Report of Investigation Task Force on Statistical Data Quality Assurance released (with photo/video)
*******************************************************

The following is issued on behalf of the Investigation Task Force on Statistical Data Quality Assurance:

     The Investigation Task Force on Statistical Data Quality Assurance (Task Force) has completed the investigation on statistical data quality of the Census and Statistics Department (C&SD) and submitted its report to the Secretary for Financial Services and the Treasury today (March 25). The Task Force set out its conclusions and put forward a number of recommendations in the report.

     The Government established the Task Force on January 10 to examine the authenticity of statistical data, the existing data quality assurance mechanism in survey data collection and the fieldwork management system in C&SD.

     The Task Force adopted an evidence-based approach to examine the available facts and findings to ensure impartial investigation, and made the conclusions and recommendations in an independent and objective manner.

     Over the past two months, the Task Force held a total of nine meetings to examine the available facts in detail. The Task Force considered that the verification exercise should be carried out in an independent and objective manner. Thus, verification of the General Household Survey (GHS) data was contracted out to a private survey research company to ensure full independence of the exercise. As for the Labour Earnings Survey (LES) and the Annual Earnings and Hours Survey (AEHS), the Task Force accepted the view that the business sector data were too sensitive such that it was not suitable for a private research company to conduct the checking. The data verification was undertaken by a different team of professional statisticians in C&SD.

     The Task Force also examined documents on the statistical data quality assurance mechanism established in survey data collection and fieldwork management system, and sought details from C&SD subject officers to clarify areas which deserved close scrutiny. Members of the Task Force (not including the Commissioner for Census and Statistics) conducted five meetings with C&SD staff of different grades to understand their views on various specific issues on data collection and fieldwork management.

     The Task Force made the following conclusions:

1. Based on the findings of the verification checks and other related parameters and arguments provided by C&SD, the Task Force is of the view that there is not sufficient evidence showing the existence of systematic data fabrication in GHS, LES and AEHS.

2. The Task Force concludes that there is not sufficient evidence to query the overall authenticity of the data collected in the above three surveys. The Task Force considers that the quality of data collected from the three surveys is sufficiently sound to provide basic inputs for the compilation of the relevant statistical figures.

3. In view of the data inconsistencies noted in the verification checks and the increasing challenges faced by C&SD's field officers in data collection work, the Task Force does not rule out the possibility that individual field officers have not strictly followed the fieldwork guidelines, such as misapplying personal judgement to handle some of the questions in the questionnaire based on their knowledge/experience, or skipping some instructions in the guidelines.

4. The Task Force is of the view that while in general the necessary processes are in place to control data quality, C&SD should introduce measures to strengthen and reinforce the existing data quality assurance system in various dimensions.

5. The Task Force considers that the fieldwork management system encompasses a host of historical, complex and deep-seated issues which would need to be addressed in an integrated and coherent manner in a separate review.

     The Task Force made six recommendations regarding the existing data quality assurance system and fieldwork management system of surveys as follows:

1. To identify data categories vulnerable to mis-classification/error and data fabrication through cross-sectional and longitudinal analyses for close monitoring of data quality;

2. To establish a Departmental Committee to oversee and co-ordinate the quality assurance of GHS in order to further safeguard the quality of the survey data collected by the various field teams;

3. To foster a sense of ownership in the mindset of all stakeholders involved in conducting GHS in order to further safeguard the quality of the GHS data;

4. To conduct research studies to assess the impact of the declining response rate and increasing rate of proxy reporting on the GHS results;

5. To consider introducing appropriate measures to raise public awareness of GHS and the work of C&SD in general with a view to enhancing the co-operation of respondents; and

6. To conduct a comprehensive review of the existing fieldwork management system. The review should at least cover aspects relating to the keeping and checking of time logs, decentralisation of GHS fieldwork, workload and work pressure of field officers, communication channels, treatment of sub-divided flats and resource situation of C&SD.

     The chairperson of the Task Force is the Commissioner for Census and Statistics, Mrs Lily Ou-Yang. Other members include Professor Chan Ngai-hang, Mr Tse Kam-keung, Mr Vincent Kwan and Principal Economist Ms Reddy Ng.

     The Terms of Reference of the Task Force are as follows:

1. To examine the authenticity of data collected by field officers in the GHS and related surveys;

2. To examine and assess the existing quality assurance mechanism in survey data collection, and to make recommendations on improvement measures; and

3. To make recommendations for improving the fieldwork management system of surveys.

     The full report and its executive summary will be available on the Government webpage (www.gov.hk/en/residents/government/policy/government_reports/reports) later today.

Ends/Monday, March 25, 2013
Issued at HKT 15:19

NNNN