"Statistics are a key source of information for decision making today and will remain so in the future digital state governance system," said deputy head of the Analytical Center Alexander Radutski at an international conference on the digital agenda for statistics in timeliness, quality and transparency of data.
However, today there is a lack of understanding of how data driven governance must be carried out and what is the degree to which statistical data must be used in decision making at various levels of state governance, the expert noted. As there is no clear link between statistics and the governance process, it creates certain limitations to the development of statistical indicators and methodologies to improve the quality of statistical information.
"Statistics organizations are going to have to do a lot of explaining, popularizing and educating, supplying a lot of footnotes and comments with their statistical reports in order to make them more useful and easier to understand and thus creating more demand for them," Mr Radutsky believes.
According to the forecasts of the US Labor Statistics Bureau, demand for statisticians and data scientists is projected to grow by 33% by 2026, which means the future of the profession is quite secure, the expert is confident. At the same time there will be more requirements for statisticians as the tasks of data analysis, data processing and data presentation will be getting more complex as time goes by. There will be a number of new competencies that will become a must such as computer programming and knowledge how to use algorithms in digital environments.
Obstacles to the development of digital statistics, according to the expert, include lack of trust in big data and controversy over how useful big data can be in solving statistical problems. One characteristic feature of big data is that they are always potentially incomplete, non-structured, come from unreliable source and can thus potentially distort statistical reports.
In order to use big data properly statisticians need to sort out a huge number of problems related to how to process big data: you need to minimize noise, eliminate random patterns and spurious correlations etc.
"For the time being the questions statisticians are facing is not how to improve the accuracy and precision of big data but rather how to measure their reliability," he summed up.