Notes from Data Visualisation and D3.js (Communicating with Data) MOOC by udacity (Highly recommend*****)
Traditional Journalism – spectrum (quantitive (numbers) vs. qualitative (context), Rigorous & Empirical vs. Anecdotal & Ad-hoc. 538 (quant and rigous) and washington post (qual and rigorous)
Correlation vs. Causation
A and B vs A –>B
Traditional Journalism used data around the narrative (data is supportive in a secondary role), Sequence of event flow, physical and/or static
Data Journalism narrative around the data (data is in front and centre), interactive allowing complex and find own story structure, interactive and open web
Common mistakes – not mostly about viz, 90% work is finding, exploring, cleaning, preparing the data before the vizualisation
Don’t lie with data – bar chart = axis show start a 0 , pie chart = no 3D,
Author bias – knowingly or unknowingly misrepresenting data through visual encoding. Your design choices should establish trust between the reader and the graphic.
Data bias – Data bias arises from the process of collecting data. Issues with data collection, sampling methods and other issues. subjective data – Feltron annual personal reports
Reader Bias – Any preconceived notions or assumptions that a reader brings to interpreting a visualization.
5min field guide data & data from web
Types of narrative structure
- intro
- rising action
- climax
- final resolution
Author driven narrative – start to end, strong order, heavy msg, need for clarity and speed, linear and pre-determined
Viewer driven narrative start multiple endings – viewer choose their story, allows ask q’s, explore, tell your own data story.
Martini glass ——< starting point, single path to then explore own story example drone strikes