Wednesday, September 12, 2007

Sure-fire ways to spoil data, Part I

How data can be misleading ...

"In his bestseller “The Situation Is Hopeless, but Not Serious”, Paul Watzlawick describes how people can pursue unhappiness (or remain unhappy if they already are). In a similar yet unrelated topic, Howard Wainer explains how people can make consumers of charts miserable as well. According to the American statistician, anyone can succeed in spoiling data by following a few, simple guidelines:

1. Show as few data as possible (i.e. Minimize the data density). If the chart looks too empty because it only contains a handful of values, fill the rest with pretty pictures that do nothing to help further explain the facts.

2. Hide what data you do show. And there are many effective ways to do so. Use a flashy grid and print the data on it – but in a subdued color. Alternatively, you can firmly abide to the “never chop the axes” rule so that the interesting differences among the data are barely visible.

3. Ignore the visual metaphor altogether. Do not sort data by size even if the chart type and data allow it. For example, use a bar chart but sort the data in alphabetical order."    (Continued via Me, myself and BI)    [Usability Resources]

Pictures Clouding Data - Usability, User Interface Design

Pictures Clouding Data

0 Comments:

Post a Comment

<< Home

<< Home
.