From Responsible Data Wiki
Jump to: navigation, search

We worked on a tool to find the most effective way to reach your goals.
That could be data vis. Or it could be not data vis.

Are you our audience?

You should go through the decision model if you are

  1. doing stuff in the field of data visualisation (because yeah, numbers, or because it looks neat), but don't know how to USE your skills for something important.
  2. working in an organisation and you've heart of data vis and you think it's the solution to ALL your problems and you want to commission one RIGHT NOW.

We want to give the data vizzer a mission, and we want to make sure that organisations are only using data vis when it helps to achieve their goals. For that, we created a decision model which focuses on formulating your mission, sub-goals, actions and strategies towards your mission.

Disclaimer: When we talk about data vis, we mean the presentation side of it. Data Vis can be part of an analytical process to find the best possible action. But in this decision tree, we will focus on the data vis that is supposed to communicate a point to an audience.
Second Disclaimer: The example is written down quickly and might be not the perfect solution to the problem.

The Decision Model

Start with the first question and work through all the main and sub-questions:
Do you want to make a data vis? Start with step 1.
Or do you want to change the world? Start with step 4.

1. Do you just want to mess around with some data, learn a new programming language or tool? If yes: Stop reading, have fun! Go to step 10.
Or do you want to achieve something that could change the world? Then:

2. We assume that you already have an idea or at least a data set in mind (aka "sketch implementation").
Example: I found that great dataset on gun deaths in the US; it seems really detailed and I can do lots of correlations with that. And immediately, I had an idea how to visualise it; in really dark colors and with red as the highlight color – will look really good!

What would happen if you would publish this data vis? What would be the best possible outcome (aka "sketch sub-goal")?
Example: I guess the gun death data vis, in the best possible outcome, could raise awareness for the topic. I hope it will get shared on Twitter a lot, so, yeah, people should see how many people actually die from guns in the US.

3. Assuming your objective was reached, what would be your next objective? What comes after success? Repeat that question and find all your sub-goals until can't get higher and find your...
Example: Once people know about gun deaths, I guess I want them to change, right? I guess I want gun owners to get rid of their guns, or to be trained more before they are allowed to buy a gun. So...yes, I want people to not buy a gun in the first place, and I want the government to make it harder to buy a gun. And if I would be successful with that goal...yes, I guess I want a world where nobody dies from a gun, not only in the US, but everywhere.

4. MISSION – the big picture

  • First, get into the right mindset to formulate your big mission: Assume success! [Here, success stories from other people might be helpful]
  • If you could achieve anything, what would it be?
  • What is your vision? What would be the best possible outcome of your project?
  • [what are other questions for that step?]

Example: My mission is to abolish gun deaths on this planet.

5. SUB-GOALS: Breaking down your big mission into smaller goals

  • What are your sub-goals?
  • What are some measurable objectives?
  • [what are other questions for that step?]

Example: Let's focus on the nation with the biggest gun death problem, or the nation where I can most effectively decrease the number of gun deaths. The most gun deaths per capita don't seem to happen in the US, but in Honduras. But then again, gun deaths might be not a problem of owning a gun: Unintentional gun deaths are not that high in Honduras, but homicide gun deaths are. So instead of focusing on the guns, I might ask: Why do people want to kill other people in Honduras and what can be done to stop that? So that might be an objective: To decrease the number of homicides in Honduras. Or, to make it measurable: To save 5,000 lives in Honduras in the next 5 years. But IF I want to focus on guns, I could look at unintentional gun deaths: And the most of them per capita happen in Uruguay. So that's another sub-goal I could have: To save 500 lives in Uruguay in the next 5 years, which would otherwise have happened by accidence.

6. ACTIONS: Translating the sub-goals into actionable steps.

  • What do you want people to do?
  • What is your action you hope happens – physical & visible?
  • What steps come after the distribution of information? [does that question belong here or to step 9?]
  • [what are other questions for that step?]

Example: Unintentional gun deaths wouldn't happen if people wouldn't own a gun. So I want to make it harder for people to get a gun. I want the government to make it harder for people to get a gun, but I also want to make people understand that buying a gun could result in the death of a family member or friend.

7. STRATEGY: Asking HOW the actions should be achieved

  • Who is the audience? Who has the power to implement your goal? What do they care about?
  • How will your action reach the right people?
  • How can you measure the action? Number of calls? People in attendance?
  • Does a strategy already exist? What are other organisations / people doing that align with your effort? How can you work together?
  • How driven are you regarding technical problems vs. social problems? Could you benefit from a collaboration with somebody on the other side?
  • [what are other questions for that step?]

Example: At this point, I need to do lots of research about why there are so many unintentional gun deaths in Uruguay. What has been tried before to decrease that, who is working on that currently? What is the government trying? What are the laws on getting a gun in Uruguay? I'm not a specialist in that, so I'm sure I'll contact NGOs during my research and ask them for their opinion on the topic and for collaborations. Anyway, I have two target groups: My first audience would be the people in the government who can change regulations for buying a gun. My second audience would be the people who are in the decision process for buying a gun. Let's focus on the second audience: How can I reach them? Well, I would need to do some research into the demographics of gun buyers – but without having this knowledge, I'm sure I will find almost all of them right in the process of buying a gun. For that, I would also need to do some research on where people get there guns from. Let's assume they buy them in gun shops: I could address them while entering the shop.

8. Will a data vis support the success of your action? Can you NOT think of any more effective way to make that action successful? If no, abort! (aka implement another strategy). If yes, go to...
Example: Yes! I think communicating the numbers could help.

9. IMPLEMENTATION: Creating your data vis

  • How should your data vis look like?
  • Is the data you have, the data you need? Does your data align with your goals?
  • What are assumptions and biases in your data? *
  • Are all the available data points important to support the success of your action?
  • What is the style / form / medium for your data vis your audience would respond to best?
  • If you would be your audience, would you be convinced by your data vis?
  • [what are other questions for that step?]

Example: Besides showing the numbers, I'm sure what would help even more are concrete stories of unintentional gun deaths in Uruguay, maybe even with shocking pictures. Do shocking pictures help? When do people change their mind the best? I need to do some research into that, too; maybe I can learn from the warnings on cigarette packages in this regard. Also, I need to do some research into communication design in Uruguay and how people get normally addressed. So it shouldn't be only a data vis. But regarding the data vis: I guess I should only show one or two numbers. It should really be more like a campaign; an advertisement. People shouldn't need to read it for longer than 15sec to understand the message.

10. FINAL DATA VIS Nice, you've made it! Now test the impact of your data vis and evaluate with step 4, your mission.
Example: I could test the idea with cameras, to see how people react to the information shortly before they want to buy a gun. Or I just position myself in front of the gun shop and take notes. If the idea doesn't work, I will need to think of another way.

In summary: Yay, I came from designing a dramatic data visualisation with lots of data points which would get shared on Twitter, to a very simple one-or-two-data-points-message in front of gun shops in Uruguay – just because I started thinking about what I actually want to achieve.

Next steps

  1. We should think of more questions, especially for step 9.
  2. We should play through the decision tree with at least three examples for both starting points, to test if the questions make sense or if we need to merge steps.


  • John Emerson @backspace
  • Steve Lambert @SteveLambert
  • Lisa Charlotte Rost @lisacrost

Resources (we <3 links!)