Measuring Trust in Official Statistics – Conceptual Framework, Measurement and Results
Siu-Ming Tam
Australian Bureau of Statistics, Belconnen, ACT, Australia

Statistics are fundamental to democracy and vital for evidence-based decisions.

Democracy is about freedom to choose. Informed choices require evidence based decisions and statistics in turn provide the evidence needed for rational decisions.

Not all statistics produced would be used for decision making, however. If they are inaccurate, not timely or unreliable, they would not be used. Also, if the statistics lack objectivity or creditability, they would not be trusted. Where statistics are not trusted, they would not be used for decision making.

It is argued that statistics are trusted if, and only if, people have trust in the products and services produced by statistical institutions (ie product trust), and trust in these institutions (ie institutional trust).

Under the product trust dimension, elements such as accuracy, timeliness, reliability, creditability, objectivity, relevance and coherence, ie elements of the ABS Quality Framework, are identified as key. Under the institutional trust dimension, elements such as confidentiality protection, integrity, openness/transparency, impartiality, and effective stakeholder engagement are recognised as key. It is noted that some of the elements are correlated with one another.

In 2009, using a framework comprising product and institutional trust, a questionnaire was developed by an international group of official statisticians, under the leadership of Ivan Felligi and the auspices of OECD. The framework recognises that there are factors outside the control of a statistical institution to engender trust, ie prevailing cultural norms, a person's awareness of statistics and the institution, and his/her past history/experience with statistics or the institution. The Framework is also amenable to subjective (ie survey of public perceptions) and objective (survey of statistical practice by statistical institutions) measurements.

Whilst the framework is designed for measuring trust in centralised statistical systems, it can be extended to decentralised statistical systems.

After cognitive testing, the international questionnaire was used in a 2010 survey in Australia to measure the trust of official statistics and of the Australian Bureau of Statistics (ABS) using a subjective measurement approach.

The 2010 survey results showed that there was a high level of trust in the ABS and its statistics. Indeed, of the 2,242 people responded in the survey, 92% said they trusted (ie trusted greatly or tended to trust) the ABS and 87% said they trusted ABS statistics. Likewise, almost all of the 137 specialist users interviewed, ie researchers, journalists etc., said they trusted the ABS and its statistics.

The survey of the general population, conducted by a market research organisation and based on telephone interviewing of respondents drawn from telephone directories, had achieved a response rate of 26%. This paper will also briefly assess possible biases in the overall results with such a rate.

Keywords: product trust; institutional trust; trust framework

Biography: As a First Assistant Statistician of the Australian Bureau of Statistics (ABS), Dr Tam is currently head of the Integrated Collection and Dissemination Services Division, and has responsibility for all ABS household and business collections, as well as ABS dissemination services.

Dr Tam has a PhD in Statistics from the Australian National University, is an elected member of the International Statistical Institute, an Accredited Statistician of the Statistical Society of Australia, a member of Steering Committee on Data Dissemination established by the United Nations Statistics Division, a past Vice President of the International Association of Official Statistics, and foundation editor-in-chief of the Statistical Journal of the International Association of Official Statistics.