Stat 515
Bad Graph or Table Redesign Project
Guidance for: Redesign Oral Presentation and Written Report
0. General limits:
Presentation lengths: 10 minutes or less
All presentations need to be uploaded on Blackboard prior to class
All team members are expected to speak a reasonable portion of the time.
Paper lengths: 9 pages or less
Get permission if there is a good reason to go over the limits.
1. Redesign Objectives
This redesign project has two main objectives:
1.1 The first objective is to gain experience applying the class design guidelines using software that is taught in class.
Remember that making comparisons is at the heart of quantitative graphics. The four general guidelines are:
Enable accurate comparisons
If the encoding is poor, consider a position along scale encoding.
If position along a scale is not option consider up grading a poor encoding to a good encoding
Adding grid lines can help.
Simplify appearance
Creating small perceptual groups at one or more scales can help.
Using easily discriminated symbols can help.
Sorting rows and/or column to put similar items close together can help. Proximity linking is often simpler than color linked labels.
Using familiar colors with common names can facilitate communication. Explicit display of changes or differences, such as the differences between observed and fitted values, help us see what is hard to envision.
Provide or increase context to support interpretation or to facilitate hypothesis generation
Include the units of measures. Don’t forget labeling and titling of the graphs! If the data has a spatial context adding a map can help.
If the data is a times series adding more recent data help.
In an exploration in modeling context including one or more potentially related variables may be very helpful.
Attract and engage the reader/analyst
Making text easier to read can help
Labeling a few points and/or providing annotation can draw attention. Colorful graphics can attract readers.
Using interactive or dynamic graphics can engage analytic
If Shiny or other R packages that use java scripts have been taught in class or you learn or have learned about them this is a possibility.
There are tensions among the guidelines. Adding context can make the graphics more complicated while at the same time more meaningful. There are almost always design constraints such as the resolution of the screen or the size of the paper. Designing good graphics often involves picking the task(s) to emphasize and making compromises.
1.2 The second main object to gain experience presenting data analytic results.
The class redesign project involves both a class presentation and a written paper. Both the presentation and paper are to include:
• A copy of the bad graph and/or table (preferably graph).
• The source of the bad graph (and/or table)
• An indication of bad graph or table context, content and intent! The disadvantages of the bad graph needs to be highlighted and suggestions for improving them need to be made.
• A comment (if appropriate) about special efforts. This can address obtaining, processing or creating data and the graphics side of production.
• A brief indication visual communication flaws or failure to show patterns
• Describe suggested improvements and their correspondence to class (or other) design and reasoning guidelines. For example: “The sorting of cases will simplify appearance by putting similar cases close together.”
• Discuss redesigned graph and the patterns that it shows.
• Make one more concluding remarks
o You might comment on the next steps that might be taken
Obtain more data, fit a different model, produce of a different plot
(this is recommended for bonus point)
o You might conclude your redesign is substantially better than the
original design providing evidence
For the redesign project I am generally opposed to using software other the R
2. The Class Visual and Oral Presentations
Presentations running much longer that the allocated time will likely be stopped. Practice and time the presentation. (The instructor may revise the time allocation a little.)
Logistics: Upload the pdf or PowerPoint on Blackboard following the deadline announced in class.
2.1 The class oral presentation
The class oral presentation can differ from the written presentation that is graded. First, it can be less formal and more entertaining. A little tasteful humor is okay. It is okay to poke fun at poor designs (not the designers). It is acceptable to have a classmate ask you a question for which you just happen to have the answer and even a picture. This means you had already considered this thoughtful or insightful question. Presentations are hopefully educational to classmates in terms of the examples or in terms useful resources or methods for producing the redesign. (The presentation is not the forum to provide an instructional lecture. You can respond to questions.) Your presentation may include contain interactive or dynamic graphics that not readily included in a written paper.
The class presentation is to provide a positive experience for both presenters and class members. The presentation setting is intended to be comfortable. For some students this may be their first presentation. For some it may be their first presentation in English. Every student is to be treated with respect! Applause is expected at the end of each presentation.
Respectful questions and constructive comments are fine when the intent is to learn, to make the presenter look good, or to gently guide the presenter in a better direction. Talking or engaging in other activities during a classmate presentation is rude!
Another, but lesser objective is for students to become comfortable with the classroom presentation environment. Setup problems sometimes happen in the redesign presentations. These are taken in stride. Just learn from the problem and move on.
Normally the oral presentation is not graded. However if the oral presentation is not given, if it is in bad taste, or if it blatantly goes against class guidelines points (perhaps all) will be deducted from the final written report score.
2.2 The Written Report
The written report is graded. The grading is based on quality and level effort factors as indicated further below. The report addresses the same topics as the presentation. It shows the bad graph and/or table being redesigned, indicate the graph and/or table source, the context, content and possibly the apparent intend. The report indicates what can be improved in term class criteria, shows an improve graph or table, includes a description of what the author(s) see in the graphic or table and may include comments what more might be done.
The written report differs is some away from the class presentation. While the presentation may have bulleted items, the paper should have clear sentences and perhaps succinct paragraphs.
I expect a writing based on academic writing standards not a bulleted point report! This is a graduate level class so the quality of your writing should meet the requirements of academic writing of proceeding or peer-reviewed articles. Every resources used needs to be references properly (preferably APA format). Failing to reference books, documents, R, R packages and codes, will result in losing points. Copy pasted text (without quoting and fully referencing) is considered plagiarizing.
The talk may mention special effort involve. The paper can also make brief mention of this and provide supportive evident in appendices that do not count again the paper page limit. I expect one slide explaining the challenges and also a written paragraph on your paper.
The initial paper writing is to be done by the project team. Once written, check the spelling and grammar with available tools such as those in Word. After the initial paper writing, in this class it is not an honors code violation to get help in revising paper from the university writing center. It is also okay to have a native English speaker to read the paper and comment.
Omit the humor and tips for classmates that may have been part of the presentation.
Don’t play page length games such as using little fonts, narrow line spacing, and tiny graphics to get more within the page limit. Don’t play the opposite game of increasing font size gaps between lines and large graphics to approach the maximum allowed page size.
Going over the target maximum size will result in losing points.
Asking your instructor about the suitability of a candidate bad graph or table for redesign is fine. The goal is to learn and to develop profession communications skills that make you a valued resource in this culture.
3. Notes
3.1 Exact data is not required on the paper or slides.
The original graph may not come with data in a convenient digital form. You need to track it down and use it for your analysis but I don’t require you to show the data on your slides. The uploading of the data is requested.
3.2 Several suggestions were made above. For example make sure to include the units of measure.
A graph about data is pretty worthless, if we don’t know what is being represented.
3.3 In general layer the information.
Make the most important information the most salient. For example, grid lines should appear in the background. Reference lines can be drawn on grid line. Confidence interval lines can be drawn on reference lines and estimate symbols plotted on top. Methods other than overplotting can also be used to support visual layering.
3.4 Resulting designs to avoid
Don’t just get a pie chart and redesign it to a bar chart!
Since this is a group project, providing multiple graphs is required to show the effort! Adding an inferential method and some analysis results to understand the data or trend is encouraged.
3.4 1 Avoid using perspective 3-D bars in the improved design.
I do not particularly like 3-D perspective bar plots. Bars in front can hide bars in back. A perspective view can complicate reading bar heights against scale.
Side note: Some of the perspective weakness can be addressed. A translucent cutting plane parallel to the bar base that touches the reference scale helps to assess bar height values and to compare the bar heights. A dynamic interactive setting the support move the reference plane can work pretty well. I have seen tools the support large matrices of 3D bars.
3.4.2 Avoid use of pie charts in the improved design
I do not particularly like pie charts. They have merit because are familiar and immediately convey that percents add to 100% (or fraction add to 1).
Angles have good perceptual accuracy of extraction, which is not bad but not as good as position along a scale.
Mostly I find pie charts hard to label. In comparison row-labeled dot plots have easy- to-read labels. The labels link by vertical position to the dots that values using position along a scale encoding, The perceptual of grouping rows can increase linking speed and accuracy.
Both designs can sort values in ascending or descending order. For pie charts this will put smallest value by the largest value.
With pie charts including additional variables in increase context is a challenge.
4. Outstanding Redesigns
Some students chose to invest considerable thought, creativity and effort in the redesign project. I give A+ to redesigns that represent serious efforts to improve a non- trivial graph or table by adapting and applying the methods taught in class, providing multiple ways of improving graphs, and going for the best options (not the easy options!).
5. Grading criteria parallels the above guidance
Level of effort: effort is recognized many areas o + Selection of non-trivial example to redesign
Converting 3 4-item pie charts into 4 lines on a line plot is at best B level work.
o + Attention to appearance details
o + Attention to comparability issues
o + Data gathering and preparation data for analysis/graphics o Development of a good new graphics design
o Discovery of misrepresentation
o Discovery of mismatch between an article and the graphic o Representing a fairly large or complex table
Use of concepts and concepts taught in class
o Perceptual grouping, layering, facilitate comparison
o Sorting and linking (example: using linked micromaps)
o Background grid lines, grid line labels. Avoid tick marks when there is a reasonable alternative.
o Context for appropriate interpretation
Units of measure, legends if necessary o Distributional summaries
o Clean appearance
No points plotted on axes
Easy to read labeling
Minimal memory burdens
Avoid too many font sizes. Use consistent fonts.
o Focus on key the issue
The graphics in some articles are off target.
o Quality of description
Indicate the nature of the data set
The bad design flaws
Indicate the goals for the redesign
Point out the improvements
Clear labeling in the graphics (for example variable names) Clarity of the text