WRT333

PowerPoint

Tufte's PowerPoint Critique

Most talks are now accompanied by what has come to be a nearly obligatory PowerPoint presentation. PowerPoint, a microsoft product, makes it easy to create title slides, to add images and graphs, and to organize text in sophisticated bullets. It gives the user the option of a wealth of visual transitional effects, allowing slides to build up a line at a time, with window-shades, fade-ins, waft down, or text flying in from left of right, at the producers discretion. A treasure chest of styles and background images creates an instant attractive layout, without any real work on the user's part. And, of course, the talk can be modified easily, with the text and images of each slide fully accessible, the order of slides completely sortable, and even a full array of visual transitional effects between slides. Wow.

There are a couple of costs that go with this amazing technology. One of the first is that the technology is now old, and overly familiar. No one is dazzled by the flying texts (or worse, the accompanying zings and zips of sound effects), and increasingly the technology itself grows annoying. In the battle between form and substance, PowerPoint form is definitely stale.

Edward Tufte's 2003 "The Cognitive Style of PowerPoint" deserves our attention as he asks us to consider other costs of using PowerPoint . Tufte is famous for his work on presenting information graphically, in books such as The Visual Display of Quantitative Information and others (here). His criticism may be over the top, considering the social function of most talks (see critical comments, here, for example), but there are at least a few points worth serious consideration.

Tufte begins by telling the story of how new IBM President Louis Gerstner, and a first-day encounter with presentations at a meeting (from Gerstner's 2002 book, Who Says Elephants Can't Dance? Inside IBM's Historic Turnaround")

I think this is the most important message in Tufte's Cognitive Style. That is, before putting together any elaborate presentation, start with a simple consideration of what you want to do. You are talking to someone, and this is the most basic form of human social interaction. You want that someone to understand something. Just what is it that is so important that you have to tell them? What do you want them to understand? What is it about your business that you want to talk about. And don't lose sight of this!

Tufte wants us to see that PowerPoint is enticing and that the enticement is presenter-oriented, "not content-oriented, not audience-oriented."

Tufte builds his contempt for PowerPoint around several arguments.

• Because of the display resolution of computer screens and projectors, PowerPoint is a very low resolution visual media. That is, it doesn't allow for many dots per inch, meaning you can't get clear and dense text or images in any projections. That is, there is little information per slide.
• The tendency to display most text as bulleted items dilutes thought. Bulleted lists imply an order, but do not state one (is the first item more important than the rest; is there a temporary order, the first item preceding the next, etc.); bullets hide order.
• The elaborate hierarchy of PowerPoint bullets tends to obscure relative importance of data, mainly by creating an actor less-timeless substitute for sentences written in agent and action form. Tufte's example of a misuse of PowerPoint in analysis of the Columbia shuttle disaster is widely known (here's a copy from Tufte's web site—note the extensive set of comments following the example, with numerous links to other presentations, critical commentary, etc.; very interesting stuff!).
• PowerPoint has created its own metaphor for talks; Tufte tries to make his point by turning Lincoln's (88 word) Gettysburg Address into a 6 slide (complete with text line builds) PPt presentation. Funny, but is there a real point here?
• Tufte, let it be remembered, is a true guru of presenting dense information. He illustrates how this is impossible in PowerPoint presentations, using a few examples. The only medium dense enough (with enough fine resolution to put many characters on a square inch) is print media. Okay, and the point of all this overkill is...?

Tufte ends with still more piling on.

You get the idea. In the end, Tufte leaves PowerPoint in tatters on the floor. "Good," he might say, "That's what this vacuous phluff deserves." But has Tufte served us well? Where do we go from here?

Suggestions for Use of PowerPoint (with Tufte still in mind)

Tufte's points about information density, stifling influences of extensive bullet lists, and perhaps even his comments on PowerPoint's ability to create a metaphor for talks, all bear reflection. Tufte serves us to the extent that he forces us to become more aware of the limits of the medium and of our audiences. But what do we do with it?

The major difference between an audience of readers and an audience of listeners is the listener's limited exposure—virtually without possibility for recall—to information packaged in slides. Tufte fails to acknowledge the fundamentally conversational and casual nature of oral presentations. The whole point of most such presentations is as much social as scientific in that it has a primary purpose of increasing awareness, and only secondary use as a vehicle for transfer of dense amounts of information. That is, the purpose of talks that use PowerPoint isn't what Tufte apparently desires it to be. For purposes of intimating awareness of what others are doing, or for opening up dialog or awareness in an audience that is less than highly versed in a particular line of research, the amount of information exchanged (more accurately, transferred) must be in the nature of a lower density exchange rate. Slides need to present simpler packets of information, given time constraints in the particular social situation of the oral presentation. That is, when your audience changes from readers to auditors, your rules must change to adopt to the greater limitations listeners face. That seems to me to be the main point that Tufte overlooks.

PowerPoint tables: If we are going to serve listeners well, we have to pay special attention to the singular messages that charts or tables contain. If we can reduce the complexity of a table into a single spoken message ("This table shows..."), then we have the option of presenting that message as text or as a simpler table. We do not, at the least, want to convey such additional information about robustness that is carried in error terms, indexes of shared central limits (i.e., two measurements that are not statistically different), etc. that are the proper content of high-resolution media such as the scientific journal article. We may still want to use a table, particularly where row and column information carries two or more messages, but beyond that, we no longer serve the listener beginning the moment we overwhelm ability to understand our messages. This is not an argument for dumbing down content for a less intellectually adroit general audience, but rather an argument that the place and vehicle for transfer of high resolution, high-density content is not an oral presentation.

The primary purpose of tables is to show the basis for relations between variables and to inform the audience of the nature of that relation, either positive or negative; i.e., in a negative response, one variable—taken to be a stimulus and labeled the independent variable—increases, the other—response, dependent variable— decreases. Tables with multiple columns can also show simultaneous relations, including showing that as one response increases, another decreases, etc. While a more detailed presentation of results also requires indicators of the precision of numbers (given as measures of variance, e.g., a standard deviation) or the robustness of the relation (statistics of slope: t or F ratios), this level of validation belongs in printed articles, and is superfluous for talks. That is, the additional level of information that statisticians require as a basis for validation of the statistical significance of any quantitative claim can, and given the limitations of talks must, wait for subsequent publication in a high-density medium such as a journal article.

PowerPoint charts: When the point of displaying data is to either illustrate how well the data conform to ("fit") a statistical model (i.e., a line or curve drawn so as to minimize the distance between the line and data points) or to clarify that the relation between dependent and independent variables is not linear (no matter the value of x, the independent variable, a given change in x will produce the same change in y, the dependent variable), a graph may be the most effective way to show both goodness of fit and curvature. We certainly would want to show the shape of a curve through the data when our interest was to convey that there was a peak in the relation or that the relation was, for example, increasing in an exponential or asymptotic way.

Powerpoint Illustrations: Photographs or drawings allow us to convey a clear understanding of things that we have seen, in the field or through magnifying lenses. They also make interesting and attractive slides for talks. The use of any image merely requires that the feature of interest be clear, and that if particular fields within the image must be highlighted or labeled, that the means to do so are highly visible from the back of the viewing room. Of course, slides with illustrations should also have a title telling in simple words what it is we are viewing.

Examples:

• Mt. St. Helens eruption (PowerPoint)
• UMass geology 105 Lecture on the eruption of Mt. St. Helens, illustrating effective integration of text and figures (PowerPoint format but stored as a PDF file, for those of you who might not have PowerPoint)
• Lecture on Mayflies for ENT561 Aquatic Entomology, a course I once offered at URI. (PowerPoint format)

PowerPoint lists: Fully mindful of Tufte's admonitions about PowerPoint lists, we do nevertheless find ourselves wanting to put our points into simple text as a aid to the audience. For lists, the fewer words used, the easier the task of reading; use 50 words per list as a norm, 100 as a maximum, adjusting display time (i.e., time for the audience to read) accordingly). Use complete, concise sentences, and parallel syntax between like points. Try to limit the use of elaborate sublists, keeping your slide to two levels at most.

Organization of a Talk

A slide (or PowerPoint) talk should follow a few basic guidelines.

• Always start with a slide that has a talk title, your name, and if possible an image that illustrates the theme of the talk. Use this so that people entering a conference room will know which talk is now beginning (schedules often get skewed).
• Present an overview of your problem, stating the nature and scope of what will follow, in a sentence or two.
• Break the body into major questions, presenting the question, methods, results, and discussion in 1-3 sequences of slides. Here, you are not following IMRAD format, but rather organizing around the question, from beginning to end for each major question.
• Make it clear that you are moving to a new question / topic with a transitional slide (e.g., statement of next question).
• Have a clear ending slide that summarizes your main findings.
• I strongly urge you to also include a slide at the end that acknowledges those who helped you with the research, possibly including a group photograph of your crew.

References

• Edward R. Tufte. 2003. The Cognitive Style of PowerPoint. Graphics Press LLC. 23p. www.edwardtufte.com
• F. Peter Woodford. 1999. How to Teach Scientific Communication. Council of Biology Editors. 199p.