2017-2 Human Computer Interaction Cognition And Usability.fex

interaction design:
	goals:
		efficiency: allow common tasks to be done fast
		safety: make human error not dangerous
		utility: is the right functionality included
		learnability: is it easy to learn
		memorability: can it be easily remembered
	design principles:
		visibility: find appropiate control for task (like correct icon)
		feedback: always provide feedback to actions executed
		contraints: restrict user to prevent malicious interaction 
		mapping: map effects of real world (like arrows)
		affordance: reuse known techniques or invent intuitive new ways
		consistence: reuse as much as possible
		
	
"Gestalt" Theory:
	A group is formed by objects which are:
		proximity: close to each other
		similar: look similar
		symmetry: symmetrical to each other
	Figure - Ground: Identify a figure from the background
	Law of closure: perceptual completing of incomplete objects
	
Gibson's Affordance Theory:
	perception is designe dfor action
	first think about the action, we look at the interface with a clear action we want to execute
	
approaches:
	top down: if you expect what you see its faster to process. Learned interfaces (such as maps) are easy to understand even if they are very complex.
	bottom down: if you just get the data, and now have to make sense from it.

seven design principles:
	most important in line of view
	most frequent in line of view 
	displays used for a single task close
	grouped infos
	close goes together
	consistency
	clutter avoidance
	
cognitive walkthrough:
	performed by analyst
	examines specific user task
	focus on easy of learning
	preparing phase:
		define inputs
		what users (backgorund, population)
		tasks (what functionalities)
		action sequences (correct action sequence for each task)
		interface description (how the interfaces reacts to inputs)
	analysis phase:
		walking through the actions
		each step ask yourself:
			will user try to archieve the right effect
			will user notice that correct action is available
			will user associate correct action with the effect it archieves
			will user see progress made towards a solution to the task
		construct success (users fulfil all requirements to fulfill task) and failure stories (one of the question is no)
	bad:
		focus on only one attribute of usability (easy of learning)
		no "real" user asked
		negative score for features which increase productivity
	difficulties:
		choosing tasks difficulties
		first question of analysis difficult
		analyist performs more of a user test than a cognitive walkthrough
	
heuristic evaluation:
	performed by user
	use usability principles to save on time and money
	general:
		interative process
	preparation:
		choose different evaluators (performance will differ)
		choose if observer present (answer questions, speed up process, formualte questions in a usable way)		
		choose appropiate heuristics (analysis of existing products, abstraction from common issues)
	steps:
		each evaluator will inspect interface alone:
			1-2h per session, multiple inspections
			use list of heuristics
			can ask questions, add other elements
		then discussion:
			specify feelings about problems (find precise explanation why something is broken)
			build up consensus of the analysis
		then build final usability report:
			written / spoken
			presented to developers
	good:
		cheap 
		lots of freedom
		explains why there are flaws
	bad:
		no systematic fixes for flaws (may miss some)
		may finds a lot of false positives (problems which are none)
	example heuristics:
		visibility of system status (red/green)
		match between system & real world (known words / tools)
		user control and freedom (can go back if necessary)
		consistency and standards (buttons always similar)
		recognition rather than recall (use icons)
		prevention of errors (make clear interface)
		flexibility and efficieny of use (can adapt to user preferences)
		aestetics and minimalist design (fibonaccy)
		help users with errors (allow to go back)
		help and documentation (explanation text)
	example heuristic (The center for universal design 1997):
		equitable use (everyone can use it)
		flexibility of use (accomodates individual preferences)
		simple use (easy to understand regardless of experience and knowledge)
		perceptile information (effective transfer of information)
		tolerance of error (minimizes hazard)
		low physical effort (efficent and comfortable use)
		size and space for approach and use (appropiate size for buttons regardles of user ability)

heuristic walkthrough:
	performed by both users and analysts
	combines heuristic evaluation and cognitive walkthrough
	phase 1:
		task oriented evaluation
		prepare set of thought-provoking questions (how can I assign events)
		prepare set of priorized tasks which should be evaluated (assign events)
		walk through product using questions and tasks
		document usability problems
	phase 2:
		free form evaluation
		walk thorugh the interface using checklists
		expert can use tools he want (heuritcs he chooses or adapts to the product)
		and will document usability problem
	phase 3:
		assign severity rating to problems
		contact developers
	good:
		replies both on users and experts
		quick, and can be made by amateurs
		probably finds local, global, minor problems
		good mis between users & expers
	bad:
		requres skilled evaluators (product dependent)
		moderator needs to be able to cover all important questions and tasks 

assessing usability experience:
	user based testing:
		formative testing: early in dev, get qualitative feedback (brainstorm) on low fidelity (less details) prototype
		summative testing: after high-level decisions made, get quantitative feedback (specific) on high fidelity prototype
		validation testing: shortly before release, get comparisations (to other products) on final product
	necessary steps:
		develop study plan
		prepare environement
		find subjects
		conduct tests
		debrief participants
		analyze data
		report findings
	most common measures:
		efficiency: time spent
		effectiveness: correctness, successful tasks
		user satisfaction: use standardized survey tool (recommended is System Usability Scale - SUS)
	likert scale: 
		state level of agreement
		questions and then agree or disagree with it
		example is scan.co.uk 
	semantic differential scale:
		presenting pairs of bipolar or opposite scale
		"the checkout was straightforward _ _ _ _ _ confusion"
	guidelines for rating scales:
		include negative AND positive statements
		use an odd number of raiting points
		use unmodified versions of adjective (not EXTREMLY or similar)
		be careful about different words in semantic scales	

GOMS:
	Goal: 
		what you want to archieve
		splitted in subgoals
	Method:
		what you actually do to archieve subgoal
		splitted in operations (like move mouse, click)
		operations can be formulated as subgoals if more granularity is needed
	
different GOMS models:
	Keystroke Level Model (KLM):
		decompose task
		measure time for each task
		sum up to determine whole length of task performed
		only good for simple tasks
	CMN:
		adds goals, subgoals, selection rules to KLM
		useful because there may be different subgoals one can use interchangebly to archive same goal
		calculate time for each subgoal
	Natural GOMS Language (NGOMSL):
		formel language with restricted english syntax
		can use selection rules (selecting different methods, recursive)
		can include working and longtime memory in model
	CPM-GOMS:
		HPM based (Human Processor Model):
			PP Perceptual Processor (perceives environement)
			CP Cognitive Processor (takes decisions)
			MP Motor Processor (perform actions, channels for hands, eyes, ...)
			different processors can be parallelized