Dynamics in Document Design by Karen A. SchriverNotable passages from
Dynamics in Document Design
by Karen A. Schriver

Wiley, New York, 1997

   Document design is the field concerned with creating texts (broadly defined) that integrate words and pictures in ways that help people to achieve their specific goals for using texts at home, school, or work. As Simon (1981) reminds us, "everyone who designs devises courses of action aimed at changing existing situations into preferred ones" (p. 129). Document design is the act of bringing together prose, graphics (including illustrations and photography), and typography for purposes of instruction, information, or persuasion. Good document design enables people to use the text in ways that serve their interests and needs. While documents must also meet the requirements of their clients, the reader's needs should drive design activity. In this way, document design is different from advertising in that advertising focuses on writing and visualizing in order to promote the goals and values of organizations rather than to promote the goals and values of readers. The challenge for document designers lies in developing courses of action that will change existing situations into preferred ones for the people who make use of our work. (pp. 10–11)


[]

   In the next section, I overview these three audience analysis models: (1) classification-driver, (2) intuition-driven, and (3) feedback-driven. Understanding them can help document designers make more perceptive choices about when to rely on one model or another.

Classification-driven Audience Analysis

Developed during the 1960s, classification-driven audience analysis provides professional communicators with methods for creating profiles of their anticipated readership, often called the "target audience." Communicators begin their analysis by brainstorming about the audience and be cataloging audience demographics … or psychograhics…. These audience profiles are then used to classify the audience into groups, for example, nontechnical or technical, general or specialized.
   Although these categories may suggest what sort of prose and graphics the audience might want, the leap between audience analysis and textual action is quite large. Authors of books about writing and design tend to skirt the issue of how professionals actually put these analyses to use….
   A strength of classification-driven models is that they prompt communicators to think about the needs and expectations of different groups for their documents….
   The weakness of classification-driven models is that they encourage a rather narrow and static view of readers. They tend to lead communicators to focus on the similarities within reader groups and to ignore their diversity. A key feature of the classification-driven models is that they "fossilize the reader" as a static compilation of demographics and psychographics that document designers somehow "keep in mind" as they compose.

Intuition-driven Audience Analysis

Described by rhetoricians and writers of fiction since the 1950s, the intuition-driven model of audience analysis is one in which communicators imagine the audience and draw on their internal representation of the audience as a guide to writing and design. In using this model, document designers look inward to "visualize the audience" or to "listen to their inner voice" as they compose….
   The strength of intuitive models is that they capture, in ways that other models do not, the phenomenon that skilled communicators are good at "doing things with words and pictures" that get the audience's attentions and keep it—that good communicators are sensitive to visual and verbal rhetorical moves the resonate with readers. The limitation of intuitive models is that they lead document designers to not question the adequacy of their own judgments about the reader.

Feedback-driven Audience Analysis

Feedback-driven audience analysis provides a view of real readers engaged in the process of interpreting texts. Studies of readers-in-action show in considerable detail that audiences come to texts with knowledge, needs, values, and expectations that dramatically influence how they interpret what they read. The image of the audience that emerges from feedback-driven methods is of people who engage with documents in order to understand, access, and use them for pragmatic purposes….
   A strength of feedback-driven models is that the representation a document designer forms about the audience is likely to be much more oriented toward real people reading and comprehending than it would if the document designer were using other models. Feedback-driven models allow document designers to get a detailed view of how particular people interpret sentences, paragraphs, illustrations, diagrams, and so on. Watching people read provides firsthand insight into what makes documents easy (or hard) to understand. Listening to readers also alerts document designers to the differences among readers and to differences between readers and themselves….
   A weakness of feedback-driven models is that like the other models, there is still a gap between forming an image of the audience and taking action based on that image. (pp. 155–162)


[]

   What we know now is that most people choose to read and to keep reading only when they believe there will be some benefit in doing so and only when they cannot get the same information in easier ways (for example, by asking someone else). in order to help readers recognize the documents (or the sections thereof) that deserve their consideration, document designers must do at least two things. They must visibly structure the document so that the main ideas catch the attention of busy readers. At the same time, they must use language (both visual and verbal) that connects with readers' knowledge, experience, beliefs, and values. (p. 166)


[]

Where Participants Assigned Blame

   Many people find that reading documents can be a kind of punishment, a loathsome activity that produces confusion, frustration, and downright resentment. The survey asked consumers to describe the difficulties they had experienced with electronics products and manuals. After asking participants what they liked and disliked about the products and the instruction guides, we posed the question, "If you experienced a problem of any sort while you were trying to use your product, where did you assign the blame?":

12%
13%
6%
63%
6%

   To the manual
   To the machine
   To the manufacturer
   To myself
   Don't remember

(p. 216)


[]

   Document designers need to anticipate those critical juncture points in carrying out a task that may lead users to a cognitive impasse—to a breakdown in completing the task. They need to give users a clear idea of the consequences of their choices, especially when choices are either nested with or contingent on other choices…. (p. 245)


[]

   The popular literature tends to focus on how technology can be used to combine prose and graphics instead of on how to choose the words and pictures that people need, want, or prefer. When document designers focus their energies on technological issues for organizing and displaying documents without first figuring out how the prose and graphics will be experienced by readers, they increase the likelihood of creating vibrantly colorful documents that are disjointed, hard to understand, and ugly. (p. 362)


[]

   The shared aspects of language are dramatically reflected in language communities: the French, the Japanese, the Navajo. Within each language community, individuals share an incredible variety of connections between signs and meanings. According to Benjamin Lee Whorf, the early twentieth-century linguist, the communal connections reflect a shared world view. He put it this way:

We dissect nature along lines laid down by our native languages…. We cut nature up, organize it into concepts, and ascribe significances as we do, largely because we are parties to an agreement to organize it in this way—an agreement that holds through our speech community and is codified in the patterns of our language. (1956, pp. 213-214) (p. 365)


[]

   Much of the research on how people read unfamiliar material, that is, stuff they've never seen before, shows that they sometimes rely more on the text to figure out the meaning (Danks & Hill, 1981) and at other times rely more on their own knowledge (Perfetti & Roth, 1981).
   So what does all this mean for document designers? By implication, we need to make textual moves in our documents that will help readers with both their knowledge-driven and text-driven constructions of the text and graphics. On one hand, most books about professional writing have suggested that the reader's understanding of sentences is the main thing to worry about. On the other, books about visual design have tended to concentrate on the relationships among design elements from the designer's point of view. Both traditional foci fail to emphasize the reader's knowledge, attitudes, values, and culture as potent resources for interpretation. Writers should not abandon their attention to style and linguistic precision. Similarly, graphic designers need not give up their interest in forum and visual tension. But both writers and designers need to widen the focus of their lens in order to learn more about how people actually interact with graphics and typography. (p. 368)


[]

   In interviews, many people mentioned hating to read hardcopy documents that had been simply "dumped" online. (p. 382)


[]

   [U]sers tended to prefer online documents when they thought one or more of the following might be the case:

  • The task would take little time and would allow them to "get in and out" of online help quickly.
  • The help file would be short and would allow them to control the size of the help window and its rate of display.
  • Displaying the help file or online document would not destroy the context of their work (the integrity of their files was more important than getting help).
  • The help would be procedural rather than definitional (task-oriented information rather than topic-oriented). (p. 385)


[]

113 participants ranked the sources [of information] in the order they would prefer to use them, assuming they could get the same information from each documentation source:

  1. Online Help
  2. Hardcopy Manual
  3. Ask Another Person
  4. Online Manual
  5. Hardcopy Handouts
  6. Informal Notes
  7. Online User Consultant
  8. User Consultant on Duty (p. 386)


[]

   Just as readers have come to expect high-quality visuals, they have come to assume they will get interesting and well-written prose. Most people want the prose to be as concise as possible, but not so short that it becomes cryptic and with ambiguous. Readers of the Virtual Tourist remarked that paperback travel guides were more detailed, often funny, gave a sense of the author's personality, and evoked a personal feeling they did not experience online. (p. 398)


[]

   Levin and Lentz (1982) evaluated the research literature that compared text only, illustrations only, or both text and illustrations. They found that of the 46 experimental studies they reviewed, in all but one learning was better with text and illustrations than with text alone. Further, in 81 percent of these cases the differences in favor of text and illustration combinations were significant. The largest benefits were found for people with poor reading skills, who on average performed 44 percent better with text and illustrations than with text only. By contrast, more skilled readers performed 23 percent better. These results clearly indicate that the text plus illustrations combination is typically superior to text along. Further, the results suggest that if the audience consists of less able readers or reluctant readers, it is especially important to combine well-designed visuals with the text and to avoid a text-only approach. (p. 408)


[]

Three key ways to integrate prose and graphics:

  • Redundant--characterized by substantially identical content appearing visually and verbally, in which each mode tells the same story, providing repetition of key ideas.
  • Complementary--characterized by different content visually and verbally, in which both modes are needed in order to understand the key ideas.
  • Supplementary--characterized by different content in words and pictures, in which one mode dominates the other, providing main ideas, while the other reinforces, elaborates, or instantiates the points made in the dominant mode (or explains how to interpret the other). (p. 413)

Redundant

   The more difficult the topic, the more likely it is that the audience will welcome redundancy. (p. 413)

   A study by Atman and Puerzer (1995) tested four ways of presenting information about the process of global warming:

  1. A continuous prose paragraph
  2. An itemized list
  3. A flowchart with a prose paragraph
  4. A diagram with a prose paragraph

   Atman and Puerzer found no significant differences in readers' comprehensive of the main points about global warming across the four ways. However, readers overwhelmingly preferred the diagram with the prose paragraph (Example 4). Participants reported feeling more comfortable with the pictures and words, preferring the more literal diagram over the flowchart. But keep in mind that their preference for the diagram did not increase their understanding. As I noted earlier, what people may prefer to read may offer them no better information, but may keep them reading long enough to get the most from a document. (pp. 414 - 415)

Complementary

   This assignment required that students document a procedure for "removing, inserting, and verifying batteries" in three ways:

  1. Words only (with a modular grid and typography to signal the textual hierarchy).
  2. Pictures only, including icons or directional symbols such as arrows (numbers and minimal headings were also included).
  3. Words and pictures (designed to instantiate a relationship between prose and graphics, such as complementary or supplementary).

   In an informal usability test of these alternatives, Sargeant found that readers preferred Example 2 (the pictures-only version) but performed more accurately with Example 3 (the words and pictures version). Notice how difficult it is to provide information about verifying the batteries using only pictures. As Bieger and Glock (1986) point out, a diagram is useful for spatial, contextual, and configural information, but prose may be a better source of procedural details. (p. 417)

Supplementary

   The literature on problem solving suggests that examples can be useful for learners, particularly if they are used actively in thinking through an idea (VanLehn, Jones, & Chi, 1992). A number of studies suggest that supplementary words or pictures may enhance learning. For example, Bernard (1990) found that extended captions can improve learning from instructional illustrations, suggesting that document designers may need to consider ways to more effectively elaborate captions to prompt readers to process the content more thoughtfully. Smillie (1985) studied the effects of illustrated job-performance aids in United States military contexts ( e.g., documents intended for enlisted personnel who carry out jobs in which the "read to do," such as reading a manual to fix an Army Jeep). He found that to get someone to understand a complex procedure that involves flipping a switch or turning a knob requires a picture that makes the action as explicit as possible, showing only the part of the object needed for the step. (p. 419)


[]

   In the mid-1980s, there was a core of writers at the CDC [Communications Design Center] who had taken part in much of the document design and usability testing research at Carnegie Mellon and who for a number of years had been using protocol-aided revision to evaluate functional texts. I observed that these experienced writers and usability testers seemed to be much better at planning and revising text than were writers who had years of on-the-job editing experience…. My hunch was that perhaps the CDC writers' repeated experiences with "catching users in the act" as they grappled with poorly written prose and badly designed graphics had increased their sensitivity to audiences' needs more generally. They seemed to know what to expect from readers, often anticipating their moves even before they saw readers actually engage with a document.
   To explore this intuition, I designed the PAM [protocol-aided audience-modeling method] teaching method, which involves a sequence of ten lessons for document designers, each with the following format:

  1. First, the document designer reads a sample document with the goal of making a list of any problems that he or she believes the prose or graphics will create for the intended audience. [Prior to the first lesson, students are provided a profile of the audience (knowledge, background, and experience).]
  2. Second, the document designer reads a transcript of a think-aloud protocol from a member of the intended audience who is trying to understand the sample document. With the help of the transcript, the document designer generates a second list of problems that he or she believes the prose or graphics will create for the audience.

   My idea was that once document designers had some practice in using PAM, they might be able to anticipate problems on the first pass (without the reader's feedback) that they could detect only after seeing reader's feedback early in the sequence of lessons…. (pp. 474–475)

Results

I compared experimental and control classes to evaluate their improvement in accurately predicting readers' problems. An accurate prediction is one in which writers predict problems that readers actually had as measured by the reading protocols collected from the freshmen. Results … indicate that writers in the experimental classes improved dramatically from the pretest to the posttest, increasing in accuracy by 62 percent, while writers in the control class remained essentially unchanged….
   In addition to asking whether writers improved in their accuracy of predictions, I was concerned with whether writers in the experimental and control classes had changed in their ability to differentiate actual reader problems from nonproblems…. I found that writers in the experimental classes were not "problem happy" but in fact had increased their ability to differentiate problems from nonproblems significantly more than had control writers. These results suggest that writers taught with protocol-aided audience modeling improved significantly in their ability to anticipate readers' needs and that the sensitivities writers developed are helpful in discriminating problems that readers actually have.


[]