Finding: An audit of six replication failures in the SCORE project reveals that many apparent failures stemmed from methodological errors (data analysis mistakes, coding errors, poor data quality) rather than genuine non-replication, suggesting SCORE's 55% replication rate is an underestimate.
Why it matters: Marginally relevant as a meta-scientific critique of replication methodology in behavioral science—important context for teaching about replication credibility, but focused on methodology rather than the professor's core topics.
Method: Audit of 12 specific claims across 6 SCORE replications, examining raw data, analysis code, and interpretation decisions.
Erin Higgins, Abdulaziz A. Aromaih, Abdullah A. Alhuzaymi et al. (5 authors) • Semantic Scholar
Finding: Scoping review mapping how Rehabilitation Science and HCI use emerging technologies (AI, robotics, digital fabrication) to develop assistive technologies, noting disciplinary silos between the two fields.
Why it matters: Tangentially relevant to HCI and AI-human interaction themes, but focused narrowly on assistive technology and disability rehabilitation rather than the professor's core interests in creativity, learning, behavior change, teams, or AI's broader societal impact.
Finding: Expertise diversity among sequential contributors to wiki-style collaborative writing increases information accumulation in a shared text, with a computational model of numerical judgment generalized to information-sharing contexts.
Why it matters: Directly relevant to team dynamics and collective intelligence — specifically how knowledge diversity affects collaborative knowledge construction, with implications for designing AI-augmented collaborative systems and team composition.
Method: Experiment with 413 participants in sequential collaboration chains, tested against a computational model originally developed for numerical judgment aggregation.
Jargon: *Sequential collaboration* — a workflow where contributors iteratively revise a shared artifact one after another (like Wikipedia editing), rather than working simultaneously; *hidden profile paradigm* — a classic group research setup where critical information is distributed across members so no single person has the full picture.
⚠️ No control group mentioned | Single study (not replicated) | Limited statistical detail in abstract
Z. Jurányi, Krisztina Liszkai-Peres, Réka Pető et al. (7 authors) • Semantic Scholar
Finding: A child-focused mobile launcher (Alphie) was designed using parental expectations and behavioral science principles to reduce solitary screen time and promote social interaction and physical activity, with positive UX testing results from parent-child dyads.
Why it matters: Directly relevant to behavior change design (persuasive technology, choice architecture), UX design for vulnerable populations, and the broader question of how technology design shapes human behavior and relationships.
Method: Three-phase study combining parental expectation elicitation, participatory/value-sensitive design, and UX testing with parent-child dyads.
Jargon: *Launcher* — a home-screen replacement app that controls which apps are accessible, functioning as a choice architecture intervention at the OS level.
Stephen Becker, Susmitha Gurram, Alisha Pradhan et al. (4 authors) • Semantic Scholar
Finding: People with dementia engage in significant "invisible work" to compensate for notification shortcomings in reminder systems, and envision AI-powered, adaptive, multimodal systems that help them follow through on tasks rather than just alert them.
Why it matters: Directly relevant to behavior change design (habit/task completion support, persuasive technology, nudges), HCI/UX design for vulnerable populations, and AI's role in supporting human agency — all teachable through the lens of how reminder systems can be redesigned using mediation theory and adaptive AI.
Method: Iterative interviews with six participants with mild-to-moderate dementia — small but methodologically appropriate for exploratory qualitative HCI work.
Jargon: *Mediation theory* — philosophical framework (from Verbeek) examining how technologies shape/mediate human experience and behavior; *multimodal reminders* — notifications delivered through multiple channels (audio, visual, haptic, etc.) to increase salience.
⚠️ No control group mentioned | Single study (not replicated) | Limited statistical detail in abstract
Finding: Question-chain prompting (structured, sequenced questions) during GenAI interaction significantly improved self-regulated learning behaviors, metacognitive awareness, and academic achievement compared to free-form querying in undergraduates.
Why it matters: Directly relevant to learning sciences and AI's impact on education—demonstrates a practical scaffolding intervention for preventing passive AI use, which connects to habit formation and engagement design in AI-assisted learning environments.
Method: Quasi-experiment with 83 undergraduates, three conditions (free-form, progressive chain, divergent chain), using interaction logs, reflections, metacognitive scales, and grades.
Jargon: *Self-regulated learning (SRL)*: learners actively monitoring and controlling their own learning process. *Question-chain prompting*: a structured instructional scaffold requiring students to ask specific, sequenced questions rather than ad hoc queries. *Metacognitive awareness*: awareness of one's own thinking processes, measured here as knowledge of cognition and regulation of cognition.
⚠️ Single study (not replicated) | Convenience sample (e.g., MTurk, students)
Finding: Computing undergraduates in Sri Lanka develop deliberate linguistic strategies (sentence variation, lexical diversification, rhetorical questioning) to simultaneously evade AI detection systems and satisfy human evaluators, while academics show conflicting responses ranging from full AI acceptance to strict prohibition.
Why it matters: Illuminates a new behavioral adaptation dynamic—students optimizing writing for dual audiences (algorithmic and human)—relevant to AI's impact on learning, academic integrity, and the evolving HCI of AI-mediated education systems.
Method: Qualitative multi-method design with focus groups (30 undergraduates), semi-structured interviews (7 academics), and textual analysis of assignments, analyzed via thematic analysis.
Jargon: *Probabilistic regularity* — the detectable statistical patterns in AI-generated text that detection systems identify; *ESL* — English as a Second Language.
Finding: Classroom mastery goal structure indirectly boosts academic engagement through self-efficacy as a mediator, with self-efficacy accounting for ~50% of the relationship between motivational climate and engagement.
Why it matters: Touches on goal-setting theory and motivation mechanisms relevant to teaching, though narrowly focused on Pakistani secondary school students rather than HCI, design, or adult learning contexts.
Method: PLS-SEM with 10,000 bootstrapped resamples to test mediation among 300 secondary students.
Jargon: *PLS-SEM* = Partial Least Squares Structural Equation Modeling, a path analysis technique; *CMGS* = Classroom Mastery Goal Structure, the degree to which a classroom emphasizes learning over performance/grades; *complementary partial mediation* = both direct and indirect paths are significant, with the mediator explaining part but not all of the effect.
Finding: An HCI researcher who stutters uses narrative inquiry to critically reflect on power, stigma, and accessibility challenges in speech technology design for people who stutter.
Why it matters: Marginally relevant to HCI and inclusive design, but focuses narrowly on stuttering/speech disability accessibility rather than the professor's core interests in creativity, teams, goal-setting, behavioral science, or AI impact.
Finding: Self-regulated learning (SRL) significantly but weakly predicts Qur'anic memorization achievement (R²=0.102), with goal-setting, strategy use, motivation control, and self-evaluation identified as relevant SRL components.
Why it matters: Tangentially touches on goal-setting and self-regulated learning constructs relevant to behavioral science and skill acquisition, but in a highly domain-specific religious education context with limited generalizability.
Method: Simple linear regression with n=44, very small sample with weak effect size.