woot woot lets go
2.7 Task analysis
- PROCESSOR: GOMS model
- strengths
- formalizing user interaction
- weakness:
- doesnt address complexity like alternatives and parallels
- assumes user is expert
- strengths
- 5 tips:
- focus on small goals - eg navigate to end of doc
- nest goals instead of operators - smaller goals. operators are smallest atoms
- differentiate descriptive and prescriptive
- assign costs to operators (usually time)
- use goms to trim waste
- PREDICTOR: Cognitive Task Analysis
- emphasis on memory and cog load
- specifically what you can’t see
- collecting preliminary knowledge
- identifying knowledge representations
- apply focused knowledge elicitation methods
- analyze and verify data acquired
- format results for intended application
- PARTICIPANT?: Hierarchical Task analysis
- large tasks are very composed of smaller tasks
- break out checkout process from purchasing process
- strenght: emphasize mental processes
- strength:
- weakness: de emphasizing context
- weakness: time intensive
- weakness: ill suited for novices
2.8 Distributed Cognition
- how a cockpit remembers its speed
- four models of context! :)
- distrib cognition: models of cognition outside the mind
- social cognition: friend driving with you. jira. but be careful of letting them knwo everything
- situated action: filming with daughter. task is what we do, not what we design. task grows out of the interface, not independently. memory is very context dependent as well.
- activity theory: predates HCI.
- why is user doing the task in the first place. not just -what- they are doing.
- users can move up and down the hierarchy of activity -> action -> operator. the more competent you are - driving a car - the more of a subconscious operator it is
- nardi - comparing activity theory and other thingies. incl permanent persistent structues
2.9 Interfaces and politics
2.10 Conclusion to principles
- Processor: External
- Predictor: Internal. Disappearing ui. gulf of execution/eval. understanding mapping, expert blindspot, learned helplessness.
- Participant: Context.
- 5 tips: on screen ui design
- use a grid
- use whitespace
- know gestalt principles: proximity, continuation, movement, similarity
- reduce clutter
- design in grayscale
3.5: prototyping
- verbal prototypes
- paper prototypes
- wiz of oz
- wireframing
- physical prototypes
5 tips:
- keep prototypes easy to change
- make clear that its a prototype
- be creative
- evaluate risks
- prototype for feedback
3.6: evaluation
- three types of eval: qual, empirical, predictive
- eval terminology:
- reliability - whether assessment is consistent over time
- validity - whether it actually affects reality
- generalizability - whether results can be extended
- precision - level of detail
- what to eval
- efficiency - how long to accomplish?
- accuracy - how many errors user commits?
- learnability - how long does it take user to hit level of expertise?
- memorability - does user rememberthe ui?
- satisfaction - careful of social desirability bias
- 5 tips: qual eval
- run pilot studies to iron out the kinks
- focus on feedback - dont spend too much time teaching one user
- use questions when users get stuck
- tell users what to do not how to do it
- capture satisfaction - just ask if they even like it
- 5 tips: empirical eval - categorical, interval, nominal, etc. see table.
- control what you can, document what you cant
- limit your variables
- work backwards - decide what qtn you want to answer, then decide what you need to find out
- script your analysis in advance, dont torture data
- pay attn to power - size of effect inversely related to size of survey
- predictive eval
- heuristic eval (take the 15 things)
- model based eval eg goms
- siulation based evaluation
3.7 - agile and HCI
- you want agile when:
- low criticality
- frequent requirements change
- small teams
- team embraces change
- framework for agile and user centered design: http://www.ime.usp.br/~marivb/ihc3.pdf
- the differences are:
- imptnce of documentation
- amount of research before getting it out to user
- high user involvement
- close team collab
- the differences are:
- live prototyping: oxymoron? optimizely. esp useful for ecom where costs are temporary but benefits can be permanent
- a/b testing with real users
- 5 tips:
- start more traditional, then shift to agile once up and running
- focus on small changes
- adopt a parallel track method. have hci one sprint ahead of dev.
- be careful with consistency. dont mess with expectations of frequent users
- nest your design cycles - keep learnings from small cycles to big cycles
3.8 - conclusion to methods
http://omscs6750.gatech.edu/fall-2018/required-reading-list/
- reading 7
- reading 8
- reading 9
- reading 10
- reading 11
- reading 12
- reading 13