Experimental Designs

Hypothesis: Increasing knowledge of a group reduces prejudice.
Independent Variable (IV): A movie that promotes empathy (e.g., Selma, 12 Years a Slave).
Dependent Variable (DV): Reduction in racial prejudice towards African Americans.
Inclusion Criteria: Participants who are not African American.

Method: Show the film to a sample of university students.
Result: Prejudice scores are lower than the norm.
Problem: Cannot confirm the film caused the reduction; students might have already had lower scores.

Method: Randomly assign university students to two groups: one watches the film, the other doesn't.
Compare prejudice scores between the two groups.
Issue: If groups aren't equal in initial prejudice levels, it can skew results.
External Validity: Using only university students limits the generalizability of findings.

Method: Compare a group of students' prejudice scores before and after watching the film.
Finding: Post-test scores are lower.
Issue: Participants may provide socially desirable responses due to awareness of the film's subject matter and testing effects.

Method: Randomly allocate students into two groups (e.g., 20 in each group).
- Both groups take pre-test and post-test.
- One group watches the film between tests.
Finding: Change scores (difference between pre- and post-test) are lower in the control group.
Problem: Pre-test may sensitize students to the study's purpose, influencing their responses.

Allows for partialling out the impact of the pre-test.
Groups:
- Group 1: Pre-test, Film, Post-test
- Group 2: Pre-test, Post-test (no film)
- Group 3: Film, Post-test (no pre-test)
- Group 4: Post-test (no pre-test or film)
Comparisons:
- Compare post-test scores of groups with and without pre-tests to assess the pre-test impact.
Can equate groups using pre-test scores (highest score to lowest, distributing across groups).
Advantage: Can compare the impact of the pre-test and different scores across time between groups.
Disadvantage: Requires a larger sample size compared to simpler designs.
- Example: 20 people per group in a Solomon design requires 80 participants, whereas a simpler design might accommodate 80 people in fewer groups, thus a more statistically powerful design.

Between-Subjects Design: Different participants in each group.
Within-Subjects Design: Comparing the same participants before and after the intervention.
Mixed Design: Examining changes across time between groups.

Suitable for intact groups: such as gender or religion.
Appropriate for irreversible interventions: such as teaching someone to swim.
Avoids carryover effects: where one treatment influences the outcome of another such as comparing Cognitive Behavioural Therapy and medication for depression.
- Solution: Compare CBT for one group, medication for another. A third group could receive both.
Useful for treatments requiring a long period of time, to reduce dropouts.

Naturally Intact Groups: Groups based on existing characteristics (e.g., gender, religion).
Artificially Derived Groups: Groups created by the researcher based on measured variables (e.g., age split at 50, religiosity based on a median split of questionnaire scores).

Independent Variable: Sex at birth (males vs. females).
Dependent Variable: Aggressive behavior.
Hypothesis: Boys will show more aggressive behavior than girls after seeing a violent movie (Guardians of the Galaxy).
Illustrates using naturally occurring groups as the IV.

Using median split to create high/low socioeconomic status groups or high/low IQ groups.

Ideal for creating control groups, but not always possible or ethical.
Ethical Considerations: Cannot force pregnant women to consume alcohol or assign individuals to male/female groups.

Used when random assignment is not possible.
Example: Studying people with head injuries (naturally occurring group).
Control for other factors: such as sex, education, and trauma of the experience.
Consult with experts (and literature) to determine which factors to control for.