For the last week or so, I’ve been talking about what Big Data is (and isn’t), the ways social sector organizations are leveraging its power, and the pitfalls they must take care to avoid. To continue this series, I’d like to move from abstract to actionable — what are the steps your nonprofit team should take to maximize the benefits of Big Data (and minimize the damage it can do)?
Knowing the potential strengths and weaknesses that are associated with it, here are six key guidelines that will help you reap the benefits of Big Data (and minimize the harms that can come with it).
Step 1: Carefully Define Success
It’s easy to get carried away with the benefits of Big Data from your team’s point of view, but the outcomes and results your Big Data machine learning is looking for must be carefully considered from a variety of stakeholder perspectives.
True success doesn’t always look the way we expect it to. It is contextual and cultural.
Designing the systems that will collect and analyze Big Data for your team requires time, consideration, and care. The humans tasked with creating your systems must examine your strategies and assumptions thoroughly at the beginning of your Big Data venture, and carefully recheck them on a regular basis as time goes on.
Step 2: Conduct Data Integrity Checks
There is no underestimating the importance of accuracy in data — no matter how big or small it may be. You must have a plan in place to ensure your data is accurate and not missing in predictable patterns. Even while we leverage the benefits of Big Data, much of our most valuable information is in tiny details.
Small data within our Big Data is often of very high value and, if it’s wrong, has the capacity to significantly alter analysis. Because it’s so small, there’s no excuse for not protecting it. If you try to perform predictive analysis on your data, missing, extra, or misstated data can throw all your numbers right off — and your results right out the window.
Step 3: Consider Your Data’s Weak Spots
Who does your data exclude? Fail? Mismeasure? Working in the social sector, we can’t just point to Big Data analysis results and absolve ourselves of responsibility for the real lives that are affected by them.
Big Data practices and related statistical modelling techniques involve the quantification, classification, and construction of individuals and populations — and the categories are never truly impartial or objective, but rather, embedded in socio-political context. Before you start celebrating the benefits of Big Data, it is your ethical responsibility as a researcher to keep these points in mind when deciding what data to use, how to get it, treat it, store it, and share it.
Step 4: Build Trust with Those You Collect Data From
But my lawyer says…
Stop. Without exception, you must ensure you have actual, informed consent from all individuals contributing to your Big Data. While legalese and loopholes might suit some private sector teams, passing the bare legal minimum is simply not enough for social sector organizations. Using your data for anything other than what you and the data producer have agreed upon is not okay.
Collecting data (which is highly valuable) from an individual or group and not providing measurable value in return is extractive — and not likely part of your team’s purpose or mandate. Be sure to keep an eye on the difference your project is making to the people you’re trying to help and avoid getting too focused on the benefits of Big Data to your team.
Step 5: Develop a Data Protection Policy
Concerns about privacy and data security are nothing new, and they’re not going away. Your organization should develop and follow a data protection policy that outlines appropriate methods for collection, storage, and sharing of personal data. It should also assign obligations and responsibilities for ensuring data protection to existing agency staff through a structure of data processors, data controllers, and data protection focal points.
In short, everyone should understand what their duties are with regards to protecting the Big Data you work with, so data is never left unguarded because everyone thought someone else was taking care of security.
Step 6: Avoid Black Boxes
Most corporate Big Data analytics —things like Google’s page rank algorithm or the Netflix recommender algorithm — are proprietary and heavily protected. That’s fine for business, but shouldn’t be the case in the social sector.
Algorithms used to make decisions that will alter people’s lives must be transparent. Social sector algorithms are used to decide:
- Whether an individual is likely to commit a crime again
- Whether or not a teacher is doing a good job
- Whether a person is at risk for homelessness or addiction
These cannot be closed, proprietary, black box affairs. There is far too high a risk that the methodological choices hidden within these black boxes are socially unjust from certain perspectives. How can we understand them — let alone adjust them — if we can’t see them?
The open data movement has made huge strides towards achieving data transparency in the social sector. The next step towards reaping the benefits of Big Data is the open algorithm movement.
Want the Benefits of Big Data?
If your organization needs some assistance leveraging Big Data or if you have questions about the benefits of Big Data that you’d like answered, our team at Datassist is happy to help. We can answer questions, provide advice or assist with your data project from inception to completion. Get in touch with us now to discuss what we can do for you.