Please enable JavaScript.

Coggle requires JavaScript to display documents.

Field Studies (Statistics (Helps us to make inferences, Inferential…

- - - - Increased realism (external validity)
      - Loss of control over users' experience
- - - - Provide a truly objective measure
      - Everywhere
      - Emerge into the background => unobstrusive
  - - - Interactions with a computer system
      - any significant events/problems during their use
      - Other aspect of their working life
    - - Date + time of event
      - Where they are
      - Information about event of significant
      - Ratings about how they feel
    - - Pen + paper
      - Alternatives
        
        tape recorder + list of questions => don't need to write things down
        
        Email address
        
        Photo camera
      - Provide a semi-structured format for recording entries
        
        Highly structured: prevent recording unanticipated events
        
        Unstructured: confusing
      - Should be followed up by in-depth interviews with participants
    - - Cheaper than location-base ethnography
      - Long term observation
      - Very good to investigate context of use
    - - Depends on users' motivation (might decrease early)
      - Not reliable: hard to find out if diary is complete or what is missing
  - - - Time-dependent trigger
      - Event-triggered
    - - In-situ
      - Reduced recall bias
    - - Carefully consider frequency and extent
  - - - Independent variables
      - Dependent variables
        
        What we measure
        
        Results are referred to as scores
        
        Measured in different scales
        
        Categorial / nominal: drunk/sober/sleeping
        
        Ordinal: low/medium/high
        
        Interval: blood alcohold concentration
        
        Ratio: time in ms
        
        Manipulate a single aspect by keeping all other factors stable -> diffcult
      - Conditions
        
        Levels of independent variables
        
        Control condition: absence of suspected cause
        
        Experimental conditions: presence of suspected cause
      - Measurements
        
        Repeated measures (within subjects) - BETTER
        
        all participants are assigned to all conditions
        
        order must be counter-balanced/randomised
        
        Pros
        
        Economical; less participants required
        
        Sensitiveness
        
        Cancelling out individual differences
        
        Cons
        
        Carry-over effects from previous conditions
        
        Conditions need to be reversible
        
        Independent measures (between groups)
        
        participants assigned to one condition only
        
        Pros
        
        Simple
        
        Less chance of practice / fatigue effects
        
        Useful when it is impossible for an individual to participate in all conditions
        
        Cons
        
        Expense (time, effort, number of participants)
        
        Insensitiveness to experimental manipulations
        
        Hybrid
      - Way of selection
        
        True experimental BETTER
        
        Full control over independent variable
        
        But independent variables cannot always be controlled
        
        Can draw stronger conclusions
        
        Quasi-experimental
        
        partial selection; selection bias
        
        Limits the conclusions we can draw from the study
        
        E.g. cannot randomly assign motorcycles to all participants (some people may not know how to ride one)
        
        Need to consider conditions when assigning subjects
- - - - Design goals reached?
      - Usability testing
      - Results not always generalisable
      - How to validate
        
        Lab studies
        
        Heuristical evaluation
        
        Usability tests
        
        High internal validity
        
        Evalutation "in-the-wild"
        
        Proof of concept
        
        Experience using a prototype
        
        High external validity
      - Human-centered design process (Diagram 19) - evaluation stage
      - Design thinking (Diagram 20) - testing stage
    - - What are user constraints?
      - What are context constraints?
      - Results often generalisable
      - Context of use and requirements analysis
      - What
        
        Understand as much as possible about users, tasks and context
        
        Product a set of requirements
      - Why
        
        Increase acceptance and satisfaction
        
        Cost-reduction
      - Human-centered design process (Diagram 19) - specify context of use and requirements stages
      - Design thinking (Diagram 20) - understand + observe stage
- - - - Uniquely identity/contact/locate a single person
      - Uniquely identify a single individual with other sources
  - - - What do participants have to expect?
      - Context in which the research takes place
    - - What kind of data will be logged?
      - Method and access, and how it will be secured?
      - What is going to be reported?
    - - Inform participants that
        
        They do not have to e.g. answer questions if they don't want to
        
        They can cancel the experiment at any time without explanation