Chapter 9: Experiments with More Than One Independent Variable

1
Chapter 9 Experiments with More Than One
Independent Variable

• Experimental Designs with Two or More Independent
  Variables
• Factorial Designs
• Mixed Designs (Between Within Factors)
• Mixtures of Manipulation Selection
• Main Effects of Independent Variables
• Interactions Amongst IVs

2
Advantages of Two IV Designs

• Gain in Efficiency--Two experiments (2 IVs)in a
  single experiment
• Interactions--Examine the combined effect of 2
  IVs on a DV

3
Do we like food hot or cold?
4
Does the light work if we replace the bulb? It
depends
5
Factorial Designs

• Two or more IVs
• Subjects assigned at random to each possible
  combination of the IVs
• Take each value of A (i.e. A1 and A2) and combine
  with each value of B (B1 and B2)
• a1b1, a1b2, a2b1, a2b2
• 2 x 2 Factorial Design

6
Building a 2 x 2 Factorial Design
7
State Dependent Memory or ESP

• Encoding Specificity Principle
• Match between Encoding context and testing
  context
• Bower (1981) Classic Study

8
Bower (1981)

• Mood induction via Hypnosis (happy versus sad
  state induced)
• Random assignment to happy vs. sad state at
  encoding
• At test half of each group tested same state
  half tested different state

9
State Dependent Memory
10
2 x 2 Factorial Design

• Encoding State (Happy vs. Sad)
• Recall State (Happy vs. Sad)

11
A Crossover Interaction
12
The results

• An interaction--a crossover interaction
• The effect of one IV (state at encoding) on
  recall (memory) depends upon the other IV-State
  (happy vs. sad) at recall
• Do subjects remember better when happy or sad?
  It depends. On what? On what mood they were in
  at Encoding

13
Godden Baddeley (1975)
14
Crossover Interaction (Godden Baddeley, 1975)
15
Example from Memory Subliminal Tapes

• Can material recognized unconsciously affect our
  behavior w/out awareness?
• Claims to boost self-esteem, improve memory,
  learn a language
• Research testing such claims--e.g. Pratkanis
  (1992) typical experiment

16
Pratkanis (1992)

• IV 1 Actual Type of Tape (Memory vs.
  Self-esteem)
• DV Memory Improvement
• IV2 Type of Tape Subjects Thought they were
  listening to (Memory vs. Self-esteem)
• Draw the Design 2 x 2

17
Pratkanis Design
18
What happened?

• No actual effect of either tape (memory or
  self-esteem) on either measure (memory or
  self-esteem)
• However, if we look at subjects self-reports
  subjects said they improved
• These reports related to what subjects thought
  they were listening to but not to what they
  actually listened to (see graph)

19
Effect of Subliminal Tapes on Self-reported
Memory Improvement
20
Summary of Results

• Illusory Placebo Effect
• No main effect of type of tape (memory or
  self-esteem) on subjects perception of (say)
  their memory
• But significant main effect of what subject
  thought the tape was for
• No interaction of type of tape actually heard and
  type of tape subject thought he/she heard

21
Mixed Within- and Between-Subjects Designs

• Factorial Design--separate group of subjects for
  each possible combination
• Other Multi-Factor Designs possible
• Mixed Design 1 or more Between-Subjects and
  Within-Subjects Factors
• Advantages efficiency, power

22
Design of a mixed between-group and
within-subjects experiment
23
An Example from Clinical Research

• Therapy study
• One group receives actual therapy other group
  receives a placebo therapy (Between-Subjects
  Factor)
• Before vs After (Within-Subjects Factor)

24
Mixed Within-subj Between-subj Design
Compare each S to him/herself
Pure Between
Or compute a difference score
Therapy Group
Placebo Group
25
Learned Helplessness Example

• Maier (1970)
• Shuttlebox--shock comes on until dog jumps over
  barrier
• Rapid learning
• What if the dogs prior experience shock cant
  be shut off?

26
A Shuttlebox
27
Learned Helplessness Experiment
Note Each dog in group 2 (helpless group)
assigned to or yoked to a pair-mate in
nonhelpless group 3--i.e. a yoked control. Why?
What does this control for?
28
Mixed Design

• Mixed Design 1st IV Type of Prior Experience
  (controllable shock, uncontrollable shock, and no
  shock)
• 2nd IV Trial or Session (1st, 2nd, etc.)
• DV Average latency enduring shock before animal
  jumps over barrier

29
Conditioned-Helplessness Exp Results
30
Follow-up Work

• Further Experiments
• Parallels between helpless animals and behavior
  of depressed human patients
• Might learned helplessness play a role in human
  depression

31
Experimental vs. Subject Variable Designs

• Pure experiment Both IVs manipulated by
  experimenter
• Sometimes an IV is experimentally manipulated but
  another variable isnt--by selecting subjects who
  differ on 2nd variable
• True independent variable vs. subject variable or
  individual difference variable or selection
  variable
• Ex Strategy Type and Gender (or Intelligence,
  age, etc.)

32
An Example Restrained Eating and
Counterregulation Study

• Eating as physiologically regulated but
  cognitive influences as well
• Herman Polivy (1985) -- restrained eaters or
  dieters
• Individual difference variable questionnaire
  used to classify intos restrained eaters and
  unrestrained eaters

33
Herman Polivy Design

• IV Preload Condition (zero, one, or two
  milkshakes)
• Indiv Diff or Selection Variable Type of eater
  (restrained vs unrestrained eater)
• DV Amount of Ice Cream Eaten

34
Herman Polivy (1985) Results
Amount of Ice Cream Eaten
35
Results

• Crossover Interaction
• How does drinking milkshakes affect how much ice
  cream is subsequently eated?
• It depends -gt On what?
• On whether youre talking about dieters
  (unrestrained eaters) or or nondieters
  (restrained eaters)

36
Explanation?

• What-the-hell-effect
• Alternative Explanation--milkshake had internal
  effects that whet the appetite (stimulated
  hunger)
• Follow-up Studies
• Polivy Hermans Account

37
Higher-Order Designs

• Extend Factorial Design from 2 to 3 or more IVs
  (e.g. 2 x 2 x 2 OR 2 x 3 x 2 OR 2 x 3 x 4 x 2)
• Main Effects
• 2-way interactions
• 3-way interactions, etc.

38
3-Factors A, B, C

• Main Effect of A, of B, of C
• Interactions (2-way) A x B A x C B x
  C
• Interactions (3-way) A x B x C

39
Three-way Design
40
Desensitization Therapy Example

• Principles of conditioning applied to human
  problems
• Phobia--intense, irrational fear
• Desensitization Therapy
• Condition a new response to the fearful situation

41
Theory of Desensitization Therapy
42
Condition a new response to fearful situation

• US (something) -gt UR (fear)CS (snake) CR
  (fear) incompatible CR (relaxation)
• Condition a new response (e.g. relaxation) that
  is incompatible with CR (fear) by successively
  closer approximations to actual CS (e.g. word
  snake, a picture, etc.)

43
Questions

• Does this work?
• Is it really due to conditioning--pairing the
  image of snake with relaxation
• Perhaps Relaxation alone is sufficient
• Perhaps Imaging the snake alone is sufficient and
  causes fear to decline

44
Davison (1998) Design 2 x 4

• DV Phobic reaction
• IVs type of treatment (4) and Session (before
  vs after therapy)

45
Davison Design 2 x 2 x 2

• DV Phobic reaction
• IV 1 Snake Imagery (present vs absent)
• IV 2 Relaxation Training (present vs. absent)
  
• IV 3 Session (before vs after)

46
Imagery x Relaxation x Session2 x 2 x 2
47
Davisons (1968) Findings
48
Davisons Findings

• Snake Imagery paired with Relaxation Training
  most effective
• Result due to Relaxation Training per se?
• Result due to Habituation to the Snake Imagery?

49
Conditioning and Awareness

• Eyelid condition (DV Percent CRs)
• IV 1 Successive trials (block 1, 2...)
• IV 2 Delay vs No Delay between CS (tone) and
  UCS (puff of air)
• Var 3 (selection) Awareness (aware vs.
  unaware) of relationship bet CS and UCS

50
Clark Squire Conditioning Exp
51
Results

• The No-delay group-- Main effect of trials-- No
  main effect of Awareness (aware vs unaware)-- No
  interaction of Trials Awareness
• The Delay Group--Main effect of Trials--Main
  Effect of Awareness (aware vs unaware)--
  Interaction of Awareness and Trials
• 3-way Interaction--Interaction of Trials
  Awareness depends on which group (no-delay or
  delay) it was.

52
Summary