week four recording RM part one

Experimental Designs in Psychology

• Different experimental designs can impact statistical power and the validity of the results.

Within Subjects Design

• This design involves using the same participants across different conditions.

  • Advantages:

    • Higher statistical power because the same group is measured multiple times.

    • Reduces variability due to participant differences (extraneous variables), offering a better match when comparing conditions (e.g., Alan vs. Alan).

    • Fewer participants are needed since each participant serves as their own control.

Matched Subjects Design

• An intermediate approach where participants are matched based on certain characteristics.

  • Benefits:

    • Provides some level of power similar to within subjects, but not as high.

    • Can control for important variables that could confound results.

  • Limitations:

    • Requires careful selection of matching variables that actually influence the study outcome.

    • If irrelevant variables are matched (e.g., province of birth unrelated to cognitive function), it may not be necessary or beneficial.

Importance of Matching

• Matching should relate directly to the outcome measure of the study, ensuring it's a relevant characteristic.

• Example given relates to a study on Alzheimer’s cognition:

  • Province of birth is irrelevant if there’s no demonstrated link to cognitive function.

Statistical Power and Effect Detection

• Power: The ability to detect an effect when there is one.

  • Within subjects and matched designs help increase power by controlling for extraneous variables.

• Inadequate detection can lead to:

  • Type I Errors: Incorrectly finding an effect that doesn’t exist.

  • Type II Errors: Failing to detect an effect that is present (especially in studies with minimal evidence).

Random Assignment and Extraneous Variables

• True experiments require manipulation of variables and random assignment to ensure balanced distribution of extraneous variables.

• Problems arising from small differences in groups may not be significant, but researchers should strive for higher detection of effects through design choices.

One-sided vs. Two-sided Tests

• Depending on the hypothesis, researchers may choose:

  • One-sided test: Predicts the direction of the effect based on strong theory or prior data.

  • Two-sided test: Considers effects in both directions, typically utilized when no direction is assumed.

Conclusion

• A thoughtful approach to experimental design is necessary to maximize power and minimize errors.

• While within subjects designs offer significant benefits, matched subjects designs can serve as effective alternatives when within subjects is not feasible.