mins vs a transit trip of 8?

Comment by Karl

Thu, 14 Jun 2012 10:24:40 EDT

Biking is always quicker the transit when no traffic signals are obeyed.

Comment by JJ

Thu, 14 Jun 2012 10:08:22 EDT

Important to remember that time is only one variable in choosing mode of transit. The reality is, if I'm at work and dressed in a suit, and need to get to a meeting elsewhere in the city, I am going to take the Metro. I'm not going take a bike and show up sweaty or with wind-blown hair for a professional encounter. On the weekends, when I can be more laid back, bikeshare is more viable.

Comment by Rob P

Thu, 14 Jun 2012 09:59:31 EDT

David C, you're right that studies that confirm pre-existing attitudes are valuable. However, I think you're conflating two different studies in this case: one that shows that CaBi users tend to use the service for short trips vs. longer trips, and another that compares a treatment group (CaBi users) against a control group (transit users) to see if and how outcomes (travel time) differ. Both of these studies are valuable, but they have to be done with different methodologies to be statistically rigorous.

Comment by Mike B

Thu, 14 Jun 2012 03:15:51 EDT

James,

If you are looking to compare trips on CaBi vs transit while controlling for walk time you could make your origin and destination points halfway between the Cabi station and the corresponding beginning and end of the transit portion. That way walk times would be roughly equal and you'd just be comparing the travel time for each mode. Alternatively, if open trip planner allows this you could subtract the beginning and ending walking times of the transit trip to get the same effect. Then, since you had a straight modal comparison, if you wanted to add in an estimate of walking time based on something like egk's idea you could apply similar logic to both trips to get a more even comparison.

Comment by Mike B

Thu, 14 Jun 2012 02:44:33 EDT

Richard Layman,

I think there is one important distinction that makes shared bikes different from owning a bike that you missed. In addition to convenient bike storage at the destination CaBi also allows for convenient mode switching. I know that when I take CaBi somewhere I'm not committed to biking back. Now, its true that you can frequently take your bike on metro buses and trains but it is so inconvenient to do so that most people do it infrequently at best(or at least I rarely see a bike on trains or buses). Whereas I frequently will take a CaBi from Adams Morgan, where I live, to Downtown(downhill) and then take the metro back(uphill).

I think that flexibility makes CaBi much more like transit than owning your own bike. When I'm going out I consider CaBi against trains, buses, and taxis. Never do I contemplate taking CaBi or my own bike. I think it is similar to difference between using your own car and taking a taxi. Same type of machine but very different mode of transportation.

Comment by Ms. D

Thu, 14 Jun 2012 00:35:48 EDT

For me, it's a quality of life issue. I use CaBi for half of my commute, even though Metro would be about the same time-wise (given that I have to walk from the Metro station but there's a CaBi dock right outside my office). Metro might even be a minute or two faster, since it's a one-seat trip instead of getting off the train and on the bike or the reverse. (I intend to bike all the way to/from my "home" Metro station once some improvements and a CaBi station go in).

So, being rational...why would I do this? It allows me to avoid the most crowded part of Metro. I no longer have the frustration of tourists, Caps fans, and other annoyances for half of my commute. Instead, I'm outside enjoying the fresh air and a little exercise. I'm generally in a much better mood biking half of my commute rather than using Metro the whole way. Yes, I also use CaBi where it saves me time or serves areas not well-served by transit, but by far my heaviest use is to keep me from climbing up into a clock tower because the 457th tourist that day shoved me/couldn't use the fare gate/screamed so loud I could hear them over my full-blast music/let their child kick me in the knee and got upset when I demanded an apology/on and on.

Also, I feel like a speed demon with these charts. Being a bit of a math nerd myself, I have my average speed calculated, and it's a good bit above average, and I only ride a significant downhill once a week for a few blocks of a many-block trip. Here I thought I was a big slowpoke. :)

Comment by xmal

Wed, 13 Jun 2012 23:12:37 EDT

Really interesting is that most of the points are clustered around the 1/2x line.

That's something that might drive policy: biking is not competitive around the break-even point, but when it is twice as fast as transit.

Comment by Lucre

Wed, 13 Jun 2012 22:45:25 EDT

@David C
Very well put. I choose to use CaBi mostly on the basis of the fact that it is incredibly convenient. When that article in the Times came out my first thought was "Smug? Metrosexual? It's just damn useful!"

So I'm glad to have someone back me up on the idea that CaBi user does not necessarily mean smug metrosexual hippie socialist snob. It might just have more to do with being rational.

Comment by Richard Layman

Wed, 13 Jun 2012 22:11:06 EDT

David C. -- you're right sort of. Given my academic inclination, I am fine with studies. But bikeshare vs. owned bikes is a non-distinction, just like Groupon/Living Social are "merely" digital-based direct marketing. To understand those businesses, you need to know DM, the digital aspect isn't all that distinctive.

The same goes for bikes. Trip behavior is trip behavior, with the only substantive distinction being bike station access and proximity.

Although there is one significant difference, as people have pointed out on this blog in other entries, it has to do with securing your bike and being on the hock for losing it if stolen. With bikeshare, as long as you use it correctly, that isn't a problem (other than lack of capacity). Once you put it in the dock, it's no longer your responsibility.

For the most part, this distinction doesn't impact trip length, trip efficiency, and competitiveness with transit.

Comment by Richard Layman

Wed, 13 Jun 2012 22:06:03 EDT

Tom C. -- buy your own bike, then you never have to worry about stations, and you'll have the benefit of getting to places more quickly. OTOH, bike parking options need to be improved. (full disclosure: I sell bike parking.)

MLD -- great photo. I think I saw something similar, but didn't have a camera. Now I can't remember.

Rob P. III -- it's not just about transfers, it's also about the difference of being able to ride directly to a place (like Georgetown) by bike vs. taking a more circuitous route via transit.

Comment by David C

Wed, 13 Jun 2012 21:33:39 EDT

I think it's important not to belittle studies that show what everyone already assumed - in part because there are so many times that studies show the opposite. So what does this show? That CaBi customers are highly rational and are making decisions based on time and efficiency. They are not weirdos, hippies or bike enthusiasts (well, they might be that too).

And as Ward 1 Guy points out this can drive policy. If we build a new bike/ped connection, how much will that increase CaBi trips compared to transit trips added by some comparably priced transit expansion? etc...

Comment by Bill Cook

Wed, 13 Jun 2012 21:03:33 EDT

Last year I regularly Cabi'd from NoMa to Crystal City, at night, and it was usually quicker than Metro due to the avoidance of a transfer.

Comment by Tom Coumaris

Wed, 13 Jun 2012 19:50:24 EDT

...but hit a full bike station on the other end and all bets are off.

Comment by Tom Coumaris

Wed, 13 Jun 2012 19:47:02 EDT

Going from 14th & S to GWU I find CaBi is much much faster. Going down 14th buses are faster a little.

Comment by Rob P. III

Wed, 13 Jun 2012 18:25:17 EDT

I take CaBi from near 12th & Newton in Brookland to my office in north Dupont Circle. The bike-ride averages 22 minutes and riding the Metro would only be 13 minutes, but by hopping CaBi I don't have to walk the last two blocks to the Brookland Metro Station nor the last four blocks from Dupont Circle to my office. I also don't have to pay for Metro every day and know that if I do go over 30 minutes the price will still be less than boarding Metro or a bus.

Also, if you're headed to Georgetown from Dupont or Mt. Pleasant CaBi is quicker than transit because there are no transfers.

Comment by Richard Layman

Wed, 13 Jun 2012 18:18:22 EDT

Um, it's pretty basic. Not even looking at your data.

Trips of 3 miles or less are likely to be faster than transit, especially if non-fixed rail. Trips of 3-6 miles may be equal or faster by bike, depending on the time of day, frequency of trains, distance to the station-stop, how long you wait, + the time required to get from the station to the final destination vs. the ability to ride by bike directly to the final destination -- of course in terms of bike share, there is the issue of station proximity both to the start point and the final end point.

This as Charlie would say, is the "killer app" of biking generally.

Longer trips cost more, that's a decision point. Topography is an issue too. And there is the health/exercise issue as a decision point. And weather as a decision point.

There's nothing really new here, other than having some data. CF. the National Household Travel Survey.

And bikeshare vs. regular bikes. But bikes are bikes at the end of the day, whether owned or shared.

Comment by Michael H.

Wed, 13 Jun 2012 17:44:28 EDT

In real-world use, many Metro trips will be delayed because of track work, sick customers holding up a train, congestion at the Rosslyn tunnel between Blue and Orange line trains, packed trains requiring a rider to wait for the next train, etc. At the same time, a CaBi rider may find the destination station to be full. This requires riding to a nearby station (if available), docking the bike there, and walking back to the area of the other station.

Buses can be delayed by traffic or malfunctions. Same thing for those driving themselves.

I generally assume that for short trips, CaBi is approximately as fast as transit or driving. For longer trips, transit will usually be much faster. At peak hours, CaBi might see more of a time advantage, although there are bike rush hours on the local trails.

One thing is for certain: A CaBi ride is usually more pleasant than taking transit or driving (excluding bad weather days).

Capital Bikeshare is a great supplement to the transportation network but I'd never rely on it as a sole means of transport. I'm glad it's here but I'm also glad to have Metro (despite their problems).

Comment by charlie

Wed, 13 Jun 2012 17:14:44 EDT

I'm finding that for me (one data point on a curve) CABI displaces:

1. Walking (lots of short 2 block trips I do by bike now)
2. Transit (Rosslyn to West End)
3. Car (I tend to park my car then cycle around)
4. Taxi (I take a taxi 3-5 times a year, so it isn't diplacing much)

Another positive is Cabi might drive bus ridership. If things are a mess, get off and grab a bike. You don't have that option on rail.

Comment by Jasper

Wed, 13 Jun 2012 17:07:49 EDT

@ dcd:Prepare to be shocked. ... I think you got caught up in assuming a straight shot via transit.

You are right. People logically want to compare biking and bus or metro on the route of the bus or metro. But they forget that a bikeshare station may be closer to your departure or destination.

It also works long distance by the way. You can bike from Beulah St in Franconia to Georgetown in 75 minutes. Take the bus, metro and walk/shuttle and it takes 90 minutes. That is how slow transit in DC is. Oddly, the preferred bike route deviates about as much from a straight line as the transit route does.

Comment by MLD

Wed, 13 Jun 2012 16:21:12 EDT

The question we should be asking is, "when is bikeshare exactly as fast as transit?" and I have the answer:

It is real, and I did take the picture.

Comment by dcd

Wed, 13 Jun 2012 16:17:08 EDT

Maybe I'm misreading the data, but I'm shocked that CaBi beats transit in any trip over about 15 minutes. For example, I can get from Metro Center to Ballston in +/- 20 minutes on Metro. That same trip takes 40 minutes on CaBi.
I'm less surprised for shorter distances, particularly if your software takes into account walking to metro station and waiting for a train.

Prepare to be shocked. I live in Columbia Heights and work in Georgetown. If you just count time on the bike v. time inside the fare gates, CaBi cuts my trip in half (30 minutes to 15). If you count total travel time, it saves an even greater percentage of time - while I am 3 blocks from the nearest CaBi station and only one from the Metro in CH, in Georgetown CaBi is across the street from my office, while I have a 10 minute walk from the Foggy Bottom metro stop. Plus, CaBi is free - well, not free, but the pro-rated annual fee-per-trip is a fraction of metro fare, especially now.

I think you got caught up in assuming a straight shot via transit. Any time you're using one mode of transit, you may be right that transit is quicker. But for any trip that requires a change of train/bus, or bus + train, it's far more likely that CaBi will be quicker. In my case, it's not close.

Comment by Ward 1 Guy

Wed, 13 Jun 2012 16:16:23 EDT

James, can you say more about how these data and analysis would be used? I can't see the connection between the data on actual usage and the decisions you would make. Not that there isn't a connection, but it wasn't made explicit.

In my opinion what you need to estimate are behavioral parameters. What would it take to get someone to switch modes? How much would a fare increase affect ridership? How would increased bikestation density reduce transit use? How heavily would a new bike path here instead of there be used?

These all depend on how people would change their behavior if things were different.

Maybe you could look at bikeshare usage between metro-adjacent stations before, during, and after a metro delay.

Comment by Ward 1 Guy

Wed, 13 Jun 2012 16:10:08 EDT

I would use survey data, because riders know their own routes.

For instance I know my trip between home and office is 13 minutes by bike, 17 minutes by walk/bikeshare, 14-18 minutes by walk/bus depending on bus timing, and 21-25 minutes by metrorail, depending on train timing. Driving is about 8 minutes with no traffic, 20 minutes with bad traffic, and a few minutes extra if I have to walk to find a car2go or a cab.

Comment by Ward 1 Guy

Wed, 13 Jun 2012 16:04:22 EDT

Do people take their bikeshare bikes on metrorail or bus? If so, they will have very short travel times over long distances.

Also, I often don't ride straight line on bikeshare. I might ride from point A to point B, pick something up, stop at point C to talk to someone, and end up at point D. This would be recorded as a slow ride from A to D.

If you pick rush hour trips between rush hour locations (residential to business district) you'll avoid this problem, but then again, you'll get a more narrow type of riding/transit tradeoffs.

Comment by Tina

Wed, 13 Jun 2012 16:02:53 EDT

@James W.-beautiful figures wonderfully presented.

Comment by Greg

Wed, 13 Jun 2012 15:57:32 EDT

Actually, Lucre, you do make sense. I probably am somewhat of a unique case in that I basically ride right down the Orange line (the exercise beats the extra time it takes). A lot of people probably do use CaBi because they don't have a convenient transit route.

Comment by Lucre

Wed, 13 Jun 2012 15:55:59 EDT

I mentioned the Benning Library trip as an example because it was the only time I can unequivocally say that I used CaBi consciously to replace rather than supplement transit service. I would not have been able to attend the meeting at the library on time without CaBi.

Like @Dave Murphy the vast majority of my CaBi trips supplement a metrorail trip though - in my case to Union Station for a distance I would rarely bother waiting for a bus.

If I'm at a meeting downtown it sometimes seems simpler to take CaBi all the way home than to take metrorail part of the way. At these times I may have taken an X2 or walked from Union Station before CaBi, but now because CaBi is already a regular part of my commute, it makes sense to make a single-seat trip of it. The sunk cost of a yearly membership also makes me feel a need to get my money's worth. But in these situations the time it takes is of little importance, so it's harder for me to say unequivocally that CaBi replaced transit; without it I might have walked.

CaBi may have replaced metrobus for me as my way to reach parts of the city not near metrorail, or get a respite from commuting in a dark concrete tube: I have not ridden the bus since getting my key.

Comment by James Wong

Wed, 13 Jun 2012 15:45:31 EDT

Lots of great feedback!
@Rob - of course we'd love to get a much better handle on the selection bias, but we were just starting with an easy-to-get dataset which CaBi was great for. Your idea of spatially separating those trips that are near transit stops/stations is neat and I'll see if we can follow up on it if we open this back up. If someone has a highly refined transit O-D database for the region, I'd love to see it.

Some other good ideas about how to generate walk time - i like @egk's a lot. Keep 'em coming.

And yes, @mm - OTP has every transit trip in the DC region modeled so we count wait time. If someone left at 8:42am on a tuesday, we simulated a transit trip that departs at 8:42 on a tuesday including any associated wait time. We assume you left when you wanted to and wouldn't defer your trip to not wait for a bus (simplifying assumption).

Comment by MM

Wed, 13 Jun 2012 15:21:05 EDT

Maybe I missed this as I'm no expert, but does the OTP data for transit account for headways between trains/buses? Seems to me that CaBi will easily beat transit during off peak hours when headways are long and during times when service is disrupted (pretty much all the time, it seems), but for actual doors-close to doors-open time the numbers probably would be in transit's favor.

Comment by egk

Wed, 13 Jun 2012 15:17:18 EDT

@ James: a simple walk-time correction could be estimated on the basis of the distance to the next CaBi station. Making the counterfactual assumption that trips are uniformly chosen in the plane, the expected walk distance at the origin (and destination) for any CaBi trip is 1/4 the distance to the next station. So to correct simply add the time it would take to walk this distance to the trip time for any trip involving that station (as origin or destination). So for any trip involving the 10th and U station you would add to the CaBi trip time the 1.5 minutes it would take to walk 1/4 of the way to 7th and T [according to Google].

Comment by Lucre

Wed, 13 Jun 2012 15:16:14 EDT

@Greg

It's a matter of when the geography of Metro service makes a certain trip take longer. For example, my CaBi trip from Union Station to Benning Road Library took 24 minutes. I can't get WMATA Trip Planner to give me a time less than 30 minutes by rail, though by the X1, I could get there in 25 minutes - a time that would still put that trip under the red line in the chart. By rail I have to go some distance out of my way to Metro Center just to backtrack in the direction of the Minnesota Ave metro. I'd also have to walk from there to the library whereas there's a CaBi station right outside.

Another example might be a trip from Howard University to Brookland Metro - I'm guessing a 15 to 20 minute ride. There's an out of the way rail transfer, circuitous bus service, but a fairly direct path for a cyclist.

Comment by Nick

Wed, 13 Jun 2012 15:08:12 EDT

@Rob P

Totally correct.

Comment by Dave Murphy

Wed, 13 Jun 2012 14:59:55 EDT

I think it's interesting that this argument frames CaBi vs. Transit, when I for one use CaBi to augment my transit trips. Like when I bike to Union Station and hop on a MARC train, or when I bike from Woodley Park to Adams Morgan. Sure, I'll skip transit altogether sometimes. But usually when I use CaBi it's only a portion of my trip.

Comment by Rob P

Wed, 13 Jun 2012 14:54:43 EDT

That's fine, my suggestion in that case is to re-run this using a different approach. Rather than pulling 1% of only CaBi trips, first establish all the O/D pairs where both CaBi and public transit both serve, then pull a random sample of CaBi trips, with a quota established for each O/D pair so that you don't wind up with predominantly short trips in the sample.

Comment by Elle

Wed, 13 Jun 2012 14:52:27 EDT

One other potential problem with the data is, I guess, a form of selection bias -- that is, the first 30 minutes of any CaBi ride are free, so it behooves riders to keep their ride under 30 minutes. Even if their destination is farther away, there is no penalty (unless it's new) for returning a bike, immediately taking out another and having a new 30 minute free period.

I don't know how popular this tactic is (I used it all over Paris), but it would artificially shorten CaBi trips even more than you are seeing with the lack of walk data.

As for the walk data, could you use the median of the assumed range-of-use for each station as a rough proxy? It's rough, but might at least give a good idea about the impact.

Comment by David Alpert

Wed, 13 Jun 2012 14:45:37 EDT

Rob: He does mention it:

This might logically suggest that CaBi riders use the service when its faster than transit, and not to use CaBi when it would take longer.

Comment by Rob P

Wed, 13 Jun 2012 14:38:28 EDT

While the 1% of trips might be randomly generated, the full sample is not random. As I'm understanding this, self-selection bias isn't mentioned.

In other words, there's not going to be a data point for my trips from Farragut West to Tenleytown, because I chose Metro for those trips instead of CaBi.

It's not at all surprisingly that most of the points on the scatter-plot fall into the "CaBi is faster" half because people pick to use CaBi when they know it is faster...

Comment by Jasper

Wed, 13 Jun 2012 14:32:05 EDT

I can biked from Georgetown to Crystal City in the same time as I can walk or bus to Rosslyn and ride metro to Crystal City.

CaBi has changed 23-27 minutes of my commute into a workout.

Comment by Adam S

Wed, 13 Jun 2012 14:31:44 EDT

Is there a way to pull out CaBi trips that start and end near Metro stations? That may address some of the walking time effect.

Comment by Greg

Wed, 13 Jun 2012 14:31:18 EDT

I'm less surprised for shorter distances, particularly if your software takes into account walking to metro station and waiting for a train.

Also, your 20 mph figure may be because the OTP software is calculating a route that is unnecessarily long. Maybe there is a sidewalk to cut over or something similar that the software doesn't recognize. Otherwise, you're right, Lance Armstrong couldn't hit 20 mph on a CaBi bike on flat land and the 20 mph'ers must be on a downhill.