Addressing Data Discrepancies - What we found and how we dealt with them

1
Ian Castles Memorial Seminar New York 23 February
2011

• Addressing Data Discrepancies - What we found
  and how we dealt with them

Francesca Grum grum_at_un.org United Nations
Statistics Division
2
The issue of Data Discrepanciesbetween national
and international data sets

• Role of Ian Castles in drawing attention to data
  quality issues
• Discussed in several sessions of the UN
  Statistical Commission (2000, 2001, 2005, 2006,
  yearly since 2008)
• Resulted in a ECOSOC resolution to Strengthening
  Statistical Capacity in 2006

3
The issue of Data Discrepanciesbetween national
and international data sets

• Countries concerns focused on the lack of
• Consultation with national statistical offices
• Transparency of data sources and methodologies
  used
• Use of existing official statistics

4
UNSD assessment Investigating Data
Discrepancies

• Assessing data discrepancies via country missions
  and workshops
• Since 2008 inventorying differences
• 20 MDG indicators/20 years (90-09)
• 40 countries
• Documenting/explaining reasons for differences
• In consultation with concerned NSOs and
  international agencies
• For matching years/indicators, we found
  discrepancies for around 50 of these instances

5
UNSD assessment Understanding Data
Discrepancies
Based on 5 project countries
Reasons for differences Percent Example
Different data sources 25 Indicator 5.2 International data from DHS MICS Country data from administrative records
Different methodologies 22 Indicator 1.1 World Bank adjusted data to account for low survey coverage, Country did not
Data not collected or compiled by NSO 22 Indicator 6.9 Tuberculosis incidence/prevalence/death rates not available in NSO
Different definitions 20 Indicator 5.2 Different definitions of skilled personnel
Different population estimates 7 Indicator 2.1 International data used UNPD estimates, country used NSO population estimates
More data available at country level than at international level 5 Indicator 1.1 Population below national poverty line-Country has more recent data
International comparability
6

Features
Data sources

• MDG indicators
• Charts and tables
• Discrepancies with explanations
• Summary table
• Download

40 countries
7
(No Transcript)
8
UNSD approach to address data discrepancies

• Consultation with national statistical offices
• Workshops with national and international
  statisticians to review data gaps, identify
  causes and reconcile inconsistencies
• Since 2008
• Covered more than 60 countries
• Transparency of data sources and methodologies
  used
• MDG Global database and website completely
  redesigned to include metadata specifying sources
  and methodologies for all data

9
UNSD approach to address data discrepancies

• Use of existing official statistics
• Initiatives focusing on data transfer mechanisms
  to improve data compilation at the country level
  and data transferring to the international
  community
• Adoption of SDMX for MDGs indicators within IAEG
  members
• Ongoing projects to Improve the Compilation and
  Dissemination of National Development Indicators,
  including MDGs

10
UNSD Long term approach to address data
discrepancies

• Promoting use of national official statistics by
  improving
• Accessibility/searchability of data to facilitate
  easy access to statistical information

• Dissemination/communication (reporting) of data
  to increase the use/understanding of national
  statistics among the widest audience possible

11
How

• Implementing a country data platform, on the
  basis of existing software, that centralizes
  country indicators agreed by national data
  stakeholders
• Linking country platforms to UNdata via SDMX
  protocol, when possible
• Improving metadata information provided to users
  (in electronic printed publications) -
  preparing metadata handbooks for development
  indicators

12
In summary what UNSD is doing to address data
discrepancies