News

Summer classes try digital tools

Educational experiment gets mixed reviews from students

While researchers were running experiments in MIT labs this summer, the Institute was conducting an educational experiment of its own, piloting for-credit summer classes through the “summer@future” initiative.

Five courses were offered, from the physics, biology, materials science, and mechanical engineering departments. As an incentive for students, MIT offered free housing and tuition for anyone who enrolled in a summer course. A total of 129 students registered.

Courses ran for eight weeks, from June 9 to August 1, but planning for summer@future began much earlier.

“The idea initially stemmed out of discussions in the Task Force on the Future of MIT Education,” said Professor Karen E. Willcox PhD ‘00, Chair of the Task Force Working Group on MIT Education and Facilities for the Future.

President L. Rafael Reif asked the Task Force to recommend “possible experiments and pilot programs” that would both “incorporate online learning tools” and “[maximize] the value of face-to-face learning for both faculty and students.”

Summer@future fits these goals by exploring a summer session to give students increased flexibility in completing degree requirements and broadening interests, and by experimenting with digital learning tools.

“It’s a chance to experiment with different ways of teaching, particularly more project-based, hands-on, more intensive ways of teaching,” Willcox said. “It’s also an opportunity to look at ways to infuse online learning and blended learning models into classes.”

Proposals were solicited from professors early in 2014. Out of seven proposals received, five were accepted and offered as summer courses. Director of Digital Learning Sanjay E. Sarma and Claudia Urrea PhD ‘07 managed the program once the call for proposals had been sent.

When the program was announced to students on March 20, the response was overwhelming. “In the first day we had 160 people reply saying they would like to take a class in the summer,” Urrea said. In total, summer@future received initial responses from 347 students, and after a selection process in which students each wrote a paragraph detailing their reasons for wanting to take the course, 165 students were accepted.

An optional survey was sent out to students during the application process asking about their motivations for signing up for courses. According to the final report released by the Task Force, 74 percent of students selected “subsidized summer housing,” 72 percent selected “exploring a new field,” 70 percent indicated that they were “already planning to be in Cambridge over the summer and wouldn’t mind taking a summer class,” and 60 percent indicated that they wanted the “opportunity to earn credits at no financial cost.” Less common reasons included “advancing towards degree” and “a friend also plans to take the class.”

The Tech sat down with professors and students in each of the five classes to find out how they went.

8.371J Quantum Information Science II

8.371, a graduate course in quantum information science, experimented with online problem sets.

“I had observed in this class which I’ve taught since 2001 that graduate students kept on dropping out of the discussions as the class went on because they would fall further and further behind by having feedback so late in the process,” Professor Isaac Chuang ‘90 said. “They would not realize they’d misunderstood something in the problems sets, and it would catch them in the next lecture, and they’d just turn quiet on you.”

In the summer iteration of the course, limited-submission problem sets, lecture videos, and open-ended research questions intended to stimulate student curiosity and class discussion were all posted online. There were, however, trade-offs when working in the online system.

“It took away some things, but it gave some other kinds of things,” Chuang said. “I couldn’t ask students to prove a theorem because the computer can’t check proofs — at least not very well, at least not the kind of hard ones that we have … But what is interesting is that you can introduce new and different problems which are above and beyond what I could write in a paper problem set.”

These “new and different problems” allowed students to construct quantum circuits using a feature of the MITx platform, and even interact with the system in a symbolic mathematical language.

He also created online proof questions by using a feature of the MITx platform which allows users to drag and drop labels onto images.

“Instead of saying ‘prove this,’ what I did is say ‘fill in these equations and these steps so that the proof is correct,’” said Chuang.

“It makes me a little sad to not be able to give open questions,” Chuang said. “On the other hand, because the problem sets were structured, it gave me more time to be unstructured in class … I let the problem sets take care of more of the technical aspects of teaching, freeing me up as a faculty member to spend more time talking with students about what is actually at the edge of knowledge, and I like that a lot.”

One element of the course which has remained unchanged from the semester to the summer version is the final project, which counts for a large portion of the grade and represents students’ original work in the field.

Student responses to online problem sets were varied. Some lamented the fact that the online grading system could not give partial credit for questions, since it was only a student’s final answer which was input to the system.

“It’s nice with problems where you kind of already know the steps and you just do it and get to the answers,” said Kevin B. Burdge ’15, “[but] on the harder problems it can be a real struggle to just have an input and then have an all-the-points-or-nothing sort of system.”

Some students felt that the lack of partial credit came to the detriment of the problems themselves.

“Every problem had to be truly handholding,” said Dan S. Borgnia ’15, “because the way the submissions worked, they couldn’t give you partial credit.”

“It’s nice to have your whole question graded instantly,” Borgnia said, “but it’s not as constructive to the learning for you to get, at every single step of the question, feedback saying ‘you got that right, you got that wrong.’” More constructive, he said, is when “you play with it for a while, you get it wrong at the end … and you start understanding ‘this part’s right, and that part’s right,’ and that’s actually where you learn … I feel like that’s a really crucial part to learning we missed out on.”

Graduate student Anand V. Natarajan described the online problems sets as “a good approximation.”

“Obviously there are some kinds of questions that you can’t do very well on online psets, like proof questions,” he said. “I think he still managed to get some of those approximated — there were some questions where you’d have to fill in a lot of numerical things and you’d end up walking through the proof by doing that.”

Jordan Ugalde ‘16 felt that the online problems, which often required numerical answers, didn’t allow students to develop “as strong an intuition for the material.”

“In my opinion, this is the type of subject in which you would need to derive proofs, do rigorous arguments, do problems where you can get partial credit … and that type of problem is not facilitated by edX, at least not as it is currently formatted,” he said. “Our professor has posted lectures on edX and I found those helpful … but online psets I do not think are appropriate for the subject.”

2.S03 Special Subject: Dynamics I

Since 2.03x had previously been available as an online course, Professor Sanjoy Mahajan wanted to use 2.S03 as an opportunity to explore in-class teaching styles and uses of online material.

“Given that all that material was already there, what could we do with it to improve education? What can we do in the classroom that’s different?” he asked.

2.S03 made use of a number of online resources — online problem sets, lecture videos, and office hour videos from when the course was offered residentially. The course’s midterms were also taken online, though the final exam was in-class. Some questions were written such that the online problem set checker gave hints if the answer submitted was incorrect.

In designing and teaching the course, Mahajan said his focus was on ways to improve the quality of education both online and in-class.

“All the stuff that’s best learned from some kind of reference material” students can learn on their own from digital learning tools,” he said. “Instead of writing equations down — I may copy them correctly onto the board, you may copy them incorrectly down into your notes — all of that is now sorted out. Instead, we’ll actually use the equations, struggle with them … increase the contact quality and not lose any of the contact time.”

With more of the technical information of the course moved online, Mahajan has tried to use class time to develop students’ intuition. But the next step, he said, is to “put more of the intuitive conceptual questions online.”

Students seemed to react positively to the online elements of the course, noting differences in the lecture format.

“I really like the hybrid learning,” said Kai Aichholz ’17. “It lets you get deeper into the technical learning and equations, and then the lectures are more conceptual.”

Said Christopher M. Knapp ’16, “the idea is good. Watching lectures and doing psets online allows one to ask more nuanced questions during physical lecture, and the redundancy in material also gives the professor the opportunity to go over stickier points and off-topic ideas as he or she sees fit. I think the execution — especially the online portion — still has room for improvement, but it was the first installment, so that is to be expected. Overall, I’d say it was a success.”

7.S390 Creating Digital Learning Materials for Biology

In 7.S390, a course offered for the first time this summer, students created their own digital learning tools, each addressing a misconception in the field of biology.

“As MIT students you’ll likely have to communicate science to the world in some way in your career in the future, so we want to help train undergrads and even grad students and postdocs if they are interested in the future to communicate science well,” said Mary Ellen Wiltrout PhD ’09, an instructor of the course along with Nathaniel S. Schafheimer PhD ’13 and Sera Thornton PhD ’14.

Students worked on a timeline to produce their final projects, first identifying a misconception and specifying learning objectives, then planning and storyboarding their projects.

“Some people are doing assessments, some are doing video, and one person’s doing a simulation … it’s fun to see what they’re creating,” said Wiltrout.

Projects addressed misconceptions about evolution and natural selection, and the widespread myth that humans use only 10% of their brains, among other topics.

Students taking the course come from a mix of departments, including EECS, Biology, and Biological Engineering. Many created educational videos, but several designed learning sequences on the MITx platform. Students took advantage of features of the platforms to ask different kinds of questions, including multiple choice questions embedded throughout longer readings, problems with boxes for numerical answers, and images and graphs to be labeled by users.

MITx currently allows three methods of assessing open-ended questions: self-assessment, where a user submits a response and receives a rubric to check their answer against; peer-assessment, where three to four other students in the course will grade the submission anonymously; and AI, where the computer will assess a student’s submission. Some 7.S390 students used the self-assessment feature to ask open-ended questions, providing users with a rubric of their own design.

In terms of 7.S390’s learning objectives, the main idea was that “at the end of the course, [students] would be able to design content based on misconceptions, design teaching materials in ways that are following the best practices, and then extend that to applying the best practices to online material and the online medium,” said Wiltrout.

7.S391 Quantitative Biology Workshop

7.S391, led by the same teaching team as 7.S390, evolved out of a workshop-based course typically offered in January.

“It was never an MIT course officially,” said Wiltrout. “The biology department ran an outreach program for four or five years, and that program invited graduate students or postdocs or faculty to come to meet the students who were visiting — the students were from universities around the U.S. … The faculty members would give a talk based on quantitative biology and something in the field, a bigger picture kind of talk. Then the graduate students and postdocs would come in and do an activity with the students for that day, and a lot of time the projects involved MATLAB or R or Python or PyMOL.”

“What changed last year is that we tried to put the workshops on the platform, so that the directions would be there, and every step, there were checks for students,” Wiltrout said. Once the content was online, Wiltrout used the summer@future program to offer this course to MIT students. “It seems like students do want that mix of biology and quantitative tools,” she said.

The course still involves graduate students leading workshops and talking to students, helping them “understand how what they’re doing in class is directly related to the research side,” Wiltrout said.

“Pretty much all of our class was online,” said Linda Wang ’16, “we’d have lecture videos that we’d watch before coming to class, and when we came to class, someone would give a small lecture, but … we do the exercises online and all of our homework is online.”

Since the course is in computational biology, Wang felt that the online medium was useful. “I liked it,” she said. “With coding it’s hard to do pen and paper stuff.”

3.S01 Special Subject: Materials Selection and Design of Nanostructured Catalysts for Sustainable Energy

3.S01 was one of two courses designed from scratch for the summer@future program, the other being 7.S390.

“We took the opportunity of this program being launched to try and do a project-based class that hadn’t been taught before,” said Assistant Professor Elsa A. Olivetti PhD ’07, who taught the course with Assistant Professors Alexie M. Kolpak and Yuriy Roman.

The class focuses on the design of catalyst materials, “in particular for hydrogen generation that would then be used downstream in a fuel cell,” said Olivetti. Materials are evaluated from a technical perspective, but also from an environmental perspective and an economic perspective.

The online content was developed specifically for this summer, with the teaching team creating new lecture recordings. The course used screen captured lectures to introduce each new topic and interspersed them with comprehension questions. A portion of the course depended on using software tools to design materials.

“The main paradigm we used was trying to see if we could focus the in-class time on using the [software] tools and running calculations,” said Olivetti.

When designing the online content and determining the distribution between online and offline material, one of the professor’s goals was “taking advantage of the fact that you have this different forum for delivering the content instead of taking the content the way you’d normally deliver it in lecture and then delivering it the exact same way,” Olivetti said.

Moving forward

Ultimately, summer@future was an experiment in several directions that still has some kinks to be worked out.

Burdge noted that, in this first iteration, some administrative details such as add/drop dates and potential for using junior/senior P/D/F were not well-communicated.

Said Urrea, “One of the lessons is, if we want to open a summer semester … registration has to open much earlier, because when we announced [which applications were accepted] in April, students had already decided whether they were going to take UROPs.” Students needed to hear back quickly about whether they’d been accepted so they could make housing arrangements, and, if necessary, make arrangements to miss UROP time for classes which were all offered during the workday.

Many of the students The Tech spoke with were involved in research or UROPs over the summer and most found it manageable to balance both a class and a UROP.

Students also expressed a desire to see HASS classes offered in any future summer program.

“I think it would be fantastic if MIT offered HASS classes over the summer,” said Burdge, partly because summer HASS classes might allow students to “delve into a HASS class” and also work on a UROP. “If you offer technicals, fine, but just be aware that a full-time UROP with a technical is much less manageable than a full-time UROP with a HASS class,” he said.

“It’s going to be different for every student,” said Willcox, referring to what students will want in any program, “and that’s why flexibility is so key — the last thing we have in mind is anything that becomes mandated. It’s really about putting options on the table for students.”

Was the experiment a success?

Ultimately, said Willcox, “there are many ways success can be measured — if it’s progress toward a degree requirement that makes room for something else in the semester, for a UROP experience, or for an extra class; or if it’s an extra class that a student takes that they wouldn’t otherwise get to take where they acquire a skill outside their major — we think these are all really positive things for students. And on the faculty or educational side: learning, trying new things, even offering a class with a different style of pedagogy which can reach different students. For example, we know that some students love hands-on learning, while some prefer the more traditional lectures. If we can get that kind of variety infused into the curriculum, it makes for this rich educational environment.”

Summer@future … in the future?

“We have to look carefully at how things go this summer, but the indications are that there are a lot of positives there, and if you look solely at the interest from the students, you can see there’s a lot of demand,” said Willcox. “We think this is something that could continue.”