Chapter 11: More on Experiments - Confounding and Obscuring Variables

Key Concepts

• Threats to Internal Validity
• Interrogating Null Results in Experiments

Section 1: Threats to Internal Validity

Review of Chapter 10 Threats

1. Design Confounds
2. Selection Effects
3. Order Effects

New Threats to Internal Validity

1. Maturation Threats
2. History Threats
3. Regression Threats
4. Attrition Threats
5. Testing Threats
6. Instrumentation Threats
7. Observer Bias
8. Demand Characteristics
9. Placebo Effects

Combined Threats

• Selection-History Threats
• Selection-Attrition Threats

Preventing Threats

• Use comparison groups
• Employ reliable and valid measurement tools
• Conduct double-blind studies
• Use counterbalancing in measuring instruments

Section 2: Interrogating Null Effects

Possible Reasons for Null Effects

1. Not Enough Between-Groups Difference
   • Weak manipulations
   • Insensitive measures
   • Ceiling and floor effects
   • Reverse design confounds
2. Too Much Within-Groups Variability
   • Measurement error
   • Individual differences
   • Situation noise

Reducing Within-Groups Variability

• Use reliable, precise measurements
• Measure more instances
• Change design to within-groups or matched-groups design
• Add more participants
• Control the experimental environment to minimize distractions

Concept of Power

• Refers to the study's ability to detect a significant effect
• Studies with more power can detect smaller effects

Summary

• Though null effects are not often reported in popular media, they provide valuable insights
• Properly designed experiments can minimize threats to internal validity and correctly interpret null effects