(c) Chris Curran, Data Display Techniques Christine R. Curran, PhD, RN, CNA October, 2001
(c) Chris Curran, Data Versus Information How does one determine which display format to use: Text, Table, Graph, Other…? How does display content / “ink” affect the amount of information obtained? rounding of numbers Labels: when and where use of white space How does color affect our ability to “see” information?
Human Cognitive Processes Humans want to organize data The human mind operates by association Humans process data through data reduction strategies Chunking of data Pattern recognition & exceptions to patterns are used to make judgments Analogy & metaphor are often used in learning & recall of information
(c) Chris Curran, Data Displays Should Facilitate Perception of salient features Comprehension of information Recall of the information
(c) Chris Curran, Data Versus Information Methods used to glean information from the volumes of data available to us: tools (calculators, computers) decision support systems data presentation
(c) Chris Curran, The data and the type of task drive the choice of display format How to Choose a Display Format
(c) Chris Curran, Types of Data Displays Words Headings Text Numbers Digital Numeric Table Analog Picture Graph Icon Video
(c) Chris Curran, Words Avoid all capital letters Use labels or symbols rather than a “key” Use Serif font for text Use San-serif font for headings
(c) Chris Curran, Text: Samples TITLE Text should be displayed in Serif font. One should avoid all capital letters. Title Text should be displayed in Serif font. One should avoid all capital letters.
(c) Chris Curran, Properties of Numerical Data Displays Digital task: symbolic data: discrete, quantitative focus:specific process: analysis display: table Analog task: spatial data: continuous, qualitative focus:holistic process: perception display: graph, icon
(c) Chris Curran, Principles of Numerical Data Displays Arrange data to convey meaning proximity of data use of white space navigation Make patterns and exceptions within the data obvious at a glance (seeing the data) rounding labeling & spacing display format
(c) Chris Curran, Digital Display: Tables Use in small data sets (20 numbers to be displayed or less) Used to display numbers
(c) Chris Curran, Rules for Table Displays Ehrenberg, 1977 Round to 2 significant or effective digits eliminate leading “0” trailing “0” does not matter Put figures to be compared in columns rather than in rows Add row & column averages (make the main effects explicit) Order rows & columns by size Show larger numbers above smaller numbers
(c) Chris Curran, Rules for Table Displays Ehrenberg, 1977 Spacing & layout White space is your friend Use white space to signal the chunks of data Single spacing guides the eye down the column Use gaps (white space) between groups (columns or rows) to guide the eye across the data & to cluster data Data meant to be compared should be close together
(c) Chris Curran, Data Rounding “Anyone who cannot learn to cope with rounding errors will probably not get much out of statistical data” Ehrenberg, 1977, pg. 282
(c) Chris Curran, Principle The Data should drive the order of the presentation. Displays should not be configured by the structure of the data collection methodology or analysis.
(c) Chris Curran, Table: Example
(c) Chris Curran, Table: Revised Example
(c) Chris Curran, Correlation Matrix: Example
(c) Chris Curran, Correlation Matrix: Example
(c) Chris Curran, Correlation Matrix: Revised Example
(c) Chris Curran, Graphical Data Display: A Form of Decision Support Goals find relevant data in a dynamic environment visualize the semantics of the domain reconceptualize the nature of the problem (Bennett, Toms & Woods, 1993)
(c) Chris Curran, The Power of a Graph Enables one to take in quantitative information in a qualitative way, organize it, and see patterns and structure not readily revealed by other means.
(c) Chris Curran, Graphical Perception The process of visual decoding of quantitative and categorical data from a graph. Cleveland, 1984
(c) Chris Curran, Analog Display: Graphs Used to display large datasets Types of Graphs: Universal - Literal Continuum
(c) Chris Curran, Universal Graph: Example
(c) Chris Curran, Literal Graph
(c) Chris Curran, Graphical Design Concepts & Principles Semantic Mapping (Roscoe, 1968; Kosslyn, 1989) Configural Displays (Garner, 1970) Chunking (Newell & Simon, 1973) Theory of Graph Comprehension (Pinker, 1981) 8 Visual Variables (Bertin, 1981) Emergent Features (Pomerantz, 1981) Data-Ink Ratio & Small Multiple (Tufte, 1983,1990, 1997) Elementary Perceptual Tasks (Cleveland & McGill, 1984) Proximity Compatibility (Wickens, 1986) Metaphor Graphics (Cole, 1988) Cognitive Fit (Vessey, 1991)
(c) Chris Curran, Design Principles for Computer Displays (Cole, 1994) Design for the analog mind and both hemispheres Design for correct encoding of information (represent the user’s model) Provide a clear context
(c) Chris Curran, Graphic Design
(c) Chris Curran, Visual Decoding of Graphs Requires Pattern Perception Pattern perception requires: detection visual grouping of a pattern estimation
(c) Chris Curran, Elementary Perceptual Tasks (ordered from most to least accurate) Position along a common scale Positions along nonaligned scales Length, Direction, Angle Area Volume, Curvature Shading, Color Saturation Cleveland & McGill, 1984
(c) Chris Curran, Position Along a Common Scale
(c) Chris Curran, Position Along Non-Aligned Scales
(c) Chris Curran, Length
(c) Chris Curran, Direction
(c) Chris Curran, Angle
(c) Chris Curran, Area
(c) Chris Curran, Volume
(c) Chris Curran, Curvature
(c) Chris Curran, Shading
(c) Chris Curran, Color Saturation
(c) Chris Curran, Elementary Perceptual Tasks Cleveland & McGill, COLOR SATURATION
(c) Chris Curran, Common Graphs by Elementary Perceptual Task
(c) Chris Curran, Recommendations: Based on Graphical Perception Parts of a Whole dot chart grouped dot chart bar charts (instead of divided bars or pie charts) Framed Rectangle Charts (instead of Shaded Statistical Maps Cleveland & McGill, 1984
(c) Chris Curran, Dot Chart
(c) Chris Curran, Grouped Dot Chart
(c) Chris Curran, Bar Charts
(c) Chris Curran, Grouped Bar Chart
(c) Chris Curran, Divided Bar Chart
(c) Chris Curran, Pie Chart
(c) Chris Curran, Framed Rectangle
(c) Chris Curran, Research Findings: Graphical Perception Perception of Change Line Graphs Grouped Bar Graphs Perception of Proportion Pie Charts Divided Bar Graphs (differs from Cleveland & McGill) Hollands & Spence, 1992
(c) Chris Curran, Cognitive Fit Vessey, 1991
(c) Chris Curran, Proximity Compatibility Principle To the extent that multiple aspects of data or information must be mentally integrated, they should be physically integrated or proximate in the display. Wickens, 1986
(c) Chris Curran, Emergent Features A property of the configuration of multiple dimensions of an object that does not exist when the dimensions are specified independently. Pomerantz, 1981
(c) Chris Curran, Innovative New Designs Metaphor Graphics
(c) Chris Curran, Clinical Data Display Cole, 1988
(c) Chris Curran, Metaphor Graphics: Database Display (Cole, 1988) Male Female
(c) Chris Curran, Clinical Data Display Patient Rectangle Ventilator Rectangle Rate (width) Volume (depth) Oxygen Alveolar Space Dead Space
(c) Chris Curran, Metaphor Icon Graph: Example
(c) Chris Curran, Metaphor Icon Graph: Questionable Example
(c) Chris Curran, Clinical Data Display Powsner, S. & Tufte,E.R., 1994
(c) Chris Curran, Tufte, 1997 Clinical Data Display
(c) Chris Curran, Color
(c) Chris Curran, Why Do We Use Color? Formatting Purposes Group data (patterns) Create focused attention to specific data (highlight data) Semantic Purpose (encode data) Create Realism Aesthetic Purpose (visual appeal)
(c) Chris Curran, Color is superior to size, shape or brightness as a mechanism to target a feature in a display When to Use Color
(c) Chris Curran, Eleven Colors That Are Never Confused White Black Gray Red Green Yellow Blue Pink Brown Orange Purple Kosslyn, 1994
(c) Chris Curran, General Guidelines: Use of Color Use warm colors in the foreground Have a large luminance contrast between the foreground and background Adjacent colors should have different levels of brightness Redundant color coding improves search tasks Color should be a secondary cue (always design for monochrome first) Travis, 1991
(c) Chris Curran, Kinds of Color Contrasts Light - Dark Cold - Warm Contrast of : hue saturation Complimentary Contrast (from color wheel)
(c) Chris Curran, General Guidelines: Colors Colors have cultural significance. Each individual sees, feels, and evaluates color in a very personal way.
(c) Chris Curran, General Guidelines: Colors Red: alert values, “warning” Blue: most easily distinguished but does not photocopy well Optic Yellow: (a greenish yellow color) most visible to humans Shades of Grey: Best for those who are color blind Use Conventional colors (e.g., forests are green)
(c) Chris Curran, Take Home Message How Data are displayed matters Displays should be configured around the data and not how it was obtained
(c) Chris Curran, Things to Consider How much data do I have? What cognitive task is needed? Are the data continuous or discrete? Am I making an exact or a relative judgment? Are the data static or dynamic? Are there display conventions about the subject area? Is the domain familiar to the audience?
(c) Chris Curran, Current Recommendations Tables: good for small datasets & to depict quantitative data where specific data are needed Pie Graphs: good for judging proportion Bar Graphs:display change or trends; excellent universal graph (better than line) Icons: good for synthesis of data and meaning; may be best for qualitative (relative) judgments Shapes / Figures: Best to display integrated data