“Frequency is freedom” is one of the slogans I’ve used on this blog and in my book. Charles Montgomery, author of the forthcoming book Happy City (Farrar, Straus, and Giroux 2012), asks “how frequent is freedom?” and goes on …
I’m writing in hopes you can answer a particularly vexing question. … I have tried and failed to find empirical evidence showing the transit frequency at which users can simply show up without consulting schedules or feeling anxious. All I have found are dense reports about ‘elasticity’ and then a range of planners’ ‘gut feelings’ suggesting anywhere from 5 to 15 minutes.
Have you found any evidence suggesting a certain headway crosses a threshold of reducing the friction and perceived difficulty of transit enough to change the game at any stop?
Elasticity is the ratio between an outcome and the variable that’s supposedly affecting it. We say that the “elasticity of frequency is 1.0” if every time you double a frequency, ridership also doubles. If ridership goes up only 80% when you double a frequency, that’s called an elasticity of 0.8. Obviously, modelers of elasticity have to sort out a lot of different variables and seek elasticities for each. You’ll find discussions of elasticity of frequency, fare, travel time, and a range of other transit variables in the literature.
But the concept of elasticity implies that there’s some linear relationship between frequency and ridership outcomes, describable by a ratio. This in turn implies that frequency affects ridership in the same way at all levels of frequency. Most transit planners know that isn’t right, that there’s a “phase-shift” in the relationship at the point where the customer stops planning their life around a timetable.
It’s very tough to make a hard case, because the only “hard” data on these things comes from observing the effects of frequency changes, and as you can imagine, other things are usually happening at the same time as a frequency change. It also takes a year to see the results of any service change fully manifest, and by then other stuff has happened that muddies the data.
So I’m in the “planner with a hunch” category. I can tell you that the top-performing all-day routes in most networks I’ve studied (top performing in riders per unit of service) are mostly high-frequency routes in dense areas. Both the frequency and the density are important. And in most cases, the threshold of “high frequency” is around every 15 minutes all day. Lower frequencies are usually associated with much lower performance.
One big caution, though, is that our tolerance for frequency varies with trip distance. A 15-minute wait for a one-hour ride feels a lot different from a 15-minute wait for a three-minute ride. So we generally aim for higher frequencies where we’re aiming for shorter trips. However, frequency is so expensive that it’s often affordable only where the capacity is required.
If any other reader has seen the definitive study on this phase-shift let me know. My hunch is the case will be made by comparing different frequencies of service through similar land use, rather than by comparing the results of frequency changes. But again, that’s a hunch, and defining “similar land use” presents its own thicket of difficulties.
Among the things I like about this post is that it avoids the phrase “tipping point”! A little breath of fresh air.
Another discussion of transit value that doesn’t take into account the fact that people aren’t just choosing transit or no transit, they’re usually choosing transit or some other form of transportation.
Your “three-minute ride” is another way of saying that the transit trip isn’t competitive with walking. The biggest factor is whether it’s competitive with driving, and there are all kinds of factors that go into that: road congestion, gas prices, parking availability and price, etc.
Not a perfect answer, but a survey-of-the-literature paper from 2004 suggests the short-term elasticity from transit service as 0.5 to 0.7 and long-term elasticity as 0.7 to 1.1. Of course, “service” can be more than just frequency.
The paper is here: http://www.nctr.usf.edu/jpt/pdf/JPT%207-2%20Litman.pdf
Here in Toronto for our subway system the magic number seems to be 6 minutes, and headways are usually much better, e.g., under 3 minutes during peak. This is the case from 6 a.m. (9 a.m. Sundays) right to 1:30 a.m., 365 days a year. Obvisouly this means running empty trains some of the time. At one point the major east-west line headways were changed to better match demand and riders complained loudly about an 8-10 minute headway. The max 6-minute headway was hastily restored.
Nowadays it’s big news if a delay of some sort means waiting 10+ minutes for a train, and in the core of the city the dominant fare medium is the monthly all-you-can-ride pass, so riders expect very convenient service even for short trips.
Personally, I think anything more than a 10 minute headway, even for surface routes, is the freedom drop-off point. That’s when people start memorizing schedules and frenetically checking their watches. That may sound like a high standard but it comes with a caveat: frequency is only as good as headway + average headway variability. If a transit service can run a bus every 15 minutes on the dot at every stop, kudos to them, but I have yet to see it in practice. So people quickly realize if they have a 10 minute trip on a route with a 15 minute headway, they probably need to allow 15 (headway) + 3 (average headway variability) + 10 (trip) + walk time = already 28+ minutes in order to never be late.
One need only look at the ubiquity of 10-minute headways in Germany, Austria and Switzerland for support for this being a marker of truly quality services.
One good rule of thumb from a retired civil engineering prof (JJ Bakker) was that the headway should be no longer than the journey time. For short trips you may be better off walking, in more ways than one.
For regional service, I’ve been starved for decent service from living in the US that I consider the Nice TER’s half-hourly trains fine. The quarter-hourly peak trains are much better, though.
For local service, the frequency has to be much higher – say, 10 minutes at the absolute worst, and only for work trips rather than short errand trips. When I lived on the Upper East Side and needed to get back from Columbia late at night, I’d usually take a taxi and pay $15, and only occasionally take the subway and walk across the park and pay zero. Part of that comes from the disutility of walking (I’d take the subway to a crosstown bus before 12-1 am, when the M72 stops running and M66 frequency slips from once every 10 minutes to one every 40). Part of that comes from the subway’s 20-minute service.
The TTC in Toronto defines “frequent service” as every 10 minutes or better. This is the point at which specific schedule times are no longer listed at the stop, but rather only the duration of frequent service (until 10pm daily, for example.) I like this threshold because while I would certainly wait 15 minutes for a bus without checking the schedule first (the standard employed by most suburban transit authorities around Toronto) the 10 minute standard allows for a bus to be quite late and still not leave me waiting more than 15 minutes.
I agree with Michael that looking for a “tipping point” is the wrong idea. Just like walksheds, everyone has their own threshold for how long is okay to wait, and it can vary based on mood, weather, whether or not they’re connecting, how far they’re travelling, how far they have to walk to get there, how likely the vehicle is to be late, and how much they’ll suffer if they miss their trip.
To illustrate the latter point, consider air travel. Many people routinely arrive at the airport in time so that they have an hour between when they get through security and when their plane is expected to depart. Can you imagine arriving at a local bus stop an hour ahead of time?
In my experience, 20 minutes is approximately the upper bound for what can be considered clockface service. If a bus comes every 20 minutes, then there will be some people, especially on a nice day, who won’t bother to check a schedule before walking to the bus stop.
On the other end of the spectrum, I’ve never met someone who wants a timetable for a service with 5 minute frequency. (But I have met many people who want a timetable for a train that comes every 7.5 minutes.)
At least in Seattle, the ubiquity of real-time bus tracking (OneBusAway) suggests that 15-minute frequency is most definitely not freedom. Then again, the relative reliability of this schedule probably has something to do with it. For example, the 71/72/73 to the University District runs every 10 minutes, but is very unreliable. Without OneBusAway, most people would treat the service as random. But with OneBusAway, you can wait inside a warm cafe or store until you know that the bus is 2-3 minutes away.
I guess the real lesson in all of this is that people hate waiting more than they hate following schedules.
One other thing to consider is transfers: frequency is at least twice as important when you’re making a transfer, because you know have two wait times to deal with, and a potentially unreliable arrival at the connection point. Having a turn-up-and-go frequency on both sides of the connection multiplies the effectiveness of the network without the compromises that are involved in having timed transfers, which include potential cascading of delays, or knowing that if you miss the transfer, you’re guaranteed a worst-case wait time.
For myself, I would say the threshold is 10 minute frequency. At that point I would stop bothering to check the schedule. That said, real-time arrival info on my phone (along with Google Maps transit directions) has removed the need for a schedule. We should remember that most people still do not own smartphones, though.
The 10 minute frequency is definitely more important in making connections. I can deal with planning my commutes and other trips around schedules within reason, but if a connection takes more than 10 minutes it just feels like a huge waste of time. If two lines cross, both with 10 minute frequency, the average wait is 5 minutes which is pretty reasonable.
I would say between 10-15 minutes.
Here the frequency is simply changed and the bus route gets signs saying it was upgraded.
Elasticity appears to be close to 1 for this case. Doubled frequency – doubled patronage. But obviously it is not linear, it is more like a 1/x decay function.
http://ses.library.usyd.edu.au/bitstream/2123/6058/1/thredbo10-themeA-Warren.pdf
What you have called ‘elasticity of X’ for short, sometimes also seen as ‘X elasticity of demand’, is more properly called ‘elasticity of demand with respect to X’ (X = freqwuency, price, speed or whatever). I would encourage people always to use the last phrase to aid the understanding of noneconomists.
This is a great question precisely because it’s so critical to network design, yet most planners rely on their gut to answer it. It would not be hard to find an answer through a well-designed survey that simply asks observed transit riders whether they consulted a timetable or not before their trip. You’d need to run the survey in more than one city to get good results, and know some other information such as the mode, perception of reliability and safety, and frequency of use.
I think there are probably three states of freedom from the timetable. The greatest freedom is a high-enough effective frequency (adjusted for reliability) to just amble onto the street with no planning, the middle ground is clock-faced headways which can be remembered without needing a timetable, and the shackled option is anything else that requires a timetable.
I’m in the majority here who sees a threshold at 10 minutes as the point when I can just wander to the bus stop. But I also agree that if I lived in Toronto where 10 minutes may seem like a long wait time my opinion would clearly be different.
And since high frequency buses are more prone to bunching, hopefully a near-future topic will be the management of bunching buses.
I’m a sucker for empirical judgement devoid of bias or emotion, but this is one case where going by the gut will (in the long run) be a better indicator than any study.
An elasticity curve is not some magical construct…it is an aggregated result of the preferences of multiple people. We could try to define this using surveys or even past experience, but we would end up with extreme generalizations that may not even hold true from one city block to the next.
In Tegucigalpa Honduras, frequencies on some routes can be measured in the seconds…and yet the inflection point for elasticity is probably in the hour range (less than 10% of the population can afford a car). In Cleveland, frequencies hover around 20-30 minutes, and yet I would place their inflection point at around 10 minutes. That is a pretty wide range.
Even comparing some routes within a city, you would find major differences in that inflection point…depending on things like the utility of the line, wealth of the passengers, density, parking availability, etc. I know people that would ride MUNI all day every day, but would rather slit their grandmother’s throat than ride on BART’s disease infested seats. Those completely frequency-agnostic preferences will affect your frequency inflection points as well! There will never be a study that can tell you those kinds of details.
If you want to find the true tipping point where ridership takes off and gives you a return on your investment, you have to test. You act on a hypothesis (a hunch), spend the money to test it out, and if it doesn’t work, you stop the experiment. If it does, you refine your hypothesis and test further. And as any good scientist or serial entrepreneur would tell you, you are never done testing.
Wow – but most of the tests people are most interested in involve rail transit lines with multi-billion dollar price tags. There’s a lot of gut feel powering many of those projects. Even reaching 10-minute bus service requires millions of dollars annually. Do we really need to just try things out to see how people behave all around us all the time?
Amazingly enough, Los Angeles Metro is doing a study of this right now! Well, I don’t think it is a scientific study, but we could treat it as one.
The Red/Purple Line subway, and Blue Line trains, have had their evening frequency doubled recently. Now, instead of 20 minute headways, each line is running at 10 minute headways on evenings. Perhaps someone can do some passenger surveys and analyze ridership data to get more information:
http://www.metro.net/service/advisories/update/rail/143/
“Blue, Red & Purple Lines: More frequent late-night service begins Sunday, November 13. From end of PM rush thru Close, trains will run every 10 minutes”
Rob –
Testing IS expensive, and it always will be. Some of the costs are inevitable, but some can be mitigated through good management practices and some can be mitigated through some creative uses of resources. Not testing, however, ensures that your existing resources are poorly utilized.
Here’s a link to some research we did in Zurich about passenger arrival rates at transit stations. Interestingly we found that people seemed to consult schedules even when service was very frequent (but also very reliable) … apparently they did not want to wait even two minutes for the bus!
http://www.andynash.com/nash-publications/Luethi2007-pax-arrivals-TRB-paper.pdf
The point where you no longer memorize departure times is probably around 10 to 12 minutes interval.
On the other hand, the interval is not the only dimension for the transit mode decision. It is IMHO even more important to have a well-connected system attached, which gets you “anywhere you want” in a straightforward manner. This means that timed connections are as important.
Transport for London assumes 12-15 minutes headway is where a bus service user changes from looking at a timetable to just turning up (hence timetable and scheduling changes from displaying and achieving a timetable to displaying and achieving a given headway).
In terms of demand elasticities – have you seen this:
http://www.demandforpublictransport.co.uk/TRL593.pdf
Professor Peter White at Westminster Uni will talk about this type of thing till the cows come home
I might be missing something here, but surely the average passengers per unit transit (eg. per bus) averaged over a time period (say an hour) will have a point where it peaks (probably around the 10-15min mark if gut instinct is correct). Of course measuring this is all but impossible without testing on a route that is already high frequency without timetables.
Vuchic’s 2005 Urban Transit text lists the following desirable headways:
– up to 6 minutes for “short urban trips”
– 6-12 minutes is “still satisfactory for urban trips up to 5-10 km”
– up to 20 to 30 minutes “for longer trips, such as regional commuting trips for which passengers use schedules”
Before looking at this, and at the risk of looking greedy, my initial thought was 5 minutes. Primarily anecdotal and based on personal experience. Failing that, 10 minutes, but even at 10 minutes I would still try to match the schedule (e.g., the bus that takes me to the subway in the morning, which runs every 9 minutes). There is a definite difference in how I view waiting for a 5-minute service vs. a 10-minute service, even if both would be classified as “frequent”.
Upon reflection, I wonder if this is because a big part of the discomfort with waiting is actually related to the uncertainty of the wait rather than the length of the wait. Here’s an example. Often you can see an approaching bus while it is still 2 or 3 minutes away from your stop. There is a definite difference in perception once you can actually see the bus — it is almost as good as being there. There used to be a saying in the “golden era” of transit service: “always a [street]car in sight”. With a bus every 5 minutes, you’re not far from that standard.
Andy, thanks for the paper.
May I propose an explanation for your observations? The curve you use for average wait time shows a sharp departure from the half-of-headway rule for 10- and 15-minute routes. Those are the routes that are easiest to run on a clockface schedule. When the routes have an easy, memorable clockface schedule and the buses run on time, people will time themselves to the buses.
I did just that when I lived on the Upper East Side and needed to use the M72, which at the time ran every 10 minutes all day, stopping at my station at xx:x5. The station was near the line’s beginning, so schedule adherence was reasonable, though some runs were outright missed.
I’d be interested to see how people behave on 10-minute routes that aren’t on an easy timetable (perhaps the buses come every 11 minutes) or aren’t very punctual.
@Alon: When I lived in Chicago, where a lot of bus routes and L lines have around 10-minute headways but don’t hold schedules particularly well, I almost never looked at a schedule or bus tracker (the bus tracker was launched shortly before I moved away), unless it was late at night.
Currently in Seattle, even during the peak, I sometimes use OneBusAway to help optimize my trips. But that’s really situation-specific. I didn’t have a Chicago analog to my morning commute, where I have a choice of using a 10-15 minute route a couple blocks from where I live, or walking two stops down the route (a little less than a half-mile), to a stop that route shares with two other less-frequent ones. Or to my evening commute, where I can choose between a handful of routes that drop me tolerably close to home; I’ll get on a 16, 26, 28, or 358 unless the 5 is within 3 minutes of arriving (based on that you can probably put a pin on a map within a couple blocks of where I live).
Your behavior on the M72 is a lot like what I did when I used the ST511, which runs a 15-minute clockface schedule and is reasonably reliable.
@Alon: Heinrich Brändli, the founder of the Institute for Transportation Planning at the Swiss Federal Institute of Technology in Zürich (I think that’s the name) was chief of the Transit system of the City of Zürich (VBZ), and he kind of made it an absolute rule that the interval time MUST be a subdivision of 60, leading to the same departure times every hour. So, that gives “legal” intervals of 60, 30, 20, 15, 12, 10, 7.5, 6, 5, 4 minutes. (in the VBZ timetables 7.5 minutes is shown as 7/8 minutes.)
I still remember a speech of Brändli, where he stated that the best way to kill a transit system is to operate it at 17 minute intervals (13 minutes is the second-best, and 11 minutes is the third-best).
For irregular operation (because of traffic overload), people nevertheless show up at the stop around the published departure times, simply getting more irritated the longer they have to wait. And in Zürich, the political pressure rises very fast to have the situation improved… be it by creating strategically located bus lanes, traffic signal priority, preemptive clearing of intersections (meaning that the transit vehicle announces it even earlier than for simple priority, so that the signalling allows to clear the intersection), etc.
It might be interesting to know that at the VBZ the communication unit (managing the communication and information exchange with the control center) displays deviations from the schedule in half-minute steps, and the control center always has an overview over the status of the vehicles on a line.
A research for the Paris Metropolitan Region implies a threshold around 10 minutes (6 veh/h) for passengers to arrive at the station without consulting the schedules. That holds mainly for guided transport modes which have a better regularity than buses.
Generally in transit modelling, we assume a high frequency service when we have no less than 5 veh/h (12 min regular HW)
Max will hopefully correct me if im wrong, but I believe Zürich has taken things a step further and operates with multiples of 7.5 minutes. The thinking is that 7.5 minutes for the whole day is a “full service”, which allows spontaneous travel (freedom). 15 minutes is a “half service”, which still offers frequent service for lighter loads and facilitates splitting a line. Key bus, trolleybus and tram routes all operate at 7.5 minutes.
30 minutes and 60 minutes are used for regional routes. These include the regional trains with timed connections with each other and feeder buses at regional hubs.
Vehicle size and network design are used to keep the frequency at the desired level. There is a wide range of street running vehicles from short buses to long trams. In really central areas 7.5 minute services will be overlaid for capacity. On the other hand too many overlapping services are avoided where possible so that the traffic light priority system can works as well as possible.
A consultant from Bern I talked to considered this 7.5 minutes real freedom. According to him travel increases significantly when a network at this frequency allows spontaneous travel from early till late. I’m sure that expectations vary in different places and for different kinds of trips. Trip length has already been mentioned and is almost certainly a significant factor. Reliability is another one.
@Lauri: the 7/8 minutes interval is indeed for during the day. However, it is not the base of the timetable; it is stil the “fraction of 60 minutes” which prevails. Of course, the 7/8 minutes interval ensures good connections with other bus lines which operate at 15 minutes intervals, or double up well with, for example the line 10 operating at 15 minute intervals to the airport (one flight ehh… vehicle to the airport, one to Bahnhof Oerlikon etc.).
There are bus lines operating at 10 minute interval, and some at 20 minute interval (one comes to my mind where the transit time is 8 minutes, which means that with a 20 minute interval, one vehicle is sufficient for the service (and, yes people memorize the schedule, or have the schedule for the closest stop pinted out an pasted on the pinboard at their office).
So much for the city of Zürich; All along the ZVV, the S-Bahn operates at 30 minute intervals (with some exceptions, where it still is 60 minutes), but in many places, more than one line are run, providing an overlay. The connectiong bus services operate, depending on the importance and traffic of the line at 60, 30, or 15 minute intervals.
For the morning and evening peaks the intervals are shortened; one streetcar line in Zürich operated at even 4 minute intervals, but that was abandoned after a second line overlapped to almost the end of its busier section. That service got the Bahnhofstrasse at capacity where that line was overlapping three other lines at 6 minute intervals.
Trip length is a factor, but I have encountered many people still taking the tram between Hauptbahnhof and Oerlikon, even if the S-Bahn were faster, and would have made it possible to catch up one or two tram services of the line. Maybe old habits… Definitely in Zürich, where the network is so interwoven, reliability has a higher importance.
Some anecdotal evidence showing the actual calculus of a real transit rider. I just moved from one city to another and my behavior changed based on the frequency of service (and some other factors).
In city A, I walked 10 minutes from home to the station. The trains went 4 times an hour with 19 and 11 minutes in between (alternating), and my ride took half an hour. I then had an approximately 18 minute walk to my office. I would time my walk to try to just make the train with a wait of 19 minutes, but if I missed it I wouldn’t feel so bad about getting a coffee or sandwich while waiting for the next train 11 minutes later.
In city B, I walk 3 minutes from home to the station. The trains depart anywhere from every 5 minutes to every 15 minutes and I still haven’t figured out the schedule 3 months later despite taking the trains every weekday. I could check online if a train is leaving soon but don’t usually bother, and the website shows only one train at a time by default. My train ride is 18 minutes, after which I have a 15 minute walk to my office. I do not try to time my arrival at the station, but get slightly annoyed if I have to wait more than 7 minutes or so (there is no cafe in either of the stations). I live in a global city, and car ownership is not an option, so I really don’t have an alternative aside from moving to the suburbs where my work is located. I’m not very interested in that option.
It seems that the difference between 15 and 20 minutes makes the difference for me in consulting the schedule. My current city has more frequent, but erratically timed service. The frequency pretty much makes up for the irregular timing. Another factor is that I have more control over exactly when I arrive at the station when my walk is shorter. If the walk is longer, I might get stuck at a traffic crossing for a couple minutes, run into someone and have a conversation on the street, etc. I can consult the schedule, but I still might have to factor in 5 minutes or more as a buffer to be sure to make it on time.
I much prefer city B, but would like it more if there was a cafe in at least one of the stations for the times that there is a bit of a wait, or for when I’m in a rush and want to buy a sandwich to eat at my desk for lunch. If my commute were shorter, I might consider using city B’s free bicycle share program instead of the metro.
This is about as anecdotal as evidence gets.
I live in Montreal, and on a bus route that changed around 18 months ago from being every twenty minutes except at peak times to every ten minutes from 6am to 10pm. It made a huge huge difference to how free I feel and how I use it. I do time myself to the bus — it runs at 06,16,26,36 etc. But I have no idea of the times coming home, I just trust that there will be one fairly soon when I get to the metro, and there is.
What it hasn’t made any difference to is how often I use it or how much money the STM get from me. I have always bought a monthly pass, I still do. And I don’t travel any more — I always used it for everywhere I go. But it has made a huge difference psychologically. Before the change, I was a reasonably satisfied transit user, and now I am a happy one. How could anyone measure that?
I think the answer to this also depends on which leg of your journey the route in question is, and how easily you can get to it.
For example, my trip to and from work involves taking a bus to the subway station and taking the subway right to my office. The bus stops right in front of my house every 15 minutes. You better believe I plan when I leave for work to that 15 minute schedule. In this case, 15 minute service isn’t frequent enough.
To get home, however, the subway ride is long enough that so many factors can affect if I catch the bus I was planning for or not. My 30 minute subway ride, even though it comes very frequently, doesn’t allow me to plan for which bus I take back home. I could leave work the same time everyday and sometimes catch my bus and sometimes not. If I leave earlier, sometimes I’ll catch the original bus I planned for and sometimes I’ll catch an earlier one. Now, 15 minutes is definitely frequent enough for me not to care.
Joe just reminded me of some trips I used to take home from UBC in Vancouver that further the point of which leg of the trip we’re talking about, and what type of service one hopes to expect from different lines.
On these trips, I needed to make a critical connection to either the last commuter train of the evening, or the last commuter bus at Waterfront Station. The route options would be to
1) Take one of two trolley buses direct to downtown, departing on 15 to 20 minute headways.
2) Take either the 99 B-line (6 or 12 minute) or 84 (20 minute) limited busses and transfer to either the 98 B-line (12 minute) or a local.
3) Take a limited bus (44) direct to Waterfront leaving on a 30 minute headway, (if it was still running).
Given transfers and walking distances, travel time was about 35 to 40 minutes for all the options with the above list decreasing in both travel time and schedule reliability. I’d generally give myself 50 minutes to make the trip. First, I’d check the 99 B-line schedule (the most frequent route), then the 84 and 44’s schedule, and if none were acceptable, I’d walk over to the trolleys.
So, what was an acceptable schedule?
I needed to consider the risks of schedule adherence and the likely wait for a transfer. So,
If 44 was leaving within 7 minutes, I’d take it, and hope it wasn’t badly delayed up Burrard St.
If the 84 or 99 were departing within 6 minutes, I’d take them, preferring the 84. The 84 had a very long walk in order to transfer to the downtown buses (almost 300 m), but it had better schedule adherence and higher frequency when transferring onto Granville. (14 local busses per hour vs. 3 local bph when transferring from the 99).
With more than 6 minutes to spare, it was worth walking to the trolley loop to check if they were departing within 5 minutes. Failing that, I’d walk further to the village bus stop where both the trolleys and the 99 stopped.
7 minute headway is sufficient. In my case, schedule reliability or speed became the dominant factor in route choice once this frequency was achieved. I definitely had a preference for a faster headway than 12 minutes, or the option of a direct but slow route.
Anyways, I always caught my train; although a had few close shaves.
TCRP #100 defines Frequency Level of Service as A for 10 minutes or less. LOS C (15-20 min) is the limit for attracting choice riders.
I have used Frequency LOS extensively and most boards and managers all want a higher grade. I have not been able to quantify increased ridership solely with improved Frequency LOS because there are usually other improvements that occur. We often solve the reliability problem and crowding problems at the same time, so there are several factors that affect ridership that are changed at the same time.
Data from anyone who has improved frequency and nothing else would be greatly appreciated.