Administrative data quality: Investigating record-level address accuracy in the Northern Ireland health register

Brian Foley, Ian Shuttleworth, David Martin

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)
236 Downloads (Pure)


Many national statistical institutes (NSIs) are seeking to supplement or replace their traditional population census with a methodology underpinned by administrative sources. Health service register data are key in this regard owing to their high population coverage; it is therefore important to improve understanding of data quality in this administrative source. This study investigated the factors associated with record-level address data mismatch between the Northern Ireland (NI) Health Card Registration System (HCRS) and the 2011 Census, using the NI Longitudinal Study (NILS). Address information in the form of anonymised Unique Property Reference Number (XUPRN) was available for circa 334,000 NILS members with census returns in 2001 and 2011, which provided a benchmark to assess XUPRN accuracy in their linked HCRS record for comparable time points. Multinomial logistic regression revealed a significantly greater likelihood of address mismatch in the HCRS for: males; young adults; individuals with a limiting long-term illness; migrants in the year prior to each census; and residents of communal establishments. Identification of population groups affected by poor quality address information in administrative sources can assist NSIs with the development and implementation of methodological improvements to ensure that official population statistics generated from these sources are fit for purpose.
Original languageEnglish
Pages (from-to)55-81
JournalJournal of Official Statistics
Issue number1
Publication statusPublished - 01 Mar 2018


  • Address data quality, census, population statistics, longitudinal data

Fingerprint Dive into the research topics of 'Administrative data quality: Investigating record-level address accuracy in the Northern Ireland health register'. Together they form a unique fingerprint.

Cite this