Chapter 8: Experimental Design Flashcards
Extraneous and Confounding Variables —Key Points—
Main point: Most outcomes are influenced by multiple variables
- Purpose of experiments is to isolate the causal factor
- Other factors that influence the DV are problematic
Extraneous Variables
Variables other than the IV that could affect the outcome/DV
Confounding Variables
An extraneous variable that that varies systematically with the IV
A confounding variable makes it impossible to claim causation from IV
Extraneous Vs. Confounding Variables
Extraneous Variables generally just add variability, while Confounding Variables can create misleading conclusions about causality.
Test Prep — Identify the Confounding Variable:
Research Question: Does explaining to someone else enhance comprehension?
* Explanation condition: Read an article about echolocation in bats (10 min). After reading the article once, attempt to explain it to someone else, with access to the article the whole time (10 min).
* No explanation condition: Read an article about echolocation in bats (10 min).
* DV: Take a test that measures comprehension (both groups get this test, obvi).
- Identify the confounding variable.
- Describe why/how that confounding variable disrupts internal validity
- Fix the experiment
Answers:
- Confounding variable: Time with the article
(The explanation group has TWICE the amount of time with the article, compared to the control group…) - Disruption of Internal Validity: If participants in the explanation condition score better than participants in the no explanation condition, we don’t know if (a) explanation improves comprehension, (b) extra study time improves comprehension, or (c) both.
- Fix the experiment:
Ensure both groups spend the same amount of time with the material. For example:
a) Explanation condition: Read for 10 minutes, then explain for 10 minutes.
b) No explanation condition: Read for 10 minutes, then engage in another active task related to the material for 10 minutes (e.g., summarizing in writing or answering questions about the article).
This adjustment controls for time and engagement differences, isolating the effect of explaining on comprehension.
Confounding Variables Relationship to Internal Validity
Confounding Variables = a THREAT to Internal Validity!!!
- Internal Validity only exists when the results of an experiment can be confidently attributed the effect of the IV.
- When a confounding variable is present, we cannot determine which variable is responsible for the effect
Basic Experiment Types:
Between-Subjects Design: Each group/participant experiences ONE condition
* Between-Subjects Design = Independent Groups Design
Within-Subjects Designs: Each participant experiences ALL conditions (only one group).
* Within-Subjects Designs = Repeated Measures Desig
Assigning participants to conditions: Between-subjects/Independent Groups Design
- Subjects are randomly assigned.
- Random assignment prevents systematic biases
Assigning participants to conditions: Within-subjects/Repeated Measures Design
All subjects experience every conditions
Assigning participants to conditions: Matched pairs Design
Participants are paired/matched based on similar participant variables (ie: age, sex, hight, etc.) before being assigned a condition.
Ensures that the groups are equivalent (on the matching variable) before introducing the IV manipulation.
Advantages & Disadvantages of Within-Groups Designs
Advantages:
* More confidence when making statistical inferences
* Fewer subjects required
Disadvantages:
* Potential Risk of Order Effects
Order Effects
- Practice/learning effect
- Fatigue effect
- Carryover effect:
Practice/ Learning Effect
Definition: Performance increases due to practice in the experiment (warm-ups)
Solution: reduced by adding pre-experimental practice trials
Fatigue Effect
Definition: Performance decreases due to fatigue
Solution: Reduced by allowing breaks during the experiment
Carry-Over Effect
Definition: Persistent influence of one level of the IV on the next level of the IV
Solution: Reduced by allowing breaks during the experiment