Within Subject Designs Flashcards
What is meant by a within subject design?
each individual receives all the experimental treatments equally - the outcomes of each treatment are compared within each individual
When are within subjects designs good? (more statistically powerful)
when there are larger differences between different groups rather than within them
What analysis is used for a within subject design
A Paired T test - because individuals are measured multiple times in different circumstances and then compared
How does within subject design differ from pseudoreplication
because each group only contributes 1 data point to the test statistics - the difference between data is independent
How to avoid confounding by time in within subject designs
have 2 groups and swap the intervention they have half way through the experiment - will prevent confounding eg hotter in second half of experiment time - called counterbalancing
Counterbalancing definition
ensuring an equal number of groups experience all possible treatment orderings
(K) = number of treatment orderings
What is carry over effect and what can be done about it?
the effect of treatment 1 still affecting indivudals once they are on treatment 2
can be tested for but not adjusted for so must be removed by design
How can carry over effects be removed by design?
introduce period with no treatment between the different treatments (washout period)
How does using the washout period change the analysis
a ‘burn in’ period is introduced so the data at the start of each treatment is excluded
What is the within and between person hybrid design
groups are randomised to control or intervention then within person design for each level of the intervention
When are within person designs not appropriate?
when the subject cannot be returned to the condition it was in before the treatment begun
