Asking Good Questions to Data.
Meenakshi (Meena) Das
CEO at NamasteData.org | Advancing Human-Centric Data & AI Equity
Welcome to data uncollected, a newsletter designed to enable nonprofits to listen, think, reflect, and talk about data we missed and are yet to collect. In this newsletter, we will talk about everything the raw data is capable of – from simple strategies of building equity into research+analytics processes to how we can make a better community through purpose-driven analysis.
?
When I started college years ago, one of my first few classes included algorithm design. I remember the first three months were very intense.?
The goal of the subject was to inform us about the fundamentals of designing good algorithms. We would take examples like creating patterns for games or designing structures for a custom application. Those first three months were intense because I understood next to nothing – the loops, coding constructs, manual debugging techniques, and, most importantly, the simple flow of it. But it was not after the end of the first mini-test (until I was almost convinced I would never learn anything) that I decided to change my technique of approaching the overwhelming number of unknowns in this subject.
?
I realized my brain is too creative to think of structures in how I was taught to look at algorithms. It was like the top chef taking me to their kitchen, giving me a tour, and then asking me to prepare the best biryani.
?
I needed to re-learn engaging with the subject before allowing my creativity to have fun. So, for almost two months, I had one job every night – picking one pre-designed algorithm to break every line and deconstruct it – all just using paper and pencil. I had my list of simple pre-designed algorithms ready. And so, I repeated – manually seeing what happens step after step. Especially what breaks, sticks, and flows in the code. By the 25th day, this became a routine – deciding my coding constructs and seeing the flow of the algorithm design on paper.?
?
I learned to engage with the subject differently through that process of extensive repetitive reverse engineering - using merely paper and pencil. I was starting to ask better questions to my algorithms.?Instead of seeking a response to a “how”, I started to chase the “why” and “why not” through “what ifs”.
?
I am bringing this for you and me today – asking better questions.
?
Between a question and an answer, we often mistake the answer for having power. The reality is that it is the question that holds power. And this extends to engaging with data. So unless you, I, and our teams learn to ask meaningful, specific questions about data, we risk missing opportunities from it.
?
Asking good questions to data can seem challenging, as it requires a combination of critical thinking, communication skills, and the ability to identify what information is needed. Some common challenges when it comes to asking good questions include the following:
?
?
So, instead of focusing on designing frameworks for asking good questions to data, let’s establish a few guiding principles for you and me to ask those questions better next time. You don’t have to be a data scientist by profession to use these principles.
?
#1: Think about the end-to-end journey, not about isolated touchpoints.
I remember a specific project a few years ago – building three to four dashboards on end-to-end operations and finance compliance for an organization’s leadership. The dashboards needed to source data from 50+ datasets sitting in a complicated data warehouse. Overall, it involved 30+ non-technical (business) users and 10 technical experts on the team. As the technical head of this work, my priority was allowing all non-technical users to see the overall flow of the work rather than having them get stuck on specific parts of the project. Our plans, team calls, strategy meetings – everything was designed so everyone could zoom out (and see the big picture) and, at times, zoom in (to discuss the details). This pattern of working welcomes many good questions from the entire team.
?
For example: instead of the finance team only asking questions on “where will we see our usual metrics like [xyz]?”, conversations became more around, “if we create a new compliance metric [abc], how does that interact with our overall compliance goals, and, can it also serve the other departments?”
领英推荐
?
#2: Incorporate your nonprofit’s context.
Before framing our questions to data, we often start with the data itself. We tend to get stuck in the metrics, KPIs, and standard reports. This limits our potential to ask questions meaningful, innovative questions. Instead of sticking with framing questions around what data we have, think of the questions that center your nonprofit’s context – questions that are focused on advancing the mission of your work. Some of the data may already exist for your questions. If they don’t, it might be worth looking into updating data collection.
Remember, this doesn’t mean asking vague questions (see below).?
?
For example: instead of only asking, “How much has our carbon footprint decreased since we implemented our sustainability initiatives?”, also think, “what is the implementation, access, and context of different locations where sustainability initiatives were implemented to realize the decrease in carbon footprints?” This framing helps you focus on not one number but break it in a way that adds more meaningful context around your mission serving.
?
#3: Ask specific questions that serve a purpose and can be tied to an action.
Vague or ambiguous questions can lead to confusion, misinterpretation, and incorrect conclusions. Specific questions tied to an objective and timeframe can help identify appropriate data to collect/gather and communicate with others, such as data analysts on the team.
?
For example: Instead of only asking, “how many views do our nonprofit’s event ad have?”, think, “what is our event sign-up rate on the current Facebook ad in the past two weeks?” The second question is actionable because then you can tweak your strategy on the event or ad pages accordingly.?
?
#4. Design questions collaboratively.
Instead of asking your analysts or data scientists, “pull me the data on…”, offer context and outcomes you intend to achieve. This creates an environment of collaboration where your context can help your technical counterparts provide the right measures and metrics in response.
?
For example: Instead of only asking, “can we create research profiles for those who are going to attend our upcoming event?”, think “To better allocate our outreach budget and reach the fundraising goal for Q2 of $[xxx], I want to know the data points [a, b, and c] of the event registrants until now, who haven’t engaged with us before in any way.” See the details of context and specificity? This can help your technical counterparts to come back and suggest – “wait, instead of only data points [a, b, and c], look for the combination of data points[a, c, and e].”
?
*********************************
Good questions to data don’t happen overnight. It starts with multiple bad and mediocre questions, consistently, until we learn what could be?our?language to data. And that requires a culture that welcomes curiosity and willingness to try.?
If I had not found my way to ask questions about algorithm design years ago, my data journey would be much different than today (something I would not want).
Good questions are not the product of degrees, leadership roles, or abilities. They are simply the outcome of imagination and curiosity. Like meeting a stranger on a train, where you ask and listen after that first hello.
It requires a belief that exploring “what ifs” are not always a dead end.
***?So, what do I want from you today (my readers)?
Realtor Associate @ Next Trend Realty LLC | HAR REALTOR, IRS Tax Preparer
2 年Well said.
Enabling Sense-Making?data processing, fusion, analysis, and contextual understanding related to intelligence production
2 年Great read. I really enjoyed the way you broke it all down and how to ask better questions and understand what we will even do with that data.