Proposal for Data Sets from Government-funded Surveys to be Published as Default Policy Stance
Prime Minister's OfficeSpeakers
Summary
This question concerns the proposal by Mr Leon Perera for government-funded survey datasets, such as the Health for Life in Singapore (HELIOS) study, to be published as a default policy stance with anonymization safeguards. Minister Josephine Teo responded that the Government must balance transparency against personal data protection, the sensitivity of findings, and potential misuse by hostile actors. She explained that anonymization is often insufficient for longitudinal surveys where re-identification risks remain high, potentially breaching respondent confidentiality. Consequently, sensitive data is shared directly with researchers under specific safeguards rather than through public release to facilitate impactful research and policy discussions. Regarding HELIOS, Minister Josephine Teo noted that research summaries will be made public and relevant datasets will be shared with the research community through the Ministry of Health.
Transcript
11 Mr Leon Perera asked the Prime Minister whether the comprehensive data sets of all surveys funded by the Government, including nationwide health surveys such as HELIOS (Health for Life in Singapore), can be published as a default policy stance, with safeguards to remove personal identifiable information and with exceptions for not publishing granted only on the basis of national security or commercial sensitivity considerations.
Mrs Josephine Teo (for the Prime Minister): The Government conducts and commissions surveys across a range of issues. Survey findings are used to improve policy formulation, service delivery and communications. We strike a balance between putting out sufficient data to inform and enable robust policy discussions against considerations such as personal data protection, the sensitivity of the survey findings, and whether the data could be used adversely by hostile actors
Many survey findings are published. Examples include the annual Labour Force Survey, the five-yearly Household Expenditure Survey and detailed Census of Population releases.
However, specific datasets may not be suitable for publication. Survey respondents often give their consent on the understanding that their responses will be kept confidential. They entrust their personal data in confidence, and we take this seriously. Removing personal identifiable information is not enough. Parts of a dataset can involve very small samples which heightens the risk of re-identifying survey respondents significantly. This is especially so for longitudinal panel surveys. Instead, we may share more sensitive data directly with researchers looking into relevant topics, with additional safeguards.
The Health for Life in Singapore (HELIOS) study which the Member raised is a longitudinal panel health study led by the Lee Kong Chian School of Medicine, in partnership with National Healthcare Group and Imperial College London. A summary of the research results will be made widely available to the public via the HELIOS website. There are also plans to make relevant datasets available to the wider research community, to enable impactful health research collaborations. Academics interested in diving deeper with the data may contact MOH to find out more about this.