Within-Subject Design

Question 1

What is factorial design?

Answer 1

• several independent variables
• 2x2x3 = 3 independent variables, with 2, 2, 3
levels respectively
Example:
• Measuring the efficacy of 2 treatments (A,B), under consideration of gender (m, f) and 3 levels of socio- economic-status (low, average, high)
• There will be a total of 12(2x2x3) groups

Question 2

When do you use factorial design?

Answer 2

• When more independent variables are of interest

* If control for carryover effects is needed, the treatment order can be treated as an additional indepedent variable

Question 3

What is single factor two-level design?

Answer 3

• 2 levels of independent variable
• All subjects receive both treatments
• Half of the group AB, half BA

Question 4

What is single factor multilevel design?

Answer 4

• 3 or more levels of independent variable
• Counterbalancing
• Latin Square (too reduce nr of orders)
• If independent variable cumulative: practice
session or breaks

Question 5

When to use single factor designs?

Answer 5

• Individual Differences main source of variance
• Can’t afford to test many subjects
• Assessing effects of increasing exposure on
behavior

Question 6

general discussion of within-subject design

Answer 6

1. If subject related variables should be controlled→Matched/Within Design
2. If carryover potential problem →Matched Design
3. If subject related variables not impactful → Randomized Group Design

Question 7

What is the mixed design?

Answer 7

• Combines between-subject & within-subject factors
Example: often time can be used as a within subject factor. For example when comparing lasting effects of treatment A and B (between factor) at the time points 1 month and 6 months (within factor)

Question 8

What are quasi-experimental design?

Answer 8

Time Series Design:
O1-O2-Treatment-O3-O4
Equivalent Time Samples Design:
Treatment-O1-No treatment-O2-Treatment
Nonequivalent Control Group Design:
O1-O2-Treatment-O3-O4 O1-O2-No Treatment-O3-O4

Question 9

What are pretest-posttest designs?

Answer 9

• True experiment (if randomly assigned)
• Pretest-treatment-Posttest
• No counterbalancing possible
• Internal Validity reduced if no control group
introduced
• External validity reduced if pretest has
lasting effects on performance

Question 10

What is the Solomon four group design?

Answer 10

• Pretest-treatment-Posttest
• Pretest-No treatment-Posttest
• No Pretest-treatment-Posttest
• No Pretest-No treatment-Posttest

Question 11

Single-Subject Baseline Design

Answer 11

• Multiple observations for change in behavior until the response is stable (stability criterion)
• Baseline interchanged with Intervention (eg ABAB)
• Intra- and Intersubject replication for better validity
• Reversal strategy: ABA

Question 12

Possible difficulties with baseline designs

Answer 12

• Finding a stability criterion: solution can be
pilot study
• Stability criterion not met: solution can be
time limit
• Volatility in data: solutions can be baseline
trends and effect sizes

Question 13

What is uncontrollable variability in baseline designs?

Answer 13

• Data can be valid even if variable
Group approach: statistical methods will take within variability into account
Single-subject: it is recommended to identify the source of variability

Question 14

What are difficulties with baselines?

Answer 14

Drift: can be subtracted
Unrecoverable: if only partially, rest/pauses can help, if completely special designs need to be applied
Unequal: Interventions to make them equal Inappropriate level: adjusting experimental settings to achieve a desired baseline