When you look at your analytics; do you see statistics or stories? This question forms the foundation of a philosophy which I practice daily. It speaks to the struggle between the logical and irrational, but also the whole of human experience.
There are two parts to my Humanity of Data philosophy:
- Data which is generated by human behavior is defined by that behavior. The numbers cannot be analyzed without an understanding of the context which generates it.
- A human mind will always be interpreting the data. This means data analysis is subject to the same biases we face when perceiving the world around us.
Desperately Seeking Humanity
Social data is a strange beast in the world of statistics. It is almost always dependent on context. This isn’t a function of the mathematics, but of the human subjects we study.
There are many feels you are interacting with when you are dealing with social media. Spend just 10 minutes on Twistori (note: NSFW). This data visualization experiment streams Tweets which have the following keywords; I love, I hate, I think, I believe, I feel, and I wish. (credit: Amy Hoy and Thomas Fuchs).
I look at our social media data about once a month, but what am I staring at? Numbers of Tweets, Retweets, Favorites, Replies, Mentions and Follower counts. We call these vanity metrics because increasing their numbers make us feel good, but they lack the substance for social media optimization on their own.
I use the median in my marketing optimization efforts. There are other ways to find averages, but it is the median which tells us about data deviating from the norm. It is a great tool for working with social data which vacillates. I am not concerned with the status quo, but the outliers and I study them intently. This is my favorite part of the process because I get to tell the story which is reflected in the data. Thus linking actions (behavior) to emotional resonance.
That is what it means to go beyond the numbers and see humanity in data. Remember, one role of statistics is to direct our attention to what is not expected under the given conditions. They point us to a truth which we may not have seen just looking at the data set alone. The role of an analyst is to explain these revelations and take action.
Understanding Human Potential for Error
The role of an analyst is also fraught with pitfalls. Our cognitive beliefs do interfere with our abilities at times to objectively view data. A study from Yale Professor Dan Kahan investigated the relationship between political beliefs and high numeracy (an ability for working with numbers). The hypothesis was that those who score higher on tasks of numeracy would be more objective when examining data related to political beliefs. The results were quite different; those who read a study which was politically charged made more errors in their analysis than the control group whose problems were presented in a non-political context.
For the specifics watch this video from Numberphile which explains the study and results in full (about 7 minutes).
This result may go against your notion that quantitative information always leads to logical conclusions. The problem in this instance was not the numbers, but how willing subjects were to perform the mathematics required to draw the correct conclusion. This has an important takeaway. If we are communicating with others, the context of how that information is presented matters. When reporting scientific facts to the general public the consequences could be dire.
I think the best tool available to confront our biases is self-awareness. Knowing that you are susceptible to these errors puts you in a position to find an objective way of looking at things. That is why I use models or frameworks which have been tested over time by social scientists. I must explain the data results in a way which is logically consist with what we know of human behavior.
I certainly have opinions, but I would rather look towards a definitive source which has been peer reviewed by those in academic positions i.e experts. There is no need to go out on a limb with your explanations when the work has already been done for you.
My colleague Joe Cardillo wrote about the differences between being data-driven and data-informed. What I understood from this is that the data alone does not point to answers, it informs them. The insights you draw from the data are where you will find solutions.
Data is Integral, but See Beyond the Digits
I recently had the pleasure of watching the film Patch Adams for the first time. In the opening scene, we witness a confrontation between Hunter and his future friend/mentor. He asks Hunter, “how many fingers do you see?” and is disappointed when Hunter simply answers four. This patient is holding on to a secret and Hunter cannot let it rest.
“See what no one else sees. See what everyone else chooses not to see… out of fear and conformity and laziness. See the whole world anew each day. Ah, the truth is, you’re well on the way. If you didn’t see something here… besides a crazy bitter old man… you wouldn’t have come in the first place.”
Data shows us what no one else sees. What those who do not rely on data cannot see. Data allows us to see things in new ways.
And if you can take away insights from the data then you will see more than the number of cranky complaints on Twitter. Instead, you will see opportunities for human connection.