The Washington, DC region is great >> and it can be greater.


What does Metro's late night service look like?

Contributors and readers have weighed in on the negative impacts of possible late-night service cuts. Ben Ross described some changing ridership patterns that underscore the vitality of this service now.

Using origin and destination data from a Saturday night in August, I mapped entrances and exits on Metrorail from midnight to closing. Additionally, I determined net passenger flow at each station to show whether stations are more weighted to entries or exits at the end of the night.

Graphic by author.

NOTE: the station traffic, total entries and exits, is indicated by the diameter of each circle, not the area. If you're interested in seeing the raw data, it is available here.

As one might expect, traffic is centralized around several major nightlife centers and drops as you move farther out of the city center. In many respects, the picture underscores all that much more the importance of late night service, while in other respects it raises interesting questions.

First, some assumptions. An entry between 12 am and 4 am most likely represents the location a person was spending their evening. An exit between 12 am and 4 am, with slightly less but still relative certainty, likely represents the final destination of a rider, i.e. home. Thus a station more weighted toward entries should be a nightlife hub, while a station weighted toward exits is likely a residential area.

On the entries side, this seems to hold true in the aggregate, though the stations with the three highest entry to exit ratios aren't exactly the epitome of nightlife. Farragut West had the highest, followed closely by Federal Triangle and Archives. Their entry to exit ratios are 7.3, 6, and 5.4 to 1, respectively.

What's telling about those is that they are near some nightlife destinations, but not ones that would be considered late-night attractions, and are also generally devoid of residences.

The biggest circles, on the other hand, are places with diverse nightlife activities, as well as more (or growing in the case of Gallery Place) residential areas. Of the four busiest stations, Dupont Circle and Woodley Park are the least red, with respective entry to exit ratios of 3.6 and 3.1 to 1. These are also the two with the most residential density nearby.

As a result, you might expect a truly livable, "balanced" neighborhood to have relatively even entries and exits. This holds true in many cases: Columbia Heights and Eastern Market are relatively busy but have ratios between 0.8 and 1.3 to 1.

More suburban activity centers are also relatively busy but tend to lean more toward exits than their center city counterparts. Silver Spring, the busiest station outside the District, has nearly 2 exits for every entry.

Bethesda and Ballston, too, have more exits than entries. Of major activity centers outside of DC, only Clarendon and King Street have more entries than exits, and King Street only by 4%.

Meanwhile, suburban stations, where there is little development around and the stations boast large park-and-ride lots, are generally bright, bright blue.

Several end-of-the-line stations are even quite busy and are also the most heavily exit-oriented station. Vienna was actually the 14th busiest station on this particular Saturday night and had nearly 14 exits for every entry.

What's more interesting are the anomalies. Eisenhower Avenue, one stop from the end of the line, and within walking distance of practically nothing, actually has more entries than exits, though just barely. Suitland, home to a sprawling federal campus and several small, low density neighborhoods has nearly 2 entries for every exit!

Largo Town Center, at the end of the Blue line, actually ties with King Street, in walkable, Old Town Alexandria, for the most balanced station, with only 4% more exits than entries. Interestingly, the more distant stations in Prince George's County are noticeably more purple than those in Montgomery or Virginia.

What could explain these outliers? I can't say for sure. Perhaps with the comparatively poorer populations served by the eastern stations, more people enter late at night to travel across the city to a night-shift job, but that is difficult to tell. What do you think explains the anomalies?

The primary point is this: DC's evening travel patterns are complex. Far more complex than they used to be, too, as Ben pointed out. Late night service serves a variety of people coming from and going to a variety of places for a likely variety of reasons.

These patterns make late night Metro service extremely difficult to replace with night buses and shuttle services. And they make cutting that service even more detrimental to way of life our region has fostered over the last few decades.

Erik Weber has been living car-free in the District since 2009. Hailing from the home of the nation's first Urban Growth Boundary, Erik has been interested in transit since spending summers in Germany as a kid where he rode as many buses, trains and streetcars as he could find. Views expressed here are Erik's alone. 


Add a comment »

Cool chart. What did you use to make it?

by erik on Feb 17, 2011 10:01 am • linkreport

The Eisenhower Avenue station is near a UPS shipping center, so maybe those entries are people getting off of a late shift.

by Tim on Feb 17, 2011 10:02 am • linkreport

I would imagine Farragut W has the mix & volume it does because it affords easy access to and from the blue/orange line to downtown and points slightly north without having to wait for a transfer. If I'm out late in the Dupont area or as far east as 15th, I'm probably going to go to Farragut West to get to NoVA, and if I'm headed to Dupont or that general area, I'll get off at Dupont rather than deal with switching at metro center and riding back in.

by aj on Feb 17, 2011 10:09 am • linkreport

Eisenhower has a multiplex movie theater which is easily walkable, so that could be part of it. On that leg of the Blue/Yellow line, I can't think of any other movie theaters that are within walking distance.

by spookiness on Feb 17, 2011 10:11 am • linkreport

@Tim There are also several large high-rise apartment complexes (my boss lives in the Carlyle) and a large movie theater at the Eisenhower station.

by Froggie on Feb 17, 2011 10:11 am • linkreport

@spookiness Until Alexandria can get the Potomac Yard station built, there are none near Metro....unless there's a movie theater in Pentagon City (I'm not sure).

by Froggie on Feb 17, 2011 10:12 am • linkreport

Very nice chart. People working might explain the anomalies. You may consider doing this same exercise for the weekend hours 7AM-9AM. You may be surprised at how crowded the Metro is early weekend mornings: people are going to work.

by goldfish on Feb 17, 2011 10:15 am • linkreport

(Walking distance of a metro)
So I would guess that carless residents of PCY or CC or other points along the Blue or the Yellow may prefer to go to Ike instead of Gallery, or having to catch a transfer elsewhere.

p.s. Why the hell is catcha giving me conformation characters in greek?

by spookiness on Feb 17, 2011 10:15 am • linkreport

What is interesting is the argument "oh, before 1999 people got home" seems a bit weaker.

Gallery Place and U Street were NOT nightlife destinations back then -- well, certainly not on the scale they are today.

Why all the traffic at Woodley?

Also a bit puzzled by Rosslyn...people getting on at midnight? there isn't much to do there. Clarendon makes sense.

by charlie on Feb 17, 2011 10:20 am • linkreport

Froggie and spooiness already touched on it, but Eisenhower Avenue metro is located directly across the street from a pretty massive 22-screen IMAX/AMC theater. Much of the traffic it generates is from cars (with massive, and free, sprawling parking lots adjacent), but its location next to metro will naturally generate the late-night metro entries. Once the rest of the Hoffman town center is completed and those lots are turned into commercial properties, that will change with it.

by Dan on Feb 17, 2011 10:22 am • linkreport

Woodley traffic is almost certainly from the massive hotels nearby.

by Alex B. on Feb 17, 2011 10:22 am • linkreport

Looking at movie times at that Eisenhower Ave cinema, it looks like they have a number of late-night showings:

by RichardatCourthouse on Feb 17, 2011 10:22 am • linkreport

Woodley is almost certainly from the Adams Morgan crowds. Someone hasn't been out to the 18th street thunderdome in a while...

Also, couldn't we use this as a great argument for closing down, say, Fed Center SW, after 10pm? Or even 8pm? Surely there would be some savings from cutting these stations late at night. Also, a time advantage to getting home.

by Sebastian Dangerfield on Feb 17, 2011 10:26 am • linkreport

Woodley traffic is probably Adams Morgan-goers, actually (if you've ever been in that station anytime between 11 and closing on the weekend, you know what I'm talking about).

Seems to me this indicates we ought to start treating the outer portions of the system during late night service as what they are: commuter rail. Keep current headways closer in and run half-hour headways (on a schedule) at outlying stations. I'm guessing Virginia and Maryland will scream bloody murder if that were put on the table, though.

by nate on Feb 17, 2011 10:28 am • linkreport

Ha. I suppose that's true. I'm never going outbound on the Red Line so my Adams Morgan metro travels always involve walking to Columbia Heights or to U Street.

by Alex B. on Feb 17, 2011 10:29 am • linkreport

Rosslyn (even Foggy Bottom) is spill over from Georgetown nightlife since there's no metro station there. Woodley Park is the closest station to Adams Morgan (another nightlife destination with no closer metro station).

by Matt on Feb 17, 2011 10:29 am • linkreport

@alexB. Right, those pesky tourists. I always forget about them. Might explain some of the Rosslyn-entering traffic as well.

by charlie on Feb 17, 2011 10:30 am • linkreport

@erik I used an opensource Illustrator clone called InkScape

@charlie The Rosslyn traffic is probably due to people in the R-B corridor walking or cabbing a short distance to Rosslyn to catch a Blue Line without have to wait for a change.

by Erik Weber on Feb 17, 2011 10:30 am • linkreport

Great work! I suspect that many of those unexpected entry-to-exit ratios around business districts are showing security guards, cleaning crews, and other night-shift workers. (And for whatever it's worth, the DC press usually considers people who work the double shift to be Gray voters.)

by tom veil on Feb 17, 2011 10:31 am • linkreport

Woodley is also the closest metro for a lot of places in Adams Morgan.

by Cbshep on Feb 17, 2011 10:33 am • linkreport

@Dan: Not really on the parking lots. The lot between the theater and the Metro station is closed and has been for at least 2 years. I'm not 100% the status on the lot in front of the Army building, but I believe it's for their use only. The large parking garage charges...though to be fair it's only $1 on weekends and weekdays after 5ish. The only nearby parking lot I know for sure is free is the relatively small parking lot (maybe 30 spaces) that sits right underneath the Yellow Line. Literally.

by Froggie on Feb 17, 2011 10:42 am • linkreport

So David and I are having a debate, and I'd be curious about everyone's thoughts:

Without looking at the key, how much more traffic would you estimate Woodley Park has compared to Dupont Circle?

by Erik Weber on Feb 17, 2011 10:42 am • linkreport

Very well done! I knew I liked this site for a reason!

Any chance you might be able to update this with quarterly averages? This would have two advantages. First, it would hopefully reduce the arbitrary selection of any given Saturday, which might not represent all Saturdays. (Or maybe not, in which case it would directly address that potential problem)

Second, quarterly data might help plan scheduled repairs.
Metro's FY ends June 30, so 1st quarter is July, August, and September. Anecdotally, lots of people are on vacation during that period. Is night utilization in Metro's 1st quarter correspondingly lower? If so, that would be a great time to shut down early on a regular basis. Plus, it would help save money if many of Metro's costs are fixed.

Also what about Fridays versus Saturdays?

Finally, is it important that the largest station is "only" about 1,900 people? What does this the utilization of the system? I'm not sure.

by WRD on Feb 17, 2011 10:45 am • linkreport

This is great!

Not at all shocked about Clarendon. That's my station, and when I come home late from DC after a night out, I'm usually shocked by the number of people on the platform headed to DC. It can be worse than rush hour!

by Max D. on Feb 17, 2011 10:46 am • linkreport

There does seem to be an argument for turning some trains around at Grosvenor, Silver Spring, RFK, and Anacostia. Ridership on the outlying portions of the Orange line is a lot higher than I thought it would be, and a bit lower in the underground Arlington portions than I would have thought...

Like many others have mentioned, it would make sense to close Federal Center and Judiciary Square (much) earlier. Both of these stations are also within very easy walking distance of another, so their loss would not be particularly hurtful, especially since the walking time to the station will still be far less than the amount of time you'd have to wait for a late-night train.

by andrew on Feb 17, 2011 10:46 am • linkreport

@Charlie - this data is from August, which means that Rosslyn probably gets a fair number of folks walking across the Key Bridge to/from Georgetown. I certainly do that any time the weather is nice enough.

by Emily on Feb 17, 2011 10:54 am • linkreport

Great anlaysis. Forgive me if someone else already mentioned this: Why August? Its traditionally the month with the least activity in DC. I think the fact that you chose an Aug weekend means these data underrepresent mean usage.

by Tina on Feb 17, 2011 10:56 am • linkreport


No, not really - these data show where people enter and exit the system in aggregate, but it does not tell you individual Origin and Destination pairs. Just because a lot of people use the central stations doesn't mean the outlying stations aren't as important - they're using those central stations because many are coming from/going to those outlying stations with smaller ridership.

I do think there would be a case to close some stations late night as a budgetary measure - that would be far preferable to shutting the entire system down early. However, this data is insufficient to make those determinations.

by Alex B. on Feb 17, 2011 10:56 am • linkreport

So David and I are having a debate, and I'd be curious about everyone's thoughts:

Without looking at the key, how much more traffic would you estimate Woodley Park has compared to Dupont Circle?

I see where you're going with this.

I didn't even read the key regarding size since the resolution is too tiny. Just from the graphic, Dupont looks to be more than double Woodley.

You scaled it according to diameter, not area...

by Alex B. on Feb 17, 2011 10:59 am • linkreport

My guess would a little more than double, but since you lead the question, I am probably wrong..

by charlie on Feb 17, 2011 11:03 am • linkreport


I think one thing missing from your otherwise excellent analysis it that some stations actually have more usage than your map suggests. Fort Totten may not have as many people using the turnstiles but as a transfer station, the platforms do see a lot of passengers. That should be a factor regarding if a station should be closed or not.

by Randall M. on Feb 17, 2011 11:06 am • linkreport

This is a really excellent graphic. Am I correct in thinking that the data shows only 12,500 riders for the whole weekend, or is that per night?

by jcm on Feb 17, 2011 11:08 am • linkreport

Given that

- It is fairly expensive to subsidize Metro operations during late-night hours

- Businesses are unhappy about proposed Metro cuts

Is there any way to work out a compromise whereby businesses benefiting from Metro's late-night operation help cover some of the costs of said operation?

(As an aside on maintenance, Metro's in the unfortunate position of appearing totally incompetent. So the best way to convince people of the need to
- close the whole system earlier on weekends or
- close certain stations earlier on weekends or
- close certain lines earlier on weekends
Would be to demonstrate exactly how specific closure options would benefit maintenance, and how this would affect safety and reliability)

by EJ on Feb 17, 2011 11:10 am • linkreport

@Randall M.

If you were going to close stations, I don't think you'd close any of the transfer stations. Period.

by Alex B. on Feb 17, 2011 11:11 am • linkreport

@Froggie: I actually live right across the street. The two parking lots south of Eisenhower Avenue and adjacent to the metro are open. The lot to the west of the metro is usually filled to 80%+ capacity during the weekends, the lot to the east rarely is filled more than 10%, but it remains open.

The lot to the north of the theater (DOD lot) is open to the public and usually at 80%+ filled during weekends. The other 3-4 lots east of that (between Stovall and Telegraph) are closed to the public.

The other lot, between the theater/parking garage and eisenhower avenue (with signs advertising development) is closed. According the master plan, that is the next parcel slated for the development when Hoffman resumes construction on the town center.

by Dan on Feb 17, 2011 11:14 am • linkreport

@Tina (and others) - I didn't choose this data. It was the data made available to me by Michael Perkins who had to parse it from large unwieldy datasets given to him by WMATA when he requested data for typical week, back in the fall.

I'd love to do seasonal averages or comparisons of Fridays to Saturday, but I need the data to do that.

Also, I'd note that the traffic on this particular Saturday was actually a bit higher than an average Saturday. There were about 12,000 passenger trips, while Saturdays typically average just under 10,000.

by Erik Weber on Feb 17, 2011 11:25 am • linkreport

@Alex B. and Charlie - actually you're right, a bit more than double. But the question David has raised and I agree with, is whether it's better to indicate the traffic with area or with diameter...

by Erik Weber on Feb 17, 2011 11:29 am • linkreport

I think you might also be able to shutter Archives, since all the nightspots along 7th are accessible to Gallery, which gives you multi-line access.

Some of the patterns might be influenced by the long transfer delays in after-midnight frequencies.

I'm guessing the numbers are not from a night that there was not a Nationals game? Nats rarely go that late, but the game definitely would put more bodies in the system later, as a few would decide to catch a beer afterwards.

Really great chart!

by GreenCityDC on Feb 17, 2011 11:29 am • linkreport

Nice job

2. Closing stations is important but would have negligible contributions for savings (some labor of the station attendant)

3. The issue of benefit is again hard to quantify. The likelihood of merchants ponying up is infinitesimal. Hell, they don't even want to pay to join merchant groups. This kind of subsidy typically comes from a local government, maybe a BID--but BIDs definitely don't have much in the way of slack resources.

by Richard Layman on Feb 17, 2011 11:43 am • linkreport

Development is enveloping the Eisenhower Metro and it is across the street from an AMC 22 screen movie theater which attracts people from all over the place.

by NikolasM on Feb 17, 2011 11:47 am • linkreport

"@erik I used an opensource Illustrator clone called InkScape "

Erik, you rock! I LOVE Inkscape and am constantly amazed at what new features are being implemented. Sure, there are still things missing when compared to Illustrator but how can you argue with the great quality and low, low price (free)?

by GPL on Feb 17, 2011 11:52 am • linkreport

I would be interested in seeing/ helping out with getting a sort of real time or at least, by the hour, data set and plug it in to Google's Motion Chart gadget. We could then see comparative usage rate and in/out rate over time.

I don't know if WMATA would release that kind of info...

by Drewh on Feb 17, 2011 11:53 am • linkreport

Seriously cool chart. Is because this data was from a single Saturday night in August that the passenger count at College Park is so small? On weekends when UM is in session, is there a lot of traffic to/from College Park late at night?

Would be interesting to see the chart for an average Saturday night spread over several months versus a separate chart for averaged Friday nights to see if the patterns are the same. Also charts for the 8 PM to Midnight period would be interesting to see where people are getting on and off.

If the Purple Line was in operation and running during these hours, any thoughts on how it would change the pattern shown on this chart?

by AlanF on Feb 17, 2011 12:01 pm • linkreport

At the end of the day, fares account for <55% of the cost of running metro. At night, that reimbursement rate drops even lower. Are the District, Maryland or Virginia going to raise their Metro subsidies or lower them to maintain existing service?

They're going to lower them. In addition, the local political class is not going to allow you to jack up the fare and exclude the late night working class.

Therefore, late night service will be dropped unless ATU employees are furloughed. Money in, money out.

by eb on Feb 17, 2011 12:06 pm • linkreport

Great chart, with the minor caveat of the use of circles as graphic comparison devices. Even though the legend explains it's about diameter (or radius), not area, the visual impression is confusable.

by KadeKo on Feb 17, 2011 12:06 pm • linkreport

Erik, what an awesome diagram. It's beautiful just as an image.

by M.V. Jantzen on Feb 17, 2011 12:07 pm • linkreport

I'm pretty sure there would be an algorithm that you could make that could infer routes that riders take and then figure out through rates along with transfer station usage.

In my free time I'm going to work on a data model for the format that WMATA uses in the link you posted to see what I can get.

by Drewh on Feb 17, 2011 12:12 pm • linkreport

First off, Great graphic! Mucho kudos!

Although I'm against eliminating late-night weekend service, data can be used to control the costs of this operation. Metro could accomplish this by:

1) Closing sparsely used stations after Midnight (for example, Judiciary Sq, Smithsonian, Fed Triangle and other stations in "dead-zones" that are close to other ones).

2) If the numbers don't come out right for that option, they should perhaps consider running shorter lines, for example Grovesnor-Silver Spring/Ballston-Stadium Armory, etc)

I have a feeling that a lot of people coming in from the suburbs to DC are coming by car anyway, so let's not punish those who are transit-dependent.

by John M on Feb 17, 2011 12:22 pm • linkreport

Yes. It is possible to build a model. In fact, somebody has already done that. Though not for late-night riders.

by Matt Johnson on Feb 17, 2011 12:30 pm • linkreport

Am I looking at this O/D data correctly? Out of 1527 people who entered at Dupont Circle, almost 1/10 of them (148) also LEFT at Dupont? What the heck is up with that? Similar numbers at Farragut North, Friendship Heights...

by MLD on Feb 17, 2011 12:33 pm • linkreport

It looks like there are about triple the riders, but I'm guessing this is because scale is done by diameter or radius instead of area.

How easy is it to replicate this for new data?

by erik on Feb 17, 2011 12:46 pm • linkreport

I'm surprised Foggy Bottom is almost entirely entries; no one returning to campus after a night out?

by Andrew on Feb 17, 2011 1:13 pm • linkreport

I love the graphic you created - I'd just like to point out that some late night Metro riders are coming from or connecting to bus transportation for their final leg home so they're neither walking home from the metro or hopping into their car. For instance, my bus runs until 2 (some bus routes go until 3). It seems an argument could be made for closing some underutilized stations late at night (Judiciary Sq, Medical Center, and Federal Center SW come to mind).
@Andrew - someone above said this data is for a Saturday night in August, so maybe GW students weren't on campus yet?

by grumpy on Feb 17, 2011 1:39 pm • linkreport

"2. Closing stations is important but would have negligible contributions for savings (some labor of the station attendant)"

I think there is a case to be made that with fewer stops you could have trains running with shorter turnbacks. If it takes less time for the train to run from end to end on the line, you can have fewer running and still have reasonable headways. Also if you are skipping one (or several) stations, it seeems that single tracking might be more manageable and expeditious when it is necessary during construction.

by spookiness on Feb 17, 2011 2:04 pm • linkreport

The main savings in closing a station would be the labor you don't have to pay in order to have a station manager on duty there during those hours.

by Alex B. on Feb 17, 2011 2:08 pm • linkreport

I was with you until your last sentence, of "And they make cutting that service even more detrimental to way of life our region has fostered over the last few decades."

First, Metro didn't start late night operations until 1999. thus the service has been operating for 12 years not decades which implies at least 20 years. Second, when late night service started it was only until 1am. Third, from what I've read half of all late night rides are betwween midnight at 1am.

I agree that closing entrances, running fewer trains, and increasing fares (higher fares are already charged) would only confuse people. If it comes down to money and the need to do maintenance then it seems logical to continue late night service but only until 1am.

By the way, I moved to the DC area in 1997 right after graduate school and lived through no late night service and then only limited (until 1am) for much of my prime going out years. I'm also darker skinned so I had issues not being able to get a taxi. It's not that I'm not sympathetic, it is just that I believe this is a hard decision that needs to be made and closing at 1am seems to be a reasonable compromise.

by Arlington Resident on Feb 17, 2011 2:16 pm • linkreport

Can we get some hard ridership numbers for Friday and Saturday nights?

What does the widely reported 13,400 figure represent? Is this average per night or average per weekend?

WMATA states that by eliminating late night service, they will save $3 million. How exactly are they coming up with this figure?

These are basic questions that need answers if we are to make an informed decisions.

by Anon on Feb 17, 2011 3:12 pm • linkreport

"I'm surprised Foggy Bottom is almost entirely entries; no one returning to campus after a night out?"

Most GW students go out in Georgetown, Dupont, or the Farragut Square area, all of which are within reasonable walking distance of campus.

by Phil on Feb 17, 2011 3:43 pm • linkreport

@Dan, thanks for the clarification. Forgot about the lots on the south side of Eisenhower.

by Froggie on Feb 17, 2011 3:54 pm • linkreport

@Andrew -- this is also an August night, which would mean fewer college students in the DC area (especially affecting Foggy Bottom, College Park, and to a lesser extent Brookland, Tenleytown, U Street).

by Jacques on Feb 17, 2011 4:20 pm • linkreport

Great map. It does a wonderful job of illustrating where people are going to and coming from. What's missing is the actual flows. I put a very broad-brush OD analysis together and posted it on PlanItMetro:

by Michael on Feb 17, 2011 5:05 pm • linkreport

Based on this information, metro should be able to realize some savings by closing some of the low-usage stations. The following stations have alternative stations nearby and do not require bus bridges:
Virginia Square
Farragut West
Cleveland Park
Federal Center SW

Close the Medical Center stop and run a shuttle between Medical Center and Bethesda metro.

Close all Orange Line stations between Stadium-Armory and New Carrollton and run timed buses to serve the closed stations from Stadium Armory.

Terminate Blue Line trains at Stadium Armory as well and serve the remainder of the line with buses.

Close all stations between Anacostia and Branch Avenue and run timed buses to serve the closed stations.

There, that's 19 stations that can close at Midnight on Saturday.

by Somebody on Feb 17, 2011 5:06 pm • linkreport


What does the widely reported 13,400 figure represent? Is this average per night or average per weekend?

WMATA states that by eliminating late night service, they will save $3 million. How exactly are they coming up with this figure?

These are basic questions that need answers if we are to make an informed decisions.

I disagree. As members of the public, all of that is interesting information. I love being an armchair analyst. But we should be questioning Metro's goals. What do they hope to achieve by closing the stations? What are the costs?

If there is reason to question the accuracy of their projections, the Board should do so. They should bite the bullet and hire a professional auditor, even. If they fail to do so, the pressure should be on them. But I do not support pretending like we the public out here in the series of tubes can solve the problem for them. We don't have the resources or the responsibility that Mr. Sarles does. We need to give him the tools and the goals. Not the methods.

by WRD on Feb 17, 2011 5:24 pm • linkreport

Awesome graphic. Is someone sending this over to WMATA, assuming that they're too stupid to just rip the image from the blog.

Also, metro needs to hire someone who turns out graphs and data like this. It will help them convince politicians of their needs.

by Jasper on Feb 17, 2011 8:33 pm • linkreport

@jasper they're quite capable of creating graphics like this, they just don't always share.

by Michael Perkins on Feb 17, 2011 9:30 pm • linkreport

Also, if you follow the link from "michael" in a comment left at 5:05p, you'll see that Metro is already paying attention to this post. A planner there has already written about it and addressed some of Erik's points.

The agency certainly doesn't get everything right, but give them some credit, please.

by Matt Johnson on Feb 17, 2011 10:05 pm • linkreport


I understand why you responded as you did. The last sentence of my post should have been worded differently. I didn't intend whatsoever to suggest that we, the public, should be trying to make decisions for WMATA management.

What I'm looking for is simple clarification of some of the figures being tossed around in the debate over late-night Metro service. I ask again. Is the 13,400 figure reported in the Post and elsewhere the number of Metro riders per night or per weekend?

My next inquiry pertains to the $3 million figure WMATA says they will save by canceling late night service. How are they deriving this number? HOW? I'd like to know. And I tend to think a lot of other people would like to know too!

Perkins? Alpert? Anyone? Who's going to step up and provide an answer or two. Come on folks, this is WMATA 101.

by Anon on Feb 18, 2011 1:05 am • linkreport

Continuing for above:

Let's say the 13,400 figure is the average number of late night riders per weekend. (Per weekend is probably correct, but I'd like definitive clarification nonetheless.) At that ridership rate, we then have 696,800 late night trips per year.

WMATA says they will save $3 million by canceling late night service. But what if a late night fare surcharge was implemented to partially offset the costs. A $1.50 per trip surcharge would bring in an additional $1,045,200 a year; a $2.00 surcharge, $1,393,600. Either way that takes a good chunk out of the $3 million estimated savings. A modest surcharge as posited will likely be amenable to most riders. In addition, as the District's population is growing, as is Arlington's, Alexandria's, and elsewhere, late night ridership will likely expand in coming years, producing even more revenue.

by Anon on Feb 18, 2011 1:58 am • linkreport

Many of you are falling into the "little used station" trap by forgetting that a)people take round trips, and b) people use two stations.

10 people might get on at dupont.
2 get off at shady
2 get off at rockville
2 get off at white flint
2 get off at medical
2 get off at glenmont.

If you say "look at all those outer stations, theyre not being used, close them!"....then all those entries at Dupont are killed.

And of course, the 9pm boardings, when those folks were going out are also eliminated because either the entire trip is canceled, or the entire trip is done by car.

by JJJJJ on Feb 18, 2011 2:58 am • linkreport

Erik -- I agree with David. It's misleading visually to represent ridership by radius. Our eyes naturally see area, so we'll naturally overestimate Dupont's importance relative to other stations. Also, many people won't read the legend...and if it's based on area, they won't have to.

by reader on Feb 18, 2011 9:34 am • linkreport

The 13,400 figure is PER NIGHT, not per weekend.

by Michael on Feb 18, 2011 9:53 am • linkreport

Getting frustrated that some people keep maintaining that a fare increase is not feasible - usually the argument is fewer people will pay a higher rate and so the net revenue change will be negative, but now someone in Arlington thinks it will also confuse people, which makes little sense since that person acknowledges that the late night fare is already higher (not to mention it's insulting). Instead of speculating, why can't we have Metro increase the fare and see what happens?

by Arnold on Feb 18, 2011 10:47 am • linkreport

@Michael 9:53 am

Thanks! So we have 1,393,800 late night trips per year.

And, yes, a late night fare increase is feasible. A modest $1.50 or $2 per trip surcharge should find acceptance among most users. Riders will readily understand what the surcharge is funding: their late night ride home. Will there be some grumbling? Of course. But what are the alternatives? Taxi? Cutting the evening short? Not going out at all? Given these choices, the extra $1.50 or $2 expended (or about 1/3 the cost of a beer) is a bargain.

by Anon on Feb 18, 2011 1:02 pm • linkreport

@ Arnold--

Fare complexity is a problem. Metro has to balance competing goals. First, they want people to pay the cost of their ride. Second, they want tie fares to demand (peak-of-peak). Third, they want to encourage Metro use more than they want to revenue-maximize.

Tax economist Howard Gleckman of Brookings-Urban's Tax Policy Center says "Metro has taken to slicing its pricing system like the local deli guy carves corned beef" and he called the results a "tour de force of obfuscation."

Now that's a bit extreme for me, but it illustrates that fares here are more complex than they probably should be. Adding a "peak-of-offpeak" fare after 12 would make a complex, confusing system worse.

Not that Metro shouldn't do it. Management needs to balance the revenue gains with the drawbacks of the added complexity. I think it probably has to happen to keep late-night service. But hopefully it will lead to a better fare structure down the line, or at least better presentation. Hopefully they can trash the ridiculous 50% subsidy to everyone over 65, too.

As to the ridership loss, perhaps Michael from Planitmetro can help illuminate the situation here. We know it should happen. We don't know the magnitude of the delta. Economists call what we're trying to estimate the elasticity of demand or how much people change their behavior in response to a give change in price.


If there are material savings in closing intermediate stations, we should do it. For example, riders might be able to take a cab or bus from Bethesda to Friendship Heights or Medical Center. Or walk. In addition, this might make it more beneficial for cabs to hang around the fewer open stations at night. How would riders respond to these changes? Alas, this is a quantitative question.

by WRD on Feb 18, 2011 1:13 pm • linkreport

WRD, isnt the whole point of metro to not take a cab? It's about Metro + walking.

by JJJJJ on Feb 18, 2011 3:50 pm • linkreport


1) I don't buy the arguments that a higher fare after midnight is too complex for WMATA or too confusing for riders. Regarding the former, they've managed to implement every other fare layer; I don't see how this is the one that would break the system. Regarding the latter, I bet most people aren't that confused by the various layers - because I don't think most people pay much attention to what the fare is; they just swipe their cards. Since this could be a significantly higher fare, WMATA would do some outreach, but I really think riders are capable of processing this information.

2) I know what elasticity of demand is. My point is, instead of us or WMATA trying to guess what it is in this case, let's do something crazy and actually find out. That's a much better alternative to arbitrarily deciding a fare increase is not feasible for one reason or another.

3) While the main goal of WMATA may be maximizing usage (and I'm not sure that's the case), the context of this discussion is the question, should/will WMATA cut late night service? The reason this question has been raised is because it recovers even less money in fares than other service.

by Arnold on Feb 18, 2011 5:02 pm • linkreport

If WMATA saves money by closing underused stations, how many people will take alternate transport from an adjacent station? I think quite a few. It could be cabs, or it could be buses, or walking.

by WRD on Feb 18, 2011 5:39 pm • linkreport

To somewhat explain the PG county data - I live in Largo, used to live near Suitland - lots of buses go to those metro stations and into the surrounding areas. And I often will choose to go to Suitland station when I want to stay on the green/yellow lines. It's only a 15 min drive for me to get there. Suitland also has a parking garage, which feels closer then going to Branch Ave and walking from their giant parking lots.

by cybrbanana on Feb 18, 2011 6:04 pm • linkreport

Thanks for the great map!
I didn't get to read all of the comments so this might have already been mentioned but the bars close earlier (last call 1:30) in VA. Might explain why King St. only has 4% more entries than exits, also might explain some of the unexpected comparisons to Maryland except I don't know when the bars close there.

by Chris on Feb 21, 2011 4:15 am • linkreport

I don't really like how August was used for some of the reasons people mentioned -- schools are out and so is Congress. A weekend in mid-September would be better, because tourists are still around but colleges are in session and Congress is back.

Actually, a lot of DOTs use Octobers as benchmarks for determining traffic counts, because construction is over but winter weather hasn't started yet, giving a more standardized result.

I bet College Park gets a few more exists during the school year, among other stations.

by Alex on Feb 22, 2011 10:03 pm • linkreport

If you want Carlyle District/Eisenhower East become real urban place, the rents for apartments, office space and retails must be lowered. #2, build some roads and bridges in the area. Like connecting Dove st with Mills st, Roth with Eisenhower ave. And build more city grid streets for people to walk and cars to park. #3, selling small unit retail spaces. #4, build pharmacy store and grocery store. #5, stop making any more office building for a while. Too much supply already. #6, relocate the train repair depot on Eisenhower, and build a new metro station over there.

by WL on Nov 25, 2012 4:11 am • linkreport

Add a Comment

Name: (will be displayed on the comments page)

Email: (must be your real address, but will be kept private)

URL: (optional, will be displayed)

You can use some HTML, like <blockquote>quoting another comment</blockquote>, <i>italics</i>, and <a href="http://url_here">hyperlinks</a>. More here.

Your comment:

By submitting a comment, you agree to abide by our comment policy.
Notify me of followup comments via email. (You can also subscribe without commenting.)
Save my name and email address on this computer so I don't have to enter it next time, and so I don't have to answer the anti-spam map challenge question in the future.


Support Us