Anyone who makes a distinction between games and learning doesn’t know the first thing about either. Marshall McLuhan
Following are summaries of two research projects sharing the same general aims. The first goal is to get more kids, especially females and certain underrepresented minorities (e.g., Black and Hispanic children), excited about and interested in science—specifically physics. Then, because interest alone is not enough, our second goal is to identify ways to facilitate and deepen science-related learning in immersive environments. Well-designed digital games represent a promising approach for meeting both goals: elevating children’s interest in STEM fields like physics, and supporting active, contextualized learning. This research aims to be both theory-driven and theory-contributing, particularly in the area of designing valid STEM learning games.
NSF Project. When it comes to math and science, American students are lagging behind many of their peers in developed countries (PISA, 2015). This can negatively impact their STEM interest—discouraging many from pursuing STEM-related majors in college, especially females and underrepresented minorities (e.g., black and Hispanic children). Our three-year project will help make STEM learning more accessible, engaging, and effective for all kids. To meet this goal, we are enhancing our 2D physics game called Physics Playground (PP) (Shute & Ventura, 2013), originally designed to dynamically assess students’ understanding of qualitative physics. The game is nonlinear where players can pick any of the 75 levels to play. To win a level, players draw objects (i.e., ramp, lever, pendulum, and springboard) on the screen to guide a green ball to hit a red balloon. In the new version of PP, we’re focusing on supporting formal physics understanding by designing and embedding various cognitive supports (e.g., animated worked examples, engaging concept-explanation videos, fun physics facts). In Year 1, we’re designing new task types and associated game levels. For instance, in addition to the existing sketching levels, we’ve added “manipulation” levels where players don’t draw anything, but instead, they manipulate physics parameters (e.g., gravity, mass, air resistance) and observe subsequent changes to the environment. The videos below show a sketching level and a manipulation level and their worked examples. Click on the provided video below to view the animated worked examples.
Video 1. The sketching level “Caveman.” The path of the green ball is blocked by a wooden club. To solve the level, players need to draw a lever to lift the wooden club that blocks the path.
Video 2. The manipulation level “Cookie Monster.” In the level, mass and gravity are set to default levels, air resistance is medium, the static blower is in full force, and bounciness is available. To make the ball hit the balloon, players need to reduce air resistance and gravity and enable bounciness.
To test these new tasks, we’re currently conducting a usability study at a local K-12 school where 25 8th and 9th graders play PP for two 1-hour sessions, thinking aloud as they do so. Observations of and interviews with the students will help us to see what works and what does not and fine-tune the game/levels accordingly. In Year 2, we’ll (a) develop an adaptive level selection algorithm, and (b) conduct a pilot study to test its effectiveness. We’ll use Bayes nets within a technology called stealth assessment to create the algorithm (Shute & Ventra, 2013). That is, when students play PP, their game data are captured and analyzed to update their student model in real-time. The estimates of students’ learning states allow PP to provide adaptive level selection and tailored supports. In Year 3, we’ll conduct a controlled evaluation to test whether students learn better under adaptive sequencing condition than in other conditions (i.e., linear and free choice—the control condition).
IES Project. While our NSF project focuses on cognitive supports in PP, our IES project additionally examines affective supports relative to their influence on learning and interest in science. Like the cognitive supports, the affective supports will be embedded in PP and are expected to fuel motivation when students succeed and encourage persistence when they fail. We’ll investigate the effects of various types of affective supports in PP including motivational messages and automated affect detectors (i.e., interaction-based detectors and webcam-based detectors). The interaction-based detectors capture data from the Baker Rodrigo Ocumpaugh Monitoring Protocol (BROMP) (Ocumpaugh, Baker, & Rodrigo, 2015) and the game log files. The webcam-based detectors record facial expressions (e.g., raised eyebrow) and body gestures (e.g., head position). Both datasets will be used to estimate students’ frustration, confusion, and boredom. We’ll conduct four experiments across four years. Around 200 7th to 9th graders from two K-12 schools in Florida and New York will participate in our experiments. During each experiment, they’ll play PP across four 50-minute sessions (spanning four days). In Experiments 1 and 2, we’ll evaluate the effects of cognitive and affective supports individually and combined. In Experiment 3, we’ll compare two support delivery methods (i.e., game- vs. student-controlled). In Experiment 4, we’ll examine the effects of using affect detectors to tailor supports on learning and engagement. Both projects (NSF and IES) will yield valuable information on the design and development of the next-generation STEM learning games for our children, especially underrepresented minorities.