Ethics, Replication & Scientific Judgment

Imagine a study could answer an important question, but participants would be secretly recorded in a private setting without consent. What scientific value might the study have, and why would that not automatically make the procedure acceptable?

• Explain informed consent, voluntary participation, privacy, confidentiality, minimizing harm, deception, debriefing, and ethical review.
• Explain why replication and converging evidence matter.
• Distinguish a single study from a scientific conclusion.
• Apply the Evidence Ladder to claims.
• Complete a responsible Week 1 scientific investigation design.

Related: reproducibility · converging evidence · peer scrutiny · uncertainty · revision

Psychology cannot separate scientific quality from ethical responsibility. A technically clever study can still be unacceptable if it violates consent, exposes private information, creates unnecessary risk, or uses people merely as means to an end. Ethical safeguards are therefore not decorations added after a study is designed; they shape what questions can be asked and how they can be investigated.

Informed consent means that people receive enough relevant information to decide voluntarily whether to participate. Consent is not meaningful if participation is coerced or if refusing carries an unfair penalty. Some research designs may use limited deception when revealing the full purpose beforehand would invalidate the study, but deception requires strong justification, ethical review, risk minimization, and appropriate debriefing.

Privacy and confidentiality are related but different. A researcher might violate privacy by collecting information from a setting where people reasonably expect not to be observed. A researcher might violate confidentiality by collecting information appropriately but later revealing a participant's identity or sensitive responses. Responsible data practices limit what is collected, who can access it, and how long it is retained.

Ethics also influences what students in this course may do. You will analyze fictional, public, academic, and provided datasets. You are not asked to secretly test classmates, manipulate family members, provoke distress, collect medical information, or create amateur diagnostic profiles. Learning research design does not require exposing real people to avoidable risk.

Scientific responsibility continues after data collection. A surprising result should not become a universal claim. Replication tests whether a finding can be observed again. Replications can be close repeats or conceptual tests using different procedures. A finding that appears across multiple independent studies, samples, and methods provides stronger evidence than a single isolated result.

Science also includes revision. A failed replication does not automatically prove that the original researchers were dishonest, and a successful replication does not prove that a theory is eternally true. Differences in samples, procedures, measurement quality, statistical power, or context may matter. The task is to investigate why evidence converges or diverges.

This is why the course uses an Evidence Ladder. Observation supports description. Association supports relational claims. Controlled experiments can strengthen causal claims under defined conditions. Converging evidence across methods supports broader confidence. Strong science is not certainty at all costs; it is disciplined updating.

Question: Does social exclusion change willingness to cooperate?

Unsafe classroom idea: Secretly arrange for real classmates to exclude one student, then observe cooperation.

Why unacceptable here: It intentionally manipulates a real student's social treatment, could create distress or reputational harm, and lacks meaningful consent.

Safer alternative: Analyze a fictional dataset, use published research examples, or present hypothetical scenarios and ask students to identify design features rather than manipulating peers.

1. Claim: “always obey” is universal and therefore extremely strong.
2. Measure: What counted as obedience in the specific procedure?
3. Evidence: What happened in the original sample and in later replications or variations?
4. Alternative: Context, interpretation, experimenter behavior, culture, and procedure may influence outcomes.
5. Limitation: One experimental setting cannot establish a universal human rule.
6. Conclusion: Authority can influence behavior under some conditions; the magnitude and form depend on context.

The evidence is mixed. A responsible synthesis does not say “replicated” or “debunked” without examining differences in design, uncertainty, and effect size.

For each fictional proposal, identify the primary ethical problem and one safer redesign.

Design—do not conduct—a responsible investigation of a public, fictional, academic, or non-sensitive behavior. Your 700–1,000 word design must include: observation, research question, at least three perspectives, variables, operational definitions, method, population/sample, predicted evidence, alternative explanation, ethics/privacy safeguards, limitation, and conclusion wording appropriate to the design.