Comprehensive Notes on Experimental vs Correlational Research (Transcript)

Key Concepts

• Transcript covers distinguishing correlation from causation, and how experimental designs control for confounds.
• Example at start: Syed thinks Diet Coke causes his mother's weight gain. Discussion reveals reverse causality possibility (weight gain could lead to Diet Coke choice, or other factors), not a third-variable problem in this case.
• polling results shown during discussion: 37% chose d, 16% chose c, 21% chose b, 26% chose a. Correct answer is d (reverse causality problem).
• Distinctions:
  • Third variable problem: a third variable causes both measured variables and creates a spurious relationship.
  • Reverse causality problem: correlation exists but the direction of causation is unclear; the outcome could influence the predictor.
  • Compounding validity effect and dependent variable effect are not real concepts in this context (not taught as valid explanations here).
• Emphasis on practice with test questions: if unsure, eliminate obviously invalid options, identify direction of causality, and recognize that some options may be made-up.
• Core concept: correlation does not imply causation; only experiments can establish causality under ideal conditions.

Correlation vs Causation: Key Definitions and Examples

• Correlation: a relationship between two variables where they tend to vary together, but one does not necessarily cause the other.
• Causality: a change in one variable directly produces a change in another, under controlled conditions.
• Example discussed: study time and GPA.
  • Best interpretation of a correlation coefficient r = 0.85 in the context given: students who study more tend to have a higher GPA (a correlational interpretation).
  • Common misinterpretation to avoid: "Study time explains 85% of GPA" (causal claim; incorrect for a correlational study).
• Why correlation alone is not enough to claim causality:
  • Potential third variables not measured could drive the relationship.
  • The direction of effect cannot be established from a correlation alone.
• Recap from earlier in the lecture:
  • Third variable problem example: a third factor explains both the predictor and outcome.
  • Reverse causality example: A correlation where the direction of influence is ambiguous.

Experimental Research: IVs, DVs, Groups, and Causality

• Experiment definition: the researcher actively manipulates a variable believed to have a causal effect on another variable.
• Independent Variable (IV): the manipulated variable that is thought to cause an effect.
• Dependent Variable (DV): the variable measured to assess the effect of the IV.
• Example 1 (discrimination and stress):
  • IV: exposure to discriminatory treatment.
  • DV: stress levels.
  • If discrimination increases stress, we infer causality under controlled conditions.
• Example 2 (sunlight and plant growth):
  • IV: exposure to sunlight.
  • DV: plant growth.
  • Demonstrates manipulation of the IV and measurement of the DV.
• Experimental group: the group exposed to the manipulation of the IV.
• Control group: the group not exposed to the manipulation; baseline for comparison.
• Placebo group (health research): a group receives a pill that looks like medication but has no active ingredient.
  • Purpose: to control for the placebo effect and isolate the pharmacological effect of the medication.
• Double-blind study: both the participant and the experimenter are unaware of the condition assignment to prevent demand characteristics and experimenter bias.
• Random assignment: participants are randomly assigned to conditions to ensure equivalence and control for participant variables.
• Random sampling vs random assignment:
  • Random sampling: how participants are selected from the population (representativeness).
  • Random assignment: how sampled participants are allocated to experimental vs. control groups (causal inference).

Experimental vs Correlational: Why Experiments Support Causality

• In correlational studies, variables are measured as they naturally occur; no manipulation occurs.
• In experiments, the IV is manipulated first, then DV is measured.
• Why this helps establish causality:
  • Temporal precedence: IV is applied before DV is measured.
  • Control of confounds: random assignment and control groups help ensure other variables do not drive differences.
  • Elimination of reverse causality: the manipulated IV cannot be caused by the DV when the manipulation occurs first.

Extraneous and Confounding Variables; Demand Characteristics

• Extraneous variables / confounds: any uncontrolled variables that could influence the DV.
• Demand characteristics: features of a study that cue participants to guess the hypothesis and alter their responses accordingly.
• Experimenter bias: the experimenter’s expectations unintentionally influence participants’ responses.
• Participant variables: inherent individual differences (e.g., baseline mood, sleep quality) that can affect outcomes.
• Ways to mitigate:
  • Double-blind designs to reduce experimenter bias and demand characteristics.
  • Random assignment to equalize participant variables across groups.
  • Use of placebos to separate psychological from physiological effects.
  • Clear operationalization of variables to ensure consistent measurement across groups.

Operationalization and Measurement in the Examples

• Operationalization: defining how a variable is measured or manipulated in a study.
  • Example: DV operationalized as reaction time measured in milliseconds.
  • Example: Resting blood pressure measured in mmHg as the DV in the mindfulness study.
• In the memory-stress study design discussions, various operationalizations were proposed:
  • Baseline memory measurements vs post-stress memory assessments.
  • Stress manipulated via time pressure, distractions, or workload.
  • Memory performance measured via a memory game performance (e.g., number of matches).

Activity: Identifying IVs, DVs, Experimental/Control Groups, and Design Types

• Several classroom examples were discussed to practice identifying:
  • Independent variable (IV)
  • Dependent variable (DV)
  • How the DV is operationalized
  • Which group is experimental vs control
  • Whether a true experiment or a quasi-experiment is described
• Example outcomes discussed:
  • Biopsychology sleep deprivation: IV = sleep condition (8 hours vs 24 hours awake); DV = reaction time (milliseconds).
  • Industrial psychology temperature: IV = room temperature (65°F vs 76°F); DV = number of packages packed.
  • Developmental psychology delay of gratification: IV = age group; DV = duration of waiting/ability to delay gratification; design discussed as quasi-experiment due to inability to randomly assign age.
  • Mindfulness and stress: IV = mindfulness meditation vs no intervention; DV = resting blood pressure.
  • Stress and memory: IV = stress level; DV = memory performance; design typically involves comparing stressed vs non-stressed groups, with various proposed methodologies (baseline tests, time pressure, distractions).
• Key takeaway from the activity:
  • When using age as the IV, the design becomes quasi-experimental because age cannot be randomly assigned.
  • In true experiments, random assignment and a clear experimental/control grouping are used to infer causality.

Quasi-Experiment vs True Experiment (Developmental example)

• Developmental example discussed as quasi-experiment: age is used as the IV, but participants cannot be randomly assigned to age groups.
• Implication: without random assignment, causal inferences are weaker; age-based comparisons are cross-sectional rather than fully experimental.
• Possible workaround if forced to design as an experiment: explicitly define older vs younger groups, but acknowledge limitations and avoid claiming strict causality.

Practical Design Principles for Exam and Real-World Research

• When evaluating a study, check:
  • Is there random assignment? If not, be cautious about causal claims.
  • Is there a control group? Is there a placebo control when appropriate?
  • Are the IV and DV clearly defined and properly operationalized?
  • Are potential extraneous variables controlled for or acknowledged?
  • Is the study double-blind to reduce bias?
  • Are random sampling and sampling bias considered for generalizability?
• Common exam pitfalls to watch for:
  • Treating correlation as causation (e.g., “85% of GPA explained by study time”).
  • Assuming a third variable is the only alternative without examining directionality or design.
  • Misidentifying the experimental vs control group or misreading the design type (true vs quasi).

Summary of Key Formulas and Notation

• Correlation coefficient: r = 0.85 (example from the lecture) indicating a strong positive relationship, not causation.
• No explicit p-values or other formulas were provided in the transcript, but the conceptual framework for interpreting r and for causal inference in experiments was emphasized.

Real-World Relevance and Ethical Considerations

• Real-world relevance: understanding when we can claim causality informs policy, clinical practice, and science communication.
• Ethical and practical considerations discussed include:
  • Use of deception in some procedures (e.g., discriminatory scenarios) requires careful ethical oversight and debriefing.
  • Placebo use in medical trials must be ethical and justified, with informed consent and safety monitoring.
  • Anonymity and confidentiality (teacher notes that responses are anonymous in class discussions) support ethical data handling.
  • Transparency about design limitations (e.g., quasi-experiments) helps avoid overstated causal claims.

Quick Reference: When to Use Which Design

• Use correlational designs to identify relationships and generate hypotheses about potential causal links.
• Use true experiments to test causal effects, ensuring random assignment, appropriate control conditions, and manipulation of the IV.
• Use quasi-experiments when random assignment is not feasible (e.g., age groups in development) and acknowledge limitations in causal inference.

Final Takeaways for Exam Preparation

• Be able to articulate the difference between correlation and causation, and explain why experiments are needed for causal claims.
• Distinguish IVs from DVs and identify them from brief study descriptions.
• Identify experimental vs control groups and understand the role of a placebo in pharmacological testing.
• Explain the purpose and implementation of double-blind designs and random assignment.
• Recognize extraneous variables (demand characteristics, experimenter bias, participant variables) and describe strategies to mitigate them.
• Understand operationalization by giving concrete examples (e.g., reaction time in ms, resting blood pressure).
• When given a study scenario, practice naming IV, DV, how DV is measured, and whether the design is true experimental or quasi-experimental due to non-random assignment (e.g., age groups).