The Cambridge Handbook of Computing Education Research  🔗 **Source:** [The Cambridge Handbook of Computing Education Research]() ## Highlights - current participation in CEd has a long way to go to achieve anything like equity of participation in terms of race or gender - Tools: Educators in science, engineering, and mathematics may use computing technology, but computing education is necessarily tied to technology. - we create tools that provide views on the program and its execution at different levels of granularity and from different viewpoints, so making computing more malleable. What tools do we use, what have we used, and how do we design tools to serve our educational needs? - Objectives: Why are we teaching students about computing? - Research Methods: How do we evaluate the effectiveness of computing education research or answer our questions about how students learn? - “sequence information,” which describes what a program will do under a given set of conditions, - “taxon information,” which describes what set of conditions would cause a program to reach a given point. - One of the issues emerging in the 1970s centered on the design of the tools to be used by novices. A central issue was the choice of the novice’s programming language and the programming environment in terms of their capability to make the workings of the underlying notional machine (the user-understandable semantics of the language) implied by the language both more explicit and more visible (du Boulay, O’Shea, & Monk, 1981). - Logo as a first programming language produced its own share of difficulties for the learner. These arose for three reasons. First, the focus was much more on what programs could do rather than on the form of the programming language itself. Second, the language was developed by experts without input from learners. Third, the language was designed at the end of the 1960s when the psychology of programming was in its infancy. - Where Papert had an objective of providing the computer as an object to think with (Papert, 1980), Kay saw the computer as a tool for expression and communication, as well as reflection and problem-solving (Kay, 1993). - Two related disciplines started to impact on computing education research. One was education research, bringing with it an emphasis on research using large cohorts over extended periods of time. The other was cognitive science, which emerged in the 1970s/1980s, and its research methodology based on modeling human cognition. - First was the use of large cohorts and longitudinal studies to study learning programming in classrooms rather than in laboratories. The second was the emergence of cognitive science to enable more detailed understanding of novice and expert programming skills, runnable models of human programming activity, and tools to support that activity to be built. The third was the use of phenomenographic methods to capture the individuality of the experience of students learning programming and other computing concepts. - The future of computational literacy will likely be of mixed modality. Students will use different kinds of programming languages at different stages (e.g., blocks-based languages as beginners, text-based languages if they become computing professionals, and perhaps domain-specific languages with graphical or textual forms for end-user programmers). - Today school pupils and college students are offered computational thinking courses that help them practice some of the thinking skills that programmers apply, without necessarily learning to program. - We have too few education researchers (or learning scientists, or psychologists), which means that we have too few people bringing new research methods into the community. We need that rich interdisciplinary background that our field used to enjoy in the past. - the Logo computer language, the first designed for children (Papert, 1980) – an event widely considered as the beginning of computing education - Other catalysts driving the mainstream acceptance of CEd include the launch of the Scratch, Blockly, NetLogo, and Alice programming environments; the launch of organizations such as the Computer Science Teachers Association (CSTA; an international body founded by the Association for Computing Machinery [ACM]), the rise of the maker movement and fablabs (Blikstein, 2013, 2018); the creation of organizations providing CS learning opportunities such as Code.org, Black Girls Code, Girls Who Code, and others2; and the rollout of national programs such as CS4All in the USA. - Is the rationale for CEd to fulfill job market needs, promote personal empowerment, teach children to code, develop students’ fluency in a new literacy, address historical educational inequalities, or some combination of all of the above? - for why CS should be taught in public education systems - The labor market rationale: Labor market changes and the need to sustain a competitive economy are the main driving forces for this rationale. Some consider that CS knowledge will be useful not only for professional programmers but also in a variety of twenty-first-century non-technical jobs, thus universally valuable for all professions. - The computational thinking rationale: The argument for “computational thinking” is that computer scientists’ ways of thinking, heuristics, and problem-solving strategies are universally important, and would transfer to a variety of knowledge domains and everyday problems. It would also support the development of students’ higher-order thinking skills. - The computational literacy rationale: Computational literacy is not a new skill or a class of problem-solving strategies, but a set of material, cognitive, and social elements that generate new ways of thinking and learning. It enables new types of mental operations and knowledge representations, creates new kinds of “literatures,” makes it possible for people to express themselves in new ways, changing how people accomplish cognitive tasks. - The equity of participation rationale: CS knowledge will be required for the best and most creative jobs, for civic participation, and for understanding the impact of computation on society. Additionally, since our cognitive capabilities will be limited by our ability to utilize computation, equity of participation in CEd becomes the central concern, and is one of the most significant gaps in research and implementation. - fixing the “leaky” STEM pipeline and driving more students to pursue CS careers. However, Grover and Horn point out that in grades K–8 especially, this concern with jobs might be misplaced: In elementary school, students and teachers are definitely not thinking about jobs. It is about what are the foundational knowledge and skills that children should have? At the middle school level, even though it is not a jobs argument, I think there is an identity argument there. This is especially relevant to computing because there are so many stereotypes associated with it. (Grover, 2017) - We have gone a little too far on the commercial end of the spectrum, we have become preoccupied with training the next generation of engineers, these economic motivations are outweighing the computational literacy ideas. (Horn, 2017) - The greatest demand would not be for professional programmers, but for other professionals who will have to use CS and programming for automating spreadsheets, programming queries, accessing online databases, using data-mining software tools, and operating physical computing devices in interactive art or home automation. - Computational literacy, conversely, is concerned with how computational media can change the way we know, learn, and think (in contrast with the focus on problem-solving or HOTS). - what students learn is clearly related to what they are taught, which itself depends on many elements: the instructional materials available in the market; the curriculum adopted locally; teachers’ content and pedagogical knowledge; how teachers elect to use the curriculum; the kinds of resources, time, and space that teachers have for their practice; what the community values regarding student learning; and how local, state, and national standards and assessments influence instructional practice. - one dominant narrative around CEd is that “we need to figure out the concepts, and teach them in the right way in a fixed curriculum.” - epistemological pluralism, in which they described a study where children engaged in programming in a variety of non-canonical ways that were all ultimately successful - The production of a quality curriculum and curricular materials is, for many interviewees, a key component for successful CEd implementations at scale. The interviewees noted that this is an area of significant and ongoing challenge despite efforts such as the K–12 CS Framework (K– - Creating comprehensive curriculum materials is especially challenging because there is a natural tension between uniformity and the potential for customization to the learners’ interests. Many interviewees noted the need to design culturally and personally relevant curricula that would cater to diverse populations - if curricula aim for the transfer of problem-solving skills to other domains, explicit time and effort should be put into it. - Teachers need time to practice inclusive CEd, and these pedagogies should be interwoven into the entire teacher preparation program. - Not only do teachers need to be introduced to the CS content, but they need to have time practicing pedagogies that are aimed at creating an inclusive CS learning environment, building on the assets, interests, and motivations of traditionally underrepresented students. Also, CS teacher PD must have equity and inclusion woven throughout everything that happens in PD, not just isolating this issue to a discrete one-hour discussion. For instance, as teachers are experiencing teaching lessons during PD, the other teachers who are in the roles as students or observers should be reflecting on their own experiences of inclusion (or not), thinking about their own students in their classrooms, and what works (or does not) to ignite the interest of all students. Also, teachers need time, and a safe learning environment, to reflect on all the biased belief systems associated with which students can and cannot excel in computing, to reflect on their own belief systems, and how belief systems impact their relationships with the students in their classrooms. Traditionally CS education has not been a place where these types of discussions or reflections have taken place, but they must if we are to broaden participation in computing. (Margolis, 2017) - Before you can formulate a question, you have to have an idea for a question. This is an inherently generative activity, and like anything creative, can require inspiration from unexpected and diverse places. - Research – of any kind – is not a solitary activity, but is constructed within a community. Therefore, ideas for questions can come from numerous places, people, and concepts. - One of the first things to evaluate is whether a question is interesting to you. - What if you might become interested in a question after further reading? Perhaps your interest is driven not by the content of a question, but more in the process of answering it or engaging with the community of people who might be interested in the answer. - Suppose you have found a question that’s interesting. The next step is to reflect on whether your question is theoretically sound. Soundness is fundamentally about whether the ideas in the question rationally leverage everything we already know about learning, education, and computing. - Because theoretical soundness concerns how consistent a question is in respect to prior work, judging soundness means knowing prior work: the more you read, the more you’ll be able to judge. - Another consideration is the extent to which your question is important. Importance can be judged by many different stakeholders along many dimensions. - A good research question is novel when we do not yet have a robust answer to it, although there may be parts of answers. To evaluate novelty, you need to understand the body of evidence that already exists to answer - In computing education research, nowhere is this more important than in measures of engagement and learning. Without strong theoretical grounding of what we believe engagement or learning to be, and strong measures of those phenomena grounded in those theories, we cannot make valid claims about either. - Most study designs narrowly focus on a tiny fraction of a much larger population of people, places, or things. External validity (also known as “generalizability”) is the extent to which the conclusions we draw about these tiny samples might generalize to other people, places, or things. Methods with high internal validity tend to have low external validity, and vice versa; this is because gaining certainty about causality usually requires one to look closely and in a controlled setting, with the trade-off that you cannot account for the other contexts not being observed. Similarly, the more diverse the contexts you observe, the less control you have. - Another criterion is ensuring your study design is ethical. For most academic research, there are legal protections for human participants against a wide range of harms, certainly including physical harm, but also emotional and psychological harms, including stress that can come from challenging tasks or distress that can come from deception. - A key strategy for successful study design is to find a way to pilot as many elements as possible – even if only with one person. This can greatly improve your procedures, data collection, and data quality. - As with research questions, an excellent approach to capturing, evaluating, and refining study designs is to prototype them into an articulation of the method and analysis plan as you might publish in the “Methods” section of a research paper. Here’s why: Having detailed documentation on the study plan can help you and others execute the plan consistently and reliably. If you wait to write it later, you may forget crucial details about the design that need to be captured to support reproducibility. In writing the full detail of a design, you’ll inevitably find more flaws, ambiguities, and details to determine before running the study. By detailing exactly how you plan to analyze the data, you’ll verify that you will have the data you need to answer the question, and that the types of analyses you have planned are appropriate for the data you’re collecting. - Treating study design as a design process furnishes you with a number of artifacts along the way: A literature review of the prior literature about your question; A series of claims that establish the importance, soundness, and novelty of a research question, refined through feedback and further literature review; A sketch of your method and analysis plan, refined through piloting into a paper outline. - How to measure students’ understanding of programming constructs like variables and loops. The impact of small group work embedded into large computer science lecture classes. Computing course design for students with numeracy challenges. Ease of marking for a new style of programming exam question. The impact of compiler error message content on debugging performance. Children’s stereotypes of computer science. The use of algorithmic visualizations as a teaching tool. The impact of collaborative formative examination on learning programming. How self-reported feelings about computing classes differ between the genders. How computing students use online discussion forums. How to identify the incidence and impact of plagiarism in coding assignments. The use of massive open online courses to train primary school computing teachers. - what the researcher wants to learn about – is called, in statistics, the property of interest (PI). - The goal of any research study is to measure the property of interest under particular conditions in order to learn something about it. - This measure must, in quantitative research, be some value that can be written down or typed into a spreadsheet. This measure is called the dependent variable (DV). - Some PIs, however, are not easy to measure. For example, one might be studying happiness, or creativity, or student engagement. There is no scale to put one’s subjects on that will give a direct and precise measure of how happy they are, or how creative they are, or how engaged they are in their studies. In these cases, the DV must be an indirect measure of the PI. That is, the researchers must define some observable, recordable measure that they believe accurately reflects the PI. This is called the operational definition (OD) of the PI. - descriptive statistics summarize and describe the data you have observed. They do not tell you anything directly about whether the same results would be seen in other subjects in other contexts. - Inferential statistics allow you make inferences beyond the people you have actually tested and measured. It is common to apply descriptive techniques first to get a picture of the patterns occurring in your specific sample. - “What are the different ways in which students conceive of programming and learning to program?” - Qualitative research methods provide the basis for interrogating existing and taken-for-granted conceptions of the world. - Three forms of empirical phenomena are identified about which qualitative researchers collect data: activities, accounts, and artifacts. - Qualitative methods continue to evolve in response to the changing priorities of a number of research communities, who raise new questions and provide new perspectives for answering them. - fields under the purview of the learning sciences are education, psychology, computer science, educational technology, linguistics, and data analytics. - Those from computer science and educational technology tend to define learning as mastery of a sequence of concepts – a list of rules that build upon each other to allow the learner to understand. - cognitive psychology and computer science. As these fields grew together, they forged a connection between how humans learn and how machines learn. - The balance in the learning sciences between the scientific study of learning and the design of environments to support learning is similar to the balance that computing education has embraced since the late 1960s - Despite overlapping values and some overlapping researchers between computing education and the learning sciences, the two fields have not been as integrated as would be beneficial - Constructivism addresses how learners cognitively build knowledge; cognitive apprenticeship addresses how instruction scaffolds learners’ emerging skills and knowledge; sociocultural theory addresses the social and environmental aspects of learning environments; and expectancy-value theory addresses the role of motivation in learning. - Constructivism is a commonly used theory, both inside and outside of computing education (see Chapters 11 and 15), about how people cognitively acquire knowledge. - constructivism states that people learn best when they construct knowledge for themselves rather than being told explicitly what to know and how to learn it (Tobias & Duffy, 2009). - definition. “Construct knowledge for themselves” means that learners are making sense of new information through reasoning and invoking their prior knowledge rather than being told how to interpret and organize new information, as is common in more direct instruction (i.e., instruction in which the instructor explicitly tells the students everything that they need to know). - students are learning concepts and skills through exploration that is guided by an instructor but not prescribed by an instructor, as it would be in more direct instruction. - The instructor can still have learning objectives, but there are multiple paths to achieve them. - “learn best” has several different meanings. It means that constructivist approaches help learners perform better on tasks and tests by increasing their depth of thought and connections to prior knowledge, resulting in better retention and transfer of knowledge (Bruner, 1973; see more about transfer in Chapter 9). It also means improving motivation and emotion by increasing student agency in learning and helping them connect knowledge to their lives (Searle & Kafai, 2015). - direct instructionists view learning as a change in the brain caused by the storage of new information, and they therefore argue that direct instruction is the most efficient and easiest method for learning. - constructivists view learning as a change in knowledge that is not worth much without a concomitant change in professional skills (e.g., solving authentic problems) and soft skills (e.g., working collaboratively). - instruction can be more direct or more constructive depending on the needs of the learner and what is most appropriate. - A core question in the research on constructivism – and learning sciences more generally – is how much guidance is optimal to support learning. - the vast majority of learning throughout history was structured as a relationship between a master, who is an expert in the domain knowledge and skills to be learned, and an apprentice, who aspires to become a master. - Cognitive apprenticeship was proposed as a means to integrate the successful practices of traditional apprenticeships with the more general knowledge and cognitive skills sought by traditional school settings (Collins, Brown, & Newman, 1989). This approach to orchestrating a learning environment holistically considers four unique components: content, method, sequence, and sociology - heuristic strategies used by experts within the domain to solve problems, the metacognitive control strategies used to monitor one’s progress while completing a task, and the more general strategies to learn new things. - A hallmark of a cognitive apprenticeship is employing a variety of pedagogical methods to synergistically achieve this goal. - Expert modeling is used by a teacher to demonstrate a particular task while voicing one’s inner thought process for direct observation by the learners. - Each component, like instructor modeling and coaching techniques, learner self-explanation/reflection, and social influences on learning, carries with it a rich body of knowledge and a set of ongoing open research questions to be explored empirically. - Productive failure is a learning design that formalizes the process of learning from one’s mistakes. - Productive failure has four central mechanisms: activating learners’ prior knowledge and experience; drawing attention to critical features of the concept; elaborating on critical features; and integrating critical features into a unified understanding of the targeted concept (http://manukapur.com/productive-failure/). - These four interrelated mechanisms are embedded in two phases: a generation phase and a consolidation phase - Learners first work in small groups to generate and explore multiple representations and solution methods (RSMs) to an ill-structured problem that is beyond their current problem-solving abilities. - In the second phase, the RSMs generated by the learners are compared and contrasted with canonical RSMs and learners consolidate knowledge, integrating the solutions they generated with the targeted concepts. - sociocultural theories of learning take a situative perspective on learning. They emphasize learning as an activity that is embedded within social, cultural, and historical context and occurs in interaction with others and with available tools and resources. - learning occurs as an individual moves from being on the edges of a community (legitimate peripheral participation) to more full participation in a community of practice. - As learners move from legitimate peripheral participation to fuller participation in the community, they begin to act like a member of that community, understanding what constitutes community membership, what members of the community do, and how members of the community talk and interact with others, both inside and outside of the community. In other words, learners begin to identify with that community. - learning is identity construction. - there is a body of work that examines scaffolding computing education for novices by providing a context for doing computing - Another significant strand of computing education research focuses on addressing the “identity gaps” that exist for women and non-dominant individuals entering into computer science (Tan et al., 2013) and finding ways to mitigate those through more approachable introductions to programming, such as storytelling - “tools are created and transformed during the development of the activity itself and carry with them a particular culture – the historical evidence of their development” (Kaptelinin & Nardi, 2006). - tools represent an accumulation of social knowledge and its transmission. - Engeström’s (1987) cultural-historical activity theory (CHAT) further elaborated Vygotsky’s model of an activity system to include subject, object, instruments, rules, community, and division of labor. - effective change, according to Engeström, requires an expanded understanding of the object that takes into account both temporal (taking a long view) and socio-spatial elements - research in learning sciences includes motivational factors, such as learner experience, attitude, values, dispositions, mindsets, and identity. - Expectancy-value theory is a motivation theory to explain choice, persistence, and performance. - Expectancy-value theory is primarily applied in K–12 education, but no research suggests that age or developmental stage impacts the predictive value of the theory - When expectancy and value are both high, the learner is highly motivated to successfully complete the task. When expectancy and value are both low, the learner is unmotivated to complete the task. When expectancy and value do not match, they can interact with each other in interesting ways to affect motivation. - researchers pay attention to different definitions of learning and measure both the process of learning (i.e., the experience during and the steps of learning) and the product of learning (i.e., the change in knowledge or experience caused by learning). - Research methods and measurements in the learning sciences also make sure to capture the learning environment and its effects. The environment includes features of the immediate surroundings, such as teachers, peers, technology, room, and time, in addition to the more abstract context, such as culture, identity, and family and friend relationships. - learning scientists engage in studies that aim to inform the intervention, regardless of whether that intervention is a piece of educational software, a set of classroom activities, or an entire informal learning environment. - learning scientists regularly engage in the iterative refinement of an intervention in order to “get it right” in the desired context. - Learning sciences and computing education research have a lot in common, yet the two communities can still learn from each other and benefit from a closer relationship. - Computing education researchers should draw from the rich literature, general learning theories, and methodologies that the learning sciences offers to guide our research, increasing the rigor within our community and contributing to education outside of computing. - Our memories are made up of overlapping patterns that are approximately recreated in the process of recall, and our reasoning is determined by heuristics, expectations, and mental models of real or imagined worlds. - Perception is the process by which we as cognitive agents receive, register, and process stimuli from the external world (and from our own bodies). - In educational terms, the concepts of attention, working memory, and cognitive load are closely related, - Sensory memory (SM) refers to the automatic and brief retention of sensory phenomena after the stimulus itself is gone. It is usually studied in the visual modality (iconic memory), but also includes hearing (echoic memory) and touch (haptic memory). - Short-term memory (STM) refers to the capacity to actively hold a small amount of information in conscious focus for a short period of time - The term working memory has been, and often still is, used as a synonym for STM. However, most researchers accept the distinction proposed by Baddeley and Hitch (1974) that STM involves the simple retention of information, while working memory involves additional systems for manipulating this information so as to perform such activities as comprehension, problem-solving, reasoning, and learning. - Long-term memory (LTM) is the encoding, storage, and retrieval of large amounts of information for significant periods of time – up to years or a lifetime. - The encoding of a memory is the process by which the brain alters its structure so as to create a lasting representation of some currently perceived or attended to stimulus or item of information. - encoding can be improved by “elaboration”: interacting with the stimulus at the time of encoding, processing it in meaningful ways, and in particular relating it to already existing long-term memories. - Note: Note-taking can be interacting with the stimulus. By taking notes on what I consume, I retain the information further. - Consolidation is a mechanism (variously considered either separate from or part of either encoding or storage) by which initial encoding of a memory is strengthened. - Note: Sleep and one’s health are vital to retaining information. - The synaptic encoding of memories can degrade over time or change as a result of new experiences or new memory formation. - Evidence suggests that more information is stored in the brain than can be successfully recalled, and that the phenomenon of forgetting is largely a failure of recall - Episodic memories are necessarily subjective (two people may remember the same event differently) and can change over time as described above. - prospective memory is “remembering to remember”/planning - The fundamental unit of knowledge in the theory is the schema, a cohesive, repeatable “action sequence” organized around a “core meaning” - Learning is seen as involving assimilation, the integration of new information into schemata, and accommodation, the process of adjusting schemata - Methodological behaviorism (Watson, 1913, 1930) holds that the only scientifically valid object of study is observable behavior; any consideration of subjective, phenomenological, or mental states or processes is excluded. Behavior in this framework is seen to arise as a result of environmental stimuli interacting with an individual’s history of reinforcement or punishment and/or current drives (e.g., hunger). - Radical behaviorism (Skinner, 1938, 1953) acknowledges the existence and valid study of “private events” (e.g., thoughts, feelings, mental states), but holds that environmental factors control these just as they control observable behaviors. - Behaviorism was rooted in earlier work on animal learning, which it developed and applied to animals and humans in two main theoretical and practical directions. - Classical conditioning involves the association of innate reflexes with new stimuli (the famous example being Pavlov’s dog - Operant conditioning is the modification of behavior using reward (stimuli that increase its frequency) or punishment (stimuli that decrease its frequency). Reward or punishment may be positive (added to the environment) or negative (removed from the environment). - of the two major learning frameworks to succeed behaviorism, cognitivism, as discussed above, grew out of psychology and artificial intelligence. Constructivism (as widely discussed elsewhere in the Handbook; see Chapters 1, 8, 10, 11, 15, 24, and 29) grew out of education and child development. - we define transfer as a student’s ability to transfer the knowledge that they’ve learned and apply it to solving new problems. - In spontaneous transfer, the student recognizes the similarities between what they already know and the knowledge required to solve the new problem without help or guidance - Knowledge domain transfer is transfer within or between different disciplines. - Near transfer – transfer within a discipline: - Isomorphic transfer – the problem-solving context (i.e., cover story) between problems is the same and only the values in the problem are different. - Contextual transfer – the problem-solving procedure between problems is the same and the contexts of the problem are different. - Procedural transfer – the type of problem is the same and the problem-solving procedures are different. - Far transfer – transfer between disciplines: - Several shared elements – the domains share several elements that make general problem-solving procedures applicable to both. - Few shared elements – the domains do not share many elements, making transfer possible at only an analogical level. - Physical context transfer is transfer among spaces - Temporal context transfer is transfer across time (e.g., weeks, months, semesters). Because synaptic connections fade over time (i.e., ability to recall information decays), the time between solving problems can affect transfer. Time transfer is also commonly called retention. - Functional context transfer is transfer among purposes (e.g., authentic web development problems, academic exercises to learn concepts used in web development). - Social context transfer is transfer among group compositions (e.g., solving problems individually, solving problems in teams). - Modality transfer is transfer between medium of instruction and application at the macro-level (e.g., auditory instructions vs. written assignment) or micro-level (e.g., instructing based on tracing, reading, or writing code vs. testing based on tracing, reading, or writing code). - The best way to help students transfer their knowledge is to teach them how to transfer their knowledge. Cognitively speaking, learning content is not the same as learning how to transfer content - instructors should tell students how to recognize similarities between novel problems and problems that they know how to solve, and they should give students practice and feedback on novel problem-solving - Helping students to identify structural similarities between problems is important in computing education because we typically rely on worked examples to help teach computing concepts. - Subgoals are functional pieces that are inherent in problem-solving processes - The worked example is visually broken down into subgoal components, then each subgoal is meaningfully labeled - Recognizing structural similarities between problems is likely the biggest barrier to transfer, but transfer also suffers when neural connections to relevant prior knowledge decays over time, causing a lack of retention, or time transfer. - In some cases, immediate feedback is the most effective type of feedback. - In more conceptually focused problem-solving, however, slightly delaying feedback can allow learners to recognize when they have made a mistake and fix it for themselves, improving retention - Learning occurs when the information from sensory input is attended to, processed within working memory, and then committed to LTM in the form of schemata or plans. - Cognitive load is the amount of work imposed on a learner’s working memory while learning a specific task - Cognitive Load Theory (CLT) is based on the architecture of the human brain, with the brain having both working memory and LTM. - Learning occurs when new information is related to knowledge already stored and organized in LTM, yielding a new, more elaborate, and extensive knowledge base - Using CLT, instructors attempt to design instructional material in a manner that will not overload working memory, so as to maximize learning. - CLT was the sum of three individual components: (1) extraneous load, the effort used to process information that is not fundamental to the concept or process that is being learned; (2) intrinsic load, the innate difficulty of the material being learned combined with the learner’s characteristics (Leppink et al., 2013); and (3) germane load, the effort used to learn the actual material or concept. - Extraneous cognitive load is any information presented to the learner that is not strictly necessary for learning the material or concept. - A topic is considered to have a high intrinsic load if the material being learned is interconnected; that is, learning requires processing several elements simultaneously to understand their relations and interactions - intrinsic load is essential for comprehending the material and constructing knowledge structures, the instructional material must provide all of the components necessary to accommodate this load without exceeding the limits of working memory. - The magnitude of the intrinsic cognitive load experienced by a learner is determined by the degree of interactivity of the essential elements relative to the level of learner expertise within the domain (Kalyuga, 2011). This begins to explain the difference between novice and expert programmers. - Worked example (problem completion) effect – students produce fewer errors and take less time to complete problems when they study expert solutions, or worked-out solutions, than when they solve the same problems from scratch with no sample solution to follow - Split-attention effect – information should be presented in a way that is as integrated as possible, both spatially and temporally - Modality effect – this occurs when multiple sources of information that cannot be understood in isolation are needed to be processed simultaneously, usually presented using spoken rather than written text - Redundancy effect – this is when unnecessary, additional information is presented to the learner - Expertise reversal effect – occurs when learning material gets in the way of learning for those with prior knowledge - Guidance fading effect – states that novices should be provided with many worked examples, followed by completion problems and then full problems - Self-explanation effect – requiring the learner to self-explain new procedures or concepts to enhance learning - Element interactivity effect – if intrinsic cognitive load is low, other cognitive load effects cannot be observed; however, if elements are highly interactive, variations in element interactivity can have profound effects on other load effects dependent on extraneous load - Boyer identified four broad categorizations of scholarship (discovery, essentially research; integration, emphasizing multidisciplinary approaches; application, generally referred to nowadays as engagement; and teaching) and conceptualized these as interacting with one another as a university professor goes about his or her everyday work. - Boyer suggested six standards: clear goals, adequate preparation/appropriate procedures, appropriate methods, significant results, effective presentation, and reflective critique. - four approaches that should in general terms be used to evaluate the quality of the scholarship involved: self-review, peer review, review of clients, and review by students. - Not all higher education disciplines have well-developed, pedagogically focused research networks. - most readers of this chapter will have some personal experience of being taught by one or more teachers who not only had little educational theory to draw on, but who also persisted in simply telling students what they needed to know to pass the exams. Most readers of this chapter will wish to be better than those teachers were. - The essence of constructivism is that learners develop their own mental models of the world around them and new information is subjectively interpreted by them, based on their prior personal experience. - Many of the modern paradigms of learning and teaching in higher education have developed from constructivist ways of thinking. These include problem-based learning, enquiry-based learning, and the broad area of social learning. - Social constructivism suggests that learning isn’t an individual process, but involves social networks, with “culture” providing mediating influences. - behaviorism is firmly rooted within positivism and objectivism and is distinctly unfashionable in many teaching and learning circles. - More enlightened advocates of behaviorist ways of doing things tend to emphasize that, in at least some circumstances, teachers do quite like to identify what it is that the student is supposed to be learning and use this “intended learning outcome” as a means to consider the most appropriate forms of teaching activity that may enable the learner to achieve these outcomes. Similarly, if learners are told what it is they are expected to learn, they may in some circumstances be more self-motivated in achieving these learning outcomes. - Behaviorism also does contribute quite substantially to traditional ways of identifying the roles and operation of feedback to students on the basis of what they’ve learned so far and what they may need in order to do better in future. - Leading theorists include Skinner, and some would identify Bloom and Krathwohl’s work on cognitive (knowledge and skills) and affective (values and attitudes) learning domains to be derived from behaviorist ways of thinking about learning (e.g., see Bloom, Hastings, & Madaus, 1971, as a useful introduction to this way of thinking). - third category, cognitivism, somewhat pragmatically attempts to straddle the gulf between these epistemological and ontological extremes. - Cognitivist interpretations of learning tend to focus on the cognitive processes that contribute to the processing of information from experience through to memory. - Constructive alignment developed from a substantial twentieth-century educational movement that nowadays is identified as outcomes-based education (OBE) or outcomes-based teaching and learning. In essence, the movement suggests that educators and education work best when the educator concerned considers what, precisely, learners will be able to do after they’ve been taught that they couldn’t do before they were taught. - Central to OBE is the intended learning outcome (ILO), and it is probably fair to say that the adoption of ILOs has contributed more angst and frustration to higher education teachers than has any other development in recent times. - constructive alignment – constructive because the ideas focus on what the students do and learn, rather than on what the teacher does; and alignment because there needs to be some sort of alignment between the intended learning (or ILO), the teaching and learning activity (or TLA), and the assessment that confirms that the students have learned. For those who appreciate it, constructive alignment has provided a great way to think about their teaching. - Phenomenography opened the way to a great deal of research that identified that students approach learning tasks in different ways and that the ways that teachers teach have some influence on the ways that students learn. - Teachers who put the students at the center of learning rather than themselves tend to think about what students are expected to learn in the course, rather than what teachers hope to teach. - At one level, they may be simply trying to memorize what is taught; a relatively lowly cognitive process. At a higher cognitive level, they may be involved in theorizing, reflecting, and abstracting about what has been taught. - Problem-based learning, for example, requires activity by learners, whereas a standard instructional lecture may not. - ask themselves if their teaching approaches will encourage learners to take responsibility for their own learning, and therefore be most likely to extend beyond the constraints of the course, or degree, both temporally and spatially, as well as to ask what forms of learning will be most valued by their students. - a major predictor of both good teaching and good learning is good feedback. - Hattie and Timperley identified four levels of feedback: task level, essentially how well tasks have been completed; process level, the extent to which learners have understood the tasks to be completed; self-regulation level, the extent to which learners have conceptualized their own learning; and self level, feedback directed at the learners’ appreciation of themselves or their generalized self-efficacy. - When it comes to assessment, those who appreciate the benefits of constructive alignment, outcome-based assessment, and the ILO do have a substantial advantage over other educators who do not. Advocates for constructive alignment and its logical consequential processes will have on their mind the nature of the intended learning outcome involved when they design their assessment. They will, for example, find it easier to describe learning outcomes in the cognitive domain than in the affective domain, and similarly, find it far easier to describe lower-order cognitive learning (such as remembering and understanding) than higher-order cognitive learning (such as application, synthesis, and evaluation). - Formal, traditional written examinations are likely unsurpassed as tools to assess what students have remembered or can explain. Practical examinations involving computers are likely good tools to assess what students can actually do with the computer. Neither are necessarily ideal tools to allow students to demonstrate their higher-order cognitive skills such as creativity and the ability to evaluate the worth of their knowledge. Nor are they necessarily tools to enable students to demonstrate a range of outcomes involving both cognitive and affective learning, such as teamwork or behaving ethically. - the majority of assessment in higher education ought to be formative in nature rather than summative. Formative assessment enables teachers to provide feedback to students in a way that does not jeopardize their final grade for any particular course. This feedback enables students to reassess their mental models of whatever is being taught and to actually benefit from the interaction that they are having with their teacher. - the evaluation of my teaching and of the consequences of my teaching is a part of being a professional and is closely allied to my commitment to scholarship. - Self-evaluation: - Reflective practitioners think deeply about what they’re doing and what they’ve done. This deep thinking provides higher education teachers with a great deal of insight into how they’re teaching and how effectively this is converted into learning. - Peer evaluation: Teaching in higher education should ideally not be a lonely pursuit. Most of us work in teams to support our students, and most of us encourage teamwork in our teaching. Peer review, or peer evaluation, is not always an easy option, but if done well, it can greatly contribute to an effective evaluation of teaching. - They can comment on our course designs or on the feedback that we give to our students. - Outcome measures: - No matter how well-intentioned, if the same university teacher designs the course, teaches the course, and assesses the course, they may not be able to maintain the level of objectivity necessary for a fair evaluation. On the other hand, in some higher education situations nowadays, higher education teachers are under a great deal of pressure to maintain class numbers and to boost retention figures. - Feedback from students: - As well as contributing to the teachers’ reflection on teaching, they also contribute in a quite direct way to promotion applications. - “What have students learned (to know, to do, or to be) that they didn’t know, couldn’t do, or wouldn’t be before?” and to expect some relatively jargon-free answers. - Good instructional design can minimize extraneous cognitive load and increase germane cognitive load - Researchers have found that experts organize knowledge differently from novices. This organization enables experts to recall information, recognize patterns, and solve problems more efficiently than novices. - adaptive expertise (McKenna, 2014). This theory posits two dimensions of expertise: efficiency and innovation. Efficiency refers to the ability to apply domain-specific knowledge with speed and accuracy. Innovation refers to the ability to identify problems and to generate ideas to achieve ambitious goals. In this theory, novices exhibit low efficiency and low innovation, routine experts exhibit high efficiency and low innovation, and adaptive experts exhibit high efficiency and high innovation. - Defined by Bandura (1977), self-efficacy is an individual’s belief in their ability to complete a task in order to achieve a desired goal. - Bandura (1977) identified four sources of self-efficacy: mastery experience, vicarious experience, verbal persuasion, and physiological states. - Students have a mastery experience when they perceive success or failure in their efforts to achieve a learning objective. - Students have a vicarious experience when they see other students succeed or fail. - Verbal persuasion consists of feedback from peers and teachers such as praise or criticism. - Physiological states are the emotions that students feel: emotions such as joy and anxiety can affect students’ self-efficacy beliefs. - Competence beliefs, which resemble self-efficacy, comprise the individual’s expectations of success in completing the task. - Value beliefs comprise the individual’s beliefs about how well the task matches their desires and goals. - Attainment value is the perception of the importance of the task for the individual’s self-concept. - Intrinsic or interest value is the individual’s interest in the task itself. - Utility value is the perception of how the task might help the individual achieve some goal other than the task. - Cost value is the perception of the time, effort, and psychological impact required by the task. - Students with the incremental theory (now called the growth mindset) believe that intelligence can improve with effort. - Students with the entity theory (now called the fixed mindset) believe that intelligence cannot be changed: they believe in innate talents – you have it or you don’t. - engineering students who held entity beliefs were less likely to engage in knowledge-building behaviors. - systematic literature reviews, which synthesize what is known in a particular research area using primary studies. - systematic reviews adopt detailed procedures for collecting and analyzing primary studies. - systematic reviews may use quantitative, qualitative, or mixed methods to synthesize quantitative, qualitative, and mixed-methods studies - flipped classroom approach, which provides lecture material or other resources as assignments to free up class time for active engagement in learning. - the motivational context for exploring the teaching and learning of programming includes academic, pedagogical, and practical factors. - it takes around ten years for a novice to become an expert programmer. - Stoppers are those who stop and appear to “abandon all hope” when confronted with a problem or a lack of a clear direction to proceed. They are likely to be those who are frustrated by or have a negative emotional reaction to errors. - Movers are those who keep trying, experimenting with and modifying their code. They can use feedback about errors effectively to solve problems and progress. - Tinkerers are extreme movers who are not able to trace their code and may be making changes more or less at random, with little chance of progress. - Learning to program involves acquiring both declarative knowledge (e.g., being able to state how a “for” loop works) and practical strategies for its application (e.g., using a “for” loop appropriately in a program) (Davies, 1993). - Fundamental ideas are those that have “wide as well as powerful applicability” and apply “at any stage of development” - horizontal (the idea is relevant across many disciplines or sub-disciplines), vertical (the idea pervades all levels from elementary through to highly advanced), time (the idea is recognized as important and it endures), and sense (the idea has meaning in “everyday life”). Schwill argues that candidate fundamental ideas in computing include algorithmization, structured dissection, and language, with more specific ideas definable under each category (e.g., under language are the ideas of syntax and semantics). - Threshold concepts are those that are key challenges in learning a given knowledge domain; if successfully acquired, they enable a qualitatively different understanding. - transformative (creating a new way of viewing, understanding, or describing), integrative (allowing new connections and relationships to be perceived), irreversible (causing a fundamental change that cannot be “unlearned”), troublesome (being problematic to grasp or difficult to integrate into current understanding), and boundary markers (helping define the scope of the knowledge domain). - Mental models are generally held to be internal models of how some aspect of the world works, an iconic representation of selected aspects of external objects and systems. Mental models have predictive power – they can be used to understand the observed behavior of the world and reason about future behavior. - Cognitive load theory is a broad framework for describing the load placed on (or the “effort” expended by) working memory during the execution of a task - intrinsic (the difficulty or required effort inherent in a specific task or topic), extraneous (effort arising from and varying with the way that information is presented), and germane (the effort required to integrate new information into permanent schemata). (Subsequent - combine cognitive load theory, fundamental ideas, threshold concepts and standard curriculum designs to propose the ideas of “anchor concepts” and “anchor graphs” as tools for curriculum planning in the CS1/CS2 sequence. - “scenario-based programming” using visual “live sequence charts” encourages abstract thinking and ordering tasks by level of increasing complexity and is an effective way of reducing cognitive load. - (Bloom et al., 1956) described six levels of increasingly sophisticated objectives for learning within the “cognitive domain,” namely: Remembering, Comprehending, Applying, Analyzing, Synthesizing, and Evaluating. - the Knowledge Dimension, defined over the categories of Factual, Conceptual, Procedural, and Metacognitive knowledge (the latter two are effectively “strategies” in the language of this chapter), and the Cognitive Process Dimension, defined over the levels of Remember, Understand, Apply, Analyze, Evaluate, and Create. - Piaget defined the following four stages of children’s development: sensorimotor (from birth to the acquisition of language at around 2 years old), pre-operational (to 7 years), concrete operational (to 11 years), and formal operational (to adulthood at between 15 and 20 years). The later stages in particular are defined largely in terms of the acquisition of logical capacities such as transitive inference. - The Bloom and SOLO taxonomies are complementary, - demographic factors – following on from early aptitude tests, a range of subsequent studies have explored the significance of factors such as age, gender, ethnicity, marital status, grade point average, mathematics background, science background, ACT/SAT math scores, ACT composite score, SAT verbal scores, high school rank, previous computing experience, and more. - Cognitive capacity – aptitude tests and related research have used a range of tasks including letter series, figure analogies, number series, verbal meaning, and tests of accuracy, mathematical reasoning, algorithmic execution, alphanumeric translation, deductive and logical ability, the ability to reason with symbols, the detection of patterns, and reading comprehension (many of these are components of common IQ tests as discussed in the next section). - Cognitive style – Is there a particular learning style or personality type that contributes to success? Tests that have been used to explore this possibility include the Myers–Briggs Type Indicator and the Kolb Learning Style Inventory. - Attitude and motivation – Similarly, is attitude/motivation critical? Studies have explored the Biggs revised two-factor Study Process Questionnaire (R-SPQ-2F), students’ self-reports, measures of self-efficacy, and factors such as perfectionism, self-esteem, coping tactics, affective states, and optimism. General conclusions are discussed in the next section. - Affective factors such as motivation, constructive attitudes to learning, positive expectations, and high self-efficacy or effort are also usually found to be correlated with success in programming - The basics of effective teaching and learning are the same in most subjects. For teachers, they include the provision of clear and relevant course materials, clear learning objectives, assessment that is well aligned with objectives, rich and timely feedback to students, fostering student engagement, pastoral care, competent classroom skills, and more. - For students, learning outcomes will be influenced by motivation and attitude, forms of engagement, IQ, attitudes to learning, time management skills, personal circumstances, sociocultural factors, and more. - Joint Task Force on Computing Curricula, - Paradigms are supposedly groups that differentiate one class of similar languages from others in some high-level way, usually focused on features that exhibit common behaviors. - Student misconceptions about programming concepts can persist into the upper-level curriculum: one recent study found juniors and seniors holding substantial misconceptions about aliasing in Java, despite taking several programming courses (and having other experiences) that used the language (Fisler, Krishnamurthi, & Tunnell Wilson, 2017). As students move beyond the first year or into jobs, they often - Today’s programming languages embrace features from across the conventional paradigms, and the problems we observe in education tie more to features and particular kinds of notional machines than to the paradigms themselves. - As a subject that is often considered vocational in nature, computing students also need to demonstrate standard skills of employability, which lacks an overall infrastructure - Over the last couple of decades, there has been a shift in focus in our education systems from teacher-centered to student-centered education and increasing use of technology. - Pedagogy is the science of how we promote learning and it consists of the learning activities, strategies, and techniques that provide the environment where learning may take place. - A pedagogy may define a specific approach (e.g., blended learning, a specific approach blending modes of learning) or a broad range of approaches (e.g., active learning, describing a number of approaches built around active student participation). - pedagogic practice when describing a specific activity within a course (e.g., peer instruction) - Behaviorists equate learning with observable behavioral outcomes and do not consider the role of mental operations in the learning process. In contrast, cognitive psychologists focus on the mind as the agent of learning. They are concerned with the internal mental constructions and activities of the learner in preference to their external behavior. - memorization, drill and practice, deduction, and induction - Behaviorism and cognitive psychology are often presented as opposing philosophies in their explanations of learning. However, there are similarities in that both seek to effect learning through the design of specific tasks. The main differences are the foci on either external or internal activities of the learner and the epistemologies on which they are based - In contrast, cognitive psychology focuses on the cognitive activity of the learner. Cognitive psychology encompasses a broad range of views on the structure of knowledge and the processes involved in learning. - From a constructivist perspective, students learn by relating new concepts to existing ones, either by combining new knowledge with existing knowledge to create new cognitive structures or through reflection upon existing knowledge. - Learning is an active process of constructing rather than acquiring knowledge. Instruction is a process of supporting that construction rather than communicating knowledge. - sociocultural constructivism, where knowledge is created through social and cultural activation - Modern pedagogic practice has tended toward constructivist or student-centered approaches over instructivist approaches. Approaches that support constructivist learning are frequently known as active learning approaches. Active learning pedagogic practices designed to support cognitive constructivism range from small-scale active learning practices that may be integrated into broader learning opportunities, such as minute papers and mind mapping (Phillips & Soltis, 2015), through to problem-based or enquiry-based learning approaches, where learning is driven by student motivation via the problem-solving process. - Collaborative learning involves students working together to informally facilitate their learning, embodying the sociocultural constructivist perspective. - Collaboration in the computing classroom can take the form of the adoption of general collaborative practices such as incorporating activities like think–pair–share or scaffolded discussions in classroom settings (Hativa, 2000), technology-assisted collaborative note-taking (Jonas, 2013; Reilly et al., 2014), and computing-specific collaborative activities, including peer instruction (Simon & Cutts, 2012), studio-based learning (Hundhausen, Narayanan, & Crosby, 2008), and peer reviews (Clarke et al., 2014; Hundhausen, Agrawal, & Agarwal, 2013; Politz, Krishnamurthi, & Fisler, 2014). - Cooperative learning (Slavin, 1991) is a sociocultural constructivist approach where students work in small groups or teams to help each other learn and are also responsible for each other’s learning (in contrast to collaborative learning). - process-oriented guided-inquiry learning (POGIL) (Moog & Spencer, 2008). In this approach, the instructor takes on the role of leader and assists learning by asking questions that prompt deeper thinking or reflection. - Pair programming (Williams & Kessler, 2002) involves students working in pairs, with one member driving the cooperation by leading the typing or writing, while the other – the navigator – observes the driver, asking questions and making suggestions. - Key features of this refinement are (Hamer et al., 2011) as follows: The instructor is explicitly using a contributing student pedagogic design. The pedagogy includes the requirement that students produce artifacts. The students are aware that the audience for these artifacts includes other students. These artifacts are shared with other students for the purposes of their learning. - Students first produce their own solution, then move to evaluating and testing other students’ work, including the production of a testing methodology report, before the final phase of each student reviewing and reflecting on their own experience and the artifacts produced by their peers in the production of a peer testing report evaluating the process as a whole. - Further examples of content creation include the collaborative preparation of learning resources, such as glossaries and knowledge bases for discipline topics, frequently supported by online tools such as wikis and blogs (Elgort, Smith, & Toland, 2008; Lutteroth & Luxton-Reilly, 2008), and the development of algorithm visualizations (Crescenzi & Nocentini, 2007). - Blended learning, implemented via the flipped classroom pedagogic practice (Bishop & Verleger, 2013), takes advantage of the availability of online or established resources, such as MOOCs, to flip the learning model toward an active pedagogic approach. - A flipped classroom pedagogy can present challenges to instructors and their students. Giannakos et al. (2014) found the main difficulties faced by instructors were that they spend more time in preparation, the students do not always respond well to the flipped classroom model, and there can be a decrease in class attendance. - A flipped classroom typically requires students to prepare before class and be actively engaged in class, which some can find challenging. - it has been argued that the difference between online learning and MOOC environments are the combination of teaching approaches course instructors use, the massive levels of participation, and their openness - MOOCs provide the potential for the development of new pedagogic approaches within the online learning space, while also facilitating leveraging existing pedagogy for adaptation to the online learning space, particularly when adopted as part of a blended learning approach. - Another game-related pedagogic approach is gamification, where game design elements such as leaderboards, badges, and constant feedback are incorporated into instruction. - Computing programs are characterized by low diversity within the student population. Concerns about equity and diversity issues such as low numbers of female students have inspired pedagogic approaches that are accessible and inclusive. - describe the implementation of this pedagogic approach in a Computational Thinking for Girls (CT4G) project and propose that showing students the social and diverse aspects of computing can be a powerful tool to engage underrepresented groups. - Learning through social media addresses increased usage of social media platforms (both specific to learning and in general) on promoting different ways of engaging with learners and different conceptualizations of the role of the teacher. - Productive failure encourages students to tackle challenging problems with potential for failure prior to receiving direct instruction. - Teachback, where learners explain back to the teacher what they have learned and what they think they know, helps the learner to reflect on and conceptualize gaps in their learning. - Design thinking is a structured approach to problem-solving that supports episodic phases of creativity and critical thinking, identifying possibilities and constraints before moving on to analysis and construction – the “solving” process itself. - Learning from the crowd uses crowdsourced information and platforms as the basis for learning, establishing pedagogic practices that use and engage with crowdsourcing platforms and approaches to solve problems and gather information. - Learning through video games builds on an extensive history of games within learning – video game pedagogies that exploit game characteristics that trigger and sustain learning. - Formative analytics expands upon the use of learning analytics to report on learning (the more traditional view of learning analytics) toward the development of analytics directly informing and assisting the learner - Learning for the future encompasses a range of pedagogic approaches that support autonomy, resourcefulness, critical thinking skills, and interpersonal skills, building on existing pedagogic approaches and practices, including studio-based learning, project-based learning, and personalized learning assisted by learning analytics. - Translanguaging recognizes the global nature of future learning environments and the subsequent need for pedagogic practices and environments that support multi-language development and encourage diversity in language and cultural context - blockchain for learning, defines new ways of storing, validating, and trading educational qualifications and evidence for learning. - Creese, A., & Blackledge, A. (2010). Translanguaging in the bilingual classroom: A pedagogy for learning and teaching? The Modern Language Journal, 94(1), 103–115. - We argue that understanding the ways in which systems of inequity produce dominant and marginalized groups is essential to understanding and addressing current patterns of underrepresentation and inequity. - “Equity” has been an umbrella term used throughout much of the CEd research (CEdR) literature, which has been conceptualized at multiple grain sizes corresponding to multiple organizing concepts. - Equality refers to the state where everyone has or is allocated the same things in the same degree, whereas equity typically refers to having access to what is needed. - there is a moral concern that the computing community reflect society writ large, particularly with respect to race, gender, dis/ability, sexuality, and other social markers. - some link diversity to the size of the workforce, suggesting that greater inclusion is needed to fill gaps in the computing labor force (National Center for Women & Information Technology, 2017). - there is a related pragmatic and economic argument that diversity in product design teams will lead to better products. One explanation for this benefit is that diverse teams have been shown to produce better results (Hunt et al., 2018). - the team may choose to make products that are accessible to a greater portion of consumers in the marketplace (K–12 Computer Science Framework, 2016, p. 29). - Within the umbrella of “equity,” another organizing concept is “access,” which acknowledges that representation and performance in computing are closely related to students’ access to the resources needed for participation in computing. - “broadening participation goes beyond issues of access to computer science (CS) learning; we also must transform CS classroom culture and teaching” (p. 71). - “social justice,” in which students use computing for the purpose of addressing injustice in their local communities and in society writ large - Rather than simply enrolling more students of color in computing majors or hiring more women software engineers, computing is conceptualized as a “discipline in service of society, its people, and their needs” - these narratives can cause harm through explicit, conscious bias referred to as “prejudice” - we focus on the less overt forms of bias that can cause disadvantage and marginalization and are fueled by these narratives – “implicit bias” or “unconscious bias” - Unconscious bias can influence split-second decisions, is fueled by narratives, and causes harm to those who are negatively stereotyped. The evidence of these biases primarily comes from what are referred to as “résumé studies” - we are interested in considering the aggregate impact of unconscious bias and the ways in which it is embedded within policies and practices within institutions - Educators may mitigate this by grading student work anonymously and using rubrics when possible. Research shows that being more aware of the biases that we have can help us reduce their impact on our decision-making (Morewedge et al., 2015). Even in our local decisions about grading we can contribute to addressing these much larger problems. - research has found that it can be helpful to communicate to students that tests are an opportunity to provide feedback about their learning, but are not a way to evaluate their potential - Active learning is sometimes defined as requiring students to do more than listen to a lecture - Research shows that Black students actually have higher rates of interest in CS fields relative to their peers of other racial backgrounds - A simple but powerful action that computing instructors can take is to systematically track the participation opportunities they are making available to students from marginalized groups. - If inequities do exist, then an instructor can think strategically about implementing specific teaching practices to include marginalized students in class discussions. - it can be productive to reflect on the interaction and revisit the interaction with students later during office hours or during the next class session. - Computing educators need to both honor students’ experiences as learners and also realize that their own positive experiences with computing can obfuscate an appreciation for their students’ negative experiences. - Beyond listening without judgment, students may want their instructors to take some kind of action. In our experience, it is okay to tell students that you will think about what they have shared and follow up with them with potential solutions. It also can be empowering to work collaboratively with the student to think of solutions that might work for them. Regardless, following up with students after an initial meeting can assure them that you have not forgotten about what they shared and are taking it seriously. - We encourage future research on equity and inequity in CEd to embrace and to develop versions of CEd that can lead to humanization and greater justice for people from marginalized groups. - What role does culturally responsive pedagogy play? The Exploring CS curriculum (Goode et al., 2012; Margolis et al., 2015) explores social, creative, and culturally responsive computing contexts. - What characteristics of the intervention and context are necessary and sufficient for interventions that have been successful at fostering diversity - Computational thinking has since been widely accepted and promoted both as the skill set that programmers develop and as the general thinking skills that should be developed by computer scientists as they learn the discipline. - Computational thinking is not primarily about the development of electronic computer systems. It is about computation and the development of systems based on computation. - Computation is not just about numeric calculation, however. It concerns symbol processing more generally. - It is specifically not just the skill of computer programming, but the much wider way of thinking that computer scientists (not specifically programmers) develop. - Programming relies directly on this skill set, but computational system design and development is about far more than just coding itself. The development of higher levels of computational systems relies on these skills, as does innovation in computing more generally. - “Thinking like a computer scientist means more than being able to program a computer. It requires thinking at multiple levels of abstraction.” - This wider view puts the emphasis not on machines or programming, but on information processing and the design and understanding of systems that do such information processing. - They give wide-ranging examples of how such language and unplugged computational thinking might be developed in an interdisciplinary way from US curricula, all concerning the students doing computation, not writing programs. The focus is on computation and information processing tasks generally, not on how they are specifically implemented in computers. - What is the skill set involved in computational thinking? (2) How does one take children on the journey to gain those skills? - Just because the young, novice developer has not provided or worked with a precise model of computation, a mathematical semantic model, or implemented a fully working, complete computational stack all the way down doesn’t mean they are not thinking computationally or about computation. - Human-centered software development also does not start with algorithms and models, but rather with understanding human and socio-technical needs. It is only later in good development processes that underlying models, precision, and completeness enter the scene. - “can understand and apply the fundamental principles and concepts of computer science, including abstraction, logic, algorithms and data representation” “can analyse problems in computational terms” - While there are differing views as to the details, there is a lot of agreement at least as to the core elements that make up computational thinking: algorithmic thinking, logical thinking, abstraction, generalization, and decomposition, for example, are generally agreed to play a part - Algorithmic thinking is the idea that solutions to problems are not limited to one-off answers - algorithms that can give answers whenever needed for general cases: instructions that, if followed blindly and precisely, are guaranteed to lead to an answer, - Algorithmic thinking is the core part of the computational thinking skill set that makes it different from the thinking skills of other disciplines such as scientific thinking, mathematical thinking, design thinking, and so on where the other building blocks of computational thinking arise. Computing overlaps and draws on many other subjects. The core of computing is centered around algorithms, however. Similarly, computational thinking draws on and overlaps with other problem-solving approaches from those disciplines. In this sense, algorithmic thinking is the defining part that makes it different. However, on its own, algorithmic thinking is not enough to be generally useful as a way of problem-solving (unless one subsumes all of the other aspects into the term “algorithmic thinking”). - Being able to think logically is a core skill that underpins all versions of computational thinking. Logic underpins the semantics of programming languages, and thinking in a logical way is needed to develop algorithms, to implement these as programs, and to verify whether or not they work correctly, either informally or formally. - Abstraction is the process of simplifying and hiding detail to get at the essence of something of interest. As part of computational thinking, it provides a way to manage complexity in order to make problem-solving easier and allow truly massive computational systems to be designed. - Generalization involves taking the solution to a problem and creating a more general version that is applicable to a wider set of problems. - Decomposition is the idea that to solve a complex problem, including writing a complex program, it can often be broken into smaller parts that can each be solved separately and much more easily. - Evaluation is a potential element of computational thinking. However, there is no complete consensus on its inclusion. Practically, it is clearly an important part of any problem-solving approach. It is also very clearly a vital part of the skill of programming, where more time is typically spent testing solutions than writing code itself. - Computational system design involves a wider set of activities than programming, and those activities include computational thinking elements. - Abstract the problem from its description (abstraction) 2. Generate subproblems (decomposition) 3. Transform subproblems into subsolutions (generalization and algorithmic thinking) 4. Recompose (algorithmic thinking) 5. Evaluate and iterate (evaluation) - having an understanding of the underlying concepts and having the vocabulary to express them by themselves can help develop the skills in a reflective way. - Programming as a tool to think with: The original purpose for teaching programming to children was to help them in their learning, which is closer to computational thinking - Programming as a tool for learning and doing: Programming can be an important medium to support learning or to support the user’s disciplinary tasks. - Most people who use programming are neither preparing to be software professionals nor are planning to use programming to learn something else. - Secondary school teachers meet the definition of people who learn to program but not to develop software. Teachers learn computing in order to teach computing. - there is value in programming as a medium for communicating, sharing, and exploring ideas, there is value in programming as a tool for learning, and there is value in using programming to achieve tasks other than professional software development. - Supervised learning refers to cases where a network is trained using data sets that include desired/labeled output from the network - backpropagation. Whereas the use of a trained model involves sending information from input nodes through the network toward output nodes, backpropagation proceeds in reverse, from output to input. Backpropagation algorithms compare the output of a network for a given input to the known correct output. - a tool can also be a definition language. Initially, the Pascal language was designed as an educational programming language. - First, there are tools and environments that support planning, designing, analyzing and constructing programs (Section 21.5.1). These tools provide a simplified environment that hides some of the complexity of programming concepts and processes as compared to working directly with professional tools. - A second category of scaffolding tools covers different types of feedback for students on their work. As an example, automatic assessment tools (Section 21.5.4) can provide feedback on students’ programs in terms of their correctness, style, structure, and efficiency. - Third, computing concepts are frequently abstract and invisible. Therefore, scaffolding student learning with visualization and animation (Section 21.5.5) is a broadly explored area of tools research. - Interclass similarity occurs when disparate activities manifest themselves in similar manners within log data. - Finally, a relatively new research area within computing education is the collection and analysis of the physiological data of programmers as they perform programming tasks. The best-known examples are studies that collect the eye-tracking data of programmers - Learning is a social phenomenon. Psychological processes develop through interactions between people within varying cultural contexts. As Vygotsky - How Students Learn (2005), Bransford and Donovan assert that “instruction must begin with close attention to students’ ideas, knowledge, skills, and attitudes, which provide the new foundation on which new learning builds” - An important component of broadening participation in computing is the enactment of a culturally responsive pedagogy in computing learning spaces. - Teachers are already doing something, and to take on a new practice means they must alter their old practice. They must stop doing something familiar and take on something initially alien. - Along the vertical axis of the grid is a series of questions or prompts intended to elicit teachers’ pedagogic knowledge of the concept under consideration: What do you intend the students to learn about this idea? Why is it important for students to know this? What else do you know about this idea (that you do not intend students to know yet)? Difficulties or limitations connected with teaching this idea. Knowledge about the students’ thinking that influences your teaching of this idea. Other factors that influence your teaching of this idea. Teaching procedures (and particular reasons for using these to engage with this idea). Specific ways of ascertaining students’ understanding or confusion around this idea (include likely range of responses). - The ability to chunk information is a well-established difference between experts and novices in a domain - The idea of chunking is that, while working memory is finite (Baddeley & Hitch, 1974), the things that are stored in working memory can be “chunks” with a high density of information. - This idea of chunking explains the otherwise perplexing result that people can remember approximately seven things, regardless of whether those things were numbers, letters, or monosyllabic words - Computing education research is focused on how to teach students to work with a computer. More than that, it is about teaching students to understand how a computer works and how to leverage that understanding to use the computer to accomplish their goals. - Motivation refers, very broadly, to the underlying psychological mechanisms that cause people to engage and persist in behaviors. - Attitudes refers to an individual’s beliefs, emotions, and judgments related to some external entity, which in the education context is typically the subject or content being learned - dispositions refers to the enduring behavior patterns that vary between individuals, which can be used to explain why they act in specific ways in specific types of situations - Performance accomplishments refers to previous experiences of success in activities closely related to the activity that the self-efficacy belief refers to - Verbal persuasion refers to attempts by others to convince an individual that they can successfully cope with an activity - Self-efficacy has an especially strong influence on students’ persistence in difficult classes. - Metacognitive self-regulation is another important component of self-regulated learning theory that consists of self-monitoring behaviors and strategies that students implement in order to overcome academic challenges - Metacognitive strategies can be distinguished from cognitive strategies, which have more to do with the specific content of a learning task, whereas metacognitive strategies have more to do with managing one’s behaviors and approach to the learning process. - Researchers should further examine how we can develop pedagogical approaches and curricula to support computing students with respect to the motivational variables discussed above. - Another important component of students’ affective experiences in computing courses are the emotions that they experience in response to the coursework. Emotional experiences are an important source of information in the formation of attitudes, so emotional responses could have the effect of pushing students in a positive or negative direction with regard to interest and engagement in computing, above and beyond their actual performance in computing courses. - It is a variation on an industry-standard practice. In education settings, students work synchronously in teams of two to create a program. Students take turns in one of two roles: the driver, who has their hands on the computing device, and the navigator, who provides instructions and guidance to the driver. Students constantly engage in communication and switch roles regularly. - For students in beginning programming courses (e.g., CS1) where pair programming was required for all programming assignments, students were more likely to pass the course and more likely to pass the next course where pair programming was not required. - Introductory programming courses have undergone the most rigorous study of their impact on retention and success in the follow-on course. However, pair programming is and can be used in many courses, including software engineering or other “programming project-heavy” courses. - Very little comparative research exists on the impact of pair programming on learning or outcomes beyond the first programming course or with K–12 students. - Pair programming refers to a two-person team working together at a single computing device to accomplish some programming-related task. The two people typically take on two roles: that of a driver and a navigator. The driver is the one with their hands on the computing device (e.g., keyboard, mouse, or trackpad) and the navigator is looking on, making suggestions, watching for errors, and otherwise providing guidance to the driver. - PI is a specific protocol by which students in a classroom setting are engaged with challenging questions designed to confront misconceptions or push their “zone of proximal development” - Studio-based learning (SBL) is a socially oriented instructional model that has been the centerpiece of architecture and fine arts education for over a century. SBL can be conceptualized as an iterative process in which learners refine their individual solutions to disciplinary problems through two key activities. In the design studio, students build camaraderie by working on solutions in a common physical space. In design crits (design critiques), students present their evolving solutions for feedback and discussion (Moog & Spencer, 2008). - Process-oriented guided-inquiry learning (POGIL) is a student-centered pedagogy that focuses on the simultaneous development of both content knowledge and process skills (e.g., problem-solving, teamwork, and written/oral communication). Students work in teams to complete inquiry-based activities that guide them to discover key concepts for themselves.