Designing and Writing to Reduce User Errors

By Michael Albers

A vast majority of documents (I consider print and online as documentation) often works to define the optimized error-free method of performing a task and provides a user with a straightforward solution. However, the user expects documentation to help solve problems and address errors. Thus, attention must be paid to potential problems users can have and how to correct them. Errors have different causes; the information designer should understand the potential types of errors since properly addressing each type requires a different approach in the design and documentation.

Errors and error recovery

As a person interacts with a situation and attempts to use an information system to gain information, they might make errors in either how information gets interpreted or what actions to perform. When an error occurs, the situation will not progress according to the user's expectations. Part of good design is to both minimize the chance of errors and make the erroneous state readily apparent. By providing error recognition information, the user knows if the right thing is happening and how to correct problems.

Regardless of how the error occurs, the system must help the person understand what error has occurred and how to recover from it. Error recovery information provides a chance for the user to correct the error and continue along their expected path to task completion (Carroll et al., 1988; Lillies, 1991). Clearly depicting the expected results versus the actual results helps people identify situation errors early, when they can be more easily corrected. Thus, in the early design, the error correction and out-of-the-norm situations should occupy a central focus and the information designer must ensure that they remain a design focus.

Error correction and out-of-the-norm situations are the major factors which draw people into using an information system and its associated documentation. In many circumstances, people don't use any documentation until they notice a strong mismatch between their expectations and the reality of the situation. For information systems which support external systems, the same pattern of waiting until a major problem is apparent before searching for information to fix it (Penrose and Seiford, 1988).

Information designers are expected to understand what users have to do, how much they need to know, and how to make that knowledge available to the users in more than just straightforward solutions (Blandford and Young, 1997; Treu 1992). A substantial part of documentation gets written with the unspoken assumption that people use the documentation from the beginning and never make errors while performing the task. Unfortunately, neither assumptions fits the reality of documentation use.

Error taxonomy

Errors can come from any of three different areas: the design, the user, and outside factors.

Design errors

The design itself may have problems which cause errors. Norman has dedicated a book, The Design of Everyday Things, to discussing this problem. Informational systems can provide access to much more information than a reader can ever process. However, even though they may have access to an abundance of information, the user is often subjected too much, too little, or ineffectively presented information.

The error recovery information is important since the information designer has minimal control over what information the user actually saw or used. Although it may seem obvious that seeing and using a particular piece of information are two very different things, many existing systems seem focused only on letting the user see a piece of information. Readers skipping blocks of text complicate designing for user interaction, but must be accepted as a fact of system use. Many of the errors that post-hoc inquires attribute to user error are really the result of poor information design that made it all but impossible for the user to fully understand the current context and the relevant information.

Too much information Too much information causes cognitive overload which results in the information being ignored. Too much information can result when the design strives for completeness over clarity, gives information that is not salient to the current situation, or gives information the user considers irrelevant (or perhaps doesn't understand that it is important).

Not enough information Not enough information can arise either when the initial analysis didn't identify the information as needed. It can also arise in a "one-size-fits-all" design mode (exactly the opposite of what most situations require). The information is constrained so it is appropriate for a subset of the potential users, such as written for low literacy readers. While an important design criteria for the low literacy group, it fails to provide adequate information to other groups that want or need more information.

Ineffective presentation Overly complex or confusing information presentation increases errors, as does making the user mentally retain too much information. A couple of examples are excessive cross-references or breaking the information into too small of chunks for the reader to mentally integrate.

Human errors

People make many errors. Norman ("Stages") suggests that errors arise when a mismatch occurs between the knowledge level used to express the intention and the knowledge level required for successful execution of the intention. These can range from not understanding how to read the document to skipping steps in the analysis process to using procedure A rather than procedure B to address problem B. The proper method of handling each type of error is different.

The following list covers the various human error categories and how to handle them.

Mistakes: Mistakes are a failure to formulate the right intention. These occur when the reader looks at the correct information but make a wrong choice. The can be caused by a combination of factors which influence the reader's perception, memory, or cognitive process. Not knowing the blue button should be pushed when event A occurs is a mistake. Training can help. Mistakes break into two major categories.

Knowledge-based mistake: Incorrect decision/action because of a failure to understand the situation. Common causes are an overly complex presentation or an inexperienced person. Because of problems in interpreting the provided information, the reader does not adequately understand the situation to make a proper decision. During user testing these errors will often be accompanied by low reader confident scores.

Rule-based mistake: Understand the situation, but making a wrong decision. These type of mistakes are best understood as following "if-then" rules. When a rule-based mistake occurs, the reader has selected and acted on the wrong rule; the choice itself is correct for the rule that was used.

Slips: Right intention is incorrectly carried out. Slips differ from mistakes because the person knows the proper action, but simply doesn't do it correctly. Math addition errors fall into this category: you know how to add numbers, but forgot to carry the one. Likewise, knowing the blue button should be pushed, but pushing the green button would be a slip. Training is not appropriate for slips. Working another 20 math problems or retaking the blue button operation training course will not prevent a reoccurrence because the user already knows the proper action.

Lapses: Failure to carry out an action. These occur when part of a sequence isn't carried out. For instance, after getting interrupted, a person skips step 8 in an instruction sequence. Lapses can be very common in design situations that involve complex information and multiple steps. If the design requires a start-to-finish execution and the reader has a high chance of interruption or mixing tasks, then lapses will be a problem. User testing normally doesn't identify lapses because the test design allows for straight forward execution of the task and a test subject is normally highly attentive.

Mode errors: Action is correct in one mode of action, but wrong in another. Modes can be when a product switches between different operational settings (such as the difference between setting the time and the alarm on a clock) or switching between related products with similar interfaces (For example, in Microsoft Word, ctrl D opens the font dialog and in Eudora it deletes the email.). The reader takes the proper action for mode 1 but is currently in mode 2, thus the action is wrong. Having distinctly different information reflected in highly similar designs can cause this problem as the reader can have trouble distinguishing which mode they are working in. Having a "Mode 1" label in the lower left corner of the screen is not sufficient to prevent mode errors.

Outside factor errors

Outside factors are everything that doesn't directly include the user or the design. It does include factors such as time pressure, less than ideal environmental conditions, and pointy-haired bosses. Users who are annoyed or unhappy, bored or fatigued, are likely to make more mistakes and to work less efficiently (Constantine and Lockwood 1999).

Causes of errors

Design often assumes the focused and successful execution of a task. Conventional task analysis often works to define the optimized error-free method of performing a task that uses a straightforward solution without outside help. The entire design and information presentation are developed from the view that nothing goes wrong. However, good information design uses the possible errors to help identify and focus on potential problems from the user's point of view. The early design analysis must identify those errors and use them as the starting place for presenting information.

Some of the potential sources of errors are:

Incorrect cause and effect: Many people do not understand the real cause and effect relationships within situation. However, they will use their incorrect understanding to try to understand and make predictions. Part of the analysis must be to uncover the common misconceptions and to figure out how to either change them or prevent them from interfering with understanding the information.

Missing information or actions:

Inadequate background to understand the information: The user may lack the technical knowledge or the reading ability to understand the information. With a dynamic information system, the information presentation can be adjusted to allow for these differences. The extent of technical background material and the level of presentation can be dynamically modified to fit the user's needs.

Not having clear understanding of situation history: Complex systems have a history that must be accounted for if the user is to understand the situation. The various starting points and how they evolve need to be examined during the analysis.

Misjudging information importance: The user focuses on less important information and ignores the information salient to the situation. The salient information may either be not clearly presented or buried in too much extraneous information. The design analyst needs to clearly understand the relative importance of information and ensure only the salient information is presented.

Conclusion

As the documentation and interface content are designed, attention must be paid to potential problems users can have with the situation and how to correct them. Understanding the types of errors and the causes of these errors are vital to being able to evaluate the root causes of the problem and fix them.

User testing can find some of the errors in the design and reader problems. However, some errors are caused by factors which can't be controlled for in a testing situation. Problems such as the reader being tired or preoccupied, the lighting being abnormally bright or dim, or social factors which only occur in actual use. Also, since people make decisions as soon as they feel confident in the answer, the amount of information and which information they look for varies. The design must identify the potential problems that can prevent or hinder a user from establishing clear user goals and properly interpreting the information needed to reach those goals.

References

Blandford, Ann and Richard Young. "Specifying User Knowledge for the Design of Interactive Systems." Software Engineering Journal (November 1996): 323-333.

Carroll, J., P. Smith Kerker, J. Ford, and S. Mazur Rimetz. "The Minimal Manual." Human Computer Interaction 3, 1988.

Constantine, Larry and Lucy Lockwood. Software for Use: A Practical Guide to the Models and Methods of Usage-Centered Design. New York: ACM Press. 1999.

Lillies, P. "Some Guidelines for Making a Computer Manual more Task Oriented." Proceedings of 38th STC Conference, 1991.

Norman, Donald. "Stages and Levels in Human machine Interaction." International Journal of Man Machine Studies 21, 1984.

Norman, Donald. The Design of Everyday Things. New York: Basic Books. 1988.

Penrose, J. and L. Seiford. "Microcomputer Users' preferences for Software Documentation: An Analysis." Journal of Technical Writing and Communication 18(4), 1988.

Treu, Siegfried. "Interface Structures: Conceptual, Logical, and Physical Patterns Applicable to Human-Computer Interaction." International Journal of Man-Machine Studies 37 (1992): 565-593.

Michael J. Albers teaches professional writing at the University of Memphis. He is the membership coordinator for the STC Information Design SIG and ACM SIGDOC secretary. He has worked for 10 years as a technical communicator writing software documentation and performing interface design. His research interests focus on dynamic presentation of complex information.