Starting a Data Visualization Project: Guidance for Getting Started

Jennifer Brown
5 min readApr 17, 2021

Before starting a new data visualization project, I ask myself a few questions. These questions help me shape the project and target it to its intended audiences.

In this post, I discuss the questions that every newbie should think about when starting a new project. I use poverty rate as a topic to illustrate my points.*

WHO IS THE AUDIENCE?

This is one of the most important questions to ask at the start of a project. Data visualization is about, duh, visualizing the data. So data is the most important thing (in this case, the poverty rate).

However, successful data graphics are targeted to their intended audience. If the graphic does not speak to the intended audience, then they risk missing the importance of the data or misunderstanding the poverty rate.

Each type of audience wants something different from the project; so it is important to make sure you get a clear picture of who the audience is. Here are a few examples of different types of audiences that may be interested in the poverty rate: The General Public, Policy-makers, Journalists, and Researchers (data analysts/statisticians).

WHAT DOES THE AUDIENCE WANT TO DO?

Each type of audience has a different purpose or goal when looking at a data visualization. This may vary by their level of interest, their understanding of the subject matter, and their experience with the data.

For example, the General Public may want to find themselves in the data compared to others. People love comparing themselves to others, so this audience may want graphics that let see the poverty rate for different demographic groups or even geography.

Policy-makers, on the other hand, may be interested in knowing if the issue has changed over time. That way, they know more about the poverty rate and if they need to address the concern.

Journalists are typically only interested in a good headline, so their interest may be different to other groups.

And finally, researchers will want to better understand the data and its distribution and normality. So their needs are different to others.

WHAT IS THEIR RESEARCH QUESTION?

This is a tricky part: write out the research question that needs to be answered. Each type of audience will have a different question in mind and it is your job to figure out that question and make sure the graphic answers it specifically.

The General Public may ask “what is the poverty rate for people like me?”

Policy-makers could be asking, “is the poverty rate changing over time?”

Journalists may be interested in, “what portion of the poor is made up of women?”

And Researchers will want to answer this question: “are the data normally distributed?”

WHAT TYPE OF GRAPHIC WILL ANSWER THE QUESTION?

The final question I ask myself is what types of charts will answer the audience questions; should the graphic compare across groups or across time, is it a proportion or a distribution that is needed?*

In the case of our specific research questions regarding the poverty rate, the General Public is looking to make comparisons across groups. So I need to give them the rates across different demographics groups, probably in bar charts or dot plots. That way, they can find their own group by race or gender or region or age or whatever, and compare their group’s rate to others.**

Policy-makers in this case want to see the rate over time. They are looking for trends across months or years, so I will give them a line chart that shows the rate at each point in time.

Journalists are looking for the proportion to a whole; that is, what portion of those who are poor are women or men? Proportion to a whole usually means a pie chart or stacked bar chart that totals 100%. That way, journalists can see what proportion of everyone in poverty is female.

And Researchers will want to answer this question: “are the data normally distributed?” in this case, the question is very specific to the quality of the data and will usually involve some more completed statistical graphics such as histograms and density plots.

GIVE THEM WHAT THEY NEED

Now that I know what their research question is and the types of charts that can help, I can start making the charts. And then get feedback from a friend to make sure I am on the right track!

_______________________________________________________________

NOTES:

*Get some help selecting the right chart for your research question

CHART CHOOSER HELP

https://www.slideshare.net/qlik_arg/choosing-a-good-chart-83071353

https://www.storytellingwithdata.com/blog/2013/04/chart-chooser

https://github.com/ft-interactive/chart-doctor/blob/master/visual-vocabulary/poster.png

**Note that graphics are illustrative. For actual rates, please see Census data:

POVERTY STATUS IN THE PAST 12 MONTHS: Survey/Program: American Community Survey

https://data.census.gov/cedsci/table?q=poverty&tid=ACSST1Y2019.S1701&hidePreview=false

--

--

Jennifer Brown

Data Scientist, Data Visualization enthusiast, cat lover, and avid reader.