Statistical modeling is typically seen as an advanced skill and rarely introduced to students at the introductory level. But in data science and its applications (e.g., public health, politics), modeling is a necessary component of data literacy. To address this need, we should teach introductory students modeling from the beginning, connecting the content to the modern world of data science. This may have the added benefit of bringing coherence to statistics. We teach this content using research-based pedagogy – the practicing connections framework – and utilize new technology (CourseKata.org) capable of conducting experiments to continuously improve instruction. We are thus able to test theories of how students learn difficult, time-consuming concepts, such as the concept of a statistical model. Together, the curriculum, pedagogical theory, and technology provide a process to make the book incrementally better at producing a modern, coherent, and flexible understanding of statistics.

Remote instruction lacks shared physical space and physical social presence. This can make it difficult to incorporate embodied learning techniques, which have been shown to strengthen learning outcomes, into our lessons, and can make it difficult to feel socially connected to peers. We propose that incorporating embodiment into a synchronous lesson by engaging students in a physical task using a shared set of materials can help strengthen social and conceptual connections. We outline a high school statistics lesson on measurement error that involves building a popsicle stick catapult and measuring gummy bear launch distances. We distributed a set of all the necessary materials to the students at the start of the course. The activity simulated a sense of shared space and brought about multimodal learning and shared physical experiences. We witnessed enhanced joint attention, the development of physically grounded understanding, and increased engagement.