54.5% of all lecture capture views happen on the day of an exam or the day before, while just 2.9% happen on the day of the lecture or the day after (Hillsley, FEBS Open Bio 2025). That single finding reframes what lecture recordings are for most students: not a replay of class, but a last-minute revision engine, often played faster than real time. Around 77% of medical students in one 2024 study usually watched lectures at 1.5x or faster (Nakhostin-Ansari, BMC Medical Education 2024), and lab experiments mostly find that 2x speed costs younger learners little. Meanwhile 95% of responding US osteopathic medical schools record their lectures (Khong et al., Medical Science Educator 2025), so the recording, not the room, is becoming the default unit of teaching. For how fast speech can be followed in general, see our speech rate statistics. We aggregated data from peer-reviewed studies in Applied Cognitive Psychology, Memory, Frontiers in Psychology, Computers and Education, Higher Education, FEBS Open Bio, Medical Teacher, the Journal of Surgical Education, NCES, and the other sources listed in the methodology.
TL;DR
- 95% of responding US colleges of osteopathic medicine record lectures (Khong et al., Medical Science Educator 2025).
- 49.6% of physician assistant programs offer lecture recordings and 50.4% do not (Herrick and Stoehr, JPAE 2025).
- 28.17% of 71 lecture capture studies found a negative effect on attendance (Banerjee, Computers and Education 2021).
- 54.93% of the same 71 studies found a positive effect on academic performance (Banerjee, 2021).
- 72.2% of students accessed at least one recording, but viewers watched only 45% of available minutes (Hillsley, FEBS Open Bio 2025).
- 54.5% of views happened on exam day or the day before (Hillsley, 2025).
- 57.9% to 63.2% of students used the speed adjustment control (Dommett et al., 2022).
- About 77% of participating medical students usually watched at 1.5x or faster (Nakhostin-Ansari, BMC Medical Education 2024).
- Post-test scores peaked at 1.25x (80.00) and were lowest at 2.0x (66.47) (Mo et al., Frontiers in Psychology 2022).
- 2x speed did not significantly lower immediate or delayed recall among 60 medical students (Kiyak et al., Medical Teacher 2023).
- Students given recorded lectures scored 71.1% on weekly quizzes vs 88.8% for live teleconference lectures (Huerta et al., Journal of Surgical Education 2023).
- Staff who found lecture capture stressful fell from 29% in 2019 to 7% in 2023 (Voelkel et al., Emerging Topics in Life Sciences 2026).
1. Deployment: Lecture Capture Is Now the Default in Many Programs
The debate over whether to record lectures is largely over in health professions education, where data is densest. 95% of responding US colleges of osteopathic medicine record lectures, 52% always and 43% most of the time (Khong et al., Medical Science Educator 2025). Physician assistant programs are the outlier: an exact coin flip, with 49.6% offering recordings, and 93.4% of all programs mandating attendance regardless (Herrick and Stoehr, Journal of Physician Assistant Education 2025). Recording and compulsory attendance are not opposites; 90.2% of programs that record still require students to show up, versus 96.8% of those that do not.
Some institutions moved years ago. The University of Liverpool has recorded all in-person lectures and made them available to students without restrictions since 2016 (Voelkel et al., Emerging Topics in Life Sciences 2026). The broader backdrop is the shift to distance education: 61% of US undergraduates took at least one distance course in fall 2021, up from 36% in 2019 (NCES Fast Facts, Distance Learning). A generation of students now expects every lecture to exist as a file, a habit that also shows up in our online learning statistics.
| Metric | Value | Source |
|---|---|---|
| US osteopathic medical colleges that record lectures | 95% (21 of 34 colleges responded, 62%) | Khong et al., Medical Science Educator 2025 |
| Record always / most of the time / often | 52% / 43% / 5% | Khong et al., Medical Science Educator 2025 |
| PA programs offering lecture recordings | 49.6% (50.4% do not) | Herrick and Stoehr, JPAE 2025 |
| PA programs mandating attendance | 93.4% overall; 90.2% with recordings vs 96.8% without | Herrick and Stoehr, JPAE 2025 |
| University of Liverpool lecture capture policy | All in-person lectures recorded since 2016 | Voelkel et al., Emerging Topics in Life Sciences 2026 |
| US undergraduates in any distance course | 36% (2019), 75% (2020), 61% (2021) | NCES Fast Facts |
| US undergraduates exclusively distance | 15% (2019), 44% (2020), 28% (2021) | NCES Fast Facts |
| US postbaccalaureate students exclusively distance | 40% (2021), 1.3 million students | NCES Fast Facts |
Context note: the NCES figures cover all distance education, not lecture capture specifically, and fall 2021 is the most recent year on the Fast Facts page (most recent available data: NCES, 2021). Herrick and Stoehr report a 41% response rate among PA program directors.
2. Attendance: The Fear Is Real but Uneven
Faculty worry about empty rooms more than any other lecture capture risk, and the evidence only partly backs them. Of 71 articles in the largest systematic review, 20 (28.17%) documented a negative impact on attendance, and 11 reported non-significant relationships (Banerjee, Computers and Education 2021). The strongest negative case is a UK research methods module: with matched cohorts before (N = 161) and after (N = 160) capture was introduced, attendance dropped substantially in three matched lectures, and attendance mediated a negative link between capture availability and attainment (Edwards and Clinton, Higher Education).
Other settings show almost no effect. Across 1,014 microbiology students over ten semesters, average daily attendance was 83% before lecture capture and 81% after (McLean and Suchman, Journal of Microbiology and Biology Education 2016). The pattern that emerges is about year of study: in medical school, 72% of first semester students attended at least somewhat often, but by the third semester 43% said they almost never attended (Gardner et al., Medical Science Educator 2022). Attendance erodes as students learn the course and grow confident in self-study.
| Metric | Value | Source |
|---|---|---|
| Studies documenting a negative attendance effect | 20 of 71 (28.17%) | Banerjee, Computers and Education 2021 |
| Studies with non-significant attendance relationship | 11 | Banerjee, Computers and Education 2021 |
| Matched cohorts before / after capture (UK module) | N = 161 / N = 160; attendance dropped substantially | Edwards and Clinton, Higher Education |
| Average daily attendance before / after capture | 83% / 81% (1,014 students, 10 semesters) | McLean and Suchman, JMBE 2016 |
| Students saying recordings encouraged them to miss class | 8% | McLean and Suchman, JMBE 2016 |
| Students saying recordings affected their attendance | 32% (14% very much so) | Voelkel et al., FEBS Open Bio 2023 |
| First-year / second-year medical students attending non-mandatory lectures | 63% / 53% | Emahiser et al., Medical Science Educator 2021 |
| Third semester medical students who almost never attend | 43% | Gardner et al., Medical Science Educator 2022 |
Context note: Emahiser et al. found daily attendance of only 26% among first-year and 15% among second-year students, and the top reason for missing class in both cohorts was the efficiency of viewing lectures online. Gardner et al. named the ability to pause, rewind and speed up videos as a factor that discourages attendance. Nordmann et al. found ‘no compelling evidence for a negative effect of recording use’ across four undergraduate years (Higher Education).
3. Viewing Patterns: Recordings Are an Exam Tool
The usage data says recordings function less like a second lecture and more like a revision archive. In a 2025 analysis of 299 students and 982 exam observations, 72.2% accessed at least one recording, but viewers used only 45% of the available video minutes and 60% of the lectures, and 54.5% of all views came on the day of or the day before the exam (Hillsley, FEBS Open Bio 2025). Just 2.9% of views came on the day of or the day after the lecture itself. The average view happened 3.3 days before the exam.
Student intent tells the same story. At the University of Liverpool, 78% of 694 survey respondents said they were very likely to watch a missed lecture eventually, but only 33% would do so within one week (Voelkel et al., FEBS Open Bio 2023). Viewing is also an evening activity: the peak window for Echo360 views in a pharmacy program was 6 PM to 9 PM, which held 19.6% of semester views in 2019 and 21.6% in 2020 (Hussain et al., Cureus 2024).
| Metric | Value | Source |
|---|---|---|
| Students who accessed at least one recording | 72.2% (216 of 299); 27.8% never did | Hillsley, FEBS Open Bio 2025 |
| Share of available minutes / lectures viewers accessed | 45% / 60% | Hillsley, FEBS Open Bio 2025 |
| Views on exam day or the day before | 54.5% | Hillsley, FEBS Open Bio 2025 |
| Views on lecture day or the day after | 2.9% | Hillsley, FEBS Open Bio 2025 |
| Students very likely to watch a missed lecture / within a week | 78% / 33% | Voelkel et al., FEBS Open Bio 2023 |
| Students re-watching attended lectures start to finish / partially | 39% / 26% | Voelkel et al., FEBS Open Bio 2023 |
| Average share of each recording viewed (two time points) | 23.7% / 18.1% | Dommett et al., IJETHE 2022 |
| Echo360 views, spring 2019 vs spring 2020 | 6,316 vs 9,135 (nearly 45% higher) | Hussain et al., Cureus 2024 |
Context note: in Voelkel’s focus groups, students said recordings took ‘around one quarter of my whole study time’, with one describing a 45-minute lecture that took ‘like two hours to get through’ once pausing and note-taking were included, a habit explored in our student note-taking statistics.
4. Playback Speed: How Many Students Speed Up
Speed control has quietly become one of the most used features of lecture platforms. In a UK module with captions and transcripts, 57.9% to 63.2% of surveyed students used the speed adjustment function, close to the 60.5% to 68.4% who used the time bar to navigate and the 71.1% to 81.6% who paused and restarted (Dommett et al., International Journal of Educational Technology in Higher Education 2022). Bookmarking, a feature designed for study, was used by only 5.3% to 10.5%.
Among medical students the habit is stronger. About 77% of participants in a 2024 study usually watched lectures at 1.5x speed or faster, to the point that the author used 1.5x as the proxy for normal speed (Nakhostin-Ansari, BMC Medical Education 2024). The time math explains why: one hour of lecture at 1.5x saves 20 minutes (Song et al., Western Journal of Emergency Medicine 2018). In a pre-registered MOOC experiment, students randomly assigned to 1.25x were more likely to get better grades, attempt more content and earn more certificates, while spending less time on videos (Lang et al., Learning Analytics and Knowledge 2020).
| Metric | Value | Source |
|---|---|---|
| Students using speed adjustment | 57.9% to 63.2% | Dommett et al., IJETHE 2022 |
| Students using pause / restart | 71.1% to 81.6% | Dommett et al., IJETHE 2022 |
| Students using the time bar to navigate | 60.5% to 68.4% | Dommett et al., IJETHE 2022 |
| Students using bookmarking | 5.3% to 10.5% | Dommett et al., IJETHE 2022 |
| Medical students usually watching at 1.5x or faster | About 77% of participants | Nakhostin-Ansari, BMC Medical Education 2024 |
| Time saved per hour of lecture at 1.5x | 20 minutes | Song et al., WestJEM 2018 |
| MOOC learners assigned to 1.25x vs 1.0x | Better grades, more content attempted, more certificates | Lang et al., LAK 2020 |
Context note: Dommett’s ranges reflect two survey time points, and only 42 students completed both, so treat the percentages as directional. How fast a recording can be pushed also depends on how quickly the lecturer talks at 1x.
5. Comprehension at Speed: What the Experiments Show
The lab evidence converges on a clear ceiling. Murphy et al. found minimal costs from raising lecture speed from 1x to 1.5x or 2x, but performance declined beyond 2x, on both immediate tests and tests one week later (Murphy et al., Applied Cognitive Psychology). Watching twice at 2x gave no advantage over once at 1x, except when the second 2x viewing came immediately before the test. The time saved only pays off if it is spent studying close to the exam.
Results diverge below that ceiling. In a Chinese study of 76 undergraduates, post-test scores peaked at 1.25x (80.00) and bottomed at 2.0x (66.47), while cognitive load rose steadily from 38.71 at 1.0x to 47.94 at 2.0x (Mo et al., Frontiers in Psychology 2022). Among 60 Turkish medical students, 2x did not significantly impair memory: immediate scores were 11.26 at 1x versus 10.16 at 2x, and delayed scores 9.66 versus 8.36 (Kiyak et al., Medical Teacher 2023). Dense content is less forgiving: on one of two emergency medicine videos, comprehension fell from 72.7 at 1.0x to 61.4 at 1.5x (p = 0.0188), while the other showed no significant drop (Song et al., 2018).
| Metric | Value | Source |
|---|---|---|
| Speed range with minimal comprehension cost | 1x to 2x; decline beyond 2x | Murphy et al., Applied Cognitive Psychology |
| Post-test score at 1.0x / 1.25x / 1.5x / 2.0x | 72.35 / 80.00 / 76.43 / 66.47 | Mo et al., Frontiers in Psychology 2022 |
| Cognitive load at 1.0x vs 2.0x | 38.71 vs 47.94 | Mo et al., Frontiers in Psychology 2022 |
| Best speed for high- / low-ability students | 1.5x (83.33) / 1.25x (76.50) | Mo et al., Frontiers in Psychology 2022 |
| Immediate recall, 1x vs 2x (60 medical students) | 11.26 vs 10.16 (not significant) | Kiyak et al., Medical Teacher 2023 |
| Delayed recall, 1x vs 2x | 9.66 vs 8.36 (not significant) | Kiyak et al., Medical Teacher 2023 |
| Comprehension, 1.0x vs 1.5x (artifacts video) | 72.7 vs 61.4 (p = 0.0188) | Song et al., WestJEM 2018 |
| Comprehension, 1.0x vs 1.5x (transducers video) | 73.8 vs 66.9 (not significant) | Song et al., WestJEM 2018 |
Context note: these are short, single-session experiments with small samples (54 to 76 students in the speed studies above), and Song’s 1.5x result is the main outlier. Most recent available data for the canonical speed studies: Murphy et al., 2021 to 2023.
6. Who Pays for Speed: Age, Attention and Grades
Speed is not neutral across learners. Across three experiments with 1,024 participants (148 + 178 + 150 + 176 + 173 + 199), Murphy and colleagues found that younger adults can watch at faster speeds without significant memory deficits, while older adults’ test performance is generally impaired (Murphy, Hoover and Castel, Memory 2023). In the third experiment, mean test performance was .59 at 1x and .51 at 2x, with the authors reporting that older adults’ performance decreased at the faster speed. Faster playback also reduced mind-wandering, which may be why younger learners keep up.
Whether recordings help grades depends on how they are used. Of the 71 studies in Banerjee’s review, 39 (54.93%) found a positive impact on performance and only 2 a negative one. Hillsley found each extra hour of viewing linked to a 1.74% higher exam score, yet within-student cramming hours were associated with a 0.84% lower score. When recordings replace live teaching entirely, the cost can be large: surgery clerkship students taught by recorded lectures scored 71.1% on weekly quizzes versus 88.8% for live teleconference lectures (Huerta et al., Journal of Surgical Education 2023).
| Metric | Value | Source |
|---|---|---|
| Participants in Murphy video speed experiments (younger / older) | 148 / 178, 150 / 176, 173 / 199 across three experiments | Murphy, Hoover and Castel, Memory 2023 |
| Mind-wandering rate, younger vs older adults (Experiment 2) | .42 vs .19 | Murphy, Hoover and Castel, Memory 2023 |
| Studies finding a positive / negative performance effect | 39 (54.93%) / 2 | Banerjee, Computers and Education 2021 |
| Exam score, recording viewers vs non-viewers | 66.1% vs 59.4% | Hillsley, FEBS Open Bio 2025 |
| Exam score change per extra viewing hour (between / within student) | +2.52% / -0.84% | Hillsley, FEBS Open Bio 2025 |
| Weekly quiz score, recorded vs live teleconference lectures | 71.1% vs 88.8% (312 students) | Huerta et al., Journal of Surgical Education 2023 |
| NBME surgery shelf exam, recorded vs live | 69.5% vs 75.7% | Huerta et al., Journal of Surgical Education 2023 |
| Engagement with mini-lectures vs internal medicine grade | r = 0.35 (p = 0.044) | Benamer and Stanley, BMC Medical Education 2023 |
Context note: none of the grade studies isolate playback speed, and Huerta’s cohorts were taught in different academic years (2020-2021 live, 2021-2022 recorded). OSCE results showed no significant difference between the two groups (p = 0.946). In the Benamer study, mini-lectures averaged 11.1 minutes (BMC Medical Education 2023).
7. Staff Attitudes and Accessibility
Faculty resistance is softening, mostly because recording has become routine rather than because staff were persuaded it helps. At Liverpool, the share of academics who felt negative about being recorded fell from 35% in 2019 to 14% in 2023, and those finding it stressful fell from 29% to 7% (Voelkel et al., Emerging Topics in Life Sciences 2026). Belief that it benefits students barely moved, from 29% to 36%, and among staff who see capture as detrimental, attendance remains the main fear (60% in 2019, 66% in 2023).
Accessibility is where recordings do work that a live lecture cannot. Among osteopathic colleges, 71% offer transcripts, captions or both, but 29% offer neither, and almost all rely on automatic speech recognition: 83.3% of transcript providers and 92.3% of caption providers use the ASR built into the lecture capture system, and only 6.3% review the output before release (Khong et al., 2025). Students with ADHD describe recorded lectures as giving them control over pace, place and time (Levenberg and Reesh, Journal of Attention Disorders 2023, 12 students). Caption quality is covered in more depth in our captions and subtitles statistics.
| Metric | Value | Source |
|---|---|---|
| Staff believing lecture capture benefits students | 29% (2019) / 36% (2023), not significant | Voelkel et al., ETLS 2026 |
| Staff feeling negative about being recorded | 35% (2019) / 14% (2023) | Voelkel et al., ETLS 2026 |
| Staff finding lecture capture stressful | 29% (2019) / 7% (2023) | Voelkel et al., ETLS 2026 |
| Detractors citing attendance as main concern | 60% (2019) / 66% (2023) | Voelkel et al., ETLS 2026 |
| Osteopathic colleges offering transcripts or captions | 71% (33% both, 24% captions only, 14% transcripts only) | Khong et al., Medical Science Educator 2025 |
| Caption providers using built-in ASR captions | 92.3% | Khong et al., Medical Science Educator 2025 |
| Colleges reviewing transcripts or captions before release | 6.3% | Khong et al., Medical Science Educator 2025 |
Context note: in a randomized comparison, adding captions did not significantly change MCQ scores (p = 0.490), though no students rated captions and transcripts as not at all useful (Dommett et al., 2022). The Liverpool staff samples were 58 (2019) and 44 (2023). For the wider open course context, see our MOOC statistics.
Summary: Lecture Capture & Playback Speed by the Numbers
| Metric | Value | Source |
|---|---|---|
| Osteopathic medical colleges recording lectures | 95% | Khong et al., 2025 |
| PA programs offering recordings | 49.6% | Herrick and Stoehr, 2025 |
| US undergraduates in any distance course | 61% (fall 2021) | NCES Fast Facts |
| Studies finding negative attendance effect | 28.17% (20 of 71) | Banerjee, 2021 |
| Studies finding positive performance effect | 54.93% (39 of 71) | Banerjee, 2021 |
| Attendance before / after capture | 83% / 81% | McLean and Suchman, 2016 |
| Third semester medical students almost never attending | 43% | Gardner et al., 2022 |
| Students accessing at least one recording | 72.2% | Hillsley, 2025 |
| Views on exam day or the day before | 54.5% | Hillsley, 2025 |
| Views on lecture day or the day after | 2.9% | Hillsley, 2025 |
| Students watching a missed lecture within a week | 33% | Voelkel et al., 2023 |
| Students using speed adjustment | 57.9% to 63.2% | Dommett et al., 2022 |
| Medical students usually at 1.5x or faster | About 77% | Nakhostin-Ansari, 2024 |
| Comprehension cost up to 2x speed | Minimal; declines beyond 2x | Murphy et al., Applied Cognitive Psychology |
| Best post-test speed | 1.25x (80.00 vs 72.35 at 1.0x) | Mo et al., 2022 |
| 2x vs 1x immediate recall | 10.16 vs 11.26, not significant | Kiyak et al., 2023 |
| Exam score per extra viewing hour | +1.74% | Hillsley, 2025 |
| Recorded vs live teleconference weekly quizzes | 71.1% vs 88.8% | Huerta et al., 2023 |
| Staff finding capture stressful | 7% (2023), down from 29% | Voelkel et al., 2026 |
| Caption providers relying on ASR | 92.3% | Khong et al., 2025 |
Methodology and Sources
Every figure above was read during research for this article, from full-text open-access papers on PubMed Central or from the published abstracts of the original studies. Vendor marketing statistics (lecture capture platform blogs and ‘state of video’ reports) could not be traced to a readable primary document and were excluded, as were market-size estimates for the lecture capture industry.
- Murphy, Hoover, Agadzhanyan, Kuehn and Castel: Learning in double time, Applied Cognitive Psychology (abstract read via the Semantic Scholar record; publisher page blocked automated access)
- Murphy, Hoover and Castel: The effect of video playback speed on learning and mind-wandering in younger and older adults, Memory 2023
- Mo et al.: Video playback speed influence on learning effect, Frontiers in Psychology 2022
- Song et al.: Does the podcast video playback speed affect comprehension, Western Journal of Emergency Medicine 2018
- Nakhostin-Ansari: Does lecture playback speed not affect memory and concentration in medical students?, BMC Medical Education 2024
- Kiyak, Budakoglu, Masters and Coskun: The effect of watching lecture videos at 2x speed on memory retention, Medical Teacher 2023 (abstract via Europe PMC)
- Lang, Chen, Mirzaei and Paepcke: Is faster better? A study of video playback speed, Learning Analytics and Knowledge 2020 (abstract via Semantic Scholar)
- Banerjee: To capture the research landscape of lecture capture in university education, Computers and Education 2021
- Edwards and Clinton: A study exploring the impact of lecture capture availability and usage on attendance and attainment, Higher Education
- Nordmann, Calder, Bishop, Irwin and Comber: Turn up, tune in, don’t drop out, Higher Education (both Higher Education abstracts read via Semantic Scholar; the publisher pages required sign-in)
- Hillsley: Exam-level analysis of lecture capture viewing and student exam performance, FEBS Open Bio 2025
- Voelkel et al.: Lecture capture affects student learning behaviour, FEBS Open Bio 2023 and staff perceptions before and after COVID-19, Emerging Topics in Life Sciences 2026
- Dommett et al.: Effects of captions, transcripts and reminders on lecture capture, IJETHE 2022
- McLean and Suchman: Video lecture capture technology in large microbiology lecture courses, JMBE 2016
- Emahiser et al.: Study of live lecture attendance, Medical Science Educator 2021; Gardner et al.: Determinants and outcomes of in-person lecture attendance, Medical Science Educator 2022
- Khong et al.: Lecture capture, transcripts and captioning in US colleges of osteopathic medicine, Medical Science Educator 2025; Herrick and Stoehr: Lecture recording and mandatory attendance policies in PA education, JPAE 2025 (abstract via Europe PMC)
- Hussain, Al-Mannai and Agouni: Video lecture capture in pharmacy education, Cureus 2024; Huerta et al.: Recorded versus live teleconference lectures, Journal of Surgical Education 2023; Benamer and Stanley: Pre-recorded neurology mini-lectures, BMC Medical Education 2023; Levenberg and Reesh, Journal of Attention Disorders 2023 (abstract via Europe PMC)
- NCES: Fast Facts, Distance Learning
- Data watch: Most lecture capture research comes from medical, pharmacy and life sciences programs (Biological Sciences is 40.84% of Banerjee’s sample, and US, UK and Australian studies are over 70%), so results may not transfer to humanities or large general-education courses. The Murphy et al. Applied Cognitive Psychology paper is sometimes cited as a 2022 Psychological Science study; it appeared in Applied Cognitive Psychology, and its Semantic Scholar record lists 2021. EDUCAUSE, Jisc, Panopto and Kaltura pages blocked automated access or returned errors, so no student-survey figures from those organizations are included; the next EDUCAUSE Students and Technology Report and Jisc Student Digital Experience Insights survey are the most likely sources of newer national data. The Nakhostin-Ansari 77% figure comes from a letter that does not report its sample size. Hillsley reports the viewer advantage as 6.2%, while the two raw means (66.1% and 59.4%) differ by 6.7 points; we show the raw means. Speed studies are mostly 2018 to 2023 (most recent available data), and the NCES distance education figures are for fall 2021.
Last updated: October 3, 2026. We update this roundup quarterly, and the next refresh is expected when EDUCAUSE publishes its next Students and Technology Report and NCES releases newer IPEDS distance education enrollment data.