# 3.4 Storytelling with data visualisation

## Highlight what’s important, tell one story

**What is a story**

A story is a connected series of events presented in a specific order such that they create an emotional reaction for a given audience.

In our case, a story is told through visuals of statistical information. Ideally, these create a compelling narrative that convinces the audience of the importance of your data insight.

**Audience**

Telling a story starts with the audience. The starting point of your figure's story is the audience's domain-relevant knowledge. The audience's context should frame your story.

However, even experts in a field take time to process complex visualisations. Assume that the reader needs as much help as possible from you to understand the key trends and relatioships shown. Keep your figures simple, avoid irrelevant or tangential information, and augment your visuals with narrative text that will help the readers infer the intended conclusions aobut your data.

**Tell one story**

Although it is possible (and far too easy with modern visualisation packages!) to tell many stories using a single chart, it is easy to overload your audience with information. Every addition to a figure increases the required mental effort of your audience. Too much information is discombobulating - the reader does not know what to focus on and your figure is diminished. Aim for _ink parsimony_, be careful to include only the visual elements that increase understanding, and embrace white space. Remember your are competing for the viewer’s time and attention!

Best practices used in data visualisation, story-telling and information design is highly influcence by the renowed statistical graphics expert Edward Tufte. An excellent visualization, according to Tufte, expresses “complex ideas communicated with clarity, precision and efficiency.” In 1980s, Tufte proposed a metric for measuring the amount of superfluous information included in a chart. He called it the data-ink ratio,  saying “the larger the share of a graphic’s ink devoted to data, the better.” 

Having said all this, sometimes a single static visualization is not enough to tell an entire story. You may want to design a sequence of simple figures that jointly create a convincing story arc. When connecting a story over multiple figures consider using different types of visualization for each distinct analysis to help the reader's focus. Furthermore, keep semantic indicators (such as colour scheme) consistent across figures. If you have many similar types of analysis an alternative is to use interactive figures and animations that enable the audience to navigate through the story.

include expandable example here -

### Titles and annotations

The first thing that your reader will see is the title of your visualisation, so is important at this point that the title not only describe what is being measured, but also why the reader should care and how to read the chart.

Depending on the audience and the story you want to tell you can use the title can adopt different styles (from [here](https://www.eea.europa.eu/data-and-maps/daviz/learn-more/chart-dos-and-donts/#message)):
- Descriptive title: this is more appropriate for technical audiences with a background knowdlege of the topic. It gives a neutral overview of the analysis being presented.
- Definitive statement: suitable for general audiences, it communicates clearly your message and the insights.
- Question: posing a clear question in your title and providing an answer to it with the data it will transmit a clear message to your audience (suitable for non-technical audiences).

You can also use subtitles to stregten your message and provide more depth on the insight obtained from the data. If your audience has no prioir knowledge on the data, annotations can improve the figure readability and provide additional detail.

### Hierarchy of the information

Consider how your audience will scan and read the visualisation.

Without any visual cues, when we’re confronted with a block of text,
our only option is to read it

we can employ preattentive attributes to create visual
hierarchy in our communications, these attributes can be colour, contrast, font, size, etc.

Figure X. Example of how to use attibutes such as colours, font size, title and anotations to create a hierarchy of information that directs the reader. Figure from the [European Environment Agency](https://www.eea.europa.eu/data-and-maps/daviz/learn-more/chart-dos-and-donts/#message).

## Emotion and context in story telling

**Emotion**

Discussing data-ink ratios (see tip above) and information can feel cold and calculating, contradictory to the idea that a good story takes the reader through an emotional journey. The emotional component of the story is that grabs your attention and makes the story memorable.

So how can you invoke emotion and minimize the data-ink ratio? In this course we do not have a clear answer for this, but we want to use an example to get you to think about it.

This comparison comes from the [Data Feminism book](https://data-feminism.mitpress.mit.edu/pub/5evfe9yd/release/3). They compare two different ways of visualising the same problem: gun violence in the US. One figure is a tipical bar chart from the Washington Post (figure X) showing a bar char with the number of active shooters incidents annually. The other figure comes from Periscope (Figure XX ), a design firm with the tagline “Do good with data,”, that took a different approach. Quoting the Data feminism book (D’Ignazio and Klein, 2020):

"But what makes Periscopic’s visualization so very different from a more conventional bar chart of similar information, such as “The Era of ‘Active Shooters’” from the Washington Post? The projects share the proposition that gun deaths present a serious threat. But unlike the Washington Post bar chart, Periscopic’s work is framed around an emotion: loss. People are dying; their remaining time on earth has been stolen from them. These people have names and ages. They have parents and partners and children who suffer from that loss as well."


Figure X.X. A bar chart with the number of "active shooters incidents" from the United State between 2000 and 2015.

vs

Figure X.X Visualisation of the "stolen years" of people killed by guns in the United States in 2013. Figure by [Periscope](https://guns.periscopic.com/).

**Context**

In the previous sections we have made emphasis in how to use data storytelling to pass a message through emotion. It is clear that tailoring a story in this way can never be neutral or objective. However, this is not only limited to a story or message, data is ever neutral or objective and there is no such thing as "raw data".

Instead of taking data at face value and looking toward future insights, data scientists must work with domain experts to first interrogate the context, limitations, and validity of the data being used. Furthermore, this should not limit itself to the stages of data acquisition or data analysis, context also comes into play in the framing and communication of results.

The following example from the Data Feminism book show how the variation in the title and framing of the chart can highlight the context in which data data was taken. and the message the author wants the audience to take home.


Figure X. Two portrayals of the sama data analysis. The data are from a study of people incarcerated for the first time in jails between 2011 and 2013. Data from Fatos Kaba et al. "Disparities in Mental Health Referral and Diagnosis in the New York City Jail Mental Health Service". Graphics by Catherine D'Ignazio, extracted from Data Feminism (D’Ignazio and Klein, 2020).

References:

- https://clauswilke.com/dataviz/telling-a-story.html#what-is-a-story
- http://www.bdbanalytics.ir/media/1123/storytelling-with-data-cole-nussbaumer-knaflic.pdf
- https://www.eea.europa.eu/data-and-maps/daviz/learn-more/chart-dos-and-donts/#message
- Data Feminism https://data-feminism.mitpress.mit.edu/