Being data-driven means being question-driven

At the Journalism Interactive 2014 conference, Derek Willis spoke about interviewing data, his advice for becoming a data-driven journalist. “The bulk of the skills involved in interviewing people and interviewing data are actually pretty similar,” Willis explained. “We want to get to know it a little bit. We want to figure out what’s here. Who are you? What are you about?” Interviewing data for a news story, therefore, has a lot in common with investigating data for a business initiative, which starts with the where, why, when, who, what and how of data usage.

“Think of data as another source,” Willis advised. “But unlike human sources, data can’t tell you that you asked a stupid question.” Unfortunately too many people – and not just journalists – don’t think to ask data any questions. Instead, data is often taken on the assumption of its quality. Before you start letting data drive your decisions, you had better give it a driving test first. “You need to adopt a posture of deep, deep, abiding skepticism,” Willis recommended. “Act on the assumption from the minute you look at data that there’s something wrong.” The great thing about data, according to Willis, is that “it encourages you think of stories as questions, not as statements.”

“Questions are more relevant than answers,” Stuart Firestein explained in Ignorance: How It Drives Science. “One good question can give rise to several layers of answers, inspire decades-long searches for solutions, generate whole new fields of inquiry, and prompt changes in entrenched thinking. Answers, on the other hand, often end the process.” Conversations about data science often focus too much on the data and not enough on the science, misunderstanding that what data science produces may not be an answer, but instead a better version of the original question – or a new question altogether. Just like all science, data science is question-driven.

“The right question asked in the right way, rather than the accumulation of more data, allows a field to progress,” Firestein explained. “Scientists don’t just design an experiment based on what they don’t know. The truly successful strategy is one that provides them even a glimpse of what’s on the other side of their ignorance and an opportunity to see if they can’t get the question to be bigger.”

It’s no question that data is getting bigger. Not only is the amount of data increasing, but the number of areas becoming data-driven is also increasing. Data-driven journalism is only one example recently getting, pun intended, more press. But no matter whether you’re interviewing data like a journalist, analyzing data like a scientist, or using data in a business context, always remember being data-driven means being question-driven.

tags: big data, Big data analytics, data quality, data science

One Comment

  1. Posted June 12, 2014 at 9:05 am | Permalink

    Thanks Jim. It's often hard to ask the right questions of people, let alone an inanimate thing like data (with no emotions or body language to read!)

    I posted this mini-guide with business requirements scoping in mind, but the approach is probably similarly a applicable to a data analytics task:

One Trackback

  1. […] Being data-driven doesn’t mean you should be driven to collect data. It means you should be driven to do something with the data you collect. Otherwise, you are a modern-day Sisyphus pushing the big data boulder up the hill just to watch it roll back down again, repeatedly. Outfitting the mythological Sisyphus with a fitness tracker to measure how fast he pushes the boulder up the hill, and how many calories he burns while doing so, wouldn’t have changed the pointless tedium of his punishment. Sisyphus didn’t need a fitness tracker since no amount of data or number of measurements was going to help him make a better decision or change his actions. […]

Post a Comment

Your email is never published nor shared. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>