Experimental design Flashcards
What are the 6 issues when creating new interaction techniques?
- Social acceptability
- Walk-up use (at the Start)
- Learning curve (over time)
- Intrusiveness
- Context of use
- Robustness and Error handling
What is social acceptability when referring to the issues when creating new interaction techniques?
Can a user use a particular interaction technique in society?
What is walk-up use when referring to the issues when creating new interaction techniques?
Can the interaction technique be used straight away with minimal setup
What is Learning curve when referring to the issues when creating new interaction techniques?
How long does it take for users to reach the crossover point?
What’s the crossover point?
Point in time when new interaction technique performs better than what users are familiar with
What is intrusiveness when referring to the issues when creating new interaction techniques?
Does the user have to wear a cumbersome piece of apparatus or markers to use the system?
What is context of use when referring to the issues when creating new interaction techniques?
How does the use context affect user’s interaction with the technique?
What is robustness and error handling when referring to the issues when creating new interaction techniques?
Is the interaction technique robust with respect to noise? How are errors handled? Can the user easily recover from errors?
How are new interaction techniques evaluated?
In controlled experiment in the lab.
Why are controlled experiments used?
Have baseline to compare against
What is the trade-off between internal and external validity?
Reliable experimental outcomes VS being able to generalise the results to actual use contexts.
What is the difficulty when evaluating with an experiment?
Asking the right questions.
What’s the purpose of a controlled experiment in HCI?
To experimentally determine if we can find a measurable difference in dependent variables between levels of independent variables.
What are threats to the validity of conclusions of an experiment?
- Reproducibility
- Study heterogeneity
Can it be compared? - Internal validity
Are there no confounding factors you do not control that influence the result. - External validity
Do results generalise to the world outside of the experiment?
What’s included in the Method section of an experimental design
Setup Participants Apparatus Material Procedure
Key: Must be written with enough detail that someone else can replicate the experiment.
What is between subjects design?
Each participant exposed to one condition
What are the pros of between subject design?
No risk of skill-transfer
What are the cons of between subject design?
Variance isn’t controlled within the participant (A better at keyboard use than B)
Each participant source of error.
When can within-subject design never be used and why?
When there’s a risk of asymmetrical skills transfer. This causes more skill transfer one way than the other, and therefore means it can’t be counterbalanced.
What is within-subjects design
Each participant is exposed to all conditions
What are the advantages of within-subjects design?
Variance is controlled within the participant
What are the disadvantages of within-subjects design?
Must counterbalance to avoid skill transfer effects.
Risk of asymmetrical skill transfer.
