Chapter 6: Experiments with One Variable

1
Chapter 6 Experiments with One Variable

• What is an experiment?
• Advantages of Experiments
• Independent vs Dependent Variables
• Control Variables
• Independent Groups Designs
• Within-Subject Designs-- Counterbalancing or
  Randomization
• Limitations of Experiments

2
What is An Experiment?

• The setting
• Manipulation
• Measurement versus Manipulation
• Manipulate A and observe/measure B while holding
  Z constant Causal Inference
• Single thread from a tangled web of possible
  influences

3
Examples of Experiments

• Nerve Impulse Speed
• Mental Rotation
• Taste Aversion in Rats and Humans -- Garcia,
  Ervin, Koelling-- Bernsteins Experiment

4
Nerve Impulse Experiment
5
Mental Rotation
6
Mental RotationSubtraction Procedure
7
Garcia Koelling Taste Aversion
8
Bernsteins Conditioned-taste Aversion Experiment
9
Anatomy of An Experiment

• Independent and Dependent Variable
• Control Variables

10
Control Variables

• Nuisance variables
• Eliminate or equalize (i.e. control them)
• Equalize groups
• Importance of comparison--e.g. a control group

11
Experimental Design

• What treatments when?
• Independent-Groups Design
• Within-Subjects Designs Concurrent Measures
• Within-Subjects Designs RM
• Matched-Subjects and Randomized Block Designs

12
Independent-Groups Design

• Two or more Groups of Subjects
• Random assignment
• Each subject in one group only
• Each subject measured once
• Advantage of simplicity
• Lacks power compared to Within-subjects Designs

13
Within-Subjects Designs

• Concurrent Measures
• Repeated Measures
• Matched-Subjects and Randomized-Block Designs (?)

14
Within-Subjects Designs Concurrent Repeated
Measures

• Only one group of subjects
• Each subject is confronted with all values of the
  Independent Variable
• Font Type Experiment (IV Lower Case vs Upper
  Case (e. g. book vs TABLE DV Recall)
• Simplicity

15
Within-Subjects Designs Repeated Measures

• Present all treatment conditions to all
  subjects--but not at the same time--repeated
  measurements
• Progressive Error Carryover Effects
• Distributing Carryover Effects Counterbalancing
  and Randomization
• Problem of Reversibility
• Reversal Designs

16
Problems with RM Within-Subjects Designs

• Progressive Error
• Carryover Effects

17
Counterbalancing
18
Problems with Counterbalancing (Reversability)

• Counterbalancing not always helpful
• Counterbalancing handles most Progressive Error
  problems (i.e. general order effects like
  fatigue, practice, etc.)
• Problem of asymmetrical order effects
• Problem of permanent carryover effects
• Reading Methods Experiment
• Onlooker or Bystander Experiment

19
A Within-Subjects Reversal Design
20
Matched-Subjects and Randomized Block Designs

• RM Powerful/Sensitive but Not Always Possible
• Cant Take Repeated Measurements on the Same
  Individual--e.g. Reading Program or Counseling
  Program
• Solution Matching--Matched subjects--Matched
  groups--Randomized-Block

21
Matched-Group Design
22
Randomized-Block Design
23
Limitations of Experiments

• Experiments not always possible
• Ethical/Practical Reasons
• Single experiment approach--sure but slow
• Scope Inferences limited to a single variables
  influence

24
Summary