Getting to grips with the National Pupil Database; personal data in an open data world

Getting to grips with the National Pupil Database; personal data in an Open Data world

Phil Booth and Terri Dowty | Open Data Institute Friday Lunchtime Lectures |15 Feb 2013

personal data ≠ open data

NPD: legislative underpinning

• Education Act ‘96 power to collect ‘school level’ data

• Amended by Schedule 30 School Standards and Framework Act 1998

• Created statutory gateway to collect personal data about pupils

• Empowered secretary of state to define data in regulations

NPD: 2

• No consent required - head teachers under duty to supply information

• Data taken directly from school MIS• Initially parents/children unaware - FPNs

Function Creep

• Original school census annual ('PLASC')• Now taken each term• Includes pre-school providers• Incremental increase in personal data• Exclusions and attendance data, poverty

markers, mode of travel to school...

The gift that keeps on giving?

NPD data tables

NPD request and data flows

TIER 4Individual pupil level: identifiable, e.g. gender, attainment, absences

TIER 3Aggregate school level: identifiable and sensitive, could have single counts

TIER 2Individual pupil level: identifiable and sensitive, e.g. ‘recoded’ ethnicity, SEN, FSM

TIER 1Individual pupil level: identifying and/or identifiable and highly sensitive

DfE Data and Statistics

Division(DSD)

DfE Data Management

Advisory Panel(DMAP)

Diagram based on NPD user guide and protocol, July 2012

REQUEST

DATA

DfE consultation: ‘widen access’ to NPD

?“Data would only be released to organisations which had been through a robust approval process and in accordance with strict terms and conditions on data security, handling and use.”

“We will achieve this through making information from the National Pupil Database available to all (with appropriate safeguards in place so individual pupils cannot be identified), and developing a new School Performance Data Portal.”

the voluntary sector

political partiesand candidates

direct marketers

profit-drivenenterprises

bullies

people withgrudges

education publishers and developers

researchers

professional bodies

consultants

educators

the media

re-identification

• relatively easy outside urban areas when combined with ward-level stats

• e.g. ethnicity + sector postcode narrow down to handful of families (at most)

• + school year group can id individual child

‘anonymisation’

• de-identification• pseudonymisation• “effectively anonymised”?• aggregate data / statistics

• ‘Differential Privacy’…

identifying people

re-identifying people

NAFIS

IDENT1

NDNAD

GCSE + A LEVEL

• personal data ≠ open data• obfuscation vs. consent• (notification ≠ knowledge)• ‘anonymisation’ vs. utility

thanks for listening

Phil [email protected]

Terri [email protected]