High-Level GOMS Model for Choosing Functionality

Description

Write interface-independent GOMS methods for users' top-level goals, choose the system functions that keep those methods simple, then elaborate to keystroke level to predict performance.

Kieras argued that functionality is usually chosen haphazardly before interface design, although no interface can make up for a missing or badly chosen function, and that task analysis such as HTA rarely connects to performance prediction. A high-level GOMS model bridges the two. Goals refer only to the task, not the interface; operators are mental (think of, decide) or invocations of system functions treated as unanalysed units (update the database, not click UPDATE); methods show the order of these steps, the information users need and how they detect and correct errors. The team then picks functions that make these methods short and simple. Because the high-level methods become the top of a full GOMS model, lower levels can later be added down to keystrokes to predict learning and execution time, giving a usability-oriented form of requirements analysis.

Attributes

Loss type
Latent
Effort
Weeks

Steps

  1. List the users' top-level goals and write high-level methods for each using only task terms, mental operators and system functions treated as black boxes.
  2. Record in each method what information the user needs and how errors are detected and corrected.
  3. Compare alternative function sets and choose the one that gives the shortest, simplest high-level methods.
  4. Elaborate lower-level methods down to keystroke level, introducing interface specifics.
  5. Use the full model to identify problems and to predict learning time (method length) and execution time (operators executed).
  6. Revise functions, methods and interface choices and re-run the predictions.

Sources

  • GOMS Models: Simplified Cognitive Architectures (Kieras, n.d.)

Also known as

  • High-level GOMS
  • Functionality analysis with GOMS

Categories

  • User Research
  • Task Analysis & Cognitive Modelling