Notes on Validity, Ethics, and Animal Research in Psychology

Validity in Research

  • Construct validity

    • Definition: Are we measuring what we say we are measuring?
    • Problem: Measuring stress vs happiness; ensuring questions target the intended construct.
    • Assessment tools: surveys, parent reports, direct observation, physiological measures; choose tools that are reliable and valid for the construct.
    • Example issue: Measuring stress with a physiological or survey tool and ensuring the item set actually taps stress, not unrelated states.

  • Reliability

    • Definition: Consistency of measurements across time, items, or raters.
    • Practical point: If a measure is reliable, repeated readings at similar times should yield similar results, unless there is a true change due to manipulation or condition.
    • BP example: Taking blood pressure readings minute-by-minute for several minutes often yields readings that vary; high variability may indicate low reliability of the measurement protocol rather than true physiological change.
    • General rule: If you administer the same survey at multiple time points, expect responses to be around the same unless you are intentionally inducing a change.

  • Measurement tools and validity considerations

    • The choice of measurement tool (survey, parent report, direct observation, physiological measure) must align with the construct and be reliable.
    • The goal is to have both construct validity and reliability for any assessment tool used.

  • External validity (generalizability)

    • Question: Do results generalize beyond the sample to the population of interest?
    • Population of interest example: Mercer students.
    • Sampling example: If you sample only from Psych 250 (mostly psychology majors) to generalize about all Mercer students, you cannot claim the results apply to the entire Mercer student body.
    • Conclusion: You can say results are true for your population of interest (e.g., the sampled group), but not necessarily for all possible populations (e.g., all Mercer students or other universities).
    • Important nuance: Not being able to generalize does not make the study bad science; it just limits the scope of inference.

  • Internal validity

    • Definition: The degree to which you can conclude that X causes Y, by ruling out alternative explanations (third-variable problem).
    • Relevance: Internal validity matters for experiments and correlations: you must ensure that the observed relationship is not explained by a confounding variable.
    • Conceptual example: If you find that Y and Z move together, and you claim Y causes Z, you need to demonstrate that there is no third variable (e.g., temperature) that drives both.
    • Confounds: The "murder in the ice cream" analogy—outside temperature is a confound that could explain the association between ice cream sales and heat.

  • Ethical foundations in research (overview)

    • Core idea: Ethics codes exist to prevent harm and misuse of participants in the name of science.
    • Everyday reminder: Curiosity should not harm people; historical abuses (e.g., Tuskegee Syphilis study) illustrate the risks of unethical research.

  • Autonomy and informed consent

    • Autonomy: Participants must have the option to say no to participating in research.
    • Informed consent: Clear information about the study, risks, and rights before participation.
    • Right to withdraw: Even after consenting, participants can leave the study at any time without penalty.
    • Vulnerable populations: Additional protections (e.g., children require parental consent and participant assent).
    • No harm principle: Researchers should avoid physical or psychological harm; benefits should outweigh risks.

  • Beneficence and risk/benefit analysis

    • Beneficence: Maximize benefits and minimize harms to participants.
    • Risk/benefit balance: If a study induces stress or discomfort, researchers must provide a method to undo or mitigate harm (undo button concept).
    • Undo/mitigation examples: Mindfulness exercises, debriefing, or cooling-down techniques after stress induction; provide resources if distress occurs.
    • Informed consent documents often include information about support resources if participation leads to negative outcomes.
    • Clinical trials example: Participants incur risk (e.g., side effects) to gain knowledge that will improve future treatments; risk must be justified by potential benefits.

  • Justice in research

    • Principle: Fairness and equitable selection of participants; populations chosen should benefit from the research.
    • Avoid exploitation: Don’t select populations merely because they are easy to access or because they are available and convenient.
    • Example caution: Studying workplace stress in prisoners is problematic if they don’t stand to benefit from the results; prisoners are a vulnerable population with added protections.
    • Aim: Distribute benefits and burdens fairly across groups.

  • Animal research and welfare (IACUC and the 3 Rs)

    • Oversight: Animal research is overseen by IACUC (Institutional Animal Care and Use Committee) to ensure humane treatment.
    • Welfare-focused rather than rights-focused: IACUC evaluates welfare, not the broader animal rights debate.
    • Three Rs framework:
    • Replacement: Use non-animal methods whenever possible; avoid animals if alternatives exist.
    • Refinement: Design procedures to minimize distress and suffering; improve welfare during experiments.
    • Reduction: Use the minimum number of animals necessary to achieve scientific aims; example discussion about reducing sample size (e.g., from 30 to 25) to achieve the same objective with fewer animals.
    • Practical nuance: Replacement/refinement/reduction aim to support beneficence toward animals while enabling scientific progress; not to override the need to answer important questions.
    • Animal welfare vs animal rights: IACUC focuses on welfare and humane treatment; some ethical debates emphasize animal rights and consent, but IACUC operates within welfare-focused guidelines.

  • Putting it together: ethics, validity, and responsible research

    • Ethical considerations influence study design, participant welfare, and the interpretation of results.
    • Validity concerns (construct, reliability, external, internal) shape how confidently we can generalize findings and infer relationships.
    • When planning studies, researchers balance methodological rigor with ethical obligations to participants and animals, aiming for beneficial knowledge with minimal harm.

Key formulas and concepts

  • Correlation (as a measure of association, not causation):
    r=extcov(X,Y)σ<em>X σ</em>Yr = \frac{ ext{cov}(X,Y)}{\sigma<em>X \, \sigma</em>Y}

  • Standard deviation (measure of variability):
    \sigma =
    rac{1}{N}
    igg(
    rac{}{}
    igg)^{ rac{1}{2}}
    = \sqrt{\frac{1}{N}\sum{i=1}^{N}(xi - \mu)^2}

  • Mean (central tendency):
    μ=1N∑<em>i=1Nx</em>i\mu = \frac{1}{N} \sum<em>{i=1}^{N} x</em>i

  • Internal validity intuition: the idea that Y is truly related to Z only if alternative explanations (like X) are ruled out; no strict numeric formula governs this in the lay explanation, but the concept is critical in experimental design and statistical controls.

  • Notation reminder (concepts to know):

    • Construct validity, reliability, external validity, internal validity
    • Autonomy, informed consent, assent (for minors), withdrawal rights
    • Beneficence, nonmaleficence, justice
    • Replacement, refinement, reduction (3 Rs)