UCAS’ end of cycle statistics released last week are rightly a major event in the data year and will be heavily used by universities. But there is a growing problem.

The twilight zone of the UCAS data, records of prior acceptance (RPAs), are on the rise. Dramatically so. This year they account for pretty much all the headline growth in numbers. But RPAs don’t score highly for statistical integrity. Their rapid growth is generating a thickening fog of uncertainty around the data. The statistics remain key to understanding the sector, but users should drive carefully.

RPAs – a field guide

A quick test of whether someone knows their HE data is to ask them about RPAs. If you get anything more than a blank look in return they probably do.

An RPA is a “record of prior acceptance”. They are the twilight zone of the UCAS data. Their distinguishing feature is that they count students who haven’t been near UCAS. They are full-time undergraduates that have arrived at the university without using the centralised admissions service.

The university will have collected some data items from these students. This is a fairly laborious and variable process, typically using a provider-specific form with the provider’s own guidance. The data is then transferred by the university to UCAS, who reward the university for their efforts by charging them a capitation fee for each student (likely to total over £1M of revenue for UCAS in 2019).

You might wonder why providers put this burden on themselves. The simple answer is that UCAS make it a condition of being in the scheme. How vigorously providers comply with this can vary depending on whether they see it as their duty to help create a complete national picture of admissions or, at the other extreme, take the view that they should only pay for admissions services when they’ve used them. It is hard to know what really goes on, but it seems likely that not every last bit of non-UCAS entry is captured, and the proportion that is captured could well vary year to year.

UCAS then blends these RPA returns with the “real” UCAS records – those people who have used UCAS’ systems to apply to higher education – and they end up as part of the overall “total” applicant and acceptance statistics.

As is natural from this type of process, data coverage and quality can be variable. Especially for harder-to-define data items. It certainly can’t match the uniform and supported data collection process that those completing a standard UCAS application enjoy. This is why many of the key UCAS statistical tables on, say, ethnic group, disability, schools and qualifications don’t include RPAs. Nevertheless, most of the key UCAS numbers you will know, like the total applicants and acceptances and the acceptance rate from the release last week will include RPAs.

What is going on with RPAs?

RPAs are not new. Their numbers have been variable across time (and were quite high in the mid-2000s) but they have been reasonably steady for most of the past decade. So any uncertainty they introduced was tolerable and didn’t affect the overall dynamic too much. But in the past couple of cycles RPAs have grown strongly. In the context of static figures from the “real” scheme, this increases the risk of distortion dramatically. We’ve looked at them from 2010, giving us a series (largely) unperturbed by scheme changes (such as the old Route B and NMAS integration). We’ve used UCAS’ new and very welcome more detailed data on entry route by age and domicile. Because of accumulating rounding errors, the totals will sometimes be a fraction off the headline figures.

RPAs have generally increased modestly in the six cycles up to 2017, taking the total from 15,000 to 25,000 (Figure 1). But over the past two cycles they have jumped. By a third in 2018 and another fifth this year, putting on an additional 15,000 to reach a total of 41,000. This now equates to around 8% of the “real” UCAS acceptances (main scheme and direct to clearing) – no longer immaterial. And the effect on the dynamic is greater. In both 2017 and 2018 RPAs have been the fastest growing entry route by some margin. In contrast, “firm” main scheme students have fallen by just under 5,000 in both years.

Figure 1: RPA totals

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

RPAs are often associated with overseas students, and the more complicated entry routes they can have. This used to be the case, but the large increases in the past two cycles have been driven by UK students (Figure 2) who now total 35,000.

Figure 2: RPA totals by student domicile

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

Looking within the UK, this elevated RPA total is underpinned by older students from England (Figure 3). And the sharp increase over the past two cycles it is concentrated in these older groups (Figure 4). For these older English age groups, the RPAs are now decidedly non-trivial. As a proportion of the “normal” main scheme acceptances, they range from a quarter (21-24) to almost three-quarters (35+). For 18 year olds, the figure is much smaller, around 1%.

Figure 3: 2019 RPA students by age and country

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

Figure 4: RPA students, change over past two cycles

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

Is the RPA increase real?

The data shows that RPAs are growing strongly, and are concentrated in particular groups. They will be shaping, sometimes dominating, the overall trends. Is this a good or a bad thing? It depends.

At one extreme we could imagine that this rapid growth in RPAs is part of a genuine trend of late-in-cycle interest in HE. This is plausible since the absence of number controls in England and Wales, together with weak aggregate demand, means many universities will be welcoming of interest very late, even as term starts. If this is what is going on, then the increase in RPAs is real, and it would be good that it is in the data as it gives a more complete picture as demand transfer from early to late in the cycle.

At the other extreme, we could imagine that nothing very much has changed on the ground. Some students have always turned up at the university, innocently bypassing UCAS. But perhaps university data capture has improved, or the return systems have got easier, or UCAS has been more vigorous in reminding providers of its contractual right to a capitation fee. Under this hypothesis the increase in total acceptances in the data is mostly illusionary. Here including RPAs adds only uncertainty to statistical trends. They would be better left out.

We can’t be sure which of these is closest to reality. We think the best data evidence is to look at the other “late interest” route, the “Direct to Clearing” acceptances. These are people who have applied to UCAS very late on after the main scheme has closed. Most apply post results. If a shift to late interest is driving the RPAs, it would be reasonable to expect it to be reflected in the direct to clearing route.

Figure 5 shows the two late interest routes for the group at the heart of the recent growth, UK students aged 21 and over. The numbers are indexed to 2011=100, when RPAs started to increase. For most of the past decade, the increase in RPAs does seem to be reflected in late interest applicants presenting themselves to UCAS. But in the past two cycles the rapid growth of RPAs has become clearly detached from late interest presenting itself to UCAS.

We think this means that the increase in RPAs up to 2017 was probably mostly “real”, reflecting greater availability and utilisation of very late entry routes. Post-2017 though it looks like something else is happening. A strong candidate is an increase in the proportion of non-UCAS students captured by UCAS. But it could be other things too.

Figure 5: RPA students, change over past two cycles

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

The thickening fog

With a combination of large numbers of RPAs, their rapid growth, and a big question mark over exactly what they are recording the scene is set for statistical problems.

At the headline level, they are introducing doubt over the direction of the sector as a whole. Figure 6 shows total acceptances split by whether they result from UCAS applicants or the non-UCAS RPAs. Last year was perceived as flat in overall recruitment. But in truth, this was the net move from an increase in RPAs offsetting a fall in real UCAS acceptances. This year the headline is a material increase in acceptances. But again, this is driven by RPAs. The like-for-like acceptances from people who have used UCAS (main scheme and direct to clearing) are flat.

Figure 6: Total students, by applicant type

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

For those groups where the RPA increases have been strongest the statistics risk being actively misleading. The figure of a 15% increase in UK students aged 35+ was highlighted in the release last week. But this is almost entirely driven by uncertain RPAs. The increase in the main scheme and direct to clearing for this group was less than 3%, over five times smaller.

Figure 7: Total students, by applicant type (UK aged 35+)

Source: dataHE analysis of End of Cycle data resources 2019, www.ucas.com

What is really happening? These are important topics and it is clearly unhelpful not to be certain. Our advice to universities for this year is to exclude RPAs from their analysis if they are including older UK students. Ideally by using main scheme and direct to clearing students. Where this isn’t possible from what UCAS publish, we would suggest just using the main scheme figures that UCAS helpfully publish as part of the data resources.

We asked UCAS last year, and again this year, whether they thought the RPA increase was generated by something they had changed in the process. They haven’t identified anything to date. Given this, we think it would be helpful to the sector if UCAS could find a way so that the figures being published over the next couple of months are always split by entry route. Or at least a UCAS applicant/RPA division. This would let users make their own decisions over which sequence to use.

Beyond this it seems to us that this could be the right time for the RPA arrangements to be looked at from a perspective of statistical integrity. Universities are rightly heavy users of any data that UCAS publish. Important decisions will be taken from it over the coming months. It isn’t easy running a university and the sector needs data owners to do all they can to provide the very best picture of the system.

9 responses to “The thickening fog of the UCAS RPAs

  1. You have to wonder how much this is costing the sector in capitation fees also. Each RPA is charged a cap fee, for a service nobody really ever used.

  2. Michelle. Its difficult to say. These students won’t have been in the UCAS system so there are no offers recorded against them. There will be a lot of variability across the sector but what we often hear from admissions people at providers (the real experts on all this) is that these students will typically be people who have responded to billboard, etc marketing very late in the cycle and will present themselves at the university wanting to get started. So the offer is likely to be ‘unconditional’ in the sense that there are no qualifications pending. But not in the sense that it is competing against other conditional offers.

  3. Michelle – all of them. That’s how an RPA works – it’s a record that a place has been confirmed. If the question is intended be “how many RPA’s were agreed for applicants who had not completed their qualifications”, then the data available won’t tell you.

    We’re not big users of RPAs (I can count the number we’ve used over the past five years on one hand) but from other providers, the value seems to be less around time of cycle and more around the fact that you can only have one offer/place at any one time with an RPA. So, whether it’s an international student at a clearing fair, or someone aged 21+ who has turned up at their local University asking about the possibility of starting a course, if you advise them to put together a mainscheme or clearing application they suddenly have a whole range of other options made available to them whereas if you sign them up, speak nothing of UCAS and just RPA them (it’s totally a verb…) then you don’t need to worry about their head being turned by your local competitor.

    Worth saying that the benefits of UCAS aren’t just about the application process. A small example, but what about those students who think they’re winning by signing up directly with a provider and bypassing UCAS, only to find that they can’t set up a student back account as quickly and easily as everyone else because they didn’t get a student status code?

  4. Hi Kim
    That I understand. What I want to know is the % of the RPAs in relation to unconditional offers made?

  5. I wonder if there are any thoughts on what courses they might be – knowing the desperation there is to get people onto nursing programmes, I wonder if it is anything like that? And they have a large over 35 component, who don’t follow the normal application pattern.

  6. £1m surcharge on universities. Surely that should attract clear data supplies and the supply of RPA and non-RPA data in EXACT requests for universities, hopefully not charging them twice.

  7. Yes, the age profile is very unusual and suggestive of a subject like nursing. And RPAs for nursing have been increasing. But they simply not large enough to account for the headline moves. RPAs increased by around 15,000 between 2017 and 2019. Nursing RPAs were up 800 or so in that period. Nursing might be contributing, but is unlikely to be the full answer.

    Lots of things could be behind the RPA trend. We’ve heard lots of good ideas on this. But trying to work the answer out from the published data is a statistical pin the tail on the donkey. We are simply arguing that the uncertainty in these data is currently too high for the importance they have in understanding the sector. UCAS are best placed to reduce this since they hold and can publish additional data at the resolution needed.

  8. Great piece Mark. I’m now the Admissions Manager at an FE College up north with a large HE provision. Last year we submitted a huge number of RPAs as a lot of our HE students are from our own Level 3, mature, local. Having reports from UCAS that split out by entry route is absolutely essential and is one of the reasons I’ve not used a lot of the UCAS reports in my analysis the past few years.

Leave a Reply