TL;DR
This paper investigates how middle school students learn computing concepts like abstraction and programming through an integrated curriculum that uses Logic Programming (LP). The study found that students performed well on foundational concepts like facts and variables, but struggled with more complex rules, suggesting that while LP is a promising avenue for integrated computing education, more support is needed for advanced topics.
What Problem Does This Paper Address?
Integrating computing into K-12 STEM education is widely recognized as crucial, yet rigorous studies on its impact, particularly on students' learning of core computing concepts like abstraction and programming, are scarce. A major challenge is the lack of a clear, operational definition of "computational thinking" (CT) in educational contexts. This paper addresses this gap by evaluating an integrated curriculum that leverages Logic Programming (LP) to provide a structured approach to teaching abstraction and programming, aiming to understand how well middle school students grasp these concepts within this framework.
Methods
The study employs a microgenetic analysis method, a technique from developmental psychology that involves intensive, trial-by-trial observation of learning over a short period to capture the nuances of cognitive change.
The core of the study is an LP-based integrated curriculum designed for middle school students. This curriculum uses Logic Programming (LP) as a tool for representing and reasoning about abstractions. The curriculum progresses through several lessons, starting with basic concepts like facts and relations in a family domain, and then moving to more complex topics like variables, queries, rules, and finally applying these to a chemistry domain. LP's text-based nature is highlighted as a key feature, offering a more direct way to learn abstraction compared to visual programming languages.
The implementation involved a teacher delivering the module over eight weeks to 317 students across grades 6-8. Data was collected through post-tests and, crucially, through one-on-one interviews with a subset of students. These interviews, conducted weekly, allowed researchers to probe students' understanding in real-time, asking clarifying questions and encouraging them to think aloud.
Datasets
Not discussed in the paper. The study focuses on student learning within a designed curriculum rather than on pre-existing datasets.
Key Insights and Ideas
- LP as a Bridge: Logic Programming (LP) offers a unique advantage by providing a formal, text-based language that naturally supports the operationalization of abstraction and reasoning, making it a promising tool for integrated STEM and computing education.
- Foundational Strength: Middle school students, when exposed to the LP-based curriculum, demonstrated a strong grasp of foundational computing concepts like abstracting knowledge into facts, understanding variables, and formulating queries. This suggests a developmental readiness for these concepts.
- The Rule Challenge: While students excelled at simpler forms of abstraction, they encountered significant difficulties with more complex concepts, particularly in formulating and understanding rules. This indicates that while abstraction can be learned, mastering more intricate logical structures requires additional support.
- Microgenetic Method's Value: The microgenetic approach proved invaluable in capturing the detailed learning processes and identifying specific points of struggle, offering a richer understanding than traditional assessments alone.
- Text-Based Abstraction Potential: The study provides evidence for the effectiveness of text-based abstraction in LP, contrasting with the often-cited challenges of evaluating abstraction in visual programming environments.
Key Results
- Facts and Variables: Students performed well in abstracting knowledge into facts and understanding variables, both in familiar (family) and new (chemistry) domains. For instance, in the post-test, students achieved 52% accuracy on familiar facts and 50% on new facts, with reasonable performance on queries.
- Rules: Performance on rules was significantly lower. In the post-test, accuracy for rules in new domains was only 16%, indicating a substantial challenge.
- Programming: Students showed fluency in using the LP software (onlineSPARC), understanding syntax for facts and rules, and using queries for testing. However, awareness of errors and debugging opportunities was less evident.
- Qualitative Findings: Interviews revealed that students could articulate reasoning processes and effectively use variables to represent general concepts, demonstrating a high level of understanding in these areas.
Limitations and Future Work
- COVID-19 Impact: The study was significantly affected by the COVID-19 pandemic, leading to rushed lessons (especially L7 and L8), a shift to online learning, and fewer opportunities for hands-on experience and class discussions. This likely impacted student performance, particularly on more complex topics.
- Time Constraints: The curriculum was compressed, especially in the later lessons, which may have hindered deeper learning of complex rules and abstraction.
- Interview Logistics: Conducting one-on-one interviews proved time-intensive and logistically challenging, leading to some missed data and potential biases (e.g., one interviewer focusing too narrowly).
- Future Work: The authors suggest that future, larger-scale studies should leverage the lessons learned from this experiment to improve data quality by implementing stricter procedures, checklists for interviewers, and more timely review of recordings. They also emphasize the need for more rigorous research to understand how to best support students in mastering higher-level abstraction and complex rules, potentially through more content, pedagogical support, and extended practice.
Summary updated