Teach For America is obsessed with data. While teachers collect data on their students to help make sure they are on track for success, the back-office teams collect data on teachers and alumni (people who have already gone through the 2-year teaching corps) to make sure they are on track for having positive social impact, particularly in the field of educational equity. I know this because for the past two months that I have been employed at TFA, I have been partly responsible for maintaining the infrastructure and integrity of this data. Call me a nerd, but it is awesome. It really isn’t surprising that I love it so much, because (as this blog shows) I truly believe that information organized into data can be used to help and improve programs, activities, and actions that make positive social impact on the world.
Some important questions should come to mind when we think about using data to improve social impact (in other words, when creating a social impact assessment): 1) what data is being collected, and 2) how is that data being utilized to inform decisions and improve programs and activities? After reading a recent FastCompany article and then doing some research, I found an organization called Kaggle that is answering the second question in a very interesting way.
We have a dataset. We have a problem. What do we do?
The idea behind Kaggle rests on the assumption that predictive modeling can actually predict the future to a reasonable degree. A very good informational video on their website explains that “predictive modeling is a way of finding patterns and relationships in existing data, and then using those to predict what will happen in spaces where data isn’t available.” If there were a way to predict the future, why wouldn’t everybody be doing it? The answer is the motivation behind Kaggle: “most organizations don’t have access to the advanced machine learning and statistical techniques that would allow them to extract maximum value from their data. Meanwhile, data scientists crave real-world data to develop and refine their techniques. Kaggle corrects this mismatch by offering companies a cost-effective way to harness the ‘cognitive surplus’ of the world’s best data scientists.”
Could data scientists help us predict what programs will be most effective, given a particular social environment, group of participants, schedule, partner organizations, background information, etc? Well, two things will have to happen before we will ever be able to test it out. First, we need to get mission-driven leaders into the game.
Check out the current make-up of the Skillbases of Kaggle users:
Perhaps non-profit leaders are in the 1.6% “social science other” category, but for some reason I doubt it. Not many people working at mission-driven organizations are employed because they are data scientists – and I’m guessing most of us aren’t being data scientists in our spare time, either. So we need to get into the game, and we need to change the culture of mission-driven organizations to value employee time being spent on data. Similarly, the culture of mission-driven organizations needs to embrace serious and studied data collection. I’m sure you’ve heard the adage “garbage in, garbage out,” meaning that your models are only as good as the data you put into them, and only good data will result in truly helpful conclusions. Why should mission-driven organizations go through such culture shifts? Because predictive modeling could be incredibly useful for leaders of nonprofit organizations who are trying to decide what programs or activities they want to do, and how effective their impact could potentially be. In the same vein, corporations could use this information to help them choose their CRM activities. I want to be clear that I do not think this is a panacea – there is never any one solution to solving the world’s problems. But this could be a powerful resource for leaders when deciding how they could have the most positive impact on communities.
As I’ve talked about before, it is no surprise that one of the hurdles to using a predictive model for social impact is the difficulty of collecting good data. There is no real model for quantifying transformational change (yet). But I bet lots of mission-driven organizations have collected tons of data, but don’t necessarily know what to do with it. If we could identify and organize the data we have, or the data we want to have, and then start a competition to find a predictive model for positive social impact, or transformational change, what might we end up with? Potentially, we could create tools (algorithms) that allow mission-driven organizations to enter data within certain parameters (size of organization, number of people being served, demographics of the environment, etc) and determine with reasonable accuracy whether current programs are truly effective, or whether a new program may have the desired outcome of positive social impact.
The improvement of social impact assessment relies on two factors: collecting good data and creating a set of tools that make use of the data. A social impact assessment exchange would be incredibly powerful if paired with a competition on Kaggle. Imagine, crowdsourcing an incredibly huge and interesting data set of mission-driven work and then finding a predictive model to help better understand the impact of mission-driven work. I can imagine it, and it gets my mind and heart racing thinking about all of the possibilities!