4: Experimental research within subjects Flashcards
selection
-a confound where systematically different
types of subjects are exposed to each level of the IV
within subjects design
-designs eliminate selection effects altogether
by exposing every subject to all levels of the IV (each subject serves as their own control)
• Concurrent-measures or repeated-measures
concurrent measures
- present all levels of the independent variable to one group at the same time
• Preference is often measured
repeated measures
-designs expose every participant to all levels of the independent variable sequentially
repeated measures design setup
- the same subjects are used in all conditions
• Subjects are randomly assigned to treatment or condition orders
-related to matched group, taken to the extreme
-Evaluate changes in behavior within subjects across multiple treatments or conditions
• Sometimes called simply within-subjects
carryover in repeated measures
common sources:
-occurs when later responses of the subject are
systematically affected by their earlier responses
-practice: in earlier treatments may enhance performance in later treatments
-fatigue may impair performance
-contrast: subjects may compare treatments to previous
treatments, which may affect behavior
-contamination: occurs when later responses of the subject are systematically affected by their earlier responses
-unbalanced carryover may lead to order effects
dealing with order effects:
-steps to minimize carryover
• Pre-training or practice
• Rest periods between treatments
• Blind design and/or concealing differences
-Counterbalancing balances
counterbalancing
- The various conditions are presented in a different order for different subjects
- Subjects randomly assigned to one of the treatment orders
- Attempts to evenly distribute carryover among conditions, eliminating order effects as potential confounds
counterbalancing techniques
- Full (complete) counterbalancing: equal numbers of
subjects randomly assigned to every possible ordering of conditions
-Partial counterbalancing: equal numbers of subjects
randomly assigned to only some of the possible orders of conditions
-Pseudo
random ordering: each subject receives multiple exposures to each condition, interleaved in a
pseudorandom order
pros within subject designs
-Avoids selection effects from nonequivalent groups
-Greater statistical power (more likely to detect an effect)
• Evaluating changes in DV within subjects eliminates error variance due to individual differences
-Fewer subjects required
cons within subject
- More demanding on subjects
- Subject attrition is a problem, especially if multi-session
-Carryover: Exposure to a previous treatment may affect
DV/behavior in a subsequent treatment, a potential
confound as order effect
• Controlled for with counterbalancing
• Counterbalancing may sometimes be impossible or impractical due to permanent/irreversible changes or asymmetrical carryover