Quasi Experiments

Introduction to Causality, Correlational, and Non-Experiments

  • Course: Psychology 204

  • Overview of content:

    • Quasi experiments

    • Correlational research

    • Interrogating association claims

    • Small n designs

Quasi Experiments

  • Definition:

    • A quasi experiment resembles a true experiment but lacks random assignment of participants to conditions.

    • Random assignment is essential for internal validity in true experiments.

Importance of Quasi Experiments

  • Situations arise where random assignment isn't feasible, necessitating the use of quasi experiments.

Types of Quasi Experiments

  1. Nonequivalent Control Group Post Test Only Design

    • Characteristics:

      • Participants are not randomly assigned, leading to potential differences among groups.

      • Only the dependent variable is measured post exposure to the independent variable.

    • Example Study:

      • Researchers hypothesized that walking past a religious landmark may bias attitudes towards outgroups.

      • Background:

      • Correlational research indicates that higher religiosity is linked with negative views toward outgroups.

      • Lab studies show priming with religious words correlates to increased prejudice toward outgroups.

      • Method:

      • 99 individuals approached at either a religious or non-religious landmark.

      • Questionnaire assessed views on social groups; dependent variable measured was warmth towards foreigners.

      • Findings:

      • Results indicated that those walking past religious buildings reported lower warmth towards foreigners.

      • Conclusion: Walking past religious landmarks may bias attitudes against outgroups.

  2. Nonequivalent Control Group Pretest Post Test Design

    • Characteristics:

      • Pretest and posttest measurements for both groups that were not randomly assigned.

    • Example Study:

      • Investigating whether telecommuting affects productivity.

      • Pretest average productivity score was around 4, none had switched to telecommuting.

      • After the option to telecommute was offered, productivity was measured again.

      • Findings showed those who telecommuted had lower productivity at posttest.

      • Internal Validity Question:

      • Is it causation from telecommuting, or could selection affects play a role?

  3. Interrupted Time Series Design

    • Characteristics:

      • Repeated measurements of a dependent variable before, during, and after an intervention without a comparison group.

    • Example Study:

      • Investigated if the Netflix show 13 Reasons Why increased youth suicide rates.

      • Suicide rates in the U.S. were measured multiple times before and after the show's debut.

      • Predicted suicide rates compared to the actual rates post-show indicated increases.

      • Internal Validity Question:

      • Can causation be attributed solely to the show, or are there confounding variables?

  4. Non Equivalent Control Group Interrupted Time Series Design

    • Characteristics:

      • Compares several groups before, during, and after an intervention or change, with no random assignment.

    • Example Study:

      • Analyzed the impact of the Affordable Care Act (ACA) in Massachusetts.

      • Data from 200,000 adults in Massachusetts compared to neighboring states without similar laws.

      • Result showed a decrease in people unable to see doctors due to cost in Massachusetts post-ACA implementation.

      • Internal Validity Question:

      • Did the ACA directly improve healthcare access, or are there other factors?

Comparison of Designs

  • Among the four designs, the non equivalent control group interrupted time series is the strongest:

    • It incorporates comparison groups and multiple measurements to enhance internal validity.

Interrogating Internal Validity

  • Internal validity concerns arise when assessing causality in quasi experiments.

    • Challenges include:

    • Selection Effects: Groups differ systematically due to non-random assignment.

    • Maturation Threats: Changes may arise naturally over time, not due to the intervention.

    • History Threats: External events may influence results during the study.

Addressing Internal Validity Issues

  • Include pretests to assess baseline similarities between groups.

  • Use matched groups to control for demographic variables.

  • Ensure constant external variables across groups to isolate the independent variable’s effect.

Benefits of Quasi Experiments

  1. Real-World Opportunities:

    • Exploits natural experiments that ethical or practical concerns prevent from being conducted in true experiments.

  2. External Validity:

    • Results often generalize better to real-life settings.

  3. Ethical Considerations:

    • Allows exploration of sensitive subjects that cannot be ethically randomized.

  4. Strong Construct Validity:

    • Real-life manipulation and measurement typically improve construct validity of independent and dependent variables.

Conclusion

  • Quasi experiments lie on a continuum of internal validity, stronger than correlation studies but lacking the robustness of true experimental designs due to absence of randomization.

  • The next unit will focus on correlational research which shares characteristics with quasi experimental designs.