Data, Heroin, HIV and Anomalies
We’ve become a little punch drunk when it comes to data, more specifically, the perceived insights that ever better analytics tools can derive. While I’ve worked with Big Data for several years on various projects, I’ve come to find that we rely on it too much to make decisions and that we’ve forgotten the humans. Google is notorious for relying on data to make critical decisions. One common refrain we here regarding data analytics is that data tells a story. It doesn’t. Humans tell a story based on what the data shows and weave the human experience into that narrative.
While good analytics can have profound benefits to society, sometimes it takes humans to uncover the truth. Data points us in a direction. It is a compass, but it is not the navigator.
A case in point is a study into variances in AIDS incidences among drug users in the United States in the early 00’s. The study was conducted in San Francisco, specifically with Injection Drug Users (IDU’s) and an issue that the data showed. That some injection heroin users had less instances of AIDS infection than others. The data showed an anomaly, but it required human ethnographic research to tell the story of what was happening. The ethnographic study, on the streets was carried out by Dr. Philippe Bourgois and Dr. Daniel Ciccarone, you can read the study here.
What they quickly noticed, was that some heroin users used a form of heroin from Mexico known as “black tar” while others used the more common powder form. The Mexican black tar is thick and oily. This meant the person had to frequently rinse the needle. This caused a dilution of the blood in the needle as it was used repeatedly amongst different users. Thus significantly reducing the likely transmission of HIV within that group. This was not the case with those using the powdered form of heroin in needles.
The data showed an anomaly. But it required the anthropologists conducting a real-world study to uncover the story of what caused the anomaly. We can too easily regard the data as having the final say.
I’ve seen a number of organizations increasingly rely too much on the data, only to be stymied when the project failed or took a different turn that was expected. But there can also be an over reliance on making decisions from the gut and the inherent bias of ones own personal beliefs.
The challenge for organisations today is to find the balance between using data and the immense benefits it offers and remembering the human. We humans are rather quirky and while we are largely predictable (hence the success of behavioural economics) we often do things that the data can’t predict.
