Instructional Resources in Programming

Khan Academy's new computer science platform is a refreshing deviation from the lecture-based curriculum resources previously offered by Khan, and has the potential to further encourage the exploration of programming as a tool for teaching creative problem solving.

Khan Academy just unleashed a new computer science platform. As somewhat of a KA skeptic, I have to admit that I was doubtful when I got the news, but on first glance it seems like this thing totally rocks. Before I get into the thick of some thoughts I've been having about instructional videos and programming, let me identify a few things that they're getting right in a most beautiful way:

• The platform is elegantly constructed, and allows for instantaneous feedback on how a change in the code has changed the behavior of the program. (...and the sliders on numerical values are awesome!)

• The activities are almost completely open-ended. Users are free to mess around with the existing program in any way that that want, or to start their own program fresh. Some of the benefits of this approach are outlined in a manifesto/blog post by KA Javascript-dude John Resig.

• Instructional narration is left off most videos. When narration is included, it's entirely optional unobtrusive for those inclined toward a more playful approach. Also, the narrator of the videos is a woman - isn't it about time we razed this CS boys' club to the ground?

I've suggested before on this blog that programming can be an ideal environment for problem solving. A programming environment offers explicit connections between cause and effect, provided the programmer is able to navigate a ruthlessly picky syntactical landscape. In other words, to succeed at programming, a student needs to both become familiar with the details of the language AND apply this language conceptually in creative ways to solve problems. The role of instructional videos in the pedagogical balance of concepts and processes is hazy, but this latest resource hits closer to the mark than anything I've seen from KA.

I found myself thinking about this balance a few months ago, when I made a few instructional videos about programming in Python. In the following video, I've tried to provide instruction on some commands required to code a "guessing game" (specifically, the use of the conditional statements if and while), but withhold some details that are necessary for completing the task. In making the video, I remember struggling with the puzzle of where to draw the line between giving student the support they need to go forward and leaving some freedom to explore the programming challenge as an outlet for creative thinking:

Rather than demonstrating outright how to complete the task, my goal was to provide what a student might need to get started programming and troubleshooting. I tried to provide a concrete foundation for why the basic design of the program is valid, and to model good programming behavior in my presentation. (For example, I put the program together in small chunks, frequently testing the individual chunks to confirm that they did what I expected them to do, and I consulted the internet for the answer to a specific question about Python syntax.) It was interesting to me to compare my video to the new KA tutorial on if statements, since the KA tutorial does some things much more effectively than mine. For example, the KA tutorial targets the if command explicitly, making it a more efficient resource than my own somewhat rambling contribution.. More importantly, the KA tutorial introduces the NEED for an if statement before the introducing the command itself. This seems similar to the Modeling Instruction tenet of introducing and defining a concept before giving it a name.

Some time ago, a math teacher in New York City created this excellent video, called "What if Khan Academy was made in Japan?" to contrast KA's "Watch. Practice. Learn." approach with a more effective "Struggle. Struggle. Learn." approach. That is, in countries like Japan and Finland, where math education programs are more successful than most programs in the US, students spend most of their class time working through difficult problems, rather than being told how to solve such problems by their teacher. At the end of the video, we're given an example of what a more effective instructional video might look like.

The main takeaway from this video is that learning takes place when students struggle with questions, not when they sit through an answer. While it's clear that instructional videos are useful for absorbing procedures, they have questionable value as tools for facilitating real conceptual learning, and for this reason most of Khan Academy still stands at odds with inquiry education. Drawing a line between concepts and processes is perhaps easier in programming than in math, but the KA Computer Science platform is a refreshing change, even from previous KA programming resources.

One hugely promising feature of the new KACS platform is that there aren't really any questions and answers to begin with, so it's up to the users of the site (teachers and students alike) to decide how to direct this sandbox experience. In the spirit of guided inquiry, I can imagine a series of "mod prompts" designed to identify unique or significant elements of the code. For example, in an assignment based on this cool little KA program that draws a radar screen from ellipses and lines in a rotating reference frame, a teacher could assign various challenges,* from "Change this program so that rotating bar is shorter in length" to "Change this program so that the trail on the line takes more time to fade away." Solving either of these prompts requires the alteration of only one value in the code, but identifying which value needs to be altered can be a non-trivial conceptual challenge. (I struggled with that second question for a little bit, but figuring it out gave me a much clearer picture of how the program was constructed. I'll leave you to figure it out on your own!!)

I'm still fascinated by the potential of programming to serve as a tool for exploring creative problem solving. I'm struck by the degree to which the environment adopted by KACS, which emphasizes the modification of existing programs over building programs from scratch, eliminates the need for a lot of direct instruction on programming language process. As KACS goes forward, I hope that we'll see more programs and tasks emerge, similar to how an educational community has been built up around Python. If the inspiration behind this branch of Khan Academy someday finds its way into the other curriculum resources they offer, Khan Academy users will be better served indeed.



*A few of the introductory programs on the KACS site include prompts to Change this Program, but this is (for the time being) left off most programs. I'm not sure whether I think it'd be a good thing for these prompts to be a more dominant part of the KACS experience. Certainly it would make for a more self-contained curriculum, but it may also excessively limit creativity. I guess we'll have to wait and see whether this expands as KACS grows further.

What We Talk About When We Talk About Physics First

Any conversation about Physics First will tend to revolve around a few distinct motives, priorities, and assumptions. In order to communicate successfully about the benefits or limitations of inverting to a Physics First sequence, it's crucial to explicitly identify what we're bringing to the table in these conversations.

I recently watched online a recording of a fantastic talk given at the recent AAPT conference by Dr. Philip Sadler, this year's recipient of the Millikan Medal for "educators who have made notable and creative contributions to the teaching of physics." A very small portion of this talk is devoted to Sadler's characterization of what he called The Physics First Hypotheses: 1) physics knowledge can inform a study of chemistry and 2) chemistry knowledge can inform a study of biology. (A slide showing Sadler's wording is shown here.) This characterization of the potential benefit of Physics First is much narrower than my own, and it got me thinking about the motives, priorities, and assumptions that have come to be associated with an inverted science sequence. Sadler is explicit in identifying his own narrow interpretation of the term, but it strikes me that we rarely do a good job of clarifying exactly what we're talking about when we talk about Physics First.

The focus of this blog has been to investigate more closely the diverse mix of ideas that get lumped together under the Physics First umbrella, so it seems useful to try to summarize some of this diversity explicitly in a single post. In the paragraphs below, I've tried to characterize a number of ideas often linked to Physics First, along with a few personal thoughts and reactions about each. I'm sure I've missed some stuff here, so please get in touch (via comments, email, or Twitter) if you can suggest anything that should be on the list!

A. Physics for All: Many schools that have implemented Physics First offer physics as a required course for all ninth graders. Rather than a physics course taught only to self-selected, highly-motivated students, physics as a required course means that physics is taught to all students. Among folks who are passionate about physics this opportunity to reach more students is one of the most popular arguments for Physics First, but it requires conceiving of the physics course as a fundamentally different classroom experience. I'm convinced that this is an extremely good thing, but it makes some traditionalists nervous. Though most Physics First schools offer physics for all, some instead offer physics as an "Honors" option for students who have passed out of a ninth grade biology course (or have otherwise demonstrated a capacity for high-achievement in science). In my limited experience such courses tend to be taught in a more traditional, lecture-based style.

B. Physics Helps Develops Mathematical Proficiency: This is a popular motivation for Physics First in districts where student struggle to achieve high scores on standardized tests in math. These schools or districts can see an early physics class as an opportunity to provide relevant context to students' study of algebra and geometry, or even simply to teach more math earlier. This motivation has positive and negative aspects, of course. In some schools, the Physics First course has led to greater communication and cooperation between the science and math departments, and students have as a result come to see the math as a powerful tool for solving real-world problems. In other schools, physics teachers are asked to take time out of their study of physics to drill algebra problems in preparation for students' upcoming state-standardized test in math, or the physics curriculum itself is centered around around drilling traditional pencil and paper problems, devoid of scientific context.¹

C. Conceptual Physics: In environments where improving performance on standardized tests is less of a priority, some ninth grade physics teachers teach a physics course that contains as little math as possible. Teachers of a Conceptual Physics course will select topics that require only rudimentary algebraic or graphical analysis. Paul Hewitt's "Conceptual Physics" text is often used as a resource in such courses.² In my opinion, any good physics course should emphasize conceptual understanding over rote memorization or blind problem solving, but teaching concepts without computational context can be quite challenging. Answering a question like, "Why does it hurt less to fall onto grass than to fall onto concrete?" requires an abstract appreciation of the relationship for impulse (F_net • ∆t = m • ∆v) that is significantly more sophisticated than that required to solve an algebra problem. In other words, removing math from a physics course doesn't always make the course easier. (The complexities raised here have been on my mind for a while, but they'll have to wait until a future post!)

D. Physics is a Foundational Science: The classic biology uses chemistry and chemistry uses physics so teach physics first line of reasoning is probably the most common argument for inverting the BCP sequence, and in my opinion the least salient. Though Sadler's data are are only questionably connected to the Physics First discussion,³ they do show that content knowledge in physics (at the novice level of an introductory student) doesn't seem to translate to greater success in biology and chemistry. Whether this is because of too little crossover of content between these respective courses or simply because of a disconnect in representation and terminology, it's clear that simply having taken a physics course will not, statistically, impact an individual's success in a chemistry or biology course.

E. Inverting the Sequence Can Spur Pedagogical Change: My most recent post identified the potential for inversion to a Physics First sequence to bring about greater change within a science department. I argued that, since inverting a sequence necessitates changes to all high school science offerings, greater cohesiveness throughout the high school science curriculum can be achieved. A similar argument can be made within the physics class itself. A transition to Physics First is an opportunity to upset the status quo at a school, and can serve as incentive for teachers accustomed to using lecture-based instruction to try something different and potentially more effective.

F. Physics Helps Develops Scientific Reasoning and Critical Thinking Skills: Some physics curricula, such as a those built on ASU's Modeling Instruction or Rutgers University's PUM (Physics Union Mathematics), are based on the notion that students can only effectively learn science by doing science. The relationships, representations, and conceptual understanding in each unit of the course are built from the ground up by the students themselves, through analysis of empirical evidence and class consensus arrived at through discussion. Aside from being an effective method of teaching physics (no small thing, of course!), this approach instills in students the unique supremacy of empirical observations. In other words, students learn that evidence matters. Physics is especially suitable for a course in "evidence-based problem solving" because the systems and relationships we study can simultaneously be both gorgeously simple and puzzlingly counter-intuitive. Not all science teaching methods emphasize this skill, but some methods do, and a transition to Physics First presents a golden opportunity to transition to using such a method (see motivation/assumption E).

In my opinion, this last motivation is the single strongest argument in favor of both inverting a curriculum to Physics First and teaching physics to all students. By designing and analyzing experiments, students learn a scientific approach to problem solving - not just a figure out how far the ball goes type of problem solving, but a broader and more relevant figure out whether this pill can do what it says it does, figure out whether this politician can do what he/she says he/she can do, or figure out how to turn this rope into a rope swing type of problem solving. Students learn to always look for relevant evidence, and to be thoughtful and critical in interpreting that evidence. They learn that failure is an essential part of any problem-solving process, and that successful problem-solvers keep an open mind as they learn from their mistakes through repeated trials and errors. They experience first-hand the immense benefits and inevitable challenges of collaborative work. In designing a syllabus for a Physics First course with this emphasis, the question is not, "What physics content is essential for my students to know?" (As much as I hate to admit it, evidence suggests that zero physics knowledge is essential for leading a fulfilling life!) Instead, the question can be, "What physics content will be effective for developing critical thinking skills in my students?" What better time to start explicitly developing such skills than in ninth grade?

Sadler points out in his talk that a crucial piece of education reform is looking for evidence that indicates whether motivations and assumptions such as I've outlined here are statistically significant. Of course, it's up to teachers and education researchers to provide this evidence, and in the case of Physics First this evidence has been particularly slow in coming. I'm convinced that this lack of evidence is partly due to the great diversity and resulting disconnect between various implementations and advocates of Physics First. Sadler's data seem to refute the notion that familiarity with physics concepts promotes success in chemistry and biology (motivation/assumption D above), but I have not yet seen evidence to confirm or refute motivation/assumption F.⁴ In advocating for Physics First, it's not enough to point to decent FCI gains in ninth graders as a reason for the switch, but with no consensus of priorities within the Physics First movement it's hard to figure out where to point. I'm hopeful that a synergy between progressive teaching methods like Modeling Instruction and the Physics First movement can provide direction toward the motivations I've advocated for here, but until we see some real numbers we'll just have to call it a hunch...


1: There is a subtle distinction to be made here. Students drilling quantitative problems involving graphs and algebra can often look similar to students using these same graphs and algebra as tools to solve a scientific problem or make a prediction. In some curricula intended for use with ninth graders, students might even develop those "drilling techniques" through small group and whole class discussions, somewhat similar to a scientific inquiry process. The difference, however, is in the students' motivation for doing this work - whether they're doing math for the sake of getting through the worksheet, or for the sake of building more sophisticated conceptual understanding. The best way to tell these apart, I'd say, is to look at what happens before and after the graphs and algebra: are students collecting data about a situation, then using the analytical techniques they've developed to make, test, and revise predictions, or are they just doing more algebra?

2: Paul Hewitt has said that he originally intended Conceptual Physics to be used with ninth graders. His publishers, he claimed, refused to promote the book as a Physics First course, since so few schools taught Physics First. Therefore, various different editions of the book have come to be used in courses for a variety of age levels, from middle school to undergraduate university students.

3: Sadler says in his talk, "We needed large numbers and there aren't large numbers of people who do Physics First. So what we looked at was how much of each of these sciences kids took in high school, and then used it to predict their college grades in these other fields." Though this approach may measure the conceptual connections between high school science courses and university level courses, it doesn't seem to me to be all that relevant to the potential benefit of Physics First.

4: I don't really know what we might use as an effective measure of gains in the rather ambiguous area I've identified as my primary focus, but I'm looking for something. A pill-purchasing or rope-swing-construction pre/post test doesn't really seem like the right way to go, so if you have suggestions, please include them in the comments below!

PS: Tim Burgess' comment below should link to the following page - a great wealth of physics first research.