Advanced Experimental Design

Advanced Experimental Design
Chapters 8 & 9
Experimental Designs

Between-subject designs




Simple randomized
design
R Group 1
T (24oz Pepsi)
M
R Group 2
T (No Pepsi)
M
Multilevel randomized
design
Factorial designs
R
Group 1
T (No Pepsi)
M
R
Group 2
T (12oz Pepsi)
Within-subject designs
M
R
Group 3
T (16oz Pepsi)
M
R
Group 4
T (24oz Pepsi)
M
Multilevel Between-Subject
Designs and Hypothesis Testing
What is the null
hypothesis for the
multilevel Pepsi study?
Group 1 = Group 2 = Group 3 = Group 4
R
Group 1
T (No Pepsi)
M
R
Group 2
T (12oz Pepsi)
M
R
Group 3
T (16oz Pepsi)
M
R
Group 4
T (24oz Pepsi)
M
How do we test the null hypothesis?
By calculating an F-ratio - Analysis of Variance (ANOVA)
If the F-ratio is significant, how do we know which groups
are different from one another?
There are two approaches for finding out
Post hoc tests
A priori tests
Post Hoc Tests
Group 1 vs. Group 2
Group 1 vs. Group 3*
Group 1 vs. Group 4
Group 2 vs. Group 3*
Group 2 vs. Group 4
Group 3* vs. Group 4
* p < .05
Group 1 = 70
Group 2 = 72
Group 3 = 83
Group 4 = 75
a priori Tests
Group 1 vs. Group 3*
Group 2 vs. Group 3*
Group 1 vs. Group 4
Group 1 = 70
Group 2 = 72
Group 3 = 83
Group 2 vs. Group 4
Group 4 = 75
* p < .05
Factorial Designs
Pepsi
No Pepsi
24 oz Pepsi
Group A
Group B
Morning
Group C
Time
Afternoon
Group D
Effects in Factorial Designs

Main effects


Interaction effects


Treatment differences between levels of 1 IV
Result of combination of two IVs
There are many null hypotheses in factorial
designs



There are no differences in test scores due to the
amount of Pepsi consumed.
There are no differences in test scores due to the
time of day the test was taken.
There are no interaction effects between amount of
Pepsi consumed and time of day the test was taken.
Graphing Main and Interaction
Effects
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
70
70
70
80
Afternoon
70
70
70
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
70
70
0
No Pepsi
No main or interaction effects
24oz Pepsi
Significant Pepsi Effects
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
70
90
80
80
Afternoon
70
90
80
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
70
90
0
No Pepsi
No time or interaction effects
24oz Pepsi
Significant Time Effects
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
Mean Score
Morning
90
90
90
80
60
Afternoon
70
70
70
`
40
20
Column
Mean
80
80
0
No Pepsi
No Pepsi or interaction effects
24oz Pepsi
Significant Interaction Effect
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
90
70
80
80
Afternoon
70
90
80
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
80
80
0
No Pepsi
No main effects
24oz Pepsi
Significant Pepsi & Interaction
Effect
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
50
90
70
80
Afternoon
70
70
70
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
60
80
0
No Pepsi
24oz Pepsi
Significant Time & Interaction
Effect
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
70
90
80
80
Afternoon
70
50
60
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
70
70
0
No Pepsi
24oz Pepsi
Significant Pepsi & Time
Effects
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
50
70
60
80
Afternoon
70
90
80
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
60
80
0
No Pepsi
No interaction effect
24oz Pepsi
Significant Pepsi, Time, and
Interaction Effects
No
24oz Row
Pepsi Pepsi Mean
Morning
Afternoon
100
70
70
70
80
Afternoon
70
90
80
Test Score
Mean Score
Morning
60
`
40
20
Column
Mean
70
80
0
No Pepsi
24oz Pepsi
Within-Subject Designs

Important Factors in Experimental Design

Groups must be equal before treatment


Use random assignment
Try to reduce within-group (error) variance

Expose same subjects to each level of treatment.
Example: Effect of beer on dart throwing
R Group 1
T (4 beers)
M
R Group 2
T (No beer)
M