Lecture+6

Experiments and Non-experimental Designs

Primary Question: Causation vs. Correlation

• Causality: Indicates a change in one variable directly affects another.

• Correlation: Indicates a systematic relationship between two variables without asserting causation.

• Importance of time order: A causal relationship demands that the cause precedes the effect.

• Need to eliminate alternative explanations to strengthen causal assertions.

Addressing Spurious Relationships

• Questions to consider:

  • Is celebrity support genuinely causing political support?

  • Could gender or other factors explain increased awareness and support?

  • Definition of spurious relationship: Apparent relationship between variables due to a third influencing factor.

Control Variables

• Control Variables: Additional factors included in research design to account for external influences and support real-world relevance of findings.

Key Points in Establishing Causality

1. Covariation: Confirmation that the alleged cause (X) and effect (Y) are related.

2. Time Order: Ensure that the cause (X) precedes the effect (Y) in time; determining order can be challenging in observational studies.

3. Elimination of Alternative Causes: Conduct research to rule out possible joint causes (confounding factors).

Implications for Causal Relationship Example

• Awareness of endorsement must correlate with Republican Party support.

• Respondents must express awareness of the endorsement before developing party support for it.

• Eliminate any confounding factors that might interfere with establishing a clear causal link.

The Classical Randomized Experiment

• Purpose: Discern direct from indirect linkages among variables.

• Sampling: Importance of selecting appropriate subjects for reliable results. Availability sampling commonly used in lab settings may affect external validity.

• Stimulus: The test factor applied to participants, crucial for measuring effects in experiments.

• Impact of stimulus on dependent variable will help answer the research question.