Source-linked AI summary
Effects of a Behavioural Commitment Scheme on Study Regularity in a Self-Paced Learning Platform
Meenakshi V., Pavani Ayinampudi, Aditya B. M. V., Jinal Gupta, Prakash Hegade, Rohit Sharma, Sakshi Sharma, S. R. S. Iyengar
TL;DR
Self-paced courses offer limited scaffolding for regular study, and their unrestricted access makes platform demand hard to forecast. The paper makes booking a study window the verified default route into the course, so the same commitment can structure study and inform capacity planning. Among committed learners, regularity improved, while the authors limit the effect claim to this self-selected group and study setting.
Problem
Self-paced courses lack fixed hours, visible peers, or an instructor who notices absence, while unrestricted study makes platform demand difficult to forecast.
Source: §1 · p. 2Method
Learners book fixed two-hour study windows, which gate course entry, are checked against watch telemetry, and generate a booking-based demand forecast.
Source: §1 · p. 2, §3.3 · p. 5Results
86.8% of bookings were honoured, and committed learners’ regularity rose from 1.01 to 1.33 active days per week.
Source: §7 · p. 13Takeaways & Limitations
The booking log made demand forecastable while the binding, verified, default scheme coincided with improved regularity among committed learners.
Source: §7 · p. 13The quasi-experimental effect claim applies to 317 self-selected committers in one course and programme over 25 days, not the whole cohort.
Source: §6.5 · p. 13
Abstract
from arXiv · showhide
Online education has enabled learners worldwide to take up courses from reputed institutions. However, a self-paced online course cannot guarantee the motivation and engagement that a learner experiences in a real-time classroom. Self-paced access also makes platform load unpredictable, which drives up the compute cost. We propose a Commitment Scheme for course access that aligns learner commitment with platform capacity. Learners book their study slots in advance; the instructor sets the budget of learning hours available for the course; and learners who use a full window earn additional watch hours. A booked window records an intention to study at a stated time, and a learner who appears in that window implements it. The byproduct is a platform load that can be forecast and bounded. The study involved two courses taken in sequence by the same learners, with the slot booking system activated only in the second. In-window study was observed on 86.8% of booked windows, and the median committer placed 95.5% of all study time inside self-booked windows. Among learners who studied across the launch, study regularity improved from 1.01 to 1.33 active days per week, with a supporting difference of +1.37 days per week against the same learners' preceding course. Commitments made on the same day as the study slot were honoured more often than advance bookings (88.8% against 62.5% two days ahead), which is consistent with the classic intention-behaviour gap.
1 Introduction
The paper addresses weak study follow-through and unpredictable platform demand with a binding, verified booking scheme that also provides a capacity forecast. It evaluates the scheme after a mid-course launch, reporting improved regularity among committed learners while bounding claims to a quasi-experimental design.
Motivation
- Self-paced courses lack classroom scaffolding, while completion records clicks rather than disciplined, distributed study; unpredictable demand also makes reactive autoscaling costly.The paper frames the intention-behaviour gap as a reason sincere advance plans can be abandoned, and notes that optional scheduling has had uneven or null results.Source: §1 · p. 2
Proposed scheme
- The scheme makes a current booking the default access route, verifies in-window activity, sets an instructor-defined hours budget, and rewards fully used windows with same-day bonus bookings.The booking log also supplies a load forecast up to two days ahead.Source: §1 · p. 2
Evaluation
- The live-course evaluation used a mid-course launch to support within-learner and same-learners-across-courses comparisons, but adoption was self-selected and the design quasi-experimental.Claims are scoped accordingly.Source: §1 · pp. 2–3
Results and contributions
- 86.8% of 1,862 bookings were honoured, and committed learners’ regularity rose from 1.01 to 1.33 active days per week, with a +1.37-day difference-in-differences against their preceding course.Booked seats explained 59% of variance in concurrent load and bounded its peak, though cost savings remain a projection.Source: §1 · p. 3
2 Related Work
Related work frames scheduling as a response to intention-behaviour gaps, distinguishes study regularity from volume, and connects learner bookings to appointment and cloud-capacity planning.
2.1 The Intention-Behaviour Gap, Implementation Intentions, and MOOC Scheduling
- Binding, verified, default scheduling occupies a gap left by prior MOOC interventions, including a randomized invitation to pre-schedule that did not improve persistence.Implementation intentions specify when and where to act; prior work also reports gains from concrete study plans and self-imposed devices.Source: §2.1 · p. 3
2.2 Habit, Regularity, and Onset Timing
- Study-time regularity measures consistency rather than volume and is the paper’s primary outcome; new behaviour may install more easily at a course’s temporal landmark.The discussion contrasts habit formation through repetition in stable contexts [30–32] with retrofitting an established routine.Source: §2.2 · pp. 3–4
2.3 The Operations Bridge: Appointments and Cloud Capacity
- A reservation log can serve as both an appointment-book analogue for estimating show-up and no-show rates and a declared-demand schedule for capacity planning.The cited operations literature links lead time to no-shows and identifies workload-prediction accuracy as a bottleneck for proactive autoscaling.Source: §2.3 · p. 4
3 The Slot-Booking Commitment Scheme
The scheme turns a learner’s booked study window into a verified commitment and a per-window demand schedule. It combines a booking-based access gate, a finite hours budget, fulfilment checks, and a same-day reward.
3.1 Platform Context
- ViBe is a MERN-stack learning platform whose video player emits watch-telemetry records about every 15 seconds of active playback.The studied course generated roughly 400,000 sessions, and the backend used reactive autoscaling by default.Source: §3.1 · p. 4
3.2 Mechanics and Their Rationale
- A learner books one of seven daily two-hour windows, with reservations available from two days before the study day until that day.The booking carries a timestamp recorded before the intended study time.Source: §3.2 · p. 4
- Course access opens only during a learner’s booked window, so the gate governs when study begins rather than when it ends.The booking is the required access path, not an optional planning tool.Source: §3.2 · p. 5
- An instructor-set account limits bookable hours; telemetry evaluates fulfilment, and a window counted as fulfilled required activity for at least 90% of its duration.A fulfilled window grants one extra booking usable the same day.Source: §3.2 · p. 5
3.3 The Byproduct, by Construction
- The bookings table records learner, study date, and window, creating a demand schedule up to two days ahead for forecasting concurrent compute load.Summing bookings by day-window cell gives the declared number of learners intending to study in that interval.Source: §3.3 · p. 5
4 Study Design and Method
The study uses a mid-course deployment and exposure-aware cohorts to evaluate booking behaviour and study regularity. It compares committed learners with their own earlier activity and uses the same learners’ preceding course as supporting evidence.
4.1 Setting and Scale
- 946 learners enrolled in the MERN course, and 317 distinct committers made 1,862 evaluable bookings during 25 observed study days.The scheme launched 37 days into the course; 317 committers represented 34% of active enrolment.Source: §4.1 · p. 5
4.2 Exposure Cohorts: The Mid-Course Launch as Instrument
- Learners were divided by exposure: C1 provides baseline behaviour, C2 is the primary within-learner cohort, and C3 is the robustness cohort.C1 was active only before launch; C2 adopted booking across the launch; C3 first appeared on or after launch. The study does not compare completion across cohorts.Source: §4.2 · pp. 5–6, Table 1 · p. 6
4.3 The Control Course
- The preceding Fundamentals of AI course provides a same-learner comparison for 931 learners, but was already 97.7% complete when booking launched.Its difference-in-differences estimate is treated as supporting evidence, with within-learner change in the committed course as the primary estimate.Source: §4.3 · p. 6
4.4 Measures
- Measures use raw watch telemetry rather than stored fulfilment labels; regularity is active study-days per week, and in-window activity figures are lower bounds because only video-player activity is captured.Regularity is the primary outcome because daily binary measures avoid duration double-counting and reflect day-spacing; show-up requires more than one minute of in-window activity.Source: §4.4 · pp. 6–7
4.5 Ethics
- Analyses use de-identified, pseudonymised telemetry and read-only access to the production database under programme data-use provisions.Source: §4.5 · p. 7
5 Results
Across the results, bookings were frequently honored and study regularity rose among learners who adopted the scheme; booked demand also tracked and bounded actual course load.
5.1 R1: Learners Kept the Commitments They Made
- 86.8% of 1,862 booked windows had in-window study, 54.9% sustained at least half-window activity, and median committers placed 95.5% of study time within booked windows.Source: §5.1 · p. 7, Table 2 · p. 8
5.2 R2: Study Regularity Improved among Committed Learners
- Among 180 C2 learners, regularity rose from 1.01 to 1.33 active days per week at booking onset; the DiD against their preceding course was +1.37 active days per week (median +1.22).Across all 810 shared learners, the DiD was approximately zero (+0.04, median −0.19); the effect claim is limited to 317 self-selected committers.Source: §5.2 · pp. 7–8, Table 3 · p. 8
5.3 R3: Follow-Through Decayed with the Distance between Intention and Action
- 88.8% of same-day bookings were honored versus 62.5% two days ahead, with follow-through declining monotonically across lead times (z = −6.8, p < 0.001).The one-day-ahead rate was 73.6%.Source: §5.3 · p. 8, Table 4 · p. 9, Figure 1 · p. 9
5.4 R4: Commitment from the Outset Outperformed Mid-Course Adoption
- From-start C3 learners exceeded mid-course C2 learners on show-up (92.1% vs 82.5%) and mean active share (55.0% vs 41.6%); the comparison is observational.The cohorts differed in journey stage and self-selected by later enrolment.Source: §5.4 · p. 9, Table 5 · p. 9
5.5 R5: The Booking Schedule Was a Measured Forecast of Actual Load
- Booked seats correlated with within-window peak concurrency (r = 0.77, R2 = 0.59); actual deployment peak concurrency was 25, within the booked-seat peak of 28.The forecast is course-scoped, not a measured infrastructure saving; provisioning remained fixed-size (35.3 vs 35.7 instance-hours/day, pre vs post).Source: §5.5 · pp. 9–10, Table 6 · p. 10, Figure 2 · p. 10
6 Discussion
The discussion interprets binding, verified, default scheduling as structural commitment, while timing and adoption shape who follows through. It treats load forecasting as promising but infrastructure savings and learning gains as unmeasured, with causal and scope limits.
6.1 Why Binding Scheduling Succeeded Where Nudged Scheduling Failed
- Binding entry, telemetry verification, and default uptake jointly distinguish this scheme from nudged scheduling, but their separate effects cannot be isolated.The comparison with the prior nudge result is interpretive rather than a controlled head-to-head; the authors identify a randomized binding-versus-nudge trial as the next experiment.Source: §6.1 · p. 11
6.2 Timing, Habit, and Who the Scheme Reached
- Booking functioned as a near-immediate start ritual, not advance planning, and C3 learners out-adhered mid-course adopters on every volitional measure.Mid-course adopters moved from below the non-adopting population baseline (1.01 vs 1.29 active days/week) to above it; the from-start advantage remains hypothesis-grade because cohort and course-position factors covary.Source: §6.2 · p. 11
6.3 The Bonus Loop, Autonomy, and the Ethics of Binding Design
- 28 learners earned 46 bonus windows, exceeding base bookings on show-up (91.3% vs 86.7%) and active share (58.9% vs 47.4%).The comparison is descriptive because bonus-window recipients were already engaged; fairness and effects on motivation beyond 25 days remain open questions.Source: §6.3 · p. 12
6.4 The Booking Log as an Appointment Book: Operations Implications
- Booked seats support an effective-load estimate of roughly 0.5–0.6× booked seats, but infrastructure savings remain projected.The measured envelope peaked at 25 within 28 seats, with 11.1% of demand invisible to the forecast, mostly overnight spill-over.Source: §6.4 · p. 12
6.5 Limitations and Threats to Validity
- The quasi-experimental effect applies to 317 self-selected committers, not the full cohort, and remains vulnerable to time-varying selection and course-position effects.The evidence comes from one committed course, one programme, and 25 days; the study measures regularity and in-window engagement, not completion or learning outcomes.Source: §6.5 · pp. 12–13
7 Conclusion
Booking study windows served both learner commitment and platform capacity planning: commitments were often honoured and study regularity improved, while booking schedules made concurrent load more forecastable. The authors identify learning outcomes, causal evidence, projected provisioning savings, and learner perspectives as priorities for future work.
7 Conclusion
- Among committed learners, study regularity improved from 1.01 to 1.33 active days per week; 86.8% of booked two-hour windows were held, and median 95.5% of study fell inside self-chosen windows.Source: §7 · p. 13
- The booking schedule explained 59% of variance in concurrent load and bounded its peak, making the same learner action useful for capacity forecasting.The schedule was known up to two days ahead.Source: §7 · p. 13
- Future work will connect kept commitments to quiz performance and learning gain, compare binding commitments with nudges randomly, and gather learner views on autonomy and fairness.The authors also plan to measure provisioning savings, which have so far only been projected, and test commitments visible to peers.Source: §7 · p. 13