how to produce effective data journalism
DESCRIPTION
Presentation given at FJUM data journalism conference in Vienna 14 June 2012.TRANSCRIPT
How to produce effective data journalism
Bella Hurrell, editor, BBC News Specials team
bbc.co.uk/newsgraphics@BBCNewsGraphics
bbc.co.uk/newsgraphics
Data and journalismData journalism roughly divides into three broad types that often overlap:
1. Traditional investigative data journalism, often called CAR - finding stories in the data – with or without visualisations
2. Using the data to tell a story or explain a complex problem – this will involve graphics or ‘visualisations’
3. Providing a service or a tool that tells the reader something personally relevant - school score cards / tables or simple financial tools and calculators
http://www.bbc.co.uk/news/uk-11333472
http://www.bbc.co.uk/news/world-south-asia-10648909
http://bit.ly/vubzYr
http://www.bbc.co.uk/news/uk-15975720
data journalism tips• It is the stories in the numbers that are interesting NOT the numbers by
themselves• Give your data a human face if you can with case studies or by making it
personally relevant • Investigative journalism can take a long time – Keep focused and work with
experts for the best outcome: investigative / data journalists, statisticians, developers, designers
• Don’t waste time fishing – have insights first – or get the user to help• Clean your data and triple check it – there are ALWAYS errors• Plan publication and partner with a range of outlets for maximum coverage• Build sharing in if appropriate to help boost reach• Always have a page where you explain your methodology• Be prepared to respond to critics and staff to cover corrections and
feedback
http://www.bbc.co.uk/news/business-15748696
http://nyti.ms/1ndlhL
data visualisation tips• Help your readers to understand something complex, don’t just make data
art• Keep your user in mind all the time. Remember you are not a normal user
so your judgement is not the best yardstick • Always test your designs with users and iterate on the feedback• Be aware some people hate graphs. You will never win them over• Circles can be perceived as more ‘friendly’• Keep the UX simple and intuitive. Avoid too many choices as can lead to
user anxiety• Sequential ‘NEXT’ options will often get more clicks than ‘EXPLORE’• Consider audio commentary or using video production tools like after
effects to give an overview
http://www.bbc.co.uk/news/business-17442946
http://www.bbc.co.uk/news/world-15391515
Data tools or apps• Provide information that users will find personally relevant and useful
• If appropriate allow users to share a key fact about themselves to boost reach and make it feel more personal
• A global dataset will be relevant to far more users and will get more shares
Tools• Excel, Google Docs and fusion tables. • Sometimes MySQL and Access databases and Solr for interrogating larger
data sets and used • RDF and SPARQL to begin looking at ways in which we can model events
using linked data. • Developers will use their programming language of choice, whether that’s
ActionScript, Python or Perl, to match, parse or generally pick apart a dataset we might be working on.
• Perl is used for some of the publishing. • We use Google and Bing Maps and Google Earth along with Esri’s ArcMAP
for exploring and visualising geographical data.• High charts javascript library for some data vis• Adobe After Effects – motion graphics software