Background: Prior research about lecturing practices in computer science (CS) has not focused on the accessibility of those practices.

College has a lot of unwritten rules, and some of our students will matriculate knowing those rules while others don't. Research has shown that revising homework assignments to make these unwritten rules explicit can help students' academic confidence and sense of belonging. This tutorial is intended for faculty who may be interested in revising their assignments to make those unwritten rules clear to students. We will focus on Transparent Assignment Design, a framework used to make explicit the purpose, task, and criteria of assignments. Come to this tutorial for a crash course on making your assignments transparent followed by hands-on help! Please bring an assignment you are interested in revising (or we will provide samples) and your choice of editing tool (e.g., laptop, pen/paper).

Background: Computer Science (CS) industry internships are valuable experiences for undergraduate CS students, and internship participation is predicted by extracurricular participation.

Students from historically underrepresented communities in computer science (CS) report being told that their successes are due to special treatment based on their gender and/or racial identity. We refer to this microaggression as the discounting-success microaggression.We analyzed survey responses from 4,327 CS majors across 221 institutions in the U.S. We found that students who identify as women, Black, and/or Asian were more likely than men and white students, respectively, to report the discounting-success microaggression. This discounting-success microaggression significantly and negatively predicts students' self-efficacy, sense of belonging, and plans to persist in CS. Our results elucidate the negative influence of the discounting-success microaggression on CS student outcomes. Efforts are needed to improve the culture and interactions in CS to eliminate the prevalence of this harmful microaggression.

Background: Instructional transparency makes a course’s learning goals, evaluation criteria, and path to success clear to students, with the goal of improving equity in higher education. Increased transparency may improve equity by bolstering students’ self-efficacy and sense of belonging in computing, both of which are correlated with persistence in the field. Purpose: We aim to understand whether there are group differences in how students perceive and benefit from instructional transparency. We are additionally interested in understanding whether perceiving instructional transparency is positively correlated with students’ self-efficacy and sense of belonging and, therefore, can contribute to the persistence of students from historically underrepresented groups in computing. Methods: To investigate these relationships, we used linear regressions to analyze survey responses from 11,046 undergraduate students from 203 institutions. Findings: We found that there are group differences in students’ perception of transparency in their CS courses: students who identify as women, first-generation college students, and/or disabled reported perceiving less instructional transparency than their peers. We also found that perceiving more transparency has a positive correlation with students’ self-efficacy and sense of belonging in computing while controlling for important confounding variables, such as prior CS experience. We further demonstrated that this relationship is different for certain groups of students: first, for Black students and first-generation college students, perceiving transparency has a larger positive impact on their self-efficacy, and second, for Hispanic students, perceiving transparency has a smaller positive impact on their sense of belonging. Contributions: Our work constitutes one of the first empirical, multi-institutional investigations of the perceptions and benefits of transparency in CS classrooms that focuses on group differences. Our work also includes a theoretical articulation of the mechanisms through which transparent teaching practices may influence students’ self-efficacy and sense of belonging in computing. Taken together, our empirical findings and theoretical argument provide important evidence for the benefits of instructional transparency in CS courses, particularly as it relates to improving equity in computing.

Broadening participation in computing (BPC) is a focus in industry and academia. Extant research focuses on what broadening participation in computing (BPC) efforts are pursued, while we propose focusing on how change happens. Our qualitative study applied John Kotter's (2012) eight-stage change framework to analyze interviews with faculty and staff engaged in BPC efforts. Illustrative examples from our interviews elucidate each of the eight stages and how they can be applied to pursue organizational change efforts that support BPC.

Background : Affirmative action programs (AAPs) aim to increase the representation of people from historically underrepresented groups (HUGs) in the workforce, but can unintentionally signal that a person from a HUG was selected for their identity rather than their merit. We call this signal the diversity-hire narrative. Prior work has found that women hear the diversity-hire narrative during their computer science (CS) internships, but women and non-binary students' experiences surrounding the narrative are important to understand and have not been thoroughly explored.

Background: The goal of affirmative action programs (AAPs) is to address the underrepresentation of people from historically underrepresented groups (HUGs) in the workforce. People who identify as women, Black or African American, Hispanic or Latinx/a/o/*, Native American, Native Alaskan, Native Hawai’ian, and/or Pacific Islander are considered to be a part of HUGs in computing. AAPs can unintentionally signal that a person from a HUG was selected for a position based on their gender or race/ethnicity rather than their merit [2, 3, 5, 6]. We call this signal the diversity-hire narrative. In computing, prior work has found that women hear the diversity-hire narrative during their computer science (CS) internships [4], but women’s experiences surrounding the narrative have not been thoroughly explored.

In this workshop, participants will learn to use physical Java memory models to help students develop a deep conceptual understanding of Java variables. These physical models use the metaphor of a "remote control" to introduce references. Using table-top versions of the physical models, participants will discuss and be prepared to use physical objects to help students understand: (1) how primitive variables and references variables behave in similar and different ways, (2) how reference variables that both reference an object or array can both modify its content, (3) how calling a method with an argument creates a local variable within that method, and (4) inside an object method, the variable this references the object we called the method on. Participants will receive a set of these physical models to use in their classroom.

Artificial intelligence (AI) and cybersecurity are in-demand skills, but little is known about what factors influence computer science (CS) undergraduate students' decisions on whether to specialize in AI or cybersecurity and how these factors may differ between populations. In this study, we interviewed undergraduate CS majors about their perceptions of AI and cybersecurity. Qualitative analyses of these interviews show that students have narrow beliefs about what kind of work AI and cybersecurity entail, the kinds of people who work in these fields, and the potential societal impact AI and cybersecurity may have. Specifically, students tended to believe that all work in AI requires math and training models, while cybersecurity consists of low-level programming; that innately smart people work in both fields; that working in AI comes with ethical concerns; and that cybersecurity skills are important in contemporary society. Some of these perceptions reinforce existing stereotypes about computing and may disproportionately affect the participation of students from groups historically underrepresented in computing. Our key contribution is identifying beliefs that students expressed about AI and cybersecurity that may affect their interest in pursuing the two fields and may, therefore, inform efforts to expand students' views of AI and cybersecurity. Expanding student perceptions of AI and cybersecurity may help correct misconceptions and challenge narrow definitions, which in turn can encourage participation in these fields from all students.

Technology should be accessible and inclusive, so designers should learn to consider the needs of different users. Toward this end, we created the theoretically-grounded CIDER assumption elicitation technique, an educational analytical design evaluation method to teach inclusive design skills. CIDER ( Critique , Imagine , Design , Expand , Repeat ) helps designers recognize and respond to bias using the critical lens of assumptions about users . Through an 11-week mixed-method case study in an interaction design course with 40 undergraduate students and follow-up interviews, we found that activities based on the CIDER technique may have helped students identify increasingly many types of design bias over time and reflect on their unconscious biases about users. The activities also had lasting impacts, encouraging some students to adopt more inclusive approaches in subsequent design work. We discuss the implications of these findings, namely that educational techniques like CIDER can help designers learn to create equitable technology designs.

Motivation: Although CS Education researchers and practitioners have found ways to improve CS classroom inclusivity, few researchers have considered inclusivity of online CS education. We are interested in two such improvements in online CS education—besides being inclusive to each other, online CS students also need to be able to create inclusive technology. Objectives: We have begun developing a new approach that we term “embedded inclusive design” to address both of these goals. The essence of the approach is to integrate elements of inclusive design education into mainstream CS coursework. This paper presents three curricular interventions we have developed in this approach and empirically investigates their efficacy in online CS post-baccalaureate education. Our research questions were: How do these three curricular interventions affect (RQ1) the climate among online CS students and (RQ2) how online CS students honor the diversity of their users in the tech they create? Method: To answer these research questions, we implemented the curricular interventions in four asynchronous online CS classes across two CS courses within Oregon State University’s Ecampus and conducted an action research study to investigate the impacts. Results: Online CS students who experienced these interventions reported feeling more included in the major than they had before, reported positive impacts on their team dynamics, increased their interest in accommodating diverse users, and created more inclusive technology designs than they had before. Discussion: These results provide encouraging evidence that embedding elements of inclusive design into mainstream CS coursework, via the interventions presented here, can increase both online CS students’ inclusivity toward one another and the inclusivity of the technology these future CS practitioners create.

Explainable AI is growing in importance as AI pervades modern society, but few have studied how explainable AI can directly support people trying to assess an AI agent. Without a rigorous process, people may approach assessment in ad hoc ways—leading to the possibility of wide variations in assessment of the same agent due only to variations in their processes. AAR, or After-Action Review, is a method some military organizations use to assess human agents, and it has been validated in many domains. Drawing upon this strategy, we derived an After-Action Review for AI (AAR/AI), to organize ways people assess reinforcement learning agents in a sequential decision-making environment. We then investigated what AAR/AI brought to human assessors in two qualitative studies. The first investigated AAR/AI to gather formative information, and the second built upon the results, and also varied the type of explanation (model-free vs. model-based) used in the AAR/AI process. Among the results were the following: (1) participants reporting that AAR/AI helped to organize their thoughts and think logically about the agent, (2) AAR/AI encouraged participants to reason about the agent from a wide range of perspectives , and (3) participants were able to leverage AAR/AI with the model-based explanations to falsify the agent’s predictions.

Although the need for gender-inclusivity in software is gaining attention among SE researchers and SE practitioners, and at least one method (GenderMag) has been published to help, little has been reported on how to make such methods work in real-world settings. Real-world teams are ever-mindful of the practicalities of adding new methods on top of their existing processes. For example, how can they keep the time costs viable? How can they maximize impacts of using it? What about controversies that can arise in talking about gender? To find out how software teams "in the trenches" handle these and similar questions, we collected the GenderMag-based processes of 10 real-world software teams---more than 50 people---for periods ranging from 5 months to 3.5 years. We present these teams' insights and experiences in the form of 9 practices, 2 potential pitfalls, and 2 open issues, so as to provide their insights to other real-world software teams trying to engineer gender-inclusivity into their software products.

Explainable AI (XAI) is growing in importance as AI pervades modern society, but few have studied how XAI can directly support people trying to assess an AI agent. Without a rigorous process, people may approach assessment in ad hoc ways---leading to the possibility of wide variations in assessment of the same agent due only to variations in their processes. AAR, or After-Action Review, is a method some military organizations use to assess human agents, and it has been validated in many domains. Drawing upon this strategy, we derived an AAR for AI, to organize ways people assess reinforcement learning (RL) agents in a sequential decision-making environment. The results of our qualitative study revealed several strengths and weaknesses of the AAR/AI process and the explanations embedded within it.

Inclusive design is important in today's software industry, but there is little research about how to teach it. In collaboration with 9 teacher-researchers across 8 U.S. universities and more than 400 computer and information science students, we embarked upon an Action Research investigation to gather insights into the pedagogical content knowledge (PCK) that teachers need to teach a particular inclusive design method called GenderMag. Analysis of the teachers' observations and experiences, the materials they used, direct observations of students' behaviors, and multiple data on the students' own reflections on their learning revealed 11 components of inclusive design PCK. These include strategies for anticipating and addressing resistance to the topic of inclusion, strategies for modeling and scaffolding perspective taking, and strategies for tailoring instruction to students' prior beliefs and biases.