Everyone who has followed a course on statistics or empirical (quantitative) research methods knows how hard it is to put together a dataset that is representative, to design research that can be replicated, to set up a questionnaire or other tools for gathering data that are neutral and unbiased and to find correlations that are valid, significant and have meaning.Not for nothing, most academic papers based on empirical research in, for example, sociology, psychology and social (behavioural) sciences, often consist for more than half of a description of the research methods, the limitations of the research and the insecurities involved with the research findings.AI, Big Data and profiling thrive on collecting, analysing and using large amounts of data and are often based on analytical tools that are grounded in basic statistics.When gathering data, the GDPR applies, at least when it concerns 'personal data'.When data-driven applications and technologies are used in practice, there is a variety of different legal instruments that apply, such as anti-discrimination law, tort law and the various human and fundamental rights instruments, at least when the applications have a direct effect on natural persons and their interests.But the methods for analysing data themselves are barely regulated and the computer scientists and programmers operating algorithmic data-analytics are not always aware of even the most basic standards of empirical research methods and statistics.
Read more