Updated: March 7th 2009
Since my previous post on this subject last week I’ve been digging a little more and received some great information both from you at other institutions and from our friends at DEEWR Higher Education Data Collection to whom I am most grateful.
Since there is so much detail I’ll just tackle Attrition in this post. The offical explanation from DEEWR is what follows:
DEEW calculate the attrition rate by processing an annual baseline enrolment dataset, a completion dataset and then an enrolment dataset for the subsequent year. I think they must also consider a completions dataset for the subsequent year too but they don’t specifically say this. Sof looking at retention in 2008 for 2007 enrolments, the approach is something like this:
If a student from the baseline file is not on the subsequent enrolment file or completions file they are classified as ‘attrited’. Some further filtering of the baseline dataset is performed by DEEWR for their own calculations so that it only includes:
- Bachelor-level students
- Commencing students
- Sole/Major course
DEEWR aren’t concerned about matching students in particular courses or awards, they just look for the same student with the same provider. They acknowledge the limitations of this but due to so many students graduating in an entirely different course from that which they were enrolled in they have no other option.
As UNE has a lot of distance education students who study part-time, this definition of attrition doesn’t make us look too good as many of our students elect to have a year out or complete their Bachelor-level study over more than the typical 3 years. DEEWR clearly recognise this when they say:
When we calculate attrition rates, we do just compare the year immediately subsequent to the baseline year. Although we are aware that a limitation of this is that students on program leave will be included as attrited students, we do not have any way of identifying them.
So there you have it, ATTRIT101 in all its glory. Next up is Progress and then Retention.
Related Posts:
A couple of years ago I had cause to analyse the legendary ‘long tail’ of students at UNE.
Yes, it was a long tail. But to my surprise it was not very thick.
With Rob’s help in the days of proto-BI I obtained figures showing for each student enrolled in a unit in semester X (sem 2 2005?) the year of their first enrolment in the course they we still studying in semester X.
There were a very few from the 70s, maybe a couple of dozen from the 80s (all from the 70s and 80s were in the BA) and (big surprise) those from the nineties only counted in the hundreds, and most of those from the late nineties at that. By a very wide margin most students enrolled in semester X had commenced in 2000 or later.
And yet the folklore was strong about UNE having vast numbers of students who take a very long time to finish a degree . Maybe this was because the few with inordinately long records also have inordinately thick files and loomed large in our minds (and on the desk of a Director of Student Admin!). Also, I think, it used to be the case that the long tail really was thicker.
As student preferences have shifted towards ‘meal-ticket’ degrees two incentives are in play:
* quicker progression means a quicker meal
* HESA 2003 made it clearer than before that HECS-HELP is a loan: you will pay it back, so minimise the government share of the meal if you can.
Of course, the figures I looked at only showed that most students who had not attritted were getting through their degrees faster than we had thought. But I wonder if the undoubted fact that the long tail is not so thick means that the DEEWR method for measuring attrition not as far off the mark for UNE as Rob fears?
Stand by Rick I aim to provide evidence to settle this matter - you see nowadays we have such a thing but note that a one year gap, for program leave, work experience or just for family reasons, would in DEEWR’s view show a lost student, even if they return the next year so this particular issue has nothing to do with a long tail, or even a medium sized tail.
Point taken, although I think it may be the case that if students are taking fewer years to complete overall then they have fewer instances of ‘program leave’ also. In an ideal world students would say “I’m just taking a semester or two or three off but I’ll be back’, so one could set a flag.
I recall graduating a student who resumed her BA in about 2001 having last enrolled in 1968! I would have thought her prior studies might have been a bit stale but the Faculty didn’t object. As I said in another context, a student is a student is a student …. and therefore load is load is load ….
Does it matter at all in this BI gathering why the student may ‘atrit’? What happens with the attrition data collected, other than inform DEEWR how much to pay the university? Do we gather any BI on the quality of the experience? BTW the unit evaluation forms do not gather ‘quality’ data than can assist to create a better learning environment. DE design in the ‘new learning’ environment requires much more targeted data collection.
Thanks for the comment Cherry, It of course does matter very much why the student decided to attrit but how do we gather that vital information? We have had some early discussions on putting an extra step into the online portal where a student notifies us of their discontinuation. This doesn’t alert us to the people who simply go missing in action however. The new unit evaluation process being launched for Semester 1 this year may well provide us with more data and therefore information and insight into student perceptions and experiences at the unit level but this data is still summative. The e-Motion data we gather is perhaps more telling and is much more of a dynamic and real-time indication of current student experience. Is there other data we should be collecting? I’m very interested in this topic and happy to discuss more.
@Rick Nelson
I just found something really interesting - http://www.wired.com/wired/archive/12.10/tail.html. It seems that, in the digital world, the long tail is actually a good thing. Ironic that it is the thing many are seeking to shorten while we make further strides into the digital marketplace where publication, distribution and marketing costs are plummeting and the differentiator is diversification… Really interesting stuff on here and Chris Anderson wrote it all in 2004
@Rob Hale
I agree with Cherry too. My comment about students = load was intended to point out that this equation is too often taken as the rationale for admission (and retention) policies, with sad personal consequences for individual students. Not to mention much red ink on assignments and in a university’s ledger.
@Rob Hale
Anderson’s work on the long tail hasn’t gone unchallenged by economists but I think there’s an obvious truth in his argument about low-cost inventories and distribution systems.
To relate his ideas to universities we would need to have a way of analysing the combination of a long tailed inventory (lots of courses and units) together with a long (or possibly medium or short) tail of customers.
A student who buys a course is also buying a certain number of credit points for successful completion. They do this by paying for the academic and administrative services that help attain those credit points. Often they buy a proportionately higher amount of services because they fail units and have to do them again or have other issues with which they need assistance. Where, I wonder, is the tipping point at which despite paying for extra services students’ costs to the university outweighs the income?
If universities have their pricing right they’ll stay ahead of the curve but this is difficult to actually show. Many other service providers cover their infrastructure costs by charging a fee for ’service availability’: electricity distributors and telcos are good examples. Then on top of that they charge per unit (of elctricity, data or whatever) consumed. It’s the same principle as the flag fall charge in a taxi, which is followed by a charge per kilometre travelled.
But at a university most charges associated with a course are bundled in the pricing of units. Since both teaching and administration are services the amount each customer consumes will vary according to their personal situation despite the fixed price they pay.
It would be truly useful to see measures that correlate per student academic performance, teaching effort and administrative cost. Since dollar values are often nebulous for these things one could start just by counting transactions.
As I’ve remarked before, this is not just about the bottom line of a university. If the subsidy from students who complete in the minimum time with minimum consumption of services drifts into too long a tail of students who need much more help then all students are disadvantaged because overall ‘quality’ falls.
But this seems a different ‘long tail’ issue to the inventory questions that Chris Anderson addresses. Are they related? After all, the inventory of courses and units is consumed by students. What if there’s a long tail of both?
to be open.. i really didnt understand most of the subject.. im really sorry.. i was just bored and was browsing and somehow landed on this page.. but it looks like its about some student enrollments of some university?