Subject Assignment Flashcards
Within-subjects design
-Every subject receives each level of IV
-Potential problems:
• Sequence or carry-over effects (minimize by counterbalancing)
Between-subjects design
-Each subject receives only one level of IV
-Test diff. groups of subjects
-Potential problems:
• Group diff. could be due to subject diff.
• Increase “bad” variance
Counterbalancing
Can eliminate treatment order confounds
- Complete counterbalancing
- Partial counterbalancing
Cannot eliminate confounds if:
- treatment produces irreversible change to subject
- asymmetrical transfer
Complete counterbalancing
All possible sequences, each used equally often
Partial counterbalancing
- Without every possible order
- Ex. Balanced Latin Square
Use between-subjects design if:
- Being in one condition makes it impossible to be in other (irreversible change)
- Asymmetrical transfer between conditions
- IV is a subject variable
Advantages:
• Subjects equally naïve
• Less time to test each subject
To minimize bad variance:
• Control irrelevant variations
• Control some subject differences
How to assign subjects to treatment groups:
- Randomized groups
2. Matched groups
Randomized groups
- Random assignment of subjects to groups
- Problem: if groups are small
Matched groups
- Select sample from population
- Measure subjects on traits to control for
- Use trait scores to pair similar subjects
- One of each pair to each group
Best if:
- Small # subjects
- Traits are expected in influence DV
- Trait is easily and reliably measured
BUT pretest must be used (might elude subjects to nature of test)
Ideal conditions for each design
Within-subjects # subjects: small *Carryover effects: absent OR easily counterbalanced* Subject differences: Likely to effect DV *Real life: Experience all conditions*
Matched groups
# subjects: small
Carryover effects: present AND NOT easily counterbalanced
Subject differences: Likely to effect DV
Real life: DON’T experience all conditions
Random groups
# subjects: large
Carryover effects: present AND NOT easily counterbalanced
Subject differences: unlikely to effect DV
Real life: DON’T experience all conditions