Abstract
The constant growth of the scope and depth of evidence-based decision making in government throughout the 19th and 20th centuries has led countries to develop the concept of statistical independence of political and other vested interests. This independence is the most important condition necessary for maintaining and increasing data quality, yet the institutional position of statistical offices as part of government makes this hard to achieve. The institutions and governance of statistics in many countries, including developed ones, still leave much to be desired. In spite of this, history of the last several decades suggests there is systematic improvement in the governance of statistics within countries, with legislation and e.g. the acceptance of codes of practice ensuring this independence, etc. Recent developments have, however, introduced a relatively new danger in the rapid development of automatisms, where evidence-based decision making is being replaced by automatic decisions that virtually, or even fully, leave the responsible human decision maker out of the process. This development, which seems to be a welcome gradation of evidence-based decision making, is in fact a grave danger for the independence and quality of statistics, because it re-orients the target of lobbying from the decision makers (who are no longer in the loop) to the data sources, thus putting the quality of official statistics under new powerful pressures. In this context it is more necessary now than ever before to protect the statistical systém from these influences by ensuring full independence in the governance of statistics.