The Dark Side of Data Journalism: Mapping for Marketing
Here is the chapter I published in Data Journalism: Inside the Global Future, available for purchase On Amazon.
Data journalism is a fast growing domain with increasing employment opportunities. If media often look for experienced applicants, marketing offers good salary prospects to young graduates. Data journalist Isabelle Marchand compiles tricks and tips to help beginners creating advertising visualisations.
Each time I introduce myself as a data journalist, the following moments are generally filled with an awkward silence. “What the hell is a data journalist?!” is the question I often read in the blank expression of my interlocutor’s eyes.
“Data journalist” seems a bit of a redundant expression, as every journalist technically handles data. The difference in my case is that instead of gathering information by interviewing people, I source it from Excel spreadsheets. The story emerges from a succession of numbers, irregularities, average or highest/lowest values. Boring, you may think. Fascinating, I would reply. Every one of us deal with statistics on a daily basis but usually, we let others tell us what they mean. People like politicians, professors, or… data journalists!
“There are three kinds of lies: lies, damned lies, and statistics.” Prime Minister Benjamin Disraeli (1868)
The truth is, a number without context is meaningless. A loophole widely used in marketing, where the story is sourced from data but interpreted following an advertising intention. Where data visualisation provides “a frame of reference to give the reader a way to understand the magnitude of the original number”[i] marketing orients that comparison in a beneficial way for the brand.
A simple scheme illustrates the difference of reasoning between a data journalist working for a news outlet, and a data journalist employed in marketing:
Employed by a creative agency, my daily task is to produce visualisations for the Ford Motor Company. Therefore my first concern when investigating an Excel spreadsheet is to convey a positive message about the brand I represent.
For instance, following a mobility project issued by Ford, a survey has recently been released about commuting in five European cities: London, Paris, Madrid, Berlin and Rome. The desired outcome was to prove that commuting is difficult in major cities and to invest time and resources, as Ford does, into multiple transportation solutions, is essential. When analysing the survey’s results, it is important to keep in mind the brand’s intention: mobility projects are crucial step for future of cities. Hence the analysis of data should demonstrate how strenuous it is to commute.
Surveys stand for “any research effort in which data is gathered systematically from a representative sample population.”[ii] They reveal to be excellent marketing tools as they offer the opportunity to define the initial variables (the questions) while providing data about topics that could not be gathered otherwise.
The reasoning differs from journalism to marketing, as the output does. A data journalist directly reaches its audience, while in marketing pick-up by media are the marks of a successful visualisation.
When both media and audience may share your work, your objective should be to maximise your impact for both categories.
As for classical journalism, title matters, as does the hierarchy of the info. Good storytelling usually follows a simple scheme[iii]:
A brief analysis of 2014 viral infographics shows that five of the top ten have “You” or “We” in the title[iv]. It is not a rule applicable at all times, however employing personal pronouns helps developing a relationship with your audience.[v]
Following that rule, the aforementioned commuting infographic is titled Is Yours The Worst Journey To Work?
Counter-intuitively, statistics that surface large quantities are not always impressive. “80% of people have used their mobile phone when crossing the road” is not surprising. “1 out of 3 have watched a movie on their mobile phone when crossing the road” is more so. The second statement has a greater chance of pick-up by media.
Don’t Sell It, Prove It
Infographics available online can be classified in 6 main categories: informative / persuasive, visual explanations, advertisements, public relations infographics and posters. Informative infographics are the most popular[vi], and represent a golden opportunity for companies to reach an audience that would not normally expect to find their products or services of interest. For instance, the probability of you promoting an infographic about Ford’s latest mobility solution is low. However, it could be tempting to share an infographic aboutcommuting in your city, two variables part of your daily life.
Informative infographics: visualisations giving valuable information without an apparent sales pitch, often issued from extensive researches[vii]. Google: Hongkiat 50 Informative and Well-Designed Infographics for examples
Spreadsheet Programmes (Excel, Google Sheets, and others)
First thing to do is to check the file’s formatting. When working on many worksheets with similar categories such as a list of countries, better be sure the same number of categories is present on each worksheet before copying the data from a page to another.
When processing your worksheet, find a way to verify your results. For instance, add up all partial values to check if you reach the total value. When calculating percentages, do not be afraid to add them together and see if they add up to 100%.
Do The Maths
Most students get scared of data journalism because of statistics. The good news is that there is no obligation to be a maths genius to be a good data journalist. Software will do the calculation. However, being aware of formulas will help to produce relevant comparisons.
Quartile 1, Median, Quartile 3
Percentage of evolution (very useful): Latest value – oldest value)/oldest value x 100
For instance, in 2010 the rate of greenhouse emissions was 0.6g by car, and in 2015 it has been reduced to 0.3g, your calculation will be:
(0.3-0.6)/0.6 x 100 = – 50% greenhouse emissions by car in 5 years!
Keep It Simple
Nathan Yau states: “Chart and graph design isn’t just about making statistical visualisations but also explaining what the visualisation shows”[viii], while Edward Tufte calls “chartjunk” the “interior decoration of graphics (…) that does not tell the viewer anything new.”[ix] Both are statisticians and pioneers in the field of data visualisation, however their approaches are slightly different. Each data journalist develops his own balance between design and data, keeping in mind viewer’s short attention span.
Inspiration, Not Imitation
Research existing infographics on the topic. Get inspiration, but do not try to adapt your data to a preformatted visualisation or you may have to drop important bits that don’t fit in.
Choice Of Graphs
A good knowledge of graph is important, as being familiar with their usage will increase your creativity. After a period of time using them, you should be able to analyse a spreadsheet and quickly make a selection of graphs that can suit your data.
A very useful website that compiles most graphs is Datavizcatalogue.com
Pie charts are widespread, even if mostly decried by the statistician community as they prove to be difficult in discerning small values. However, they are useful to show contrasts.
Do Not Add Extra Useless Info
Each graph should be a part of your argument. Ask yourself: “Will my visualisation still be clear without it?” If the answer is yes, get rid of it.
Where, as in a press release, you can embellish a story with details and explanations, it is not possible with graphs. The visualisation itself proves the point. If statistics do not back up the message, adding a caption explaining why seems pointless.
Newbie Common Mistake
When designing a proportional area chart (meaning using a shape area to represent the data), it is a common error to use the length of the shape to determine the shape’s size. In order to avoid missing on proportionality, it is necessary to calculate the square root of the initial value and to define it as a length.
Let’s say you want to show 8 out of 10 people find commuting stressful. Instead of taking 8 cm and 10 cm as the length of your squares, calculate the square root of 8 and 10, and use that number as a length. There’s no reason to fear the calculation, it is a formula to enter in the spreadsheet and the software will do it for you. In Excel, enter =SQRT(number)
“Create Your Own Visual Style… Let It Be Unique For Yourself And Yet Identifiable For Others”
As part of the same culture, we have a common library of symbols. To use the most recognisable ones helps reaching a wider audience. From my experience homogeneity encourages visual attractiveness. To mix different icon styles may damage the general effect. Once set on a simplistic/complex design, keep it.
Do Not Get Carried Away
A successful visualisation is a subtle balance between an attractive design and selected content. Fall into one of those extremes and the general effect will be impaired. A beautiful visualisation with poor content is as ineffective as an appalling design with strong content.
It gets easy to forget about proportions when designing with software such as Illustrator, where it is possible to zoom in and out. Your audience however may use a smaller screen than your. Export your work at various stages of the design process to check the text remains readable.
… AND TRICKS
Unfortunately, it also happens that the information retrieved does not fully back up the message you wish to convey. Tricks may apply. Don’t get me wrong, there is no magical solution to transform a negative dataset into a positive one. However it is possible to set up a visualisation that reduces negative aspects.
If your goal is to display an increase or a decrease over a long period of time but data shows small irregularities over time, you can group days, months and years together to extract the average or median until the curve proves your point.
If the data is lacking, add projections to show what should happen following your forecast.
Colours will help your audience to interpret the data in the way you would like them to. Red is bad, green is good.
Wording counts! Captions should be positive or negative to back up the point made by the graph. Another example of this is how saying ‘1 out of 3’ can have more impact than measuring by percentage and saying ‘30%’.
WHAT TO KNOW BEFORE SUBMITTING
What Is Obvious For You Is Not For Everyone
On social networks there was a picture of Steven Spielberg sitting next to a dead dinosaur circulating. A bunch of people made comments, outraged that the director could “slaughter such innocent animals” and boast about it. Following the event, my editor-in-chief told me something I will never forget: “Remember that our readers believe dinosaurs are still alive!” He meant that we should never assume that everyone holds the same knowledge as we do.
“If Confidence Is One Key To Success, Enjoying Your Work Is Another”- John Hegarty
Journalists should be confident to stand by their articles. The same is true of data journalists and their visualisations. Usually the employer provides initial statistics. However, comparisons may be based on third party information. Consider the worst-case scenario where inquisitive media may question the work. Make sure sources are reliable.
Journalism is synonymous with long hours, low pay and little recognition. The good news is that data journalism, especially related to communications, is a well-paid area, and it is in the interest of your employer to publicise your productions. Although there is no escaping the long hours, instead of complaining about a weekend spent analysing surveys, be glad to learn new things and your mood will significantly improve.
The First Shot Is Rarely The Good One
With submitting a visualisation comes the never-ending approval process. You will have to modify parts of your work. Do not hesitate to explain why you have selected a graph instead of another when asked. However standing firm where changes bring a real possibility of improvement is not the wisest behaviour. Hippos (highest paid person’s opinion) are not an unusual phenomenon in the corporate world, patience and determination will be your best virtues.
The first years of training seem to be the hardest ones as learning is an everyday process. Being familiar with graphs, comparisons and design softwares takes time. Data journalists such as Simon Rogers, who launched and edited the Guardian Datablog and Datastore, or David McCandless, author of ‘Information is Beautiful’ (2010), have developed their own analysis and design styles. It is what is called “voice” in written press. Failures are also part of the game, but can be avoided by relying on a team, or you may end up on websites such as http://viz.wtf/.
Isabelle Marchand is Data Journalist at PRISM, part of the WPP group. She produces visualizations, infographics and interactive infographics for the Ford Motor Company. She completed a Masters in International Journalism at Brunel University, as well as a Masters in Written Press at the French Ecole Superieure de Journalisme de Paris.
[i] Krum, Randy (2013). Cool Infographics: Effective Communication with Data Visualization and Design. Indianapolis: John Wiley & Sons. p19.
[ii] Panda, Tapan (2006). Marketing Management . 2nd ed. New Delhi: Excel Books. p116.
[iii] Krum, Randy (2013). Cool Infographics: Effective Communication with Data Visualization and Design. Indianapolis: John Wiley & Sons. p27.
[iv] Kristi Hines. (2014). What You Can Learn from the Top 50 Infographics of 2014. Piktochart.
[v] Harvard Business School Press (2005). Power, Influence, and Persuasion: Sell Your Ideas and Make Things Happen. U.S.A.: Harvard Business Review Press.
[vi] Krum, Randy (2013). Cool Infographics: Effective Communication with Data Visualization and Design. Indianapolis: John Wiley & Sons. p69.
[vii] Krum, Randy (2013). Cool Infographics: Effective Communication with Data Visualization and Design. Indianapolis: John Wiley & Sons. p69.
[viii] Yau, Nathan (2011). Visualize This: The FlowingData Guide to Design, Visualization, and Statistics. Indianapolis: John Wiley & Sons. p2.
[ix] Tufte, Edward (1983). The Visual Display of Quantitative Information. 2nd ed. USA: Graphics Press. p109.