Make or Take: How Students Navigate Self-Created and Instructor-Provided Cheat Sheets
Source: arXiv:2607.24736 · Published 2026-07-27 · By Helen Weixu Chen, Victoria Sakhnini, Lesley Istead
TL;DR
This paper addresses the understudied question of how students choose between self-created and instructor-provided cheat sheets during exams, and how these choices influence their preparation and perceived exam support. Conducted as a longitudinal study in a senior undergraduate software requirements course, students were given the option to use either cheat sheet format for both midterm and final exams. Through three sequential surveys with up to 53 respondents each, the authors investigated students' rationales, usage patterns, and changing preferences over time. Their findings reveal that students’ cheat sheet choices are shaped by tensions around trust in instructor expertise versus personalization, cognitive efficiency, perceived coverage, and exam preparedness. While self-created sheets involved more initial effort, they were often seen as better aligned to individual needs and correlated with a trend toward higher final exam scores. The study highlights how providing choice transforms the cheat sheet from a fixed tool into an active site of student agency and learning strategy negotiation.
Key findings
- 70% of students chose self-created cheat sheets for the midterm, increasing to nearly 80% for the final, indicating strong preference for personalization.
- Midterm preparation time was significantly longer for self-created cheat sheet users than for instructor-provided sheet users (Chi-square 𝜒2=12.58, p=0.0056), but this difference was not significant for the final exam (𝜒2=2.34, p=0.51).
- Over 60% of participants reported only occasional cheat sheet use during both exams, with about 30% using them frequently; less than 20% rarely or never consulted them.
- 57% of self-created cheat sheet users reported coverage of 51–75% of final exam content, compared to a more even and lower distribution among instructor-provided sheet users.
- Most students (73%) continuously referenced their cheat sheets during exams instead of only scanning at the start or checking uncertain questions.
- Majority perceived cheat sheets as improving confidence and speed: 55% said cheat sheets helped them work faster and more confidently, and 57% said cheat sheets reduced time recalling definitions or formulas.
- No statistically significant performance difference was found at midterm; however, a medium effect size showed a non-significant trend toward higher final exam scores for self-created cheat sheet users (mean 69.5 vs 61.8, Cohen’s d = -0.63).
- Students’ cheat sheet preferences evolved over time, with some switching from instructor-provided to self-created due to perceived misalignment of instructor sheets with exam content.
Threat model
The adversary is modeled as a student faced with exam preparation choices under the constraint of permitted cheat sheet use, who must select and use either an instructor-provided or self-created cheat sheet to optimize exam performance. The adversary does not have the ability to obtain unauthorized information or bypass exam rules; the focus is on decision-making and agency within allowed options.
Methodology — deep read
- Threat Model & Assumptions: The study assumes a classroom adversary model where students have the option to freely choose between cheat sheet types without instructor preference or interference. The instructor did not provide the cheat sheet before exams, removing bias toward either format. The adversary here is the student navigating exam preparation strategies, with no explicit cheating considered.2. Data: The data comprised three waves of surveys administered in a senior undergraduate software requirements course during Winter 2025: before midterm (53 responses), after midterm/before final (50), and after final (44). A longitudinal cohort of 41 students completed all three. Surveys collected both closed- and open-ended responses on cheat sheet choices, rationales, preparation time, cheat sheet usage frequency, perceived coverage, and exam experiences. No experimental manipulation was performed.3. Architecture / Algorithm: This is a qualitative and quantitative educational study, so no ML models or algorithms were applied. The analysis combined deductive and inductive thematic coding of open responses to identify rationales and emergent themes. Quantitative analysis used chi-square tests for preparation time distributions, t-tests for score differences, and descriptive statistics of usage patterns.4. Training Regime: N/A5. Evaluation Protocol: The authors evaluated: - Preparation time differences by cheat sheet type, using chi-square on ordinal time ranges - Frequency of cheat sheet use per exam, with descriptive stats - Coverage of exam content by cheat sheet type from survey questions - Patterns in usage strategies (continuous referencing vs skimming) - Perceived confidence and efficiency improvements - Exam performance differences between cheat sheet groups using independent t-tests for midterm and final. No adversarial or distribution shift evaluation was done. Respondents self-selected cheat sheet types.6. Reproducibility: The paper does not mention public release of code or detailed datasets. Since the data is student survey and linked to exams, it is likely closed. The detailed survey instruments are included in an appendix, and the study design and analysis approach are described thoroughly. A single concrete example can be traced: a participant initially used the instructor-provided sheet but switched to self-created for the final after finding the instructor sheet lacking coverage, illustrating how preferences and preparation strategies evolved over time. Overall, the methodology is robust for a naturalistic observational study emphasizing student agency and reflection on cheat sheet choice in a realistic educational context.
Technical innovations
- Framing the cheat sheet as an active site of student choice and negotiation rather than a fixed instructional tool, filling a gap in prior literature that treated cheat sheets as fixed interventions.
- Longitudinal mixed-methods study design capturing evolving student attitudes and preparation strategies across multiple high-stakes assessments with free choice of cheat sheet type.
- Identification of three core tensions shaping cheat sheet preference: trust in instructor versus self-knowledge, coverage versus clarity, and effort versus payoff.
- Detailed synthesis of qualitative rationales demonstrating students’ strategic reasoning beyond passive use, revealing cheat sheets as instruments of personalized learning and cognitive engagement.
Datasets
- Survey data from senior undergraduate software requirements course — ~53 responses per wave across 3 survey waves (Winter 2025 cohort) — collected via course platform surveys (not publicly released)
Baselines vs proposed
- Midterm exam preparation time: instructor-provided users had significantly shorter prep time than self-created users (Chi-square 𝜒2=12.58, p=0.0056) vs final exam where differences were not significant (𝜒2=2.34, p=0.51)
- Midterm exam scores: instructor-provided mean=91.7, self-created mean=92.7 (t=−0.76, p=0.46) vs final exam scores: instructor-provided mean=61.8, self-created mean=69.5 (t=−1.65, p=0.13, Cohen’s d=−0.63)
Limitations
- Non-randomized design with self-selection into cheat sheet types limits causal inference about impact on performance or preparation time.
- Moderate sample size (max 53 per wave, 41 longitudinal) in a single senior-level software requirements course limits generalizability to other domains or levels.
- Lack of direct performance measures linked to cognitive or anxiety states limits interpretation of cheat sheet benefits beyond self-report.
- No adversarial or deception scenarios explored; study focused on legitimate use and student attitudes.
- Instructor-provided cheat sheet was not revealed before exam, possibly limiting the extent to which students could fully evaluate or adapt to that format.
- Survey-based self-reports can suffer from biases such as social desirability or inaccurate recall.
Open questions / follow-ons
- How do these findings generalize across different disciplines, education levels, or types of assessments?
- What is the impact of revealing the instructor-provided cheat sheet earlier on student choice and preparation strategies?
- Can integrating instructor-provided scaffolds with opportunities for personalization improve both fairness and learning?
- How do cheat sheet choices affect cognitive load and anxiety in controlled experimental settings?
Why it matters for bot defense
For bot-defense and CAPTCHA practitioners, this paper offers a nuanced view on how users engage with customized versus generic aids under time pressure and cognitive constraints. Analogously, bot-challenge design might consider providing users with options that balance personalization and standardized support to optimize human engagement and trust. The observed tensions between trust in authority, autonomy, coverage, and efficiency suggest that defense mechanisms should not be one-size-fits-all but should allow user agency while preserving security objectives. Understanding how users negotiate tool choices could inform adaptive challenge systems that reflect user preferences or skill levels, enhancing usability without compromising bot defense efficacy. However, unlike academic cheatsheets, bot-defense challenges operate in adversarial security contexts with risk of misuse; thus, trade-offs in offering choice must be carefully calibrated.
Cite
@article{arxiv2607_24736,
title={ Make or Take: How Students Navigate Self-Created and Instructor-Provided Cheat Sheets },
author={ Helen Weixu Chen and Victoria Sakhnini and Lesley Istead },
journal={arXiv preprint arXiv:2607.24736},
year={ 2026 },
url={https://arxiv.org/abs/2607.24736}
}