Study Notes on Quasi Experiments

Introduction to Quasi Experiments

  • Definition of Quasi Experiment: A quasi experiment resembles a true experiment but lacks random assignment of participants to conditions, which is crucial for establishing internal validity.

Outline of the Unit

  • Overview of four video topics:

    • Quasi experiments

    • Correlational research

    • Interrogating association claims

    • Small-n designs

Types of Quasi Experiments

1. Non-Equivalent Control Group Post-Test Only Design

  • Definition: Measures the dependent variable after exposure to one level of the independent variable, without random assignment.

  • Example Study:

    • Research Focus: The effect of walking past a religious landmark on bias towards out-groups.

    • Hypothesis: Walking past a religious landmark leads to bias against other social groups because religious affiliation is correlated with negative views of out-groups.

    • Process:

    • Researchers approached 99 people passing by either a religious or non-religious landmark in two different cities.

    • Participants completed a questionnaire regarding their views on different social groups.

    • Dependent Variable: Warmth towards foreigners measured as a rate of positive sentiment, plotted on the y-axis.

    • Findings: Participants passing religious landmarks reported lower warmth towards foreigners, supporting the research hypothesis.

2. Non-Equivalent Control Group Pre-Test Post-Test Design

  • Definition: Groups are tested before and after an intervention without random assignment.

  • Example Study:

    • Research Focus: The impact of telecommuting on productivity.

    • Process:

    • Measured productivity of employees before they had the opportunity to switch to telecommuting (pre-test).

    • After the switch, productivity was measured again (post-test).

    • Findings: Productivity among telecommuting employees decreased post-switch, raising questions about causation.

3. Interrupted Time Series Design

  • Definition: Measures the dependent variable several times before, during, and after an intervention without comparison groups.

  • Example Study:

    • Research Focus: The impact of the Netflix show "13 Reasons Why" on youth suicide rates.

    • Process:

    • Suicide rates were measured several times before and after the show's debut in April 2017.

    • Findings: A substantial rise in suicides post-show debut compared to predicted rates based on previous years.

4. Nonequivalent Control Group Interrupted Time Series Design

  • Definition: Combines interrupted time series design with variation in comparison groups without random assignment.

  • Example Study:

    • Policy Focus: The rollout of the Affordable Care Act (ACA) versus prior state-level health care legislation in Massachusetts.

    • Process:

    • Collected data on health care access from nearly 200,000 adults in Massachusetts (under new ACA) and neighboring states without similar laws.

    • Findings: ACA led to a reduction in the percentage of people unable to see a doctor due to cost, supporting its effectiveness.

Internal Validity in Quasi Experiments

  • Issues of Internal Validity:

    • Quasi experiments are vulnerable to several internal validity threats because of the lack of random assignment.

    • Selection Effects: Non-equivalence between groups; different characteristics may influence results.

  • Solutions to Enhance Internal Validity:

    • Matched Groups: Ensuring similarity on relevant demographic variables.

    • Pre-Test Measurements: Assessing equivalent baselines before interventions, as in the telecommuting study.

Potential Threats to Internal Validity

1. Maturation Threats

  • Changes over time unrelated to treatment or intervention, possibly influencing dependent variable.

  • Example: Employee productivity may decrease naturally over time, necessitating a comparison group to confirm changes.

2. History Threats

  • External events occurring simultaneously with the intervention may confound results.

  • Example: The suicide rates after the "13 Reasons Why" may be influenced by external factors like economic downturns or societal changes.

3. Selection History Effects

  • External events that impact only one of the non-random groups, further complicating causal inference.

Advantages of Quasi Experiments

  • Real-World Applications: Ability to study situations where random assignment is not feasible.

  • Enhanced External Validity: Results from real-world settings are more likely to generalize to other situations.

  • Ethical Considerations: Avoidance of unethical experimental manipulations, allowing for exploration of significant social issues.

  • Strong Construct Validity: Real-life manipulation of independent variables enhances the relevance of findings.

Comparison to True Experiments

  • Quasi experiments offer intermediate internal validity levels compared to true experiments due to their lack of random assignment but enhance validity through other designed measures.

Conclusion

  • Quasi experiments provide valuable methodologies for studying causal relationships in contexts where traditional experimental designs cannot be employed. They offer insights while maintaining ethical considerations and real-life relevance.

  • Next Steps: Further exploration in the next unit will cover correlational research, highlighting similarities and differences with quasi experimental designs.