Dependent variables - variables that the experimenter measures in the experiment.
Internal validity (kehtivus) - if i would fun the same test again would you see the same result
External validity - does it apply to only specific group or the whole world?
You have to use fair comparisons. You cant compare apples to oranges. The IPhone keypad example.
Vacuum cleaner example
Manipulation: vacuum cleaner
Measures: cleanliness, speed
Between subjects design - half of the participants use one vacuum cleaner the other half the other one.
Its hard to say if one vacuum cleaner is better then the other one. Maybe the ones we assigned to one vacuum cleaner are just better cleaners?
Worry about individual differences.
Within subjects design - everyone uses both interfaces.
Worry about ordering effects - in which order the participants use the different vacuum cleaner.
Half of your participants could test one vacuum cleaner first end the second afterwards and the second half of the group could do it in vice verse order. Its also good to vary the tasks environments.
Three Major Strategies
- Within-subjects: everyone tries all the options. Good when you’re not worried about learning/practice/exposure issues (that trying one version will ‘pollute’ the date from another version)
- Between-participants: each person tries one. Requires more people, and more attention to fair assignment. Has the beneļ¬t that each participant is uncorrupted (at least by the study...)
- Counterbalancing can help minimize variation in a between-subjects design
No comments:
Post a Comment