In your data visualizations make sure the point you are trying to make is clear. Your point should be obvious even if the people who read your chart aren’t experts. Start your design process by identifying the decisions leaders can make based on your data and identifying how specific points can help them make those decisions. This analysis will lead you to the key point you want your chart to make. Focus on highlighting that single point, and reinforce that point with a clear title and concise caption. What do you want people who see this chart to take away? While it’s true that people’s attention tends to dance around the elements of a chart, we’ll focus on the change in the trend line or the visual outlier first. Take care that those visual outliers align with your central point. In other words, what’s most important about your chart should also be what’s most interesting about your chart.
If you’re considering adding an element that doesn’t contribute meaning or understanding, leave it out. Include just the visual information that is needed for the intended purpose of the graphic, avoid excessive detail, visual embellishments, and dense layouts. If you have multiple points to make, break them into separate charts, each of which should provide enough information for the message or intended task.
To further underscore what you want a decision maker to take away, consider using your main point as the chart’s title. Use the caption to provide context, including factors that may be leading to this data point. A well-written caption can direct your reader’s attention, help them understand, and influence their interpretation. The language you use will influence how your readers think about the graphic (Herald, et al., 2015; Coventry, et al., 2013), and can provide key knowledge needed to interpret the graphic (Kosslyn, 2006; Herald et al., 2016). Captions are useful because it’s important to keep the graphic and the language that explains it together. If the graphic and the associated text are spatially distant, your reader’s attention may be diverted.
2. Use emotion with intention
The relationship between emotion and charts is an area of recent exploration. While data are considered cold or free of emotion, the people viewing it are not. The emotions you elicit affect how people will respond to your chart and the actions they take. For example, someone who has decided what they think about a topic may respond to being affirmed for those decisions.
Emotions matter–especially for charged issues like forced labour or child labour. Kennedy and Hill (2018) note that “[o]ur research showed that when our participants encountered data in visualisations, responses are not confined to the rational” (p.839). Their research found that emotions, especially about the subject matter, play a role in interpretation of visualized data.
Like other forms of information, data visualizations can elicit emotions and their impact can be tempered by demographics, and one’s trust in various data sources. Credibility (trust) of source, or brand, also plays a role, interacts with design interpretation and understanding, in that if the viewer doesn’t like the source of the information, they won’t trust the chart. (Kennedy and Hill, 2018; Allen, 2017).
It’s also true that there’s an “arithmetic of compassion” Dr. Paul Slovic has written extensively about the paradox of caring–we feel for individuals what we can’t feel for groups. So using terms like “one in four” may be more compelling than “one fourth” or “25 percent. ” Some chart designers eager to remind viewers that data are the aggregation of people’s lived experiences have successfully used scatterplots to represent the individual lives whose data is being represented.
3. Flex your credibility
The ILO is a trusted organization and the foremost expert in data related to child labour and forced labour. Clearly labeling each chart with ILO as the source will increase their credibility, especially when the charts are well-designed and speak to a particular decision facing those in a position to address the problem.
But even when people trust the source of the data they’re looking at, if the chart contradicts their existing beliefs or understanding, they may look for ways to discount it. To increase the likelihood that your charts and reports are trusted, it will be useful to prepare a brief statement about how the data were collected, cleaned and prepared.
One more note on credibility–research suggests that people find charts that are easier to read more credible. Perhaps more importantly, they see the information itself as more reliable if they can easily understand the chart depicting it.
4. Know who you are creating for
We can’t assume everyone who sees your charts will have education levels needed to understand complex charts/relationships in data (Kennedy and Hill, 2018; Allen, 2017). When people feel confident that they’re understanding your charts, they’re more likely to engage with them. If the people who are looking at your charts have prior experience and knowledge of the issue, they’ll be more likely to focus on it.
Choose and design graphics with your viewers’ familiarity and knowledge of using graphics and their knowledge of the domain, that is, knowledge about what the data represents in mind. (Kosslyn, 2006; Herald et al., 2016). If your viewers are less familiar with the topic and the data, provide additional direction to them about which features of the graphic are important to look at, for example, in a caption or nearby text, (Herald et al., 2016).
As you choose your format, think about how people in cultures different from your own may draw different meanings from signals than you do. These kinds of differences are part of the reason it’s so important to include people on your project team who have lived experiences that are different than your own, and bring a diversity of perspective to the work.
5. Make visual choices with care
The paradox of visual communication is that while we require visuals to comprehend complex ideas, our visual attention is selective. This means you should present only the visual information you need for communicating the goal at hand, and strip out anything that’s not relevant.
Start by choosing the appropriate plot type, such as a bar plot, scatter plot, or line plot. Make sure you are using the best kind of display for your data to highlight the takeaway–for instance a bar chart is better for comparing among sections of a whole. While it can be tempting to use a pie chart, multiple studies suggest that pie charts can be difficult to read and draw conclusions from.
Use color with care and meaning. For example, using warm or “hot” colors like red, orange and yellow can signal danger, while green and blue are more soothing colors that signal growth. You’ll want to check in with regional experts to ensure that the colors you’ve chosen have the meaning you intend in the culture or region the chart will be used in. Our brains associate other kinds of meaning with color. For example, similar colors signify similar things, and color saturation will signify a progression of whatever it is you’re measuring.
For example, COVID-19 has produced countless charts that many find confusing. This data visualization effectively displayed risk using a single metric: daily cases per 100,000 of population. The chart employs an easy to understand color scheme.
6. Know that different cultures use different metaphors
Conceptual thought often makes use of cultural metaphors. You’ll need to consider differences in direction, color visual perception and linearity in working populations in different global regions. Match the visual representation of data to metaphors that aid conceptual thinking, for example, ‘up’ is associated with ‘good’ and ‘down’ is associated with ‘bad’ (Lakoff and Johnson, 1980); data with negative connotations may be easiest to understand if presented in a downwards direction (Meier and Robinson, 2004; Herald et al., 2016). Provide a legend for color to explain for all audiences (Schaap, 2012).
7. Incorporate context
Data needs context. Data alone cannot move someone to act or shift their perspective. In fact, when people see data without context, they are likely to see that data as a manifestation of what they already believe. Numbers mean nothing in and of themselves. We assign them meaning.
Research–notably that of behavioral economist and psychologist Kahneman and Twersky– suggests that providing context to data in the form of narrative is critical to others’ interpretations of that data. Without that narrative context, people will draw conclusions about the meaning of the data that support their existing worldviews and moral values.
Their insights have been further supported by recent research by Rebecca Hetey of Stanford University. A study by Hetey suggests that when data is offered without context, people will use that data to affirm their existing beliefs. In her study, she explored how context might overcome biases. Earlier work showed that white Americans, hearing that Black people were overrepresented in prisons, used that statistic to support heavier policing of Black communities. Her work points us toward highlighting data that undermines stereotypes. Instead of citing a statistic that 60 percent of traffic stops are of black drivers, it is more constructive to point out that the stops are made for less severe offenses than the ones made of whites. It is also helpful to describe the role existing policy plays in leading us to this place–in the case of forced and child labour, this may be an opportunity to describe the importance of investing in enforcement of existing polices.
It’s also true that without context, using data can be counterproductive in two important ways.
Large numbers trigger pseudo-inefficacy. While large numbers may shock or horrify us, they also have the strange effect of making us believe that we ourselves can’t make a difference. Mother Theresa famously said, “if I look at the mass, I will never act. If I look at the one, I will.” Research backs her up. Research conducted by renowned scholar Paul Slovic suggests that when we are asked to help more than even just one person, we are less likely to. So, larger and larger numbers diminish our personal sense of agency.
Without context, data affirms our existing beliefs. Hetey’s research shows that using statistics to inform the public about racial injustice can backfire, leading them to support policies that exacerbate rather than ameliorate the problem. In one of her studies, she and a fellow scholar found that white Americans did not support criminal justice reform when they read statistics about how Black people are over-represented in prison. Nearly 40 percent of the prison population is African-American, but blacks are 13 percent of the US population. After they heard this data, study participants were more supportive of policies like California’s Three Strikes law and New York City’s stop-and-frisk policy that profile people by race and lead to higher rates of incarceration–exactly the opposite shift that advocates hoped for.
Sharing data about the prevalence of forced labour and child labour is crucial for documenting the challenge and showing trends. It is important, however, to connect that data to what’s working to reduce the numbers in similar regions or sectors. Where you can, draw on evaluations of other efforts that show the connection between specific changes and changes in the numbers of people experiencing forced labour and child labour.
Context helps people understand the significance of statistics, helps them understand what it means for people who experience this condition, and–most importantly–helps them decide how to use the data to make choices.
Some forms of context to consider:
Trends: How has this indicator shifted over time? How might that window capture significant time spans, like shifts in government, treaties or high-level meetings?
External factors that explain the change: What kinds of things happened that caused those shifts? In providing this context, it’s important to be as specific as possible, so instead of saying, “COVID-19 has caused these numbers to increase,” you’d want to be specific about how that happened, by connecting to how lockdowns may have led to loss in family income, or illness may have caused lasting health effects that made it difficult for other earners in the family to return to work.
Images: Should you choose to include images in your documents, it’s critical to examine those images carefully. The images should support the point that’s made by the data.
Story: What stories are available that might help your reader or viewer understand how the experience that the data documents unfolds for real people and communities?
Solution: How have others addressed the problems the data documents? While there are of course no easy answers, providing context in the form of progress other countries have made on ending forced labour and child labour may stimulate ideas or point toward solutions in another country.
Scale: Our brains aren’t good at processing or differentiating among huge numbers, so putting the number into a scale or schema that our brains can process can help people remember and understand concepts.
8. Give yourself time to create, solicit feedback and reflect
As you create your charts, step back and look at them. Do they convey what you’d hoped they would? Try drawing them out and sharing them with a colleague or, even better, someone you trust who has insight about the people who will use them. You may want to share them with the team you established in Step One. Listen to their feedback, apply it and share it again to ensure that you’ve captured and demonstrated what you’ve intended to. Ask for feedback about each element of the chart: whether the font is legible, what meaning they took from the colors you chose, and finally ensure that the meaning they took from the chart is the one you intended.
