John D. Gould

Designing for usability: key principles and what designers think

This article is both theoretical and empirical. Theoretically, it describes three principles of system design which we believe must be followed to produce a useful and easy to use computer system. These principles are: early and continual focus on users; empirical measurement of usage; and iterative design whereby the system (simulated, prototype, and real) is modified, tested, modified again, tested again, and the cycle is repeated again and again. This approach is contrasted to other principled design approaches, for example, get it right the first time, reliance on design guidelines. Empirically, the article presents data which show that our design principles are not always intuitive to designers; identifies the arguments which designers often offer for not using these principles—and answers them; and provides an example in which our principles have been used successfully.

How to design usable systems

Publisher Summary This chapter describes the way in which usable systems can be designed. Usability is a combination of many factors, each of which is often developed independently. User-interface code is becoming an increasingly large percentage of the total system code. Standards are beginning to emerge for user interface design. Establishing standards for software aspects of the user interface is probably premature. There are lots of guidelines for a good system design. However, these are not enough for the design of good systems. One should at the beginning and throughout development focus on prospective users and their work. It is often heard that people buy computer systems for the functions in them. One is unlikely to figure out what the functions should be without talking with users. One should continuously measure each aspect of usability, and then iterate in a hill-climbing way toward a better system. All aspects of usability should begin evolving from the beginning and should be under one focus.

Reading from CRT displays can be as fast as reading from paper

Experiments, including our own (Gould and Grischkowsky, 1984; Gould et al., 1987), have shown that people read more slowly from CRT displays than from paper. This series of six experiments shows why. The explanation centers on the image quality of the characters. Reading speeds equivalent to those on paper occur on CRT displays containing character fonts that resemble those on paper (rather than dot matrix fonts, for example), that have a polarity of dark characters on a light background, that are anti-aliased (i.e., contain grey level), and that are shown on displays with relatively high resolution (e.g., 1000 × 800). Each of these variables probably contributes something to reading speed improvement, but the trade-offs have not been determined. The results of these experiments, together with our earlier studies, exclude some other general explanations such as possible inherent defects in CRT technology itself or personal variables such as age, experience, or familiarity with reading from CRT displays.

Making usable, useful, productivity-enhancing computer applications

Almost a decade has passed since we started advocating a process of usability design. This article is a status report about the value of this process and, mainly, a description of new ideas for enhancing the use of the process

Reading is slower from CRT displays than from paper: attempts to isolate a single-variable explanation

People read more slowly from CRT displays than from paper (see, for example, Gould and Grischkowsky, 1984). This report summarizes ten experiments and several more analyses that seek to explain the cause of this reading-speed difference. Typically, each experiment isolates one variable and studies whether it explains the difference. Results show that no one variable studied (e.g., experience in using CRT displays; display orientation; character size, font, or polarity) explains it. The tentative conclusion is that the difference is due to a combination of variables, probably centering on the image quality of the characters themselves.

An experimental study of people creating spreadsheets

Nine experienced users of electronic spreadsheets each created three spreadsheets. Although participants were quite confident that their spreadsheets were accurate, 44 percent of the spreadsheets contained user-generated programming errors. With regard to the spreadsheet creation process, we found that experienced spreadsheet users spend a large percentage of their time using the cursor keys, primarily for the purpose of moving the cursor around the spreadsheet. Users did not spend a lot of time planning before launching into spreadsheet creation, nor did they spend much time in a separate, systematic debugging stage. Participants spent 21 percent of their time pausing, presumably reading and/or thinking, prior to the initial keystrokes of spreadsheet creation episodes.

Some psychological evidence on how people debug computer programs

Ten experienced programmers were each given the same 12 FORTRAN listings to debug. Each listing contained a non-syntactic error in one line. Median debug times (7 min), number of bugs not found (11% of the listings), and number of incorrect assertions about the location of the bug (less than one per listing) all replicated earlier results ( Gould & Drongowski, 1974 ). Although subjects were given the opportunity to use the interactive debugging facilities of an on-line computer, they rarely did so. Bugs in assignment statements were about three times as hard to detect as array or interaction bugs. Debugging was about three times as efficient on programs subjects had debugged previously (although with a different bug). A number of basic concepts relating to debugging are identified and a gross theory of debugging is described.

The 1984 Olympic Message System: a test of behavioral principles of system design

There was more than athletic talent being pressed to peak performance at the 1984 Olympic Games in Los Angeles. Behind the scenes, a multilingual Olympic Message System ran round-the-clock keeping more than 10,000 athletes and officials in contact with families and friends, both far and near.

Pattern recognition and eye-movement parameters

Pattern perception was studied by recording eye movements while Ss visually scanned nine simultaneously presented patterns of asterisks for target patterns. Pattern parameters studied were: similarity of target patterns to non-target patterns (absolute difference in the number of elements), number of target elements, and frequency of targets. Systematic correlation between the first two pattern parameters and eyemovement parameters were found. Mean duration and mean number of fixations on targets and also on non-targets increased with increased similarity. Mean duration and mean number of fixations increased on targets with an increase in the number of target elements. Non-target patterns were perceived more auicfely than targets. Fixations of longer duration were required to perceive the original target than to identify the other target patterns subsequently. The eye-movement results provide the basis for developing inferences about higher order processing of visual stimuli.

Eye-movement parameters and pattern discrimination

Pattern discrimination was studied in a visual-search task by recordingan O’s eye movements while he determined how many of eight patterns, arranged in a square around a standardpattern, matched the standard pattern. The results demonstrate the role of eye movements in visual search and human pattern discrimination. The mean duration of an eye fixation on a pattern, the probability of fixating it, the probability ofrefixating it, and the sequence in which patterns were fixated were all systematically related to various pattern measures. Multivariateanalyses showed modest correlations between the duration of individual eye fixations and various pattern measures. Relative characteristics of patterns influenced performance more than absolute characteristics of patterns. Patterns that matched a standard were fixated more often and longer than patterns that did not match a standard. The order in which patterns were fixated depended upon their relative characteristics. The results were consistent with a model ofpattern discrimination consistingoftwo stagesin which (1) features of a fixated pattern are abstracted and encoded, and (2) these features are then compared with the features ofanother pattern.