Transcript PSY2005 Week 8 - Complex Experimental Designs

Lecture 8
Complex
Experimental
Designs
PSY2004 Research Methods & Ethics in Psychology
PSY2005 Applied Psychology Research Methods &
Ethics
Experimental
Designs
Experiments
• Characterised by random assignment
of groups or participants to conditions.
• At least one variable is manipulated by
the experimenter(s).
• Usually employs a control group
QuasiExperiments
• Groups or subjects not randomly
assigned (e.g., sample of convenience)
• May not have a comparison group
The Principle of Analysis Of VAriance
(ANOVA)
Total Variability
Systematic
Variance (caused by IVs
and found in differences
between means
Unsystematic variance
(caused by things we
do not know about)
3
Types of Variable
Independent
Variables
Dependent
Variables
• what is manipulated in the
experiment
• what is measured
Control
Variables
• what is held constant
Random
Variables
• what is allowed to vary randomly
Confounding
variable
• correlated with an independent
variable
Design Terminology
Between
Groups
Repeated
Measures
Mixed
• An “independent groups” design
• Different groups
• Dependent measures taken one time
• A “within groups” design
• One group, different conditions
• Dependent measures taken multiple times
• Between and repeated
Design Complexity
Single
factor
(one-way)
Factorial
• Studies one independent variable
• Studies multiple independent variables
• May have several levels
• Examples:
• Two-way (e.g., 2 x 2)
• Three-way (e.g., 2 x 2 x 2)
Factorial Design
The Principle of Analysis Of VAriance
(ANOVA: mood by intelligence example)
Total Variability
Systematic
Variance
mood
intelligence
Unsystematic
variance (caused
by things we do
not know about)
Interaction
8
Main Effect
Example Main Effect of
Intelligence on Creativity
Conclusion: people that are high in IQ score higher on creativity
Tests regardless of mood. No effect for mood
Interaction Effect
Main Effects &
Interactions
High
High
High
Low
High
Low
Lightness
Seasonal Affective Disorder
No Seasonal Affective Disorder
Lightness
A Detailed Example
Godden and Baddeley (1975) Memory
Context Study
• Participants (who were divers) learned lists of
words either on the beach, or beneath 15 feet of
water and then recall in the same or opposite
environment.
• 2 IVs:
• Place of learning (2 levels: Beach or
Underwater)
• Place of recall (2 levels: Beach or Underwater)
• DV: Number of words recalled
A Detailed Example
Four Conditions:
Learn on Beach
Recall on Beach
Learn Underwater
Recall Underwater
Learn on Beach
Recall Underwater
Learn Underwater
Recall on Beach
A Detailed Example
Advantages of Factorial
Designs
More than
one
hypothesis
can be
tested.
Potentially
confounding
Enables
variables can interaction
be built into effects to be
the design
tested.
as factors.
Randomised Controlled
Trials
An experimental design used to evaluate the
effectiveness of psychological interventions. Participants
assigned randomly, to one of at least two conditions
Pre & post tests
Control group
• Outcomes
measured before &
after intervention
• Includes no
treatment; treatment
as usual; and ‘wait
list’ sample.
17
Example
Does Psychological Therapy improve outcomes
for People with Mild/Moderate Depression?
Psychological Therapy
(CBT + TAU)
Sample
Ramdomise
Treatment as Usual TAU
2 IVs: 1) Time of measurement (pre- or post)
2) Treatment Group (CBT or TAU)
DV: Self-reported mood/symptoms
Results of Study
•
•
•
Main effect of Time? Yes
Main effect of type of
treatment? Possibly but
not important
Interaction between time
and type of treatment?
Yes
– Simple main effects?
Conclusion: These results show a significant interaction between time
and intervention. Whilst the participants reported lower depression following
Treatment this effect was stronger in the CBT Treatment
Complex Designs 1
• Complex designs represent research which
has more than one independent variable.
• Complex designs have more than one
outcome: main effects and interaction effects
• Example complex designs
– Two independent groups IVs and one DV
•
•
•
•
Place of learning (beach/underwater): Main effect 1
Place of recall (beach/underwater): Main effect 2
Learning by recall: interaction effect
DV: memory recall
Complex Designs 2:
Independent Groups
• A independent groups designs is one in
which the participant is in a single level of
one IV.
• The role of attraction and type of crime in
determining prison sentence
– IV1: Attraction: Attractive/Ordinary
– IV2: Type of Crime: Robbery/Swindle
– DV: Time spent in jail
The role of type of crime and attraction
on determining prison sentence
ME Attraction: NS
ME Type of Crime: Swindle > Robbery
A x TofC: Significant cross-over effect
Conclusion: Although attractive people are likely to be given lenient
sentences, when they use their attractiveness to commit a crime they
are given longer sentences.
Complex Designs 2: Repeated
Measures Design
• A repeated measures designs in which the
participant takes part in all of the
conditions in the experiment
• Can mood effect performance on visual
and verbal forms of creativity
– Mood (positive/negative)
– Type of creativity (visual/verbal)
– DV: creativity task performance
Results
Type of Creative: main effect fully
explained by interaction
Mood: main effect fully explained by
interation
T x M: Significant effect that explains
the systematic variance found
Conclusion: that people perform better on visual creativity tasks
When they are in a positive mood.
Complex Designs: The mixed
model
• A mixed model is one in which there is at
least one independent groups IV and at
least one repeated measures IV.
• Example: The effects of sex differences on
types of intelligence
– Sex (male/female): IG variable
– Intelligence (spatial/verbal): RM variable
– Sex by type of intelligence: Interaction effect
– Intelligence score: DV
Outcome for the mixed model
ME Intelligence: No effect
ME sex: No effect
I x S: Significant crossover effect
Conclusion: both within and between women are better at verbal tasks
and men are better at Spatial tasks