Notes on Experimental vs Non-Experimental Methods
Experimental vs Non-Experimental Methods
- Experimental methodology
- Systematic, controlled conditions
- Tests a hypothesis to establish a causal relationship between independent (IV) and dependent variables (DV)
- Non‑experimental methodology
- Used when control or ethics prevent an experiment
- Describes behavior; cannot infer causation
- Definition: Causation – a relationship where one variable directly influences another.
- \text{Causation} \equiv \text{direct influence of IV on DV}
- In contrast, correlation does not imply causation.
- Types of Non‑Experimental Designs
- Case Study – intensive study of a single individual, group, or event
- Risk: Hawthorne effect – participants alter behavior because they know they’re being observed
- Correlational Study – examines the strength and direction of the relationship between two variables
- Important: Correlation ≠ Causation – only a controlled experiment can determine cause‑and‑effect
- Susceptible to the third‑variable problem (an unmeasured variable influences both studied variables)
- Meta‑analysis – statistical technique that combines results of multiple studies on the same topic to reach a more robust conclusion
- Feature comparison (Experimental vs Non‑experimental)
- Control of variables: High (Experimental) vs Low (Non‑experimental)
- Ability to infer causation: ✅ (Experimental) vs ❌ (Non‑experimental)
- Typical designs: Lab experiments, field experiments (Experimental) vs Case studies, correlational studies, meta‑analysis, naturalistic observation (Non‑experimental)
- Naturalistic Observation
- Observing behavior in its real‑world setting
- Challenge: limited contextual knowledge can lead to misinterpretation (e.g., observing a school during a temporary COVID‑19 setup)
Designing a Study
- Formulating the Hypothesis
- A specific, testable prediction about the relationship between variables
- Must be falsifiable (capable of being proven wrong)
- Example: “Students who use the Ultimate Review Packet will score higher on the AP Psychology exam than students who do not use it.”
- Operational Definitions
- Precise description of how each variable will be measured or manipulated
- Enables replication
- Sleep Study Example:
- More sleep: ≥8 hours of continuous sleep (tracked with Apple Watch)
- Less sleep: <8 hours of continuous sleep
- Performance: Score on the AP Psychology National Exam, reported on a 1–5 scale by the College Board
- Identifying Variables
- Confounding variables reduce internal validity; the more control, the fewer confounds, but over‑control can create an inauthentic environment
- Variable Type → Role → Example (Sleep Study)
- Independent Variable (IV) – Manipulated cause – Amount of sleep: $S \ge 8\text{h}$ vs $S < 8\text{h}$
- Dependent Variable (DV) – Measured effect – Exam score
- Confounding Variables – Uncontrolled factors that may affect DV – Study habits, stress level, overall health
Participants: Population vs. Sample
- Population – the entire group of interest
- Sample – a representative subset of the population used in the study
- Sampling Techniques
- Sampling bias – systematic error where the sample does not reflect the population, limiting generalizability (the extent findings apply to the broader population)
Experimental vs. Control Groups
- Experimental group – receives the IV (e.g., review packet)
- Control (placebo) group – receives a neutral version lacking the critical component of the IV
- Random Assignment
- Participants are randomly assigned to experimental or control groups, ensuring equivalent groups and reducing bias
- Note: Random assignment ≠ random selection (the latter pertains to how participants enter the study)
Quasi‑Experiments
- Used when random assignment is unethical or impossible (e.g., assigning depression)
- Lacks true randomization → cannot definitively establish causation
- Key Principle: Only true experiments with random assignment can reliably infer cause‑and‑effect relationships
Procedures: Blind Designs
- Single‑blind procedure – participants do not know whether they are in the experimental or control group
- Reduces social desirability bias and the placebo effect
- Double‑blind procedure – both participants and researchers are unaware of group assignments
- Mitigates experimenter bias as well as social desirability bias
- Biases and effects to be aware of
- Social desirability bias – Tendency of participants to answer in a way that will be viewed favorably by others
- Placebo effect – Improvement in a participant’s condition that occurs because they believe they are receiving an active treatment
- Experimenter bias – Unconscious influence of a researcher’s expectations on the outcome of a study
Measurements: Qualitative vs. Quantitative
- Qualitative measures
- Collect non‑numerical data for rich, descriptive insight
- Example: Structured interviews with open‑ended questions
- Strength: Captures participants’ thoughts, feelings, and experiences in depth
- Limitation: Subjective, harder to replicate
- Blind Type – Single‑blind: Participants only; addresses social desirability, placebo
- Blind Type – Double‑blind: Participants and researchers; addresses experimenter bias, social desirability
- Limitation: Subjective, harder to replicate
- Quantitative measures
- Gather numerical data amenable to statistical analysis
- Example: Likert scale (rating agreement from strongly disagree to strongly agree)
- Strength: Objective, facilitates replication and statistical testing
- Limitation: May oversimplify complex phenomena
- Qualitative vs. Quantitative vs. Mixed methods (not explicitly stated, but implied)
- Data Forms and Typical Tools (summary)
- Qualitative: Non‑numerical data; Structured interviews, observations; Pros: Rich detail; Cons: Subjective, less reproducible
- Quantitative: Numerical data; Likert scales, surveys, physiological metrics; Pros: Objective, statistical power; Cons: May miss nuance
Protecting Participants: Ethics & Consent
- Informed consent
- Participants receive complete information about the study’s purpose, procedures, risks, and benefits, allowing a rational, voluntary decision to take part
- Informed assent
- Used when participants cannot legally give full consent (e.g., minors); requires parent/guardian permission plus the participant’s agreement
- Ethical research practices
- Create a trusting environment, ensure no harm, and aim for societal benefit
- Measurement & Data Considerations
- Informed consent – Process by which participants are fully briefed on a study’s details and voluntarily agree to participate
- Typical Tools, Pros, and Cons (conceptual overview)
- Qualitative data – rich, context; Pros: depth; Cons: subjectivity and reproducibility concerns
- Quantitative data – numerical; Pros: objectivity and power; Cons: may oversimplify
- Informed assent – Consent process for individuals (often minors) who cannot legally consent on their own; requires guardian approval
- Historical Milestones
- Researchers must debrief participants after the study, explaining the true purpose and any deception used
Conclusions: Peer Review & Replication
- Peer review – Independent experts evaluate a study’s methodology, data, and conclusions before publication
- Replication – Independent researchers repeat the study to verify findings
- Together, peer review and replication uphold scientific standards, ensuring findings are reliable, valid, and generalizable
- Timeline highlights (Significance)
- 1892 – American Psychological Association (APA) founded; Governing body for psychology research
- 1947 – APA’s first Ethical Committee established; research standards
- 1974 – Creation of Institutional Review Board (IRB); formal protection of human participants; Institutional Animal Care and Use Committee (IACUC) oversees ethical treatment of animal subjects
Key Terms Glossary (quick reference)
- IV: Independent Variable – manipulated cause
- DV: Dependent Variable – measured effect
- Confounding Variable: Uncontrolled factor that can influence DV
- Hawthorne Effect: Participants alter behavior due to being observed
- Placebo Effect: Improvement due to belief of receiving treatment
- Random Assignment: Randomly assigning participants to groups to create equivalent groups
- Random Selection: How participants enter the study; different from random assignment
- Naturalistic Observation: Observing behavior in real settings without manipulation
- Meta-analysis: Statistical synthesis of multiple studies
- Quasi‑Experiment: Non‑randomized design that cannot definitively infer causation
- Blind Design: Single‑blind or double‑blind procedures to reduce bias
- Informed Consent: Legal and ethical authorization to participate
- Informed Assent: Assent from participants unable to provide legal consent (with guardian approval)
- Debriefing: Post‑study explanation of true purpose and any deception used
- IRB/IACUC: Regulatory bodies protecting human and animal subjects in research