quiz Flashcards
independent variables are manipulated (two groups, two levels, or condition for each IV)
feature of experimental design
control of individual variability (random assignment to groups in between-subjects/independent group designs; each participant serves as their own control in within-groups designs and matched random assignment)
feature of experimental design
sacrifice of ecological validity (need to consider experimental and mundane realism)
feature of experimental design
problems: demand characteristics, reactivity, experimenter effects
feature of experimental design
experimental variance: due to manipulation of IV; variance variance is due to what we put our participants through
extraneous variance: uncontrolled, confounding variables that affect the group as a whole; confound; occurs at the same time as the manipulation (like hx, if it only affects one group)
systematic between-groups variance
error variance: random factors that affect only some participants in a group; random assignment helps to mitigate this
nonsystematic within-groups variance
one IV with two groups/levels/conditions
simple randomized
one IV with three or more groups/levels/conditions
multi-level, posttest only
DV is measured before and after the intervention/tx
pretest-posttest control group
participants matched on a given variable, compared to each other after treatment/intervention
match-subjects
sensitive to time-related effects like history, maturation, statistical regression; lasting effects of one level or condition may affect subsequent levels or condition
within-subjects
as many conditions as there are groups
factorial
practice effects–train participants at the same level of performance
habituation
fatigue–give participants a break
carryover–allow enough time between levels/conditions for a washout period
sensitization–hard to control for; best not to use within-subjects design if this is going to be a concern
order effects and how to control for them
participants matched precisely on a given characteristic (all groups have the same IQ score)
precision matching
participants are matched based on a range (all group members’ IQs fall within the same range)
range matching
for large number of conditions, this is giving every group/person a different order of questions/tasks
complete counterbalancing
if there are 720 orders and only 60 participants, randomly select 60 of those orders and assign these to the 60 participants
partial counterbalancing
each condition appears at each place in the order
Latin square design
would use this for independent (between-groups design) and paired (within-groups) design
t-test
would use this for one IV, two IVs, and three IVs
ANOVA
use when you need to analyze results to make unplanned comparisons
post hoc statistical tests
planned before the study, based on what you expect to see
a priori statistical tests