ARGGGHHHHHHHHHH…….
[Had a link to last year’s numbers release and didn’t notice (that’ll teach me to type whilst on the phone! And that’s why it’s much safer sticking to F1 data… it doesn’t really matter if I get that wrong ;-)]
UCAS released their latest figures for university applications for 2013 today UK Application rates by country, sex, age and background (2013 Cycle, January deadline), along with data files for data charted in the report. Details of the actual number of applicants is also available: 2013 cycle applicant figures – January deadline.
Relates to 2012: (Data reported for applications considered on time for 15 January deadline). Headlines report that “Total applicant numbers at this stage of the cycle are 7.4% lower than at the same point in 2011”, with 18 year old England domiciled applicants down 4.1% and the largest drop in terms of actual numbers year on year coming the the age 19 year group (down over 17,000). The percentages are based on differences between the actual number of applicants year on year.
[Other parts of this post are also thrown off course now, eg after spotting in UCAS reports 3.5% increase in applications to higher education that “Application rates, which take population changes into account, show that the proportion of English 18 year olds applying in 2013 has increased by one percentage point. The application rates of 18 year olds across the UK are at, or near, their highest recorded levels”. Rates for last year were also released last year. I do wonder a couple of things though – why two separate releases, rates and actual numbers, (that can catch the casual user unaware… ahem… ;-) And why doesn’t there appear to be much consideration of the possible effect of demographic changes (population by age) on actual numbers applying, “all other things being equal”…?]
I wondered whether demographics might account for some of the change, or even work against it, assuming that the actual percentage of individuals within a year group that applied to university was consistent. [It turns out that information about population application rates was also release.] A quick peek at the ONS stats reported the following counts for age by single year in the 2011 Census for England as follows (2011 Census: QS103EW Age by single year, local authorities in England and Wales (Excel sheet 1045Kb) <- don't ask how I found that on the ONS website. I have no idea and could not recreate the steps… [Via @paulbradshaw, 2011 Census, Population and Household Estimates for England and Wales, Table P02 2011 Census: Usual resident population by single year of age and sex, England, which is differently identified to the data I found… I wonder if the numbers are different too?!]):
| Year | Count |
| 17 | 665,447 |
| 18 | 674,980 |
| 19 | 700,335 |
| 20 | 727,389 |
| 21 | 711,470 |
| 22 | 715,130 |
| 23 | 728,435 |
| 24 | 712,897 |
(A 10,000 change on 700,000 is about 1.4% of the 700,000.) If a fixed percentage of 18 year olds from England are applying to university each year (30%, say), then demographic factors could account for some of the change in actual numbers of applicants.
One of the things that surprised me slightly about the mechanics of the UCAS data release was that they didn’t make downloadable files containing the data available (although it is easily scraped from the data tables on the announcement page; an additional, slightly more expansive breakdown of one dataset – applications by subset – is provided though). I was surprised because the precursor announcement, December’s End of Cycle report 2012 which announced the publication of a figure filled PDF based report, also included a set of datafiles containing raw data used to generate the figures in the report, so it seems as if this is now UCAS’ standard way of releasing data referred to in reports? (See also: Press Releases and Convenient Report Publication Formats for Data Journalists where I introduce the notion of “view source for data” and describe a few other examples of how public bodies have released data in support of formal reports.)
If you wanted to try to make sense of the university application numbers, and try to get a feel for what sort of effect policy changes might have had on university applications and uptake over the years, you’d probably want to look at some longitudinal data.
One of the criticisms raised about the ONS website by the Public Administration Committee over its website as part of an inquiry on Communicating and publishing statistics (see also the session the day before) related to the availability of longitudinal datasets. One dataset that I tried to looked for related to the number of students in Higher Education over the last 40 years. My first guess at the natural home for this on the ONS site was Higher Education Enrolments, and Qualifications Obtained, at Higher Education Institutions in the UK, but this only appears to go back to 2006.
I recalled struggling to find historical workforce data on the ONS site before now, but recall finding deeper historical data on nomis, the ONS official labour market stats site – but that doesn’t really do education…
However, I did manage to turn up a research briefing from the House of Commons Library (this was also referred to by Michael Blastland in his submission to the Public Affairs Committee about the ONS, I think?): Education: Historical statistics – Commons Library Standard Note. Unfortunately, the actual data referred to in that note is not available as a dataset.
(As an aside, whilst looking around the parliament site to see what else might be there, I came across pages for searching through research briefings as well papers deposited to the House libraries (for example, in response to official questions). The deposited papers include a whole range of document types, including spreadsheets, so FWIW, I started building a scraper to try to index them: scraperwiki: Parliamentary deposited papers.)
So.. the data’s out, and also in a form where it can be played with. So, has anyone played with it?! I also wonder if there’s historical data on the UCAS website detailing application numbers going further back than 2006, ideally as a nicely packaged longitudinal dataset…?
OUseful.Info, the blog... 