Recently I have become fascinated with the research around how students learn though dialogue. My favorite piece of quick evidence is Derek Muller’s TED application video where he presents his research about videos for learning.
You really should take the six minutes to watch the video, but the summary is that he tested two types of instructional videos; direct instruction and instruction through dialogue. Students who watched the direct instruction videos said they were clear and easy to understand, yet their test scores did not increase. Students who watched the dialogue videos said they were confusing and didn’t like them, but their scores increased significantly. Interesting.
Similarly interesting to me is the recent obsession in the education world with the ‘flipped classroom.’ There seems to be some evidence that flipping the classroom does indeed increase learning; my question is why. The article on flipping linked above has an entire section on how student-student and student-teacher interactions significantly increase with the flipped model. Is this the primary reason flipping succeeds? If so, then why the obsession with video lectures and programs like Khan Academy? Is the video piece even necessary? Before I dive into this I want to give you a picture of where I am coming from with all of this.
I have taken a long road to get to where I am today as a teacher. I started teaching physics in the fall of 2005 with very little knowledge of how students learn, particularly the vast amounts of Physics Education Research (PER) that has been conducted in the last 30 years since the development of the Force Concept Inventory (FCI). I started a Masters degree in 2007, and through the research for my thesis on inquiry in physics I stumbled upon the FCI. I pre-posted my students for the first time in the 07-08 school year. Though my average postest score of 47% is above a national average for traditional teaching of 42%, I was pretty dismayed. Really? After a whole year of physics my students can’t even answer half of the FCI questions correctly? Not ok.
My research showed slightly higher student gains with inquiry, and, particularly interesting, that the standard deviation of the scores shrunk. My interpretation was that the high end learners gained about the same, while the low end learners gained more with inquiry. That’s good. But it wasn’t enough. In 5 years, my scores never got above 50%.1
I knew my kids weren’t really getting it, but I didn’t know what to do about it. Enter grad school #2. I decided in the spring of 2010 that I wanted to learn more about Educational Technology, so I enrolled in online courses at Mankato State University. I decided to research clickers (student response systems) for one of my papers, and I stumbled upon Eric Mazur’s work on Peer Instruction (PI). PI is a technique developed primarily for large lecture clases. The idea is that a multiple choice conceptual question is posed, and students answer via clickers (though this can work with low-tech solutions like raising a piece of paper with the answer on it). Particularly if the distribution is evenly split, the instructor has the students talk to each other, and then re-answer. More often than not (in my own experience) the distribution shifts towards the correct answer. Mazur has some great research out there about how students are able to reason to each other better than an expert, thus their explanations often make more sense. More importantly, the process of the discussion is another form of the dialogue used by Muller, and my suspicion is that in this lies the reason for understanding gains.
The following summer a colleague from another school in Minnesota mentioned Modeling Instruction (MI) to me. Dialogue and Inquiry are both central to MI. The modeling cycle typically starts with a paradigm lab where students use guided inquiry to investigate a phenomena. From there the phenomena, or Model, is expanded and refined, often through White Boarding. The idea is that student interaction, questioning, and revising of ideas drives the learning. And it works.
So we have Muller and his video instruction with dialogue, Peer Instruction with dialogue in large lecture classes, Modeling with dialogue in the form of white boarding, and the general idea of flipping the classroom. Most of the praise I have heard about for flipping is that it provides more time for projects, problem solving, and other more interactive methods of learning than when the teachers ‘had’ to lecture during the hour. I have to wonder if the problem is simply that lecture doesn’t work, period? Does flipping work only because teachers who flip are using techniques during class that actually do help students learn? Do the videos really have anything to do with it, if they are just direct instruction?
I will say that with both PI and MI require that before the conversation takes place students should be familiar with the problem at hand. I recall research (but can’t find at the moment) that showed gains in understanding when students worked on a problem before it was used as an example in class. The standard MI white boarding process involves students first working on the problems on their own (often as homework), then comparing in their group, then presenting their agreed upon solution to the class for more dialogue. PI requires them to first answer with their own reasoning, then compare that to another’s. Do out of class videos serve this same purpose?
I don’t feel like I have an answer to lots of the questions I have posed above. However, the main point I want to get across is that I think it is silly to focus the flipped classroom conversation on what takes place outside of class; the power of flipping (which I would then argue is really the power of quality instruction) is the changes that can be made inside the class to promote student learning. Let’s just focus on how students actually learn, then
teach them guide them to understanding using effective methods.
UPDATE: Here’s another resource that discusses the use of dialogue in Physics classes, though some of the information is the same as those listed above. The Art (and Science) of Questioning via Clickers (podcast).
1 This is for the general level physics classes. It is noteworthy that my advanced classes have scored significantly higher. In the two years I have been testing them they have posttest averaged around 70%. Though this number is much higher, I am not satisfied with what would equate (in a standard grading scale) to a C- average, particularly with advanced kids. I do think it is interesting, however, that with essentially the same type of instruction these kids score so much higher. It is probably a combination of three things, in my estimation. 1) Higher scientific reasoning skills, which makes me wish I had given Lawson’s Classroom Test of Scientific Reasoning. I don’t want to over-test though. 2) More depth, both mathematical and conceptual, in the advanced class. 3) The idea that students who make it to the advanced classes are those who are able to have more internal dialogue and compare what they are learning to their own understanding without the need for the external dialogue. This may correlate to number 1, though.