Remote Learning Cheating: What Works and What Doesn't (Evidence Review)

Key Takeaways

Online exam cheating surged dramatically during the pandemic: self-reported rates jumped from ~30% before COVID to nearly 55% during remote learning peaks, and the increase has persisted even after institutions returned to in-person formats.

The most effective prevention strategies aren’t surveillance tools. Evidence consistently points to authentic continuous assessment, structural changes (like lowering stakes), and academic culture shifts as the strategies that actually reduce cheating.

Remote proctoring and AI detection work for deterrence but create serious collateral effects—privacy concerns, test anxiety, and false positives—that can undermine the very learning environment educators want to protect.

The 2025 consensus among researchers is that institutions should pivot away from punitive tech stacks toward authentic assessment design, scaffolding, and honor-code culture.

No single strategy eliminates cheating. The evidence shows a layered approach works best, combining assessment redesign, process documentation, and targeted monitoring only where high-stakes licensing requires it.

The Real Scale of Remote Learning Cheating

Let’s start with what the data actually says, because the numbers are both sobering and clarifying.

A systematic review by Professor Patrick Newton (2024) analyzed 19 surveys covering over 4,600 students and found that 54.7% of respondents admitted cheating in online exams—compared to roughly 30% before the pandemic. That’s not a temporary spike; it’s a structural shift in how assessment integrity operates online.

The 2024 study published in Higher Education confirmed online exam cheating was self-reported by 44.7% of students overall, with pre-COVID rates sitting at 29.9%. During the pandemic lockdown period, the rate climbed significantly.

A separate University of Cologne study (2024) found the single biggest driver of online cheating: “Cheating is more likely to occur if the exam content does not appear to have an apparent use in future professional practice.” When students view the exam as irrelevant, the moral cost of cheating drops.

Why do students cheat? Two behavioral economics models explain most of the variance:

Gary Becker’s Rational Crime Model (University of Chicago): Students weigh the rewards of cheating against the probability of getting caught and the expected punishment.
Dan Ariely’s Lying Study (Duke University): Cheating is an internal tug-of-war between the desire to get ahead and the desire to see oneself as a good person. When the stakes feel irrelevant, the “good person” side loses.

This matters because it means prevention strategies that only focus on catching students—without addressing motivation—will always be incomplete.

What Actually Works: The Evidence-Backed Strategies

Research from multiple institutions and peer-reviewed journals has converged on a set of strategies that consistently reduce cheating. Here’s what the evidence supports.

1. Authentic Assessment (Highest Evidence)

The single most effective prevention method is replacing easily-cheated assessments with authentic, applied tasks. This isn’t a preference—it’s a finding replicated across dozens of studies.

The University of Pittsburgh’s Center for Teaching and Learning recommends:

Authentic tasks: Require students to create realistic products (presentations, arguments, analyses) that mirror real-world professional work. These cannot be “looked up” because there is no single correct answer.
Low-stakes frequent assessments: Replace a few high-pressure exams with many small, formative checkpoints. Assign an aggregate value of 10% or less to these self-assessments.
Process documentation: Ask students to submit drafts, research logs, annotated bibliographies, and evidence tables showing how they arrived at their answer.

A 2025 Emerald publishing study (Journal of Research in Innovative Teaching) confirmed that continuous formative assessments dramatically reduce the motivation to cheat because there’s less to gain from a single incident.

Why this works: Authentic assessments remove the incentive. When the assignment requires original thought and personal synthesis, cheating becomes pointless—you can’t outsource a creative project or a uniquely applied analysis.

2. Structural Assessment Redesign (Strong Evidence)

Professor John Cantiello (George Mason University, 2024) reviewed the literature and found that changing the format of assessment is more effective than any technology stack.

Recommended structural changes:

Randomized question pools: Pull questions from a larger bank so no two students gets the same exam.
Sequential scaffolding: Break high-stakes assignments into pieces (proposal → outline → draft → revision). At each stage, the instructor evaluates individual progress, making it difficult to fabricate work.
Conceptual over recall questions: Design questions that require application and synthesis, not memorization. These can’t be searched or copied because there’s no single answer to find.

The University of Pittsburgh’s instructional design guidelines echo this: “Scaffold assignments. Start with assignments that focus on lower-level skills and then have students work up to higher-order work.”

3. Culture and Policy Interventions (Strong Evidence)

The Emerald study (Journal of Research in Innovative Teaching, 2025) found that clear institutional honor codes decrease cheating more effectively than surveillance alone.

Research-backed culture strategies:

Explicit policy communication: Define what academic dishonesty means in your course, including generative AI use. Share these parameters transparently.
Honor code acknowledgment: Require students to sign and return an academic integrity statement at the start of the term.
Values-based conversations: Teachers at Chicago’s Oriole Park Elementary built an “honor code” culture by discussing character and integrity explicitly—not as a punishment threat, but as a community standard.

The Turnitin blog (2020) confirmed that defining AI boundaries explicitly (“tell students what constitutes dishonesty in the context of generative AI”) reduces opportunistic cheating.

4. Technology That Complements (Not Replaces) Design (Moderate Evidence)

Technology tools work best when they supplement authentic assessment, not replace it.

Effective technology strategies:

Low-stakes daily assignments: Turn in classwork daily during remote learning. This wasn’t always possible in-person because teachers could observe students; online, daily submission creates the same visibility through a procedural requirement.
Camera verification: Require cameras on during tests. A 2020 Education Week report noted this as an effective deterrent.
Lockdown browser + AI monitoring: When licensing or accreditation requires high-stakes exams, proctoring tools like EduLegit deter cheating and raise the perceived cost.

The ScienceDirect study (Dendir, 2020, cited by 386 researchers) found that “relatively simple, technology-based tools can be used to significantly mitigate cheating in online courses.” However, the same study acknowledged the technology alone is insufficient.

What Doesn’t Work (or Backfires)

Here’s the part most guides don’t tell you: some commonly-used strategies are either ineffective or actively harmful.

1. Punitive Technology Stacks

Relying exclusively on AI detection tools and proctoring software creates significant collateral damage:

Privacy concerns: TEQSA (Australia’s Tertiary Education Quality and Standards Agency) reports that AI proctoring raises serious data privacy questions. Student trust erodes when monitoring feels invasive.
Test anxiety: The International Journal of Advanced Scientific Research (2025) noted that advanced deep learning monitoring (gaze tracking, background noise) creates significant psychological stress during exams. Anxiety degrades performance, making exams less valid assessments of learning.
Algorithmic bias: Proctoring systems have documented false-positive rates. The EduLegit FAQ on false positives addresses this directly.

2. High-Stakes, Low-Value Exams

A 2025 study published in Innovative Higher Education by Sophie Leaton Gray (cited by 59 researchers) argues that universities must move beyond detection-based strategies toward ethically grounded assessment. The research shows that when exams are both high-stakes and perceived as low-value, cheating becomes rational.

The University of Cologne (2024) confirmed: students cheat when exams feel irrelevant. Punishing them for cheating while also running irrelevant exams is counterproductive.

3. Surveillance Without Culture

Monitoring tools raise the probability of detection—but they don’t change the underlying motivation. Dan Ariely’s behavioral research showed that cheating is also a moral decision, not just a cost-benefit calculation. Without a culture of academic integrity, surveillance only pushes cheating to more sophisticated methods (peer messaging, unauthorized source referencing, and GenAI tools).

4. Over-Relying on AI Detection

The 2024 article in Journal of Academic Integrity (Leaton Gray) found that AI detection is unreliable for authorship verification. False positives affect honest students, and false negatives miss sophisticated cheating. The article advocates for process monitoring (watching how students work) over output detection (analyzing what students produce).

A Practical Framework for Choosing What to Implement

Here’s a decision framework educators can use:

Strategy	Best For	Evidence Strength	When to Avoid
Authentic assessment	All courses	Very strong	When course objectives require standardized testing (e.g., licensure exams)
Scaffolding	Project-heavy courses	Strong	When course schedule is too compressed for incremental feedback
Honor code + culture	All courses	Strong	When institutional policy conflicts with academic integrity emphasis
Randomized questions + process docs	Exam-heavy courses	Moderate-strong	When question bank creation requires excessive instructor time
Remote proctoring	High-stakes licensure exams	Moderate	When privacy concerns or tech anxiety outweigh the need
AI detection tools	Suspicion cases	Weak	As a sole strategy—never

What we recommend: Start with authentic assessment and scaffolding. Layer structural changes (randomized pools, daily submissions) around that foundation. Use technology only where licensing, accreditation, or institutional policy demands it.

The Emerging Threat: Generative AI Cheating

The 2025 ScienceDirect study by Huang et al. (cited by 38 researchers) used the modified theory of planned behavior to track GenAI cheating among undergraduates. Key findings:

Students use GenAI not just for content generation, but for citation fabrication and research synthesis manipulation.
The perceived ease of AI generation lowers the “moral cost” of cheating—students rationalize it as “using available tools.”
Detection-based AI tools are ineffective against GenAI because the generated content is indistinguishable from legitimate research when combined with proper citation.

The 2025 Springer article by Leaton Gray argues that institutions must evolve their assessment philosophy to address GenAI specifically. This means moving toward oral defenses (viva voce), process-based authorship verification (like draft history analysis), and real-time engagement monitoring—the very tools EduLegit was built to provide.

Bottom Line: A Layered Approach

The research is unanimous: no single strategy eliminates cheating in remote learning. The most effective approach combines:

Assessment redesign (authentic tasks, scaffolding, low-stakes frequent checkpoints)
Culture and policy (honor codes, explicit AI boundaries, values-based conversations)
Targeted technology (only where licensing requires it; never as the sole strategy)
Process verification (draft history, oral defenses, engagement monitoring for high-stakes exams)

This layered approach doesn’t just reduce cheating—it actually makes the learning experience better, because authentic assessments are more engaging, low-stakes frequent checkpoints improve student feedback loops, and process documentation reinforces learning.

Ready to Implement?

If you’re looking to strengthen your academic integrity practices, contact our team to discuss a tailored plan. We specialize in evidence-based proctoring, AI detection, and process verification tools that complement assessment redesign rather than replace it.

Remote Learning Cheating: What Works and What Doesn’t (Evidence Review)