Districts across the country have faced pressure from families and community members to shift school start times so that high school students can start school later. With pediatricians recommending that adolescents get more sleep and some emerging research showing that shifting times can improve learning outcomes, proponents of delaying school days for teens have made a strong case for such changes.
But recent efforts to accommodate these requests and respond to the research demonstrate just how complex transportation planning is at a time when students have growing options on where to attend school.
Two years ago, the Boston Public Schools (BPS) challenged computer scientists, mathematicians and researchers to take on the complex problem of rearranging school schedules. But with more than 600 buses traveling 45,000 miles per day, the district’s transportation department wanted to carefully consider the implications that shifting start times — and therefore bus routes — would have on students, families, drivers and the budget.
“We were trying to understand what would happen if we changed our times,” said William Eger, a strategic project manager for BPS. “We were not quite sure how to do it in a way that would keep transportation costs the same or lower.”
Existing software programs, he said, did not have the capability to map out what could be trillions of bus route scenarios. And even though he’s a former math teacher, Eger — and the district’s other transportation team members — didn’t have the mathematical and operations expertise needed to solve such a massive puzzle.
The winning proposal came from researchers in their own backyard — doctoral students Arthur Delarue and Sébastien Martin, who work with Professor Dimitris Bertsimas at the Massachusetts Institute of Technology’s (MIT) Operations Research Center.
In math terminology, they created a mixed-integer optimization algorithm — what Delarue calls a “giant calculator” — to generate start times that would have resulted in roughly three-fourths of the district’s high school students beginning their school day at “a more developmentally appropriate time,” he said.
But the new start times were never implemented. The scheduling changes that many families would have experienced were viewed as too disruptive and sparked significant backlash. Eger said leaders decided that more “community engagement” is necessary for parents to understand the tradeoffs involved.
Continuing efforts to shift start times statewide in California are sparking debate over whether it’s an issue that should be left to local school boards. And opposition to proposed changes has erupted in other communities, with parents concerned about issues that include the impact on after-school activities, transportation needs and meal times.
“When your kids are affected negatively, it is hard to see the big picture,” Bertsimas, the operations research professor at MIT, said recently when he, Delarue and Martin were recognized as finalists for the 2019 Franz Edelman Award from the Institute for Operations Research and the Management Sciences (INFORMS).
Tabling the start time changes, however, didn’t mean the end of BPS' partnership with the researchers. The MIT team has continued to work to increase transportation efficiency for a district in which some schools draw students from 20 zip codes, John Hanlon, BPS chief operating officer, said during the INFORMS presentation. Bertsimas is also is working to implement his software program in other districts across the country.
The role of transportation policies
Transportation is critical to many of the challenges facing districts, such as integrating schools, allowing families to choose the schools they want their children to attend and getting students with special needs the educational services they need. With increasing attention to equity, district leaders are also considering which students are most affected by long bus rides or the least-preferred start times.
A report published last fall by the Urban Institute discussed how expanding choice policies have made school transportation more complex than ever — for districts as well as families.
“Travel time, distance to school, safety en route to school, the availability of busing and transportation costs may all influence whether families can consider sending children to school outside of their neighborhood,” wrote Carolyn Sattin-Bajaj of Seton Hall University. “Transportation policies also play a role in how schools operate, including recruitment strategies, budgeting and even decisions about whether to open new schools.”
Think tank Bellwether Education Partners also issued a report in 2017 calling for more modern approaches to transportation planning.
“School districts need more information about their own transportation systems, and states should invest in tools and technology that enable districts to collect, analyze, and use this data to improve operational efficiency as well as to provide a higher level of customer service to families,” the authors wrote.
When considering complex and potentially controversial issues such as school start time changes, district leaders typically get input from families, business owners, after-school providers and other interested parties. But they’re finding that advanced data tools and machine learning can also inform decisions that affect students’ focus in school, their safety after school, what time thousands of parents get to work and how many vehicles are on the roads during morning rush hour.
‘Losing a level of efficiency’
Denver Public Schools (DPS) was also receiving requests from individual schools to change their start schedules. Because they support local school autonomy, district leaders were trying to accommodate many of those requests. But they were also “losing a level of efficiency as to how we could maximize the utilization of school buses,” said Nicole Portee, the executive director of transportation for DPS.
So Portee also turned to an operations researcher to develop a formula for routes that would create what she calls “the best pairing.” That involves reducing “reposition mileage” — the distance from the end of one route to the beginning of the next, explained Amanda Chu, a doctoral student at the Georgia Institute of Technology. She designed the bus route scheduling analysis tool that DPS is using in collaboration with Monica Villarreal, a senior planning analyst with the district, and Professor Pinar Keskinocak in Georgia Tech's School of Industrial and Systems Engineering.
“With the model, the user can define different costs they may want to rank in importance,” Chu said, adding that she’s also received interest in the tool from the Atlanta Public Schools and the Tulsa Public Schools in Oklahoma.
Now in the second year of using the tool, Portee said she is able to approve some schools’ requests for changes but has to turn down others. Drivers, she said, have also grown a little more competitive over who gets which route. Those decisions are usually based on seniority.
During the 2018-19 school year, the use of the tool led to 34 fewer routes and a 12% drop in school buses on the road, according to Chu's data.
The more options the better
In BPS, the MIT team’s method creates multiple routing solutions for each school — within the constraints given by the district — and identifies which of those “interfaces the best with other schools,” Martin added in an interview.
Hanlon further explained the approach during the INFORMS event.
“The more options the algorithm has to choose from, the better,” he said, “The best solution is not the one with the fewest number of buses for each school, but is the one that best reuses buses from school to school.”
One of the district’s priorities, for example, was reducing the walk-to-stop distance for younger students and for those living in unsafe neighborhoods. Historically, the maximum distance students were expected to walk to their stop was half of a mile, regardless of their grade level. When the algorithm showed that reducing that for younger students could be accomplished at minimal cost, a policy change was implemented so that most students in BPS now walk no more than one-fifth of a mile to their stop.
Additional efforts have focused on “improving the underlying map” and addressing issues such as keeping buses off small side streets and making sure drivers aren’t taking routes that require u-turns. The new routing system has allowed the district to reduce its fleet and redirect $5 million a year to other priorities.
“Our technology process took a big step forward,” Eger said.