tl;dr A book detailing the architecture of data-based graphics

For my first review I wanted to start with one of my favourite books, The Visual Display of Quantitative Information by Edward Tufte.
This book covers, with excellent detail, the world of 'graphs' in particular graphs that deal with numerical (quantitative) information, hence the name.

Anscombe's quartet

Tufte argues that a well designed and presented visualisation or graphical display of information delivers far more value than a table of values would or could ever produce. For example Anscombe's quartet is a set of 4 datasets with nearly identical descriptive/summary statistics but which are very clearly widely different when visualised (mean, variance, linear regression equation, and correlation co-efficient are identical to 2 decimal places)

The book pursues graphics from the perspective that the ultimate objective is to present an accurate representation of the information that transfers the largest amount of information in the shortest amount of time, using the least amount of space to a reader or viewer. This is explored by investigating and uncovering subtle tricks that can be used in the manipulation of graphics for an ulterior motive, such as the framing for a story or political point of view.

After providing an in depth view into good and bad practice of graphs, Tufte elaborates and presents some theories, that I affectionately refer to as the architecture of graphs, for designing, improving, and avoiding regarding graphs.

Some powerful concepts that have stuck with me are:

  • Data / ink ratio
    The ratio of ink in a graph that adds useful/numeric information (e.g. data points) to that which adds no extra information (e.g. grid lines)
  • Lie factor
    The disproportionate amount that physical quantities/lengths are varied with varying numerical quantities
  • Chart junk
    Visual clutter that can be added that obscures and makes a graphic more difficult to interpret
  • Sparklines
    High density graphics, about the size of a word, used to convey a numerical trend.

Finally a favourite quote of mine from the book:

"if the statistics are boring, then you've got the wrong numbers. Finding the right numbers requires as much specialised skill - statistical skill - and hard work as creating beautiful design or covering a complex news story."

The Visual Display of Quantitative Information is available from Book Depository
Note this is an affiliate link, which means that if you click through the link and make a purchase, I will receive a very small percentage of the purchase price. This does not increase the cost to you, in any way.