chapter 11: within-subjects experimental design Flashcards
within-subjects design
AKA repeated-measures design
Same participants are observed one time in each group/level of within-subjects IV
strengths/weaknesses of within-subjects design
can manage sample size and observe changes in behavior over time
BUT lack of randomization - cannot demonstrate cause and effect
for a within-subjects to qualifiy as an experiment:
must manipulate levels of the factor and include a comparison/control group
AND make added efforts to control for order and time-related factors
time-related factors
maturation, testing effects, regression toward the mean, participant fatigue
order effects = Order of participation in treatments causes changes in DV
carryover effects = Participation in one group changes performance in a second group
controlling order
counterbalancing - offsetting the order participants recieve different treatments
complete = all possible order sequences
partial = representative subset of order sequences (ex// latin square: # of order sequences = # of treatment, Each treatment appears equally often in each position, Each treatment precedes and follows each treatment one time)
controlling timing
interval between treatments/groups - minimize testing and carryover effects
total duration of experiment - minimize demands placed on participants + likelihood of participant fatigue/attrition
between-persons variability
individual differences across all groups (SAME IN WITHIN-SUBJECTS)
between-groups variability
mean difference in a dependent measure caused by manipulation of the levels of IV
within-groups variability
variability in participant responding within each group
two related samples
related-samples t test
multiple related samples
one-way within-subjects analysis of variance (ANOVA)
repeated-measures design
related sample:
one sample from one population - same participants are observed in each group (level of IV)
used for experiment only if we manipulate the levels of IV + control for order effects
ex// music improves mood among children in grief therapy:
one sample of children 2 sessions (one with music, one without music)
matched-samples design
related sample:
match participants based on preexisting characteristics/traits they share then split into groups
natural (twins)
experimental (iq)
quasi-experimental
related samples t test (paired samples t test)
1 IV with 2 levels
1 sample doing 2 levels of IV OR 2 paired samples doing 1 level of IV EACH
compares differences between two related samples means (between-group variability / within-group error) *** difference in samples bc of TREATMENT, not other factors via individual differences
ex// students who read in ebook will recall more content than they will when reading from print book
one-way within-subjects ANOVA
1 IV with 2+ levels
1 sample doing multiple levels of IV
*only repeated-measures design
ex// adults will judge taste as wine based on its price (inexpensive, moderate, or expensive) – ANOVA determines if manipulation caused mean ratings of liking to vary between groups
post hoc/pairwise comparisons determine which pairs of group means were different
