Coursera Iteration 1 – Survivor analysis

The slope of the log survival function is a measure of the probability the transition from one state to another; for our purposes here it is an indication of the probability of withdrawal from the course.  In passing, the definition of “withdrawal” here has simply to be the point at which the participant last connected to the course website, and can say nothing about any ongoing engagement with “the course” by other routes or resources.

In general terms then, if the curve were convex, then the probability of withdrawal, for any individual participant still active on the course would be increasing as time went on. A concave curve suggests that the longer a person persists on the course, the longer they are likely to persist. The probability of withdrawal decreases with time into the course.

Key: BLUE line signifies the course content end, ie. 5 or 7 weeks after launch. RED line signifies final course deadline, ie. the point after which submission of assessment would not contribute towards successful completion and potential award of a Statement of Accomplishment – a dotted red line signifies the first stage deadline of a peer assessment final exam.


From left to right: Equine Nutrition, AI Planning, Astrobiology, Critical Thinking, Introduction to Philosophy and E-learning and Digital Cultures (EDC)

sa-eqnut1 sa-aip1 sa-astro1 sa-crit1
sa-introphil1 sa-edc1


From left to right: Equine Nutrition, AI Planning, Astrobiology, Critical Thinking, Introduction to Philosophy and E-learning and Digital Cultures (EDC)

sa-aip2 sa-eqnut2 sa-astro2sa-crit2 sa-introphil2 sa-edc2


Breakdown by course


Equine Nutrition

sa-eqnut1 sa-eqnut2

There is quite a strong suggestion of an “S-Shaped” (sigmoid) curve here for Equine Nutrition. That is, noting the 5 week duration of the course, the section of the curve up until the end of the 5 week period is quite distinctly concave, suggesting that people remaining with the course as time passed would be increasingly likely to stay until the end.  Retention is strong. There is then a downturn in the curve after the five week period suggesting (not surprisingly) that people  start to fall away increasingly after the end of the course, as they complete and submit (or do not submit) the final assessment.

The more concave the curve, the more that one might characterise the course as “sticky” – the more it can be seen to retain students once they choose to commit. That would seem to be in keeping with what we know about the Equine Nutrition course.


AI Planning

sa-aip1 sa-aip2

This curve here is pretty flat. This would suggest that the probability of withdrawal is fairly constant over the life of the course, and rises at the end of the course, as the period of the final assessment is entered, and passes.



sa-astro1 sa-astro2

Again here we have a clearly sigmoidal curve. The 5-week period of the course is characterised by concave shape, suggesting greater commitment on the part of participants as the course unfolds. Again, there is an increase in the probability that people will fall away after the end of the course, and drops more acutely as the end point of the final assessment period passes.


Critical Thinking


Again, a relatively flat curve over the duration of the course, then a rapid falling off in attention directed to the course after the end of the 5-week period.

The final deadline for the course was Monday 11th March (day 43) and there is a very rapid drop over days 40 – 45, suggesting many made their final submission (or final checks) over the weekend period and did not revisit the site after this event, perhaps due to the nature of the course being more designed for individual rather than community development.



Introduction to Philosophy

sa-introphil1 sa-introphil2

The shape of the log survival function for the Philosophy course suggests a slowing of the rate of withdrawals after the early days of the course, and a shallow slope over the rest of the course (in this case, seven weeks) showing strong retention.  The probability of withdrawal then increases across the period of final assessment submission, and increases rapidly at the end of that period.


E-Learning and Digital Cultures

sa-edc1 sa-edc2

Here again we see an early drop-off in participation followed by a relatively steep falling off in participation that is relatively steady over the period of the course and beyond.  The final assignment for this course fell due during the final (fifth) week of the course, and there then followed a period in which participants were asked to provide feedback on the work of their peers.  There thus existed an extended period beyond the end of that fifth week which demanded the engagement and interest of participants while assessment outcomes were being decided and declared.  This perhaps explains why there is no clearly synchronised downturn in retention at that point; the activity on the course tailed off, rather than ending abruptly.


Creative Commons License
This work, unless otherwise expressly stated, is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Leave a Comment

Your email address will not be published.