In January, I wrote a post about Hennepin Avenue arguing that we should Make Hennepin Lovable. (The conversation is ongoing; please share your hopes for Hennepin through this survey.) I argued that the number of human beings using the street is almost as high as the number of cars (with drivers in them) — so let’s share the space in a way more balanced to the number of people.
I spent hours trying to calculate those numbers, but that’s not my day job, so I included a lengthy coda of disclaimers. I was sure the number of people was a lot – I came up with about 32,000 compared with about 20,000 cars on weekdays. I was also sure the real number was not the exact number I offered.
A few weeks later at a public meeting about the project, the City asked, “How should we allocate the space on Hennepin Avenue to create the street we want in our future?” I was surprised when the presentation offered a number of transit users one third the number I had calculated. I was especially bothered because I’m passionate about using good data and never manipulating data to make your point. I have no desire to become a statistician, but I DO think it’s important to be able to read statistics intelligently and to see whether they’re being manipulated or misused.
Did I manipulate numbers? Or were the City’s numbers wrong? Were we using totally different kinds of numbers? Do I know even less about manipulating Excel spreadsheets than I thought?
So, I asked one of my favorite public employees, the one running that meeting (Simon Blenski, if you are wondering), where his number came from. As an example of why I like Simon so much, he asked me to send my blog post and responded to me, looping in his contact from Metro Transit (Michael Mechtenberg). He confirmed that we were using different kinds of numbers, and explained what sorts of numbers they report:
- Boardings (or an alternative total ridership activity boardings + alightings)
- Max load (point along a corridor or route with the greatest number of riders on the bus)
- Mode split (transit use as a percent of all transportation)
- Or even something like bus volumes relative to total traffic (for example, transit carries X% of corridor users while only making up Y% of corridor traffic)
I had used boardings + alightings (every person who got on and off a bus in the corridor), and Simon had reported max load.
Now, I wanted to understand how using those options affect project decision-making, and I wanted use Hennepin Avenue as an example.*
Max Load is the point along a corridor or route with the greatest number of riders using transit.
Talking with Mike, he shared a slightly different concept. He compared it to the bike and pedestrian counts tallying the number of people using a particular way of getting around as they cross a screen line — an imaginary line across the street. This is the kind of number Simon had — the number of people on a bus crossing 12th and Hennepin in both directions a day.
Metro Transit has sensors at both doors of a bus that count people getting on and off. The sum of people getting on and off counts how many people are on the bus at a given point. Metro Transit has samples of this number at every stop along each bus route for every run of the day in each direction.
Mike pointed out to me that counterintuitively, “The load tends to be highest just before entering downtown. That’s the point where everyone who is heading for Downtown has boarded but people haven’t yet started disembarking. Southwest Minneapolis is the strongest market, so counts are highest on the southern end of downtown.” That’s why 12th and Hennepin is the location cited.
What are the pros and cons?
- It’s intuitive! I totally get this number! (My young niece totally gets this number.)
- It’s apples to apples comparing it to counts of cars (or people on bikes or walking) where there’s a counter.
- It can undercount users. If you have “21” at 12th and Hennepin, and you have “21” at Washington and Hennepin, are those the same 21 people? Or are there 6 of the same people and 15 people got off/15 people got on? Who knows!
- It doesn’t capture activity on perpendicular routes that have destinations on the same corridor. That includes the light rail, and — to name a few — routes 3, 5, 14, 19, and 22. There’s no situation in which a person riding the number 5 can be counted.
- The best location for the Max Load count may vary based on project or situational need.
How does this number get used?
Metro Transit uses it to assess whether they have the right level of service on a street. A hypothetical example Mike offered, “We look at loading on the Route 5 as it enters downtown. If it’s consistently overloaded, if the max load is above the capacity, that would be a red flag for adding new service or an additional trip before or after that one.“ It’s also useful for building the case for the importance of transit in a corridor.
The City also uses it to assess and communicate who is using streets and how, the way Simon used it for the Hennepin Avenue meetings.
What is the Max Load for Hennepin?
|Location||Max Load Total|
|12th and Hennepin||8,100|
Total Boardings counts all the people who get on the bus in a given geography.
Here, too, those bus sensors count how many people get on and how many people get off at each stop. Instead of using the counts to calculate how many people are on the bus at a given time, it is the sum of the boarding counts.
Total Boardings shows the activity in the core, unlike max load. The locations with the biggest counts are the destinations people riding the bus are coming from. This is closer to my original data goal, but I used boardings plus alightings — that’s every counted transit-related activity.
What are the pros and cons?
- It captures the number of people who use a given bus.
- It highlights riders’ origins.
- It gives a breakdown of ridership for different times of the week: average weekday, Saturday, and Sunday.
- You get a sense of total transit use in a corridor.
- It’s complicated to understand what the numbers actually mean.
- It’s intuitive — and dangerous — to think of these numbers as people riding the bus, but boardings can’t be equated to people.
- It can overcount people, as riders boarding twice due to transfers are counted twice.
- It doesn’t count through riders – for example, someone boarding in Uptown heading to Northeast.
- It doesn’t count riders boarding on routes perpendicular to that corridor.
How does this number get used?
Metro Transit uses it when identifying where to place shelters, or, in the case of Downtown where every stop merits a shelter, whether to light and heat them, and for sizing them. It can also build the case for improvements for people walking, like curb bumpouts or benches.
What are the Total Boarding numbers for Hennepin?
Including all the stops between 12th and Washington — the reconstruction segment of Hennepin for the project, Total Boardings for Hennepin are 5,637. I was curious about Total Boardings by bus riders using Hennepin stops but that don’t travel along the corridor (like the light rail), and that totals 5,661.
This map shows what they are at different points on the route. They are color-coded by corridor, so orange dots are buses that travel along Hennepin.
Mode Split is transit use as a percent of all transportation, counting people.
This is probably what I was actually looking for when I was working on my first post. This is calculated using City vehicle data counts.
What are the pros and cons?
- It gives a clear sense of what portion of people currently using the street choose to drive, bus, bike, or walk.
- People are very comfortable with percentages in this context.
- There are many ways to look at the numbers. They could be taken at the edge of downtown, an average of every block, or in the middle of downtown. They all yield different numbers, and all are valid.
- Max Load is imperfect as a count of transit riders, so all the “cons” of Max Load are hidden in this number.
- It requires high quality data that isn’t as reliable as the Metro Transit sensor data.
- It doesn’t convey how space gets used — if most people are traveling by bus, but only 3% of the vehicles on the street are buses, that is obscured.
- It represents the status quo, rather than what is possible. Using me as an example, I avoid Hennepin when I can. When I travel by bike it is often the most direct way to get where I’m going, but it’s so scary to navigate the many cars and wide lanes that I prefer to ride twice as far around the edge of downtown instead. On a bus, the slow and unpredictable timing through downtown on Hennepin means I’m better off taking cross-town routes. In other places, there might be encouragement for driving, like free street parking or lots of surface parking lots. Design choices change behaviors, and this number assumes nothing changes.
How does this number get used?
This is a way to explain how people currently use a street. In a corridor like Hennepin, it’s a way of highlighting its importance for people riding the bus.
What are the Mode Split numbers for Hennepin?
Here’s the calculation for 12th and Hennepin.
Step 1: (Total motor vehicles x Average # passengers) + Max Load transit riders + walkers + bike riders = total people
(23,969 x 1.15) + 8100 + 4,340 + 630 = 40,634
Step 2: Max Load [transit riders] / All people = % transit riders
8100 / 40,634 = 20% mode share for people riding buses
4340 / 40,634 = 11% mode share for people walking
630 / 40,634 = 2% mode share for people riding bikes
Which numbers go into this formula determine what comes out of it. For example, should you use the same location for all ways of getting around, like I did here with 12th and Hennepin numbers? Or in this context where we are exploring corridor design, should you use the highest counts along the corridor for each option? (Biking and walking is 4,000 just a few blocks north of this point.)
Bus Volumes Relative to Total (Motorized) Traffic
Bus Volumes relative to Total [motorized] Traffic is used as a percent of all vehicles, not accounting for how many people each vehicle carries.
This one is similar to the Mode Split, but it’s got greater precision because it is easier to accurately count vehicles than to accurately count the people IN vehicles.
What are the pros and cons?
- It highlights the value buses offer in a congested corridor.
- It ignores that not everyone uses motorized traffic.
- It’s only as current as your traffic count data, which in this case is six years outdated.
How does this number get used?
Metro Transit uses this to make the case for the importance of transit to ease congestion in a corridor, and it shows they have an oversized role compared to their footprint on the streets.
What are the Bus Volume Relative to Total [Motorized] Traffic numbers for Hennepin?
The calculation at 12th and Hennepin, counting buses traveling in both directions:
Bus/Motorized vehicles = 464 / 26,939 = 1.9%
The calculation at 3rd and Washington (using the higher vehicle number from south of Washington):
Bus/Motorized vehicles = 538 / 21,176 = 2.5%
Putting Mode Split and Bus Volumes Relative to Motorized Traffic together, you know that at 12th Street and Hennepin, transit carries 20% of the people with only 1.9% of the vehicles.
So What Did I Learn?
I’ve learned a few lessons writing this up:
- These numbers feel precise and meaningful, but they miss important details. When I was learning how each number is different, I kept asking, “But what about…” None of them capture the complexity of what happens on our streets. I’d argue that none of them give a picture useful for informing design decisions like what we’re exploring on Hennepin today.
- Each shows and hides different things. Having marched through all these options, none do what I wanted in my initial Let’s Make Hennepin Lovable post.
- Sometimes the numbers are used in ways different than intended. For example, Metro Transit uses Max Load numbers to assess service levels. The City used them to communicate transit ridership.
- While there is a lot of data, getting and understanding it is really, really hard. As a streets.mn writer and masters degree holder (I took stats), I’m better prepared for this than your average Minnesotan. It took me a month to write because I got so overwhelmed by the 36,586 row spreadsheet of raw data. I was able to access that data only due to lots of direct communication with Minneapolis staff. And I had 7 experts volunteer to help me navigate it.
- Choosing what number to share in a public meeting or on a project website matters. Offering specific numbers without context creates a false sense of understanding.
The next time I’m in a public meeting, I’m going to ask what numbers are being shared, where they come from, how reliable the underlying data is, and when it was collected. And, I’ll still listen critically.
*All numbers in this post were collected prior to the Nicollet Mall detour and reflect normal transit operations on Hennepin Avenue.
Thank you to the people who helped me turn my post concept into an actual post. In particular, Michael Mechtenberg and Metro Transit offered a couple of hours of his time to orient me, get me data, and help me use it responsibly. Simon Blenski of the City of Minneapolis answered my questions and connected me to Metro Transit. Aaron Isaacs helped me to understand how transit planners think, and Alex Cecchini offered comments on the draft post.
Technical question: How do the sensors distinguish between people boarding versus alighting? I probably see 2/3 of people getting off the bus using the front door.
I think they add up the total movements through both doors and divide by two.
Recently learned an interesting fact that layovers points have unreliable data because the counters can’t distinguish between drivers getting on/off for a bathroom break and passengers boarding/alighting.
There are multiple beams across the doorway. The order that the beams are broken indicates the direction of motion. This works for single stream (narrow) doors, but is less effective for wide doors, like on light rail cars.
Light rail passenger counters use an entirely different technology. They are based on a combination of active IR (bouncing light off objects) and passive IR (sensing the heat given off by objects) and are mounted above the doors to create a virtual curtain detection zone.
AADT doesn’t really capture “Number of People in Cars Using the Street” either. I’d speculate there’s not a lot of people driving across the Hennepin Ave Bridge and then continuing into Uptown, so the users on the north end of the street are likely to be different than the users on the south end. Also, more than 1 person can fit in a car, the numbers used for the St. Croix Crossing related studies found 1.2 occupants per car, I don’t know if that would also apply to Hennepin or if it might be higher because of incentives to carpool.
A more accurate way would be to do original/destination studies on both ends of the street and then figure out how many people are in an average car.
That’s true, Monte.
Re: # of occupants per car — In the calculation of total people (under Mode Split above), you’ll see that there’s a multiplier assuming 1.5 occupants per car. Mike said 1.1-1.2 was a good rule of thumb, I couldn’t find a better number, so I ussed that.
Re: whether AADT is a good number or not given those entering and leaving the corridor, the same logic applies to bus and train users, so using a screen line calculation for everything incorporates those same problems for people traveling no matter how they get around. Bruised apples to bruised apples is better than apples to oranges, I think. I share your dissatisfaction with those numbers, though!
Great work, Janne!
Seconding this. Always a appreciate a smart, clearly written article!
Thanks! I struggled a lot with this one, and I’m happy with how it turned out.
*Wanna-be streets.mn writers, often those with less expertise are better at making expertise accessible — that’s why we want YOU writing here, too.
Unfortunately, the 3rd Avenue protected bikeway project demonstrates that often numbers don’t matter for decision makers. Minneapolis Public Works’ own traffic studies found that a 3-lane would work just as well as a 4-lane concept. However, a majority of Council Members and downtown business leaders felt differently and prevailed.
I had an outstanding inquiry to the city about updated traffic count numbers, that I just received this morning. My question was about vehicle counts for the last two numbers (mode share and bus volumes relative to motorized traffic).
Below is the quoted update info from the City of Minneapolis (Thanks, Simon!). For a summary of what it means
–the traffic counts from earlier this year are quite a bit lower than the ones available on the City’s website. That means that the mode share for people walking, busing, and biking are higher than listed in the original post.
–the City used the highest volume for vehicles and the highest volume for buses (unlike using a single point as I did in the original post)
–the estimate of rider/vehicle is higher than what I used, but it is also 10 years old and may or may not be accurate.
Updated mode share numbers, using the highest point motorized vehicle count and the higher passenger per vehicle number:
-22% people riding transit
-12% people walking
-2% people riding bikes
Updated bus volume relative to motorized traffic:
-Buses are 2.5% of vehicle traffic and carry 22% of people at the highest traffic volume point on Hennepin in downtown.
-Buses are 3.7% of vehicle traffic and carry (we’re not sure what % of people) at Washington.
The City collected additional motor vehicle data in the fall of 2015 – after the completion of the LaSalle Ave reconstruction project and prior to the Nicollet Mall detour being routed to Hennepin Ave.
· The highest daily count was 18,600 between 6th St and 7th St
· The lowest daily count was 15,600 between Washington Ave and 3rd St.
· The rest of the blocks were all in the range of 16,000-18,000.
This counted vehicles, not people. To estimate the number of people in personal vehicles, we typically multiply by 1.26, which is the average occupancy of a personal vehicle entering downtown Minneapolis. I should note that the 1.26 factor is about 10 years old and may not be representative of current conditions.