I’m going to try something new here and go by chapters in my summary so this should be updated often as I get through the book.
At their best, graphics are instruments for reasoning about quantitate information. Often the most effective way to describe, explore, and summarize a set of numbers – even a very large set – is to look at pictures of those numbers. Furthermore, of all methods for analyzing and communicating statistical information, well-designed data graphics are usually the simplest and at the same time the most powerful. intro
The premise of Tufte’s book, graphics at their best are a simple and powerful means of illustrating information.
The first chapter show the readers many different kinds of statistical displays and illustrates their effectiveness.
Graphical displays should:
Tufte runs through milestones in the history of graphical data displays, showing how they evolve from time-series graphs to “narrative graphics of space and time.”
Graphical Excellence is the well-designed presentation of interesting data – a matter of substance of statistics, and of design.
Graphical Excellence consists of complex ideas communicated with clarity, precision, and efficiency.
Graphical excellence is that which gives to the viewer the greatest number of ideas in the shortest time with the list ink in the smallest space. 51
I particularly like the last one, usability! Simplicity and efficiency are always great principles to aim for.
Here Tufte begins to examine what makes graphs flawed – examples in which the display doesn’t fairly represent the data.
Show data variation, not design variation. 61
Data, when represented visually in two or three dimensions, is often grossly misrepresented. Because the data has only one dimension, but it is being represented visually as an area or volume of space, the relation between various data points skews.
Context is essential to integrity.
Six principles which help ensure graphical integrity, from page 77:
Data has been dumbed down by media outlets. It’s a pervasive means of transmitting information, but in it’s current state it conveys little information and is often corrupted somehow. Media outlets in the U.S. are particularly low brow, with very little sophistication present in news graphics.
Graphical competence demands three quite different skills: the substantive, statistical, and artistic. Yet now most graphical work, particularly at news publications, is under the direction of but a single expertise – the artistic. 87
Data graphics should show the data, noting else. Draw attention to the numbers and figures, not unsubstantial artistry.
Data Ink is the non-erasable core of a graphic, the non-redundant ink arranged in response to variation in the numbers represented. 93
The Data-Ink ratio then is the amount of Data-Ink in the graphic divided by the total amount of ink in the graphic.
So to maximize the data-Ink ratio of a graphic, one must:
Above all else show the data. Maximize the data-ink ratio. Erase all non-data-ink. Erase redundant data-ink. Revise and edit. 105
Vibration: the unneeded and obnoxious optical effects which add noise to a graphic. Moiré vibration is bad and clutter. Don’t use hatching and various levels of color to key your graphic – label important areas with words and not hatching.
The Grid: Mostly unnecessary. If possible lose the grid behind the plot, if not possible, make it grey so it falls behind the graphics.
The Duck: What a graphic becomes when it is decorated over-purposefully.
When a graphic is taken over by decorative forms or computer debris, when the data measures and structures become design elements, when the overall design purveys graphical style rather than quantitate information then the graphic may be called a duck in the honor of the duck-farm store, “Big Duck.” 116
Lose all the chartjunk!
Here Tufte works out new versions of various graphics according to the rules he has already laid out. Lose excess ink, within reason, and let the data show itself. This graphic might be beautiful, but it isn’t art. But don’t go overboard. Display as much information in as little space as possible, within reason. Edit and revise your graphics. Don’t confuse your audience, and explain the graphic in text. Newer designs may look odd, but this is because they aren’t common like the thousands of graphics which have come before them. As they grow more popular this will change.
Maximizing data ink (within reason) is but a single dimension of a complex and multivariate design task. The principle helps conduct experiments in graphical design. Some of these experiments will succeed. There remain however, many other consideration in the design of statistical graphics – not only of efficiency, but also of complexity, structure, density, and even beauty. 137
Use the same ink to achieve more then one result!
Mobilize every graphical element, perhaps several times over, to show the data. 139
Instead of just a dot or a line, use a number or a picture. This enables a complex comparison to be simple visually, and results in a high data ink ratio. One shouldn’t have to look all the way across the graphic to find the scale, for example. Incorporate breaks in a bar graph at even spaced intervals so that the height of the bar can be quickly computed by the viewer.
The data density of a graphic equals the number of entries in the dat matrix divided by the area of that data graphic 162.
The Higher the data density the better. The larger the data matrix the better. The size of the graphics should also be minimized, our eyes can handle it.
Small multiples are a set of varying data set onto the same background, illustrating the change from one frame to the next.
Graphical elegance is often found in simplicity of design and complexity of data. 175
Beautiful graphics do not traffic with the trivial. 175
Put words and pictures together. Words on a graphic are data ink. Explain the graphic with words.