One of the things that drew me to data 25 years ago was its very intimate, non-obtrusive way of telling the human story. It seems weird given how we perceive data engineering as mathematical or alchemic magic. But, for anyone who has spent hours culling and wrangling data, we get a glimpse into the human experience that is truly intriguing. We are storytellers responsible for data-driven behavioral changes.
Intellectually, I think that makes sense. I’m asked to create data enterprises in each industry I support that will predict and portend human behavior. We call this practice Advanced Analytics, Machine Learning, and AI to put some skillset and distance between us and our spy-like activities. But in most industries, what we’re really doing is assessing human behavior and interactions with our products, services, and our companies.
Honor the human story
In the spirit of honesty and disclosure, I would say what makes me good at my job is that I am keenly aware of what I am doing. I spend time with my teams explaining the need for discretion, governance, contextualization, and temperance in our daily work. We must honor the human story as we transcribe it into algorithms, graphs, and pretty visualizations.
And, most importantly, we must be aware of assigning importance to characteristics that are outcomes, not indicators.
What exactly do I mean by that? Let’s look at our current pandemic. Black Americans were disproportionately affected by COVID-19. That sentence suggests that melanin must be a factor in the communicability and the resiliency to the virus (it isn’t, by the way). The data suggests that individuals who lack access to healthy food choices, healthcare, and education compounded by the systemic trauma in economic sustainability for at least one generation are more susceptible to this (and other) pandemics. Race and ethnicity are algorithmic shorthands for telling that story.
We shape the future
So many people have asked why SingleStone Consulting would merge the roles of Data and Diversity into one person for progress accountability. Those who shape the story we tell ourselves shape the future state. In our industry, we call it data-driven behavior change. Anyone who has looked at their Apple watch in an effort to “close their rings” has experienced data-driven behavior change. In 1896, it was statistically justified that Black Americans were uninsurable. This data-driven decision led to financial redlining which led to trauma in economic sustainability.
Data-driven behavior change when done nefariously or neglectful can have catastrophic effects on the collective. Consider some of the impacts Affirmative Action has had on institutions and people. There are anecdotes of Black and Brown Americans and Women not being deserving of their positions—only having been accepted or hired to satisfy as many variables as possible. I would know, I checked a few boxes in the 90s.
So, how do we responsibly use data to affect societal, collective, and institutional change?
Establish Context
We are ultimately telling stories. Check and double-check the context from which you are culling your data. Please do not tell me Black, Brown, and Female technologists do not exist. I am not a unicorn from faery land. I’m also not standing around waiting for you to find me and hire me. I’m choosing to work for companies and stakeholders who accept me with equity, and who will benefit from my skills, experience, and insight. In other words, find out where we are instead of telling us where we are not.
Right Variable Definition/Right Policies
There are centuries of variables, features, and nomenclature that has been codified sans validations, with wrong definitions and bias included. This has been exacerbated by volumes of data collected. We have an opportunity to contextualize and rename (dare I suggest eradicate) these variables.
For example, maternity leave implies that fathers do not want to and are not needed in the bonding and nurture process of new family members. It doesn’t consider the newborn’s needs at all. Instead, it’s a fiscal algorithm for forecasting the impact of losing resource productivity for 2 to 6 weeks. Furthermore, the decision to make it more inclusive and equitable between men and women in the workplace will be considered as an increased expense in the financial data sets.
Right Behaviors to Change
Most societal-isms are fostered in fear. If we are going to become equitably diverse, we must positively incentivize inclusion, particularly through the eradication of fear in the workplace. Increased communication and engagement eradicate fear. That doesn’t sound like data-driven behavior change, does it?
Ironically, data can only impact behavior after a framework for impact has been established. The interactions and engagements must be defined and executed to capture the data. Your desired healthy steps must be determined or capturing the total number of your steps is meaningless. If we create frameworks for communication and engagement that include measurable as well as qualitative factors, the interaction model will collect data points that will indicate progress, such as promotions. If you have a positive outlier, figure out what enabled that and build it into your framework. If you have a negative outlier, figure out what didn’t work and adjust accordingly.
I suppose what I am suggesting from experience is that data practices are not mathematical alchemic magic. What we do, when done properly, will explain our human experience and interaction within a context of change. As executives, we define the frameworks of change. We must invest and we must commit to the framework in order for the data to indicate progress over time. Then, and only then, will I cease to be an anomalous data point and I won’t need the Diversity title adjacent to my Data title.
