Search for: All records

How to integrate foundational and advanced computer science content at the high-school level 

This guide presents the "MENTORS in CS" program, a comprehensive model for providing sustained support to K-12 computer science (CS) teachers, particularly those new to the field. It details the program's foundational structures, including mentor-mentee partnerships, a community of practice, and continuous research and refinement through design-based implementation research (DBIR). The guide offers actionable insights and reproducible resources, such as program timelines, recruitment materials, and a mentor toolkit, to facilitate the replication and scaling of similar equity-driven mentoring initiatives. Key learnings regarding participant outcomes, mentor training, and adapting the program for diverse educational contexts are shared to aid widespread dissemination and impact. 

Rapid advancements in artificial intelligence (AI) necessitate changes in what AI content is taught to K-12 students. These changes will ensure that students are prepared to be smart consumers and competent creators of AI, as well as informed citizens. To meet this need, CSTA, in partnership with AI4K12, spearheaded the Identifying AI Priorities for All K-12 Students project. The project gathered experts – including teachers, researchers, administrators, and curriculum developers – to articulate priorities for AI education. This report summarizes the result of that effort. 

This toolkit offers a structured, year-long framework to support peer mentoring partnerships in computer science education. It provides (1) a program calendar outlining monthly activities, scheduled mentor–mentee meetings, and three mentorship cycles aligned with key CS teacher standards; (2) guided self-reflection tools to help teachers identify professional strengths and areas for growth; (3) a partnership agreement to establish norms for communication and collaboration; (4) goal-setting templates with illustrative examples to scaffold targeted professional learning; and (5) mentoring logs to document bimonthly meetings and track progress across three 2.5-month cycles. 

This report examines the similarities and differences between K-12 CS standards across seven international locations. This analysis includes background information on each of the locations included in the study, trends in content, and comparisons by topic area. 

Introduction: Learning standards are a crucial determinant of computer science (CS) education at the K-12 level, but they are not often researched despite their importance. We sought to address this gap with a mixed-methods study examining state and national K-12 CS standards. Research Question: What are the similarities and differences between state and national computer science standards? Methods: We tagged the state CS standards (n = 9695) according to their grade band/level, topic, course, and similarity to a Computer Science Teachers Association (CSTA) standard. We also analyzed the content of standards similar to CSTA standards to determine their topics, cognitive complexity, and other features. Results: We found some commonalities amidst broader diversity in approaches to organization and content across the states, relative to the CSTA standards. The content analysis showed that a common difference between state and CSTA standards is that the state standards tend to include concrete examples. We also found differences across states in how similar their standards are to CSTA standards, as well as differences in how cognitively complex the standards are. Discussion: Standards writers face many tensions and trade-offs, and this analysis shows how – in general terms – various states have chosen to manage those trade-offs in writing standards. For example, adding examples can improve clarity and specificity, but perhaps at the cost of brevity and longevity. A better understanding of the landscape of state standards can assist future standards writers, curriculum developers, and researchers in their work. 

Introduction: Recent AI advances, particularly the introduction of large language models (LLMs), have expanded the capacity to automate various tasks, including the analysis of text. This capability may be especially helpful in education research, where lack of resources often hampers the ability to perform various kinds of analyses, particularly those requiring a high level of expertise in a domain and/or a large set of textual data. For instance, we recently coded approximately 10,000 state K-12 computer science standards, requiring over 200 hours of work by subject matter experts. If LLMs are capable of completing a task such as this, the savings in human resources would be immense. Research Questions: This study explores two research questions: (1) How do LLMs compare to humans in the performance of an education research task? and (2) What do errors in LLM performance on this task suggest about current LLM capabilities and limitations? Methodology: We used a random sample of state K-12 computer science standards. We compared the output of three LLMs – ChatGPT, Llama, and Claude – to the work of human subject matter experts in coding the relationship between each state standard and a set of national K-12 standards. Specifically, the LLMs and the humans determined whether each state standard was identical to, similar to, based on, or different from the national standards and (if it was not different) which national standard it resembled. Results: Each of the LLMs identified a different national standard than the subject matter expert in about half of instances. When the LLM identified the same standard, it usually categorized the type of relationship (i.e., identical to, similar to, based on) in the same way as the human expert. However, the LLMs sometimes misidentified ‘identical’ standards. Discussion: Our results suggest that LLMs are not currently capable of matching human performance on the task of classifying learning standards. The mis-identification of some state standards as identical to national standards – when they clearly were not – is an interesting error, given that traditional computing technologies can easily identify identical text. Similarly, some of the mismatches between the LLM and human performance indicate clear errors on the part of the LLMs. However, some of the mismatches are difficult to assess, given the ambiguity inherent in this task and the potential for human error. We conclude the paper with recommendations for the use of LLMs in education research based on these findings. 

Introduction: State and national learning standards play an important role in articulating and standardizing K-12 computer science education. However, these standards have not been extensively researched, especially in terms of their cognitive complexity. Analyses of cognitive complexity, accomplished via comparison of standards to a taxonomy of learning, can provide an important data point for understanding the prevalence of higher-order versus lower-order thinking skills in a set of standards. Objective: The objective of this study is to answer the research question: How do state and national K-12 computer science standards compare in terms of their cognitive complexity? Methods: We used Bloom’s Revised Taxonomy in order to assess the cognitive complexity of a dataset consisting of state (n = 9695) computer science standards and the 2017 Computer Science Teachers Association (CSTA) standards (n = 120). To enable a quantitative comparison of the standards, we assigned numbers to the Bloom’s levels. Results: The CSTA standards had a higher average level of cognitive complexity than most states’ standards. States were more likely to have standards at the lowest Bloom’s level than the CSTA standards. There was wide variety of cognitive complexity by state and, within a state, there was variation by grade band. For the states, standards at the evaluate level were least common; in the CSTA standards, the remember level was least common. Discussion: While there are legitimate critiques of Bloom’s Revised Taxonomy, it may nonetheless be a useful tool for assessing learning standards, especially comparatively. Our results point to differences between and within state and national standards. Recognition of these differences and their implications can be leveraged by future standards writers, curriculum developers, and computing education researchers to craft standards that best meet the needs of all learners. 

Introduction: Because developing integrated computer science (CS) curriculum is a resource-intensive process, there is interest in leveraging the capabilities of AI tools, including large language models (LLMs), to streamline this task. However, given the novelty of LLMs, little is known about their ability to generate appropriate curriculum content. Research Question: How do current LLMs perform on the task of creating appropriate learning activities for integrated computer science education? Methods: We tested two LLMs (Claude 3.5 Sonnet and ChatGPT 4-o) by providing them with a subset of national learning standards for both CS and language arts and asking them to generate a high-level description of learning activities that met standards for both disciplines. Four humans rated the LLM output – using an aggregate rating approach – in terms of (1) whether it met the CS learning standard, (2) whether it met the language arts learning standard, (3) whether it was equitable, and (4) its overall quality. Results: For Claude AI, 52% of the activities met language arts standards, 64% met CS standards, and the average quality rating was middling. For ChatGPT, 75% of the activities met language arts standards, 63% met CS standards, and the average quality rating was low. Virtually all activities from both LLMs were rated as neither actively promoting nor inhibiting equitable instruction. Discussion: Our results suggest that LLMs are not (yet) able to create appropriate learning activities from learning standards. The activities were generally not usable by classroom teachers without further elaboration and/or modification. There were also grammatical errors in the output, something not common with LLM-produced text. Further, standards in one or both disciplines were often not addressed, and the quality of the activities was often low. We conclude with recommendations for the use of LLMs in curriculum development in light of these findings.