I was recently asked this question by a senior medical researcher who has been teaching research methods for more than two decades. She was truly puzzled with the term, and went on to ask, “don’t all experiments need data?”
She was absolutely right – there is no scientific process without data. Nevertheless, the terms “data science” and “data experiment” have become so accepted by swathes of techies. The conversation revealed the interesting, sometimes awkward, juxtaposition of the venerable institution of scientific research and the relative upstart of data analytics.
I was prodded to explore the relationship more deeply – how has the burgeoning industry around data been influenced by traditional research? Is analytics the truant child of scientific research? How do scientific research and the analytics industry converge and how do they diverge, hereby resulting in the field of data science?
I found a fascinating perspective in a book titled “A Survival Guide to the Misinformation Age” by David J Helfand, a professor for more than four decades and former chair of the Department of Astronomy at Columbia University. As suggested by the title, Professor Helfand espouses a well fortified skepticism of mass media, and cautions the world against blindly succumbing to the avalanche of misinformation – largely fed through our search engines and news programs. He exhorts people to develop “scientific habits of mind” that are built on certain pillars that define science.
One of these is “falsifiability” – paraphrasing from his explanation, this means that science is “not a search for the Truth but a mode of thought that seeks falsifiable models of nature.” Science demands that a theory be presented as a carefully constructed model which is supported by evidence. The model can then be tested and proven wrong if any experimental results shows evidence that refutes the model. In this sense, according to Dr. Helfand, scientific understanding must be tentative.
By contrast, the business world is accustomed to demand certainty in decision making. C-Suite discussions typically don’t have a high threshold for tentative models. As the science of data increases its impact upon the business world, will executives need to learn how to reframe decisions in probabilistic terms? Will organizations embrace a culture of iterative discovery rather than sequential project planning?
These are broad questions that may not have ready answers – but it does give us some food for thought as we see the words “data”, “decision” and “science” used together with increasing frequency.