The Voice of Data Quality: Neither an Echo nor a Form of Confirmation Bias

As published on www.dataversity.net on December 02, 2016

In the aftermath of the U.S. presidential election, and amidst cries about “the day data died,” it is fitting to respond to the purported demise of data and questions about the value of this subject in general. Let me, then, state at the outset that data are alive and well: It is the interpretation of data – selective by many and prejudicial by many more – that makes it seem that this material is irrelevant; that it has no voice, so to speak, except the one we choose (often erroneously) to give it; that the numbers are meaningless because, as Donald Trump’s victory over Hillary Clinton allegedly demonstrates, we should not trust this or any other kind of data.

In point of fact, the election should put an end to confirmation bias, not a stop to data. For the latter has a voice – it is the signal that separates itself from the noise – and it must be our responsibility not to confuse that sound for something it is not. It is our duty, not unlike that of a translator who seeks to best preserve the integrity of a document that he rewrites in a separate language, to stay true to the letter and spirit of the information before us.

If we take liberties with data, if we use the thinnest of pretexts to commit the most egregious of mistakes, if we choose to hear a portion of that signal while we ignore the entirety of its message, then it is very easy to lose your way. It becomes deceptively convenient to convert a few pings into a symphony of your own preference, one that says your candidate will win or your product with flourish or your business will thrive.

Our job is not to critique the sound, nor is it to muffle, distort or remix it. Rather, our task is to identify – and retransmit, in an accessible and intelligible manner – the totality of the things we hear; to know what the overall expression is, so we may respond to it with a campaign that resonates with voters or consumers or filmgoers or television viewers, or some other audience. 

Remember, too, the words of the late physicist and Nobel Laureate Richard Feynman: 

“The first principle is that you must not fool yourself – and you are the easiest person to fool.” 

Put another way, data does not create fools; but fools create their own data. The latter is a grievous wrong because it proves another maxim by Feynman, this one having to do with the problems of social science. He says: 

“Because of the success of science there is a kind of a … I think a kind of pseudoscience, social science is an example of a science which is not a science. They don’t do scientific … they follow the forms … you gather data, you do so and so and so forth but they don’t get any laws, they haven’t found anything, they haven’t got anywhere yet, maybe someday they will but it’s not very well developed, but what happens is… even on a more mundane level we get experts on everything. They sound like a sort of scientific experts. They are not scientists.” 

Understand that data are not partisan. The sounds are what they are.