Assessments provide a critical diagnostic tool for educators and can provide actionable feedback to students so they can act to improve their understanding. In the vast literature on defining actionable feedback, there is a lack of studies on actionable feedback in computing contexts. Feedback Engagement and Analysis in Assessments (FEAST) aims to define actionable feedback in computing and provide a guide for educators on how to provide actionable feedback in computing education.

About 15 years ago, education researchers conducted a systematic literature review (SLR) on change strategies for improving undergraduate STEM education instruction. Analyzing 191 articles from 1995 to 2008, the researchers identified four broad categories of change strategies through a comprehensive interdisciplinary literature review: (1) disseminating curriculum and pedagogy, (2) developing reflective teachers, (3) enacting policy, and (4) developing a shared vision. With recent developments in STEM education practices, it is imperative to conduct a follow-up SLR comparing the effects of change strategies and associated student success. Similarly, with the influx of scientific articles published in recent decades, it is time consuming to conduct comprehensive SLRs without the assistance of machine learning (ML) analysis techniques. Working with qualitative researchers, we investigate the impact and ability of using ML to assist in these analyses. While most ML approaches can be easily written with a few lines of code, using them to extract meaningful information for literature reviews is challenging. In this work, we describe our experience with integrating machine learning techniques into the analysis pipeline of SLRs. Specifically, this poster will: (1) share results from clustering analysis to identify themes of the chosen abstracts, (2) explore the effects of data bias on found clusters, (3) present the challenges of adding machine learning into SLRs, and (4) assess if the addition of ML in SLRs aligns with the expected goals of qualitative researchers.

Over ten years ago, our collaborators conducted a NSF-funded project identifying four broad categories of change strategies used to improve undergraduate STEM education through a comprehensive interdisciplinary literature review of articles from 1995 to 2008. Since this first iteration, there have been many major developments in undergraduate STEM education; particularly, the rapid development in sophisticated technology tools. These developments affect the nature of classroom instruction as well as expand the bounds on analyzing a corpus of articles. Thus, it is crucial to repeat this review to better understand the changes in STEM education from the more recent past. Our goal is to use machine learning to identify, potentially new, themes in the recent literature. We plan to compare and contrast both AI-assisted modeling and traditional, human-qualitative coding approaches in an effort to: (1) identify the benefits and faults of using AI verses human coding, and (2) portray a comprehensive story of change instruction literature from 2010. This lightning talk will describe the data extraction process and preliminary results from machine learning models. In addition to sharing the beginnings of our work, we hope to gain new perspectives and ideas from computing education scientists regarding our data representation and modeling choices.

As computing becomes increasingly intertwined with other disciplines, research that centers computing education in an interdisciplinary context (and the challenges surrounding it) is increasingly relevant. In particular, writing modular code (i.e. using functions) is a fundamental part of scientific programming. However, functions have been identified as challenging for students to learn. This assignment leverages a fun, real, and approachable dataset from the 2018 Central Park Squirrel Census as well as experiences authentic to developing scientific programs to introduce the concept of functions in Python. The assignment is intended to be delivered in a "flipped classroom" format, where students are first introduced to concepts in videos and short problems prior to coming to class. Once in class, the students work collaboratively in groups of 4-6 with a hands-on programming activity. By the end of the assignment, students have not only had the opportunity to learn about functions in an authentic context, but they have also been able to make calculations and propose their own problems to learn about the data.

Students struggle with how to loop through lists in Python, often mixing and matching their strategies between looping by value and looping by index. At the heart of this issue is the concept of indexing, which can be a challenging abstraction for students. This work uses Resource Theory, a cognitive framework based on DiSessa's Knowledge in Pieces, to identify concepts and procedures that students activate when solving problems involving indexing.

At a large Midwestern university, students with a variety of prior knowledge and experience with respect to computation engage in our interdisciplinary introductory computation courses. We strive to create a curriculum that brings students' disciplines and experiences together with computer science concepts, data analysis, and computational modeling. In this work, I present two comparative case studies exploring how students with varied prior computational experiences approach computational tasks. The data generated consist of student interviews where they describe their thought processes aloud while engaging in tasks presented in a Jupyter notebook. In particular, I compare the characteristics of the pathways that students take to arrive at a solution to a particular task. The insights gained from this work will drive curricular change to better facilitate student learning.

Few would argue that modern careers across a wide range of disciplines can be performed in the absence of computing in one form or another. As such, it is becoming increasingly important for our education system to appropriately prepare students for the modern world by integrating computing and computational thinking into how students learn disciplinary content (i.e. disciplines outside of computer science). However, how to best perform this integration is not yet known, nor is there likely to be one ''best'' method. In this session, we will facilitate a discussion of the variety of ways in which post-secondary institutions are actively using computing to support disciplinary learning by highlighting examples while also considering mechanisms not yet explored. We invite anyone who wants to learn more about current efforts, share their own experiences, and contribute ideas for future endeavors. In an effort to maximize the diversity of perspectives in this discussion, we've included discussion leaders who span a variety of roles within curriculum development and classroom instruction and encourage participants from all levels and backgrounds.

Data science is often heralded as a key learning goal for students in STEM classrooms. There are also myriad efforts to integrate data science into these classrooms, and many dedicated research efforts for identifying the best ways to do so. However, the problem is that there is little agreement on how to introduce data science to students, whether it be through computer science courses where students can learn programming, through STEM courses where students can learn disciplinary knowledge, or through newly designed data science centric courses. Furthermore, best practices for teaching data science require an understanding of what data science is from students' perspectives, and how they experience it. This poster explores this problem by showcasing an interview study of an undergraduate course offered at Michigan State University, which focuses on computational modeling and data analysis. Students in this course learn data science via problem-based group work and apply it to several disciplinary contexts. The interview study examines how students perceived what they learned, and how their self-efficacy developed over the course of the semester. In effect, we demonstrate a course where students are learning data science, identify the key features of the course that students perceive, and build an understanding of data science self-efficacy, which can be used to help design positive, effective experiences in data science courses.