The research results confirmed the majority of th e hypotheses . Overall, high as well as low skill subject s performed better using the expert system as compare d to not using the system . Main effects were found for skill level, task uncertainty, and knowledge presentation format with respect to problem-solving time and accuracy . When measuring solution time an d accuracy, significant interactions were found for skil l level and knowledge presentation format, as well a s skill level, task uncertainty, and knowledg e presentation format . Confidence ratings were positively correlated with decision speed. Ratings of satisfaction with procedural vs . declarative knowledg e presentation formats varied with skill level and tas k uncertainty . Based on the study's results, conclusion s are presented emphasizing links between the variable s in the framework . Future research issues focusing on intelligent interface design and the presentation o f knowledge in expert system explanation facilities i s discussed . Donna M. Lamberti Donna M. Lamberti received a doctorate degree in Information Systems and Decision Sciences, an d a masters degree in Cognitive Psychology/Huma n Factors from Rensselaer Polytechnic Institute i n 1987 . She is a technical staff member for IBM Scientific Center, Cambridge, MA . Donna's doctoral thesis focused on the development of an expert system to support programmers at IBM. This work was supported by an IBM Fellowshi p for Information Systems Research . Her interest s include intelligent interface design for decision support technology, AI-based advisory systems , and knowledge representation and cognitive modeling . simplicity, clarity, and consistency for all the icons. Finally, we report the effect of this manual on th e production team responsible for integrating and improvin g the user interface of the CADDS workstation . Aaron Marcus, Michael Arent, Bruce Brown e Aaron Marcus and Associates, Berkeley, Calif ., is a consulting firm specializing in user interface design and information graphics design . Aaron Marcus, principa l of the firm, has presented tutorials and papers a t SIGCHI, SIGGRAPH, and NCGA . Michael Arent serves as design director and Bruce Browne is senio r designer . Carl Jacobse n Carl Jacobsen is director of human interface design o n the CADDS Project at Computervision, Bedford , Mass . REFINING EARLY DESIGN DECISIONS WITH A BLACK-BOX MODE L ANDREW MON K The poster describes a simple procedure for evaluating th e potential usability of a user interface at the earliest stage s of product design . A black-box model of the user interface is generated which can be examined for inconsistency , unnecessary complexity and unpredictability . The black box model is specified in terms of action-effec t rules relating any of the actions a user might take with it s effect . The actions and effects must be specified in term s relevant to the user rather than in terms relevant to the designer. Thus examples of actions might be "pressing th e key" or " selecting icon X", rather than "input X " or "input Y given machine state Z" . Similarly, example s of effects would be in terms of changes in the display or t o some assumed object such as a document rather tha n changes in program state or outputs . The steps to be taken in applying this procedure are described by means of an example. The poster illustrates how the procedure can be used to identify potential problems and make explicit the decisions taken in th e process of design. Andrew Monk Andrew Monk is a part of the Human-Computer Interaction Group at the University of York, U .K . He has a Ph . D . i n Cognitive Psychology from the University of Bristol .The project, within which this work was done, is supported b y the U .K . Alvey Directorate (Grant GR/D 0231 .7) and is concerned with empirical and formal approaches to the design of interactive systems . He , has edited two books: "Fundamentals of Human-Computer Interaction" (Academi c Press 1984) and with Michael Harrison "People & Computers : designing for usability " (Cambridge University Press 1986) . His address is Dr A .F. Monk, Department of Psychology, University of York, York YO1 5DD, U .K . AN ICON DESIGN MANUAL PAGE FOR THE USE R INTERFACE OF A CAD/CAM/CAE WORKSTATIO N AARON MARCUS, MICHAEL ARENT, BRUCE BROWNE, CARL JACOBSE N For all the modules of Computervision's computer-aide d design, engineering and manufacturing workstation, th e CADDS station, the number of icons representing the various applications and tools is greater than 5000, wit h the new icon designs being added daily. This number o f symbols is greater than the number of characters in the basic Chinese system of ideograms, according to one introduction to the language . Computervision had no icon design manual to ensure legibility, readability, an d consistency among these signs, which were bein g developed by software groups in four different countries . We report on the approach, organization, and content of an icon design manual intended to govern the creation of these signs . We also describe the development of an icon "toolkit" of visual syntax elements that help to promote SIGCHI Bulletin 55 October 1987 Volume 19 Number 2

/lp/association-for-computing-machinery/refining-early-design-decisions-with-a-black-box-model-bq6oygWiD6