Tuesday, 10 July 2018

Piss poor use of statistics of the week

From the BBC:

Research for the report, carried out by the University of Exeter, shows the extent of the challenge - with figures showing how few places in 2016 were awarded to applicants from areas with few young people going to university.

Meaningless. This boils down to: "few people from areas where few people go to university go to university".

The University of Cambridge had only 3% of entrants from such "low participation neighbourhoods", the University of Bristol 3.7%, Oxford 4.6% and Exeter 5.3%.

Meaningless. We are not told what percentage of potential students live in a "low participation neighbourhood". If about 4% of potential students live in such areas, we'd only expect 3% - 5% of students at any university to be from such areas.


pen seive said...

Statistics show that the majority of people who stick their hands in a fire report an uncomfortable feeling of heat.
Statistics also show that the majority of voters polled think our politicians should be strung up, which may be why you can't find piano wire and old lampposts on eBay any more.
Anyway, statistics show that 78.4% of statistics are made up.

ontheotherhand said...

The underlying report from Exeter is here https://static1.squarespace.com/static/543e665de4b0fbb2b140b291/t/5b4457fb70a6ade52de65f16/1531205646268/Research+into+the+use+of+contextualised+admissions_University+of+Exeter+report.pdf

They do give a benchmark of sorts in the Table 1.2 in that the % from Low Participation Neighbourhoods for All Providers (England) is 11.4%. In the note to the table they mention 'location adjusted benchmark' to the numbers, which I guess means that universities that are far from Low Participation Neighbourhoods get some adjustment?

My general questions comments:
1. What the heck does the euphemism "Contextualised Admissions" mean? Is that Orwellian speak for Positive Discrimination?

2. Just by eye the strongest correlation on table 1.2 seems to showing that the higher the UCAS score required, the lower the entry from Low Participation Neighbourhoods. The commentary to the table says, "The institutional case studies focused on high-tariff HEIs ... with strong demand for places, and international research records. Competition for places and high entry standards imply these institutions are more likely to be prohibitive for educationally disadvantaged students who are less likely on average to reach the required entry level in terms of qualifications..."

Mark Wadsworth said...

PS, yup.

1. Yes.
2. I'm sure the original research was a bit more specific, it is the BBC's write up which is piss poor.