Five Communication Tips for Talking with Data Newbies

December 1, 2017 3:49 pm PDT | Data as a Service

People in every discipline can benefit from a better understanding of quantitative analysis. There is an urgent need for government organizations to not only collect and process large amounts of data, but to be able to articulate the story inside that data in an effective way to a variety of stakeholders.

Below are five communication tips for being a catalyst for an even stronger data culture in your organization.

1. Be a data champion. Being a data champ is more than just “doing data.” Be an enthusiastic advocate for the power of data to uncover hidden solutions, achieve civic goals, and transform decision-making processes. Talk to your colleagues and educate them on how to address problems with data. Share your process, your progress, create a consistent vision, and lead by example. It’s a long road to shifting the culture of an organization to be data-oriented. Start with communication and building relationships.

2. Question the question. Imagine someone asked you to plot data on a map. They have geolocated data points about police activity and they want to visualize them. At that moment, you have a choice.You can simply address the question at face value and give an accurate, but potentially shortsighted answer. Or, you can dig deeper and try to understand the problem and the objective behind the request, asking a few thoughtful questions of your own to see if there might be a different solution.

For example, let’s say it turned out that any police initiated activity, including leaving their vehicle to buy a coffee, was mixed with shots-fired data. What story does it tell to have all this information mixed together and visualized on a map? As a data scientist, it’s good you’re naturally skeptical and constantly test assumptions in your own work.

Bring that same level of curiosity to your peers’ inquiries. Vigorously engage and question their questions, breaking it down to what everyday people need to know.

3. Answer the question. As you know, there are many potential pitfalls in working with data. And it’s easy for those without as much experience to fall into traps like assuming correlation means causation, or to get sidetracked by data that ultimately isn’t relevant to the issue at hand.

Use your skills to go beyond pointing out why the claims don’t hold up and explore alternative approaches to answering the question, and what datasets could be used to better test the hypothesis. Helping your colleagues examine problems from different angles will not only yield better analysis in the end, it will help condition them to be more thorough in their own research.

4. Understand how your output will be used. Providing too much detail or going too far in depth when it’s not warranted can also be a risk. To avoid overkill and wasted effort, find out the level of precision the problem warrants, and don’t overwhelm your colleagues with a ton of very technical information. It’s enough to make sure they understand your conclusions, and you can keep standard deviations and confidence intervals in your back pocket as supporting evidence when called for.

5. Understand what your output means to the person doing the work. There are lots of factors involved in correlations, including those that are less obvious. As a data scientist, you’re no doubt tuned into this reality as it pertains to your own projects. But it can be harder to bring this level of scrutiny to one-off requests from colleagues, especially when you’re busy.

For example, it’s easy to see that garbage trucks are only 25 percent utilized and your city has a lot of missed pickups and think, “What are those drivers doing? Taking five-hour lunches?!” If you dig deeper, you may find out that these trucks often break down because of how heavily they are used, and that the city has a contract with a single mechanic to fix them, and that mechanic can only fix one at a time. Trucks break down before the ones that are being serviced are fixed, so even though you have number of trucks, it’s only possible to drive 25 percent of them.

The conclusions you help generate influence decisions with the organizations that your colleagues are evaluating. More to the point, they can have a direct impact on the people who work for those organizations.

Take Time to Teach and Communicate

Because your skills as a data analyst are so highly sought after, it may seem like there’s no time for extended communication with colleagues across different departments. But promoting open dialogue and building rapport with contributors and managers outside your domain is vital to strengthening your organization’s data culture.


Ready to grow your data skills? Socrata Education offers live and on-demand courses. See our course catalog.


Previous Articleperson with computer
Data as a Service
User Experience Checklist for Government Data Programs

December 5, 2017

Next Articlecode4PA logo
Data as a Service
How to Host a Successful Hackathon in 4 Steps: The Pennsylvania Story

November 27, 2017