CS252. User Interface Design

Lecture Outline for Feb. 25
Non-Empirical Methods in UCS

1. 	Overview of Non-empirical Methods

	also called predictive evaluation:

	model-based prediction (Jordan's "task analyses")

	expert-based prediction (Jordan's other three)

2. 	Model-Based Methods:  Task Analyses
	Method identifies the physical and cognitive steps needed to accomplish "typical" user tasks

	We'll look at the GOMS notation as summarized in [Preece 1994]
	GOMS: goals, operators, methods, and selection rules



	Example: comparing basic file and directory tasks PC-DOS and Macintosh
		Specifically, we'll look at how GOMS models can capture principle of consistency

		Results of comparison:

			Macintosh  methods:       total length:

			PC-DOS     methods:       total length:   

	Consider how it changes if we also look at user goals of
			Move directory
			Delete directory
		Use slide to show structure

	What GOMS does

	But GOMS analysis can be misleading

3. 	Expert-Based Methods
	Property checklist:

	Expert appraisal:

	Cognitive walkthrough:

4. 	What You'll Be Doing
 	Last assignment: non-empirical (predictive) usability evaluation of parallel performance tool
		assess what users' response would be overall, if were to see tool/demo in present form;
			identify occasions where the presentation interferes with or obscures the 
			intention of the tool
		evaluate the accuracy and intelligibility of text strings
		evaluate the accuracy and intelligibility of visual displays of data
		determine any color schemes that would cause problems for older and/or color-impaired users
		identify any operational elements that are likely to cause confusion for users
		identify inconsistencies within the tool structure
		identify where the tool is incompatible with other tools, real-world experiences
		identify any elements that will cause problems when tool is used in a 
			typical Windows environment (i.e., with other applications active)