A couple of months ago, I shared a few ideas for applying the idea of “crowdsourcing” in the classroom.  Last month, Dan Cohen, director of the Center for History and New Media at George Mason University, experimented with crowdsourcing using Twitter, asking his Twitter followers to work together to solve a “historical puzzle,” which they did in just under half an hour.  Cohen’s experiment started me thinking about similar uses of Twitter in the classroom, and I’ll share those thoughts below after summarizing the experiment.

You can read all about Cohen’s experiment on his blog, but here’s the short version: He posted an image of an object found at a Victorian archaeological dig on his blog at the start of a talk he gave at a recent conference.  At the same time, he “tweeted” an invitation to his 1,600 Twitter followers to help each other identify this object.  His Twitter followers jumped right in, making observations about the object.  They also “retweeted” Cohen’s invitation, which allowed their followers to enter the conversation, too.  Everyone contributing to the conversation was asked to tag their tweets with a particular hashtag so that anyone could search Twitter for that hashtag and follow the conversation.  Cohen, meanwhile, displayed the live Twitter conversation on the screen behind him during his presentation.

It took the “crowd” all of nine minutes to come up with a preliminary answer to Cohen’s puzzle and 29 minutes for “a fairly rich description of the object to emerge from the collective responses of roughly a hundred participants.” (I’m quoting Cohen’s blog entry here.)  It’s worth noting, as Cohen does on his blog, that some participants in this crowdsourcing experiment engaged in what Cohen calls “scholarly discussion,” drawing on their own knowledge and skills to contribute to the conversation.  Others took a “Google knows all” approach (to use Cohen’s term), relying on their search skills to ferret out information on the image in question.

How might crowdsourcing-via-Twitter be used in the classroom?  I can imagine giving students a challenge along the lines of the one Cohen used and letting the students be the “crowd” that tackles the challenge.  (We’ll assume each student has access to Twitter during class via a laptop or smart phone.)  It’s worth noting that Cohen already knew the answer to the question he posed in his experiment.  A question posed to students would likely be one with an answer known to the instructor, but a more open-ended question would be potentially useful, too.

One of the key ingredients in Cohen’s experiment was what he calls the “Twitter multiplier effect”–not only were his Twitter followers invited to contribute, but as they retweeted his invitation, their followers were invited, as well.  Since many Twitter users read and respond to tweets regularly throughout the day, this multiplier effect can happen very quickly.  In a classroom situation, this effect would mean that students might be able to draw on their own social networks to help tackle their instructor’s challenge.  Some instructors would welcome that since that’s how problems are often solved in the “real world.”  Others wouldn’t be comfortable with it, preferring that students work “on their own.”  Twitter is designed to be a public forum, however, so instructors really bothered by this would need to seek an alternate, closed system.

The fact that Twitter is a public forum has implications for students, as well.  A student might not want to have his or her contributions to a class discussion made public to his or her social network on Twitter.  For that matter, one of the advantages of teaching with clickers is they allow students to voice their perspectives anonymously during class.  Contributions to a Twitter discussion during class would not be fully anonymous were students to use their usual Twitter accounts.  Instructors could, however, instruct students to create new Twitter accounts for the purpose of contributing to class discussions.  This would allow students to retain some degree of anonymity while also reducing the impact of the Twitter multiplier effect described above (for better or worse).

The “Google knows all” approach Cohen describes would also be an issue.  If the goal is to create a “scholarly discussion” among one’s students, what role should Google jockeys play in that discussion?  Some instructors prefer to keep student access to resources beyond course resources limited, perhaps because they want a more accurate assessment of what students themselves bring to the table or because they want to better scaffold their students’ learning experiences.  Others, like Michael Wesch for instance, prefer to have students bring the outside world into the classroom and to help students make sense of that outside world.  I suspect the launch of Wolfram|Alpha (a “computational knowledge engine”) tomorrow will only complicate this issue.

A few other issues, largely inspired by lessons learned from teaching with clickers, come to mind:  How might an instructor monitor, respond to, and participate in a Twitter discussion during class?  (There’s no bar chart to aggregate results.)  What kinds of students would be more likely to participate in a Twitter discussion than in more typical class discussion?  Clickers allow students to respond to questions before they hear their peers’ perspectives, which encourages independent thinking–would that benefit be lost in a Twitter discussion?  And what kind of questions, challenges, or tasks would best motivate students to engage in meaningful dialog via Twitter?  I think it’s important that Dan Cohen didn’t just ask for general discussion in experiment; he presented his Twitter followers with a very concrete task.

I’m glad Cohen tried this experiment and that he shared its results on his blog.  Given the increasing use of Web-enabled smart phones by students, crowdsourcing (via Twitter or other systems) in the classroom is an ever more possible opportunity for instructors to engage students.

The New York Times recently published an article describing MIT’s TEAL (Technology Enhanced Active Learning) classrooms used in introductory physics courses.  The classrooms enable an interactive class format in which students work problems in small groups during class as professors and TAs circulate and answer questions, replacing the traditional lecture model used for these courses.  Classroom response systems are used to monitor student progress and learning during class.  I like this quote from Harvard physics professor Eric Mazur:

“Just as you can’t become a marathon runner by watching marathons on TV,” Professor Mazur said, “likewise for science, you have to go through the thought processes of doing science and not just watch your instructor do it.”

The TEAL classrooms cost $2.5 million each, but, of course, many aspects of the pedagogy used with the classrooms can be implemented for much less.  See the SCALE-Up site at North Carolina State University for more information on effective classroom design.  Classroom design can have a significant impact on the learning experiences of students, so it’s important to consider pedagogy when designing (or redesigning) classrooms.  That being said, one of the reasons I’m a fan of clickers is that they can be used to facilitate aspects of the TEAL pedagogy (like peer instruction) in any kind of classroom.

According to the NYT article, MIT is having great success with the new model for teaching physics.  According to some of the comments on the article, students at MIT aren’t all happy with the new approach.  Here’s a quote from Chris, a recent MIT graduate:

“Personal response clickers”? Ask any student how they feel about them and discover that they’d much rather hurl them into the Charles than actually use them, if not for the fact that participation points are oftentimes given out as inducements for clicking.

As one of the other comments points out, this kind of negative response might be related to the fact that prior to the change to the TEAL classrooms, attendance rates in these courses would often drop below 50 percent.  As productive as active learning approaches can be, I think it’s important to remember that some students will resist these approaches.  I wish Chris would have elaborated on why students might want to throw their clickers in the Charles River.  I have some hypotheses, but I’m curious to know (a) if you’ve experienced student resistence to active learning and (b) why you think students sometimes resist these approaches.

Reference: Crossgrove, K., & Curran, K. L. (2008). Using clickers in nonmajors- and majors-level biology courses: Student opinion, learning, and long-term retention of course material. CBE-Life Sciences Education, 7(1), 146-154.

Summary: The authors report on their study of the impact of clickers in two courses, an introductory biology course for non-majors and a genetics course taken by sophomore biology majors. Both authors came to these courses with experience using active learning techniques and had experience with clickers prior to this study.

The authors surveyed their students and according to survey responses, students were generally very positive about the use of clickers. The non-majors in the introduction course were more positive than the majors on some points, including the usefulness of clickers in helping students score higher on exams.

Students’ average performance on final exams during the semester prior to the use of clickers was compared to average performance during the semesters in which clickers were used. There was no statistically significant difference in performance, although, as the authors note, the cohort of students changed each semester.

Within each course that used clickers, however, student performance on exam questions covering topics that were treated in class with clicker questions was statistically significantly better than student performance on exam questions covering topics that did not involve clicker questions. This was true for all question types–factual recall, conceptual understanding, application, and analysis.

The authors also asked for student volunteers to take tests on the course material four months after the course ended. Although only a few students did so (14 or 15 for each course), the non-major students retained information taught via clickers at a significantly higher rate than they retained information not taught via clickers (dropping from 88% to 83% for clicker material, 92% to 75% for non-clicker material). Major students did a poor job of retaining all material, whether or not it was taught via clickers (dropping from 86% to 60% for clicker material, 87% to 61% for non-clicker material). It’s worth noting, however, that most of the clicker-material questions for the majors were “harder,” application-level questions, which may have contributed to the poor showing.

Comments: To explain the finding that the major students weren’t as positive about the use of clickers as the non-major students, the authors cite the different class sizes of the two courses (large in the case of non-majors, smaller in the case of majors). Other possibilities that occurred to me include the following:

• Different exam formats, perhaps because of a greater match between in-class clicker questions and exam questions in the introductory course,
• Different active learning techniques, perhaps because clicker questions didn’t enhance the problem-based learning approach used in the genetics course to the degree that they enhance the peer instruction approach used in the introductory course
• Affective domain issues, perhaps because the mostly pre-med genetics students were more sensitive to grading issues than the non-major students, or
• Differences in expectations for learning, perhaps because the genetics students expected to spend time memorizing information, not interacting with information via clicker questions.

I would be interested in seeing more research (by the authors or by others) into the use of clickers with upper-level students exploring these issues.

I found it an interesting result that the students’ performance on exam questions related to topics taught with clicker questions was better than their performance on other questions and that this result didn’t depend on the kind of exam question.  However, it would be useful to know what kinds of clicker questions were asked during these course. Are there certain kinds of clicker questions (conceptual understanding, application, etc.) that lead to better exam performance or retention of knowledge?

The authors noted that the standard deviation on questions involving topics taught via clickers was smaller than on other questions. They point out that this was noted in a previous study, as well. This statistic might be worth examining in future studies about clickers.