Most online courses are a solitary experience for learners. Students lack the ability to strike up an impromptu conversation about last week’s homework or compare notes with whoever’s sitting next to them in class. The lack of social interaction could be one reason behind high dropout rates in online classes.
Instead their interactions are relegated to stale chat forums, where questions go unanswered or where few students regularly visit. Sure, there are sophisticated online lectures where learners can see one another on screen and break out into small groups to chat via video. But these are still the minority online classes. For large introductory courses, even if students can interact with their professors there are still those who “sit” in the back and neglect to raise their virtual hands.
Students enrolled in online classes at community colleges drop out of class at a higher rate than peers in face-to-face environments, according to the Community College Research Center. Those in primarily online classes are also less likely to attain a degree or transfer to a four-year institution.
Several California community colleges are hopeful that adding a way for learners to interact with each other in online classes will help them complete their coursework. This fall, students taking introductory statistics courses at six colleges will pilot using a tool to complete lab exercises in teams, working in sync with partners who are miles away.
“Often in online courses students disappear and drop,” says Barbara Illowsky, dean of basic skills and online educational resources (OER) and a professor of mathematics and statistics at De Anza College in Cupertino, California. “I’m hoping this helps them to stay vested in the course as they’re working with their classmates.”
Carolyn Rosé, an assistant professor in the Human-Computer Interaction Institute at Carnegie Mellon University, has been exploring ways to add social engagement to MOOCs since 2013. She and fellow researchers developed Bazaar, the tool that California community colleges will test in online statistics courses this fall.
After students complete a lesson, they can choose to enter an online chat group where they discuss and apply what they learned. A computer agent—a chatbot—prompts them to discuss amongst each other and reflect on the material they just covered.
Bazaar automatically groups the students into teams using algorithmic matching process that Rose and her colleagues have carefully considered. “We’re not grouping good students and bad students,” she says, adding that ideally the tool matches students who have somewhat different opinions. If a student chose to enter the chat but no peers were around, he or she would still reflect in conversation with the chatbot.
Rosé and her colleagues found that when students talk to each other, they tend to stick around and show up for their next class more than peers who chose not to complete the activity or those who only talk to the robot.
The human element makes a difference. “The way people interact with an agent fundamentally changes when there’s another human around,”Rosé says. Students who chatted with other students had a 70 percent lower rate of attrition than those who chose not to complete the activity or who could only talk with the bot.
In summer 2016 Rosé and her team used Bazaar to run a team-based project for a MOOC offered by the Smithsonian Institution. The class, “Rise of the Superheroes and the Heroes of the Future,” included 400 students who were grouped into teams of four. During the three-week course, they collaborated to design a superhero story together. Rosé says that nearly all of the teams completed their project—an impressive figure given that the typical completion rate for a MOOC is in the single digits.
Bazaar’s California debut
Illowsky estimates that 200 students in introductory statistics classes at De Anza and five other California community colleges will use Bazaar this semester. She and fellow statistics faculty colleagues worked closely with researchers at CMU to build interactive labs for students to complete in teams. They had to rewrite their statistics lab exercises so that Bazaar’s computer agent can prompt groups of students with appropriate questions and directions.
As students are learning about regression, they’ll collect pieces of data to analyze and plot on a graph, Ilowsky says. Bazaar will prompt students to share their graphs with each other and combine their data while asking questions including “what’s similar and different?” Participation in the computer-mediated discussion is mandatory, and will be included in their lab grade.
In addition to De Anza, Lake Tahoe Community College, Skyline College, the College of San Mateo, Pasadena City College and Sierra College will roll out the tool in courses this fall.
Illowsky says that the statistics faculty and CMU researchers will evaluate Bazaar on its ability to improve success rates, as determined by course completion and grades. “I’m hoping that they learn how to collaborate online more, as opposed to just discussions where they’re posting a question and answering, posting and answering,” she says. “The bottom line is improved understanding of the course.”