Volume VII, #25
Despite what I sometimes hear in the Bay Area, the set of things that technology makes more useful is not equal to the set of all things. In other words, there are things that don’t become more useful simply from digitization. Bathrooms rank near the top of this list.
I was thinking of bathrooms last week, talking with my former roommate Dave. In college, Dave spent a lot of time thinking about bathrooms. Junior year, for the second issue of the college tabloid publication we started, Dave compiled a ranking of Yale’s Best Bathrooms. A parody of the U.S. News & World Report college rankings, Dave established an arbitrary set of categories and weightings, recognizing that once a ranking is in print, everyone treats it as authoritative and definitive. Categories such as # of sinks, urinal-to-toilet ratio, cleanliness, ventilation, décor and safety were rated and weighted: “odor counted three times, because we figured bad odor was the least tolerable situation one could encounter in a bathroom.”
The resulting rankings spanned from the bathroom in the newly renovated William L. Harkness Hall (“the Taj Mahal of Yale bathrooms”) and in the basement of the British Art Center (“This place smells like sweet fruit – as if somebody were boiling a pot of Hawaiian punch”) all the way to the bathroom on the third floor of the Yale Film Study Center (“If you are shooting a horror film, this is the place to come”) and the bathroom in the basement of Sterling-Sheffield-Strathcona Hall (“To describe this bathroom in two words is easy: indoor outhouse”). Dave concluded the bathroom rankings with the following U.S. News-ish comment: “Sure, statistics like ‘urinal-to-toilet ratio’ are valuable bits of information to consider when conducting your bathroom search, but there’s a human element to bathrooms that a ranking like this simply cannot convey. The only way to really find out what a bathroom is like is to pay a visit.”
In stark contrast to bathrooms, consider lectures. They can be grand (RIP Vincent Scully) or plebian, but either way there’s little way to measure (let alone guarantee) that learning is occurring. With technology, it’s a different story.
Last month, I had the opportunity to sit in on an anatomy class at Ponce School of Medicine, part of Ponce Health Sciences University. I walked in about 15 minutes after the class had started and was struck by the sound: not the stentorian sage-on-the-stage, but rather the hum of student chatter. My first impression was that students sounded distracted. But it was actually the sound of learning.
Ponce employs a flipped classroom model, but Ponce President David Lenihan goes further. He calls it a “dynamic classroom.” Ponce spent the last two years recording all lectures for its MD program. The lectures are digitized, indexed and coded to learning objectives. When students come to class, they don’t have to listen to a lecture. Instead, each class is spent on five to ten multiple choice questions based on the lecture. For example:
An easy one
Professors introduce each question while students sit in groups, reference materials (ranging from 3D anatomy diagrams to the digitized lecture) on their iPads, the question itself on their phones. After reading the question as a class, students have two minutes to make their selection. It turns out the hum I heard is the sound of students discussing the question. Then the professor goes through each option in a 5-10-minute back-and-forth with students, exploring why each answer might be right, but 80% of the time why it’s wrong. (By the way, the answer to the above question is B.)
According to Lenihan, this is the power of active learning: “much of the learning students achieve at Ponce comes from understanding why the wrong answers are, in fact, wrong.” It seems to be working. Since implementing the dynamic classroom, USMLE Step I Board pass rates have significantly increased from 69% to 88% and improving. So while both lectures and multiple choice questions feel very last century (if not the century before), technology-powered multiple choice questions are at the core of Ponce’s dynamic classroom – a better lecture.
There are two salutary byproducts of the better lecture. The first is data. In each class, Ponce receives five to ten data points on each student i.e., responses to the multiple choice questions. While these answers aren’t graded, they’re used to determine who needs interventions. Lenihan says Ponce knows within the first few months which students are unlikely to pass the Boards without intervention.
The second is scale. Previously, lectures were thought to scale via something like a Massive Open Online Course (MOOC). Record the lecture, throw it online, slap an assessment at the end, refrain from granting a meaningful credential, and bingo, you have the recipe for < 5% completion and the higher education fad of the decade. But Ponce’s model produces both learning outcomes and scale because the digital curriculum can be deployed at any university.
As a result, Ponce has created a company – Tiber Health – to productize and market its better lectures. In September, a cohort of 50 students at Texila American University in Guyana started medical school via Tiber lectures. Lenihan says Tiber’s curriculum can enable the creation of a new medical school in months, not years, and for one-tenth of the cost of a new medical school in the U.S. Tiber believes it can power 50 new medical schools by 2023. With a shortage of 100,000 doctors in the US and almost a million globally, scale in training doctors means lives saved.
I want to make one additional point about why the dynamic classroom produces better learning outcomes than the lecture, and it concerns adaptive learning – a buzzword in education for at least a decade but still a long way from viability, let alone scale. Knewton – funded in part by Pearson – claimed to have developed the first adaptive learning platform for a mere $150M. Amplify – the News Corp. company led by Joel Klein – dropped a cool billion. Both promptly went the way of MOOCs.
Other than failure, what these and other adaptive efforts have in common is an unrealistic view of how learning works. Learning is less likely to occur in a 1:1 interaction between man and machine than in a social setting. The failure of adaptive learning to date is a product of a monomaniacal focus on individual adaptivity, disregarding that the best learning experiences – certainly the ones most likely to be retained – are social in nature, not individualized. As Dan Meyer, America’s “most famous math teacher,” says, “there’s limitations on what kinds of work can be done on a computer without a teacher.”
Meyer’s dissertation was on how conversation is instrumental for learning, and this is why the dynamic classroom is so effective: students discussing questions and answers, struggling to attain understanding. Moreover, the dynamic classroom provides a new way of looking at adaptive learning. Because where Lenihan plans to take Tiber is adaptivity at the level of the cohort. In a year or so, the multiple choice questions in Tiber dynamic classrooms will adapt to the cohort. “The goal for the dynamic classroom,” Lenihan says, “is to aim for 60% correct. More than that, there’s not enough learning, and the system will serve up more advanced questions. Less than 60%, it’s too challenging, and the system will dial back.”
Last month, Susan Dynarski wrote in the New York Times about how laptops and lectures are incompatible: “The research is unequivocal: Laptops distract from learning, both for users and for those around them.” Ponce’s dynamic classroom suggests that Professor Dynarski’s critique says more about her lectures than her students' laptops. It’s not that she has too much technology in the classroom. In fact, she has a suboptimal amount.
By doing the work to shift from traditional lectures to a dynamic classroom, Professor Dynarski and millions of other postsecondary and secondary school educators can produce better learning outcomes with better lectures. And that might just be something worth ranking (and not parodying).