Encoding Values and Designing Figures

The effective display of quantitative information involves two fundamental challenges

  1. Best medium (table/graph type).

  2. Designing graph to convey message.

Encoding Values in Graphs

Data graphics ~ verbal language

The rules of graphical communication are rarely arbitrary

Usually based on an understanding of visual perception:

  1. How we see,

  2. Visually encode information for easy and accurate decoding by audience.

Most Graphs Are ...

  • 2D

  • 2 axes (horizontal, x; vertical, y)

Value-Encoding Objects

  • Points

  • Lines

  • Bars

  • Boxes

Points

  • Pinpoint specific location on graph,

  • Encode values associated with scale on each axis.

  • Shapes (dots, squares), filled/open.

  • YES: dot plot, strip plot, scatter plot.

  • NO: time series.

plot of chunk unnamed-chunk-1

Lines

  • Lines connect points.

  • Encode values by location.

  • Ends of line segments mark values.

  • Overall shape of data values.

  • Slope is meaningful.

  • YES: time series.

plot of chunk unnamed-chunk-2

Points & Lines

  • Overall shape & individual values.

  • Multiple time series.

  • Lines should only connect points that are related/connected.

plot of chunk unnamed-chunk-3

Bars

  • Encode values 2 ways:

    • location of end point,
    • length/height

  • Easy to compare lengths to determine relative magnitude.

  • Discrete values, not connected.

plot of chunk unnamed-chunk-4

Bars

  • Bars can be used for connected values (e.g., time series) when:

    • you want to compare individual values at specific times, or
    • histogram

plot of chunk unnamed-chunk-5

Bars need to start at zero

plot of chunk unnamed-chunk-6

plot of chunk unnamed-chunk-7

In this case, points are better

plot of chunk unnamed-chunk-8

plot of chunk unnamed-chunk-9

Relationships

Time series

Values change through time (e.g., $ per mo.)

Ranking

Values ordered by size (e.g., sales, population)

Part-to-Whole

Values represent parts/proportions of a whole (e.g., relative cover, regional sales)

Deviation

Values represent the difference between two sets of values (e.g., income vs outgoing)

Distribution

Counts of values per interval/bin (e.g., number of trees per size class)

Correlation

Comparison of two paired sets of values (e.g., height vs weight)

Geospatial

Values displayed on a map (e.g., population per city, species richness per site)

Nominal Comparison

Values are compared for unordered categories (e.g., regions, fruit type)

Graph Selection Matrix

plot of chunk unnamed-chunk-10