Information Processing Fall 2003 IST 331 Lecture 3

Information Processing Fall 2003 IST 331 Lecture 3

Human Performance u Enhanced through use of machines u Consider interface connecting human and machine u Success of system depends on the design of the interface

Human Performance u Apply human information processing (HIP) to design of computer interfaces resulted in philosophy composed of: – Task analysis – Approximation – Calculation

Human Performance u Card, Moran, and Newell – Concentrated on common situation of skilled users performing tasks within their area of skill – Approximating actual performance by error-free performance – Produced models called GOMS u Goals u Operators u Methods u Selection rules

Models u. A model is a simplification of reality u Useful to help design, evaluate or understand complex behavior such as interaction with a computer system u Think of models on a continuum – Analogy and metaphor at one end – Mathematical equations at the other end

Predictive Models u Allow measurement of human performance in an analytical method – No need for time-consuming experiments – Allow a hypothetical design scenario without implementing a real system

GOMS The fox jumps over the lazy quick brown dog.

GOMS u GOMS is a method for describing a task and the user’s knowledge of how to perform the task – Goals – the user’s goals; what does he/she want to accomplish? – Operators – actions that the software allows the user to take u Could be a command u Menu selection u Button press u Most common operators for analysis are at the “keystroke level” such as moving the cursor, clicking mouse button keying information

GOMS – Methods – well-learned sequences of subgoals and operators that can accomplish a goal – Example: 1. 2. 3. 4. 5. Highlight the text to be moved Type CTRL-X Move the cursor to the desired location Click the mouse button Type CTRL-V – If there is more than one method to accomplish the same goal, then selection rules are required

GOMS – Selection rules – personal rules that users follow in deciding what method to use in a particular circumstance u GOMS analysis applies to situations when users perform tasks they have mastered - Cognitive Skill. u Users are not problem solving!

GOMS u Has been applied to: – Text editors – Spreadsheets – Information browsers – Operating systems – Flight-management computers in commercial airplanes – Etc.

GOMS Quantitatively – good predictions of performance time and relative learning time u Qualitatively – used to design training programs, help systems, and actual systems u When GOMS uncovers a frequent goal supported by a very inefficient method, the design can be changed to be more efficient u

GOMS Case Study u Project Ernestine – 1988, NYNEX, telephone company considered replacing workstations – Toll and assistance operators (TOA) – Factor driving decision making was improved productivity of TOAs – Decrease of 1 second in work time per call would save ~ $3 million per year

GOMS Case Study u GOMS modeling took 2 personmonths using CPM_GOMS technique – Quantitative predictions, using expert call-handling time as benchmark – New workstations would be an average of 0. 63 seconds slower than the old workstations – NYNEX would need an additional $2 million a year to operate

GOMS Case Study u Field Trials – Took 18 months involving scores of people – Supported CPM-GOMS predications – New workstations were 0. 65 seconds slower on average than the old workstations

GOMS Case Study u GOMS models helped explain why the new workstations were slower – CPM-GOMS based on critical path of a task – Whole task, the faster features of the new workstations failed to shorten the critical path to complete each TAOs task

GOMS u Three critical restrictions on kinds of tasks GOMS models can be used for 1. Task in question must be analyzed in terms of “how-to-do-it” or procedural knowledge required 2. GOMS analysis can represent only skilled behavior 3. The designer must start with a list of toplevel tasks or user goals u u u Comes from interviews with potential users, Observations of users of similar systems Intuitions of the analyst

GOMS u The strength of GOMS is in its quantitative predictions of performance time and qualitative explanations for those predictions

Keystroke-Level Model u KLM – Prediction is the sum of times and overhead – Four motor-control operators u. K = key stroking u P = pointing u H = homing u D = drawing – One mental operator u. M – One system-response operator u. R

CMN-GOMS u GOMS model presented by Card, Moran, and Newell u Strict goal hierarchy u Operators are executed in strict sequential order u Methods are represented in pseudocode-like notation

Problem Each group form 2 sub-groups u One group study KLM, the other study CMN-GOMS u Each sub-group apply their method to the following problem u – In an Excel spreadsheet, enter 2 numbers into cells then compute the sum of the numbers in an adjacent cell u Share your method with the other part of your team