Things unfold over time. One thing I remember from my high school history lessons is that a particular historical event, e.g., a war, could be triggered by one single incident, but might also be already predetermined by a broader historical context. In the time dimension, various factors come into play, interact with each other, and frame historical events. So interpretation of any event needs to take various factors and players involved. And that’s what makes history so interesting.
Cooking is another interesting context to think about time. While a perfect dish requires the right amounts of different ingredients and them being cooked for a certain amount of time, in many cases it also depends on ingredients been added in a certain order so they could interact and produce desired flavors.
Time is important for learning and teaching. Educators ask all sorts of questions about time: In which age should students be taught what and what? How much time a pupil should spend in school each day? At which time should students be doing what types of activities? In a specific lesson, what should a class do in a specific moment? By asking these questions, we care about (1) the optimal temporal moment learning should happen, (2) the amount of time learning requires, and (3) the desired sequence of events to achieve a learning goal.
My recent work in the area of learning analytics has focused on the third aspect—the sequence of learning events—which I think is under-represented in educational research.
In classrooms where most of my research happens, students work together to build explanations about scientific phenomena or historical events, following a pedagogy called knowledge building. In the process, students carry on communal, constructive dialogues, with a goal to advance their shared understanding. They contribute to their dialogues in many different ways, such as asking a thought-provoking question, proposing a theory, introducing information, making synthesis, and monitoring discussion. We call them ways of contributing to knowledge-building dialogues, and have been working to understand what a “good” dialogue looks like based on the composition of different contribution types, just like figuring out the right ingredients for a good dish.
However, research efforts so far relying descriptive metrics of contribution types have largely failed. They failed because of the failure to recognize the fact that a delicious dish is not only a function of the amounts of required ingredients, but also how they get added. Like many educational studies, our efforts thus far applied a “coding and counting” strategy, collapsed the temporal structure of events, and forced the important temporal aspect of learning to be lost.
My LAK14 paper attempts to (start to) address this issue. By applying Lag-sequential Analysis, I uncovered what really matters for effective explanation building dialogues. While descriptive metrics failed, I identified a few transitional patterns among contribution types distinguishing effective dialogues from ineffective ones. More specifically, effective dialogues require students to work more constructively with resources, engage in increasingly deepened questioning and theorizing, and problematize proposed explanations. More details of this research could be found in my presentation and paper.