Early assessment submission and course outcomes among students of an open online university

Introduction: Students who withdraw or fail online courses often disengage from graded work early. It is uncertain whether early submission of continuous assessments predicts completion and passing, and whether this depends on prior study experience. Methods: Retrospective cohort of registrations in the Open University Learning Analytics Dataset (presentations from 2013 and 2014). Students still registered at day 60 were classified by submission of at least one tutor-marked or computer-marked assessment by that day. Logistic regression with module-presentation fixed effects gave adjusted odds ratios and standardized risk differences, with interaction tests by previous attempts and education. Results: The cohort comprised 23,943 students, 91.8% of whom submitted early. Completion was 85.3% with and 72.6% without early submission (adjusted odds ratio 1.30, 95% CI 1.10-1.53); passing was 62.4% and 7.9% (adjusted odds ratio 4.06, 95% CI 3.36-4.89). The adjusted risk difference for completion was larger with previous attempts (6.9 vs 2.9 percentage points) and rose with education. At a 90-day landmark, the association with completion included no effect. Conclusions: Early submission was strongly associated with passing and modestly with completion, and both varied with prior study experience. Because assessments count towards the mark, early submission is better used as a signal for outreach than as a causal lever.