Exploring the Design and Impact of Interactive Worked Examples for Learners with Varying Prior Knowledge DOI: XXXXXXX.XXXXXXX Conference: The ACM CHI conference on Human Factors in Computing Systems; 13–17 April, 2026; Barcelona ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computer Science Education, Intell

Sutapa Dey Tithi, Xiaoyi Tian, Ally Limke, Min Chi, Tiffany Barnes · 2026

Read Exploring the Design and Impact of Interactive Worked Examples for Learners with Varying Prior Knowledge DOI: XXXXXXX.XXXXXXX Conference: The ACM CHI conference on Human Factors in Computing Systems; 13–17 April, 2026; Barcelona ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computer Science Education, Intell by Sutapa Dey Tithi, Xiaoyi Tian, Ally Limke, Min Chi, Tiffany Barnes (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-16806.txt.

Introduction

Sutapa Dey Tithi and co-authors test interactive worked examples in a logic tutoring system. Tutoring systems lift learning through tailored help such as worked examples. Yet they often show an aptitude-treatment snag: low prior knowledge learners gain more than others. The team used ICAP learning theory to design two new example types with varied cognitive demand.

Buggy examples ask students to fix bugs, while Guided examples ask them to fill in missing rules. In a controlled trial with over one hundred fifty undergraduates, both groups beat passive worked examples on the posttest. Buggy problems aided high prior knowledge learners, while Guided problems aided low prior knowledge learners. Behavior study found Buggy spurred explore-revise cycles, while Guided spurred help-seeking with fewer errors.

Catalog record

CREATORSutapa Dey Tithi, Xiaoyi Tian, Ally Limke, Min Chi, Tiffany Barnes
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-16806

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Exploring the Design and Impact of Interactive Worked Examples for Learners with Varying Prior Knowledge DOI: XXXXXXX.XXXXXXX Conference: The ACM CHI conference on Human Factors in Computing Systems; 13–17 April, 2026; Barcelona ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computer Science Education, Intelligent Tutoring Systems, Cognitive Load Theory, Worked Examples, Human-computer Interaction Sutapa Dey Tithi, Xiaoyi Tian, Ally Limke, Min Chi, Tiffany Barnes Abstract . Tutoring systems improve learning through tailored interventions, such as worked examples, but often suffer from the aptitude-treatment interaction effect where low prior knowledge learners benefit more. We applied the ICAP learning theory to design two new types of worked examples, Buggy (students fix bugs), and Guided (students complete missing rules), requiring varying levels of cognitive engagement, and investigated their impact on learning in a controlled experiment with 155 undergraduate students in a logic problem solving tutor. Students in the Buggy and Guided examples groups performed significantly better on the posttest than those receiving passive worked examples. Buggy problems helped high prior knowledge learners whereas Guided problems helped low prior knowledge learners. Behavior analysis showed that Buggy produced more exploration-revision cycles, while Guided led to more help-seeking and fewer errors. This research contributes to the design of interventions in logic problem solving for varied levels of learner knowledge and a novel application of behavior analysis to compare learner interactions with the tutor . Figure 1 . The tutor interface with the integrated three-panel design: student workspace (left), domain rules (center), and contextual information/instructions panel (right). The hint system and feedback message appear in the bottom-left corner. The tutor interface with the integrated three-panel design: student workspace (left), domain rules (center), and contextual information/instructions panel (right). The hint system and feedback message appear in the bottom-left corner. 1. Introduction Intelligent tutoring systems (ITSs) provide adaptive scaffolding and personalized interventions, demonstrating significant positive impacts on learning outcomes ( 48 ; 43 ) . However, the effectiveness of instructional interventions is not uniform across all learners. Research has found evidence of aptitude-treatment interaction (ATI) ( 17 ; 70 ) , meaning that the effectiveness of specific instructional strategies varies significantly based on individual learner characteristics. These aptitude treatment interactions manifest across multiple dimensions of learner differences, including prior experience level, prior working memory, and incoming self-regulated learning ability ( 34 ; 40 ; 20 ; 82 ) . ITSs often incorporate two primary instructional approaches: Problem Solving (PS), where students independently construct solutions, or Worked Examples (WE), where the tutor presents step-by-step solutions for students to study. With regard to learners’ abilities, research within ITSs has shown that high-ability learners can benefit from less guidance or more independent problem-solving practice. In contrast, low-ability learners benefit from more specific and direct guidance ( 3 ; 42 ) . Worked examples reduce the intrinsic cognitive load on learners and improve their learning ( 53 ) . Moreover, examples can save learners time without reducing their learning ( 47 ) . However, they may not be beneficial or could even be detrimental for learners with high prior knowledge, a phenomenon known as the expertise reversal effect ( 50 ; 35 ) . Thus, designing instructional interventions faces the challenge of creating opportunities so that learners of any incoming proficiency can master the same material and improve their learning outcomes in a similar amount of time. The ICAP (Interactive, Constructive, Active, Passive) framework provides a theoretical lens for addressing this challenge ( 13 ) . This theory links students’ cognitive engagement to their learning outcomes. It states that learning outcomes systematically increase as cognitive engagement increases, from Passive (receiving information), to Active (manipulating materials), to Constructive (generative), to Interactive (co-generative). Traditional worked examples, requiring only passive observation, fall into the lowest level of this hierarchy. There remain critical gaps in understanding: (1) how to design worked examples that promote higher levels of cognitive engagement while maintaining their scaffolding benefits, (2) empirical evidence of how worked examples with different levels of cognitive engagement help learners. To address these gaps, we designed and empirically evaluated two types of interactive worked examples within an intelligent logic tutoring system: Buggy (students identify and fix bugs in worked solutions, promoting active engagement through debugging and fixing), and Guided (students reconstruct missing parts in worked solutions with step-by-step hints, promoting active reconstruction of proofs). The tutor also provides a baseline Passive WE and a Constructive PS intervention. The student-tutor interaction design and desired cognitive engagement in Buggy and Guided examples follow the ICAP framework and aim to increase interaction with the tutor relative to passive WE while maintaining more structure than independent PS. Propositional logic proofs are open-ended with multiple valid solutions. Our Buggy examples present expert-designed erroneous solutions without marking the incorrect steps, requiring students to understand both local rule applications and global proof coherence to successfully debug and fix incorrect steps. Guided examples decompose the solution into manageable chunks or subgoals, providing step-specific hints to help reconstruct the proof. Both Buggy and Guided examples are Interactive WEs, since they provide immediate feedback on student work. These two interventions are novel in that intelligent problem solving environments are often custom-built to enable the tracking of student work, and rarely have the capability to provide erroneous examples or subgoal-focused, partially-worked guided examples. We integrated these interactive worked examples (Buggy, Guided) in an intelligent logic tutor along with passive Worked Example (WE) and constructive Problem Solving (PS). We investigated the impact of these worked examples on students’ learning of propositional logic problem solving in a controlled experiment with 155 undergraduate Computer Science (CS) students. Students in different conditions practiced interleaving PS with one of: WE, Buggy, or Guided examples. We recorded students’ clickstream interactions in the tutor and analyzed the log data. We investigated students’ performance in posttest problems after training with three different interventions and analyzed further to answer the following research questions: • RQ1 (Learning Outcomes): How do worked examples targeting different engagement levels (i.e., passive WE, Buggy examples, and Guided examples) affect students’ performance in posttest problems? • RQ2 (Moderation by Prior Knowledge): How do these effects vary with students’ prior knowledge? • RQ3 (Problem Solving Behavior): How do these interventions impact students’ problem solving behaviors during training? The contributions of this work are: (1) two theory-driven novel interactive worked example interface scaffolding designs in propositional logic learning promoting higher cognitive engagement; (2) empirical evidence of differential effectiveness of worked example interventions based on prior knowledge, informing adaptive logic tutoring; (3) a Markov model based visualization to interpret problem-solving behavior patterns; (4) design implications for worked examples that support guided proof construction, evaluation and debugging practice. 2. Related Work 2.1. Instructional Support in Intelligent Tutoring Systems Multiple approaches have been explored within intelligent tutoring systems (ITSs) to facilitate student learning ( 69 ) . These include Problem Solving (PS), where students solve the problem themselves, and Worked Example (WE), where the problem is solved for them step-by-step. WEs have been shown to reduce intrinsic cognitive load and improve learning ( 53 ) . However, the research has shown that WEs have mixed efficacy; while they may not be beneficial to students with high prior knowledge when problems are structured ( 50 ) , they have also been shown to be most

Read the complete text