Notes for Exam: Correlation, Causation, and Experimental Methods in Psychology

Overview and purpose of the optional lectures

• The lecturer frames these sessions as a way to bring psychology to students in a context that feels relevant and helpful for university life.
• Two core parts of psychology are highlighted as especially useful for success, belonging, and wellbeing:
  • Understanding how correlations are related but not necessarily causal.
  • Using the scientific method and experiments to establish causal relationships.
• The sessions are optional; the instructor anticipates many online participants and aims to provide space for interested students.
• Administrative updates discussed include:
  • Indigenous welcome as a social bridge to help students make friends on campus.
  • An issue with connecting with fellow students via the class platform (PureScholar/Outlook integration); the dev team is working on a fix and an announcement will follow.
  • SONA (psychology participation platform) is used for recruiting participants for experiments; people may see delays before experiments appear, and early term slots fill quickly.
  • SONA experiments often release over time (e.g., October) as programs are programmed and deployed.
• The session closes with a reminder that this material will feed into Friday’s lecture.

Key concepts introduced

• Correlation vs. causation
  • Correlation: A relationship or association between two variables.
  • Causation: A directional relationship where one variable directly influences another.
  • Important takeaway: Correlation does not imply causation. Seeing a relationship does not tell you which variable causes the other, or whether a third variable explains both.
• The role of causality in science
  • Scientists seek to establish causal relationships to make reliable predictions.
  • The move from correlation to causation typically requires experimental manipulation.
• The scientific approach in psychology
  • The emphasis is on testing explicit hypotheses that follow from theories.
  • Theories are stepwise and lead to testable predictions; empirical testing is essential.
  • The goal is falsification: to potentially reject theories if data contradict them, rather than proving them forever correct.
• Philosophical contrast: theory vs. hypothesis vs. data
  • Theories describe general relationships and mechanisms.
  • Hypotheses are explicit predictions derived from theories, designed to be tested.
  • Data from experiments inform whether a theory remains viable.
• The Vienna Circle and the scientific method (historical note)
  • The Vienna Circle contributed to formalizing scientific reasoning: theory -> hypotheses -> experiments -> data-driven conclusions.
  • The scientific method is presented as the most reliable way to determine causal mechanisms.
• Ethical and practical implications in science communication
  • Peer-reviewed publishing serves as an important quality control; however, a notorious example (vaccine-autism controversy) shows that poor-quality papers can mislead public discourse when they pass peer review.
  • Emphasizes the need for critical evaluation of scientific claims and sources.

Core terminology and definitions

• Correlation
  • A relationship between two variables where changes in one are associated with changes in the other, but not necessarily causal.
  • Example discussion: teacher ratings vs. student performance – correlation does not tell you which direction causality lies, or whether a third variable explains both.
• Causation
  • A relationship where a change in one variable directly produces a change in another.
  • Establishing causation requires experimental manipulation and control of confounds.
• Independent variable (IV)
  • The variable the experimenter actively manipulates to assess its effect.
  • In the example study, the manipulation is to induce different self-esteem states (high vs. low).
• Dependent variable (DV)
  • The outcome measure that is observed and recorded.
  • In the example study, prejudice (as measured by endorsement of prejudicial attitudes) is the DV.
• Operational definition
  • Precise, replicable procedures used to manipulate an independent variable and to measure a dependent variable.
  • Example: “knocking over slides” to create low self-esteem; “survey indicating prejudicial attitudes toward French Canadians” to measure prejudice.
• Manipulation check
  • A post-hoc assessment to verify that the intended manipulation actually affected participants as planned.
  • Helps rule out alternative explanations (confounds) by confirming participants experienced the manipulation.
• Confound
  • A variable that differs between experimental groups other than the intended manipulation, potentially explaining observed effects.
  • Example: momentary anger or distrust toward the researcher could become a confound if it changes responses independently of self-esteem manipulation.
• Internal validity
  • The extent to which a study establishes a causal relation between the IV and DV without influence from confounds.
  • Peer reviewers scrutinize for confounds and whether the manipulation causes the observed effect.
• External validity
  • The extent to which study results generalize beyond the specific experimental conditions to broader populations, settings, and times.
• Replication
  • Repeating a study (possibly with variations) to see if results hold across different samples and methods.
• Cultural validity (cultural generalizability)
  • The degree to which results apply across different cultures and contexts.
  • Noted: much early psychology data came from Western, largely white, university students; cultural context can influence psychological phenomena.
• Theory, hypothesis, and experiment
  • Theory: broad explanation of phenomena.
  • Hypothesis: explicit, testable prediction derived from a theory.
  • Experiment: a controlled test to determine whether the hypothesis holds under manipulated conditions.
• Random selection vs. random assignment
  • Random selection: helps ensure the sample represents the population.
  • Random assignment: ensures that groups are equivalent on average on all variables not manipulated by the experiment (minimizing preexisting differences).
• Control group vs. experimental group
  • Control group: does not receive the manipulation (baseline condition).
  • Experimental group: receives the manipulation.
• Operationalization in practice
  • The “how” of definitions is essential for reproducibility and interpretation.
  • Example: whether “low self-esteem” is induced by a task, a feedback condition, or a social manipulation.
• Internal validity checks in peer review
  • Confounds, manipulation checks, and measurement validity are scrutinized.
  • Reliability and accuracy of prejudice measures are evaluated; implicit attitude tests are an option.
• External validity checks in peer review
  • Assess whether findings support broader theory and real-world applicability.
• Replication and extension
  • Demonstrating effect across multiple methods and settings strengthens theory.

A worked example: prejudice and self-esteem experiment

• Theoretical background
  • A historical perspective links prejudice to contextual economic hardship and to ignorance about out-groups.
  • In-groups vs. out-groups: people derive identity from groups and may react to perceived threats with prejudice toward out-groups.
• Goal of the experiment
  • Test whether lowering self-esteem increases prejudicial attitudes.
• Experimental design (two-group randomized design)
  • Random selection of participants into two groups: control vs. experimental (random assignment).
  • Independent variable: self-esteem state manipulation (high vs. low self-esteem).
  • Dependent variable: level of prejudicial attitudes (measured via a survey).
  • Control group: does not experience the self-esteem manipulation.
  • Experimental group: experiences the self-esteem manipulation.
• Operationalization details (how the manipulation is done)
  • Example manipulation: participants go through a situation that manipulates self-esteem (e.g., a task, feedback, and whether they feel they performed well or poorly relative to peers).
  • Example measurement: attitudes toward French Canadians via a survey with endorsement choices reflecting prejudice.
• The specific study (slides-knockover paradigm)
  • Participants come in for an experiment; a graduate student presents slides for a presentation.
  • Half the participants experience a scenario where the slides are knocked over by the experimenter with a foot pedal (low-status manipulation), causing perceived blame and lower self-esteem.
  • The other half experience no incident (control condition).
• Hypotheses derived from the theory
  • H1: Participants in the low-self-esteem condition will display higher endorsement of prejudicial attitudes toward French Canadians than those in the control condition.
• Operational definitions in the study
  • Self-esteem manipulation: presence or absence of the “knocked over slides” incident (manipulation of self-esteem).
  • Prejudice measure: survey items assessing prejudicial attitudes toward a target group (French Canadians).
• What counts as evidence of causality
  • Because the groups are randomly assigned and the only systematic difference is the manipulation, any observed difference in prejudice can be attributed to the manipulation of self-esteem.
• What the findings imply
  • The experiment shows a causal link between lowered self-esteem and increased prejudicial attitudes in this context.
  • Demonstrates the value of experimental methods in establishing causation rather than relying on correlation alone.
• Additional methodological notes
  • The study illustrates the use of manipulation checks and potential confounds to ensure internal validity.
  • It also highlights the importance of measurement validity for prejudice (survey methods, potential use of implicit tests).
• Why this matters for psychology as a science
  • Clear demonstration of causal mechanisms aligns with the scientific goal of falsification and theory refinement.
  • Emphasizes the role of random assignment, control conditions, and explicit operational definitions.

Peer review, validity, and the scientific community

• Peer review as quality control
  • Papers go through at least three peer reviewers who assess: internal validity, external validity, measurement quality, and overall rigor.
  • Reviewers look for potential confounds, manipulation checks, replication opportunities, and cross-context applicability.
• Internal validity concerns
  • Potential confounds and alternate explanations must be ruled out.
  • If a manipulation could plausibly cause the observed outcome through another path, the causal claim is weakened.
• Manipulation checks and measurement concerns
  • Researchers should assess whether participants experienced the intended manipulation.
  • Measurement validity: are the instruments (e.g., prejudice surveys, implicit tests) accurately capturing the construct?
• External validity concerns
  • Does the explicit finding apply beyond the laboratory setup to real-world settings and broader populations?
• Replication and cross-method validation
  • Multiple experiments with various manipulation methods strengthen confidence in a theory.
  • If three different manipulations yield the same directional effect on prejudice, theory support increases.
• Cultural validity
  • Acknowledges that results derived from Western, largely homogeneous samples may not generalize globally.
  • Calls for cross-cultural studies to test whether the same mechanisms hold across cultures and contexts.
• Practical takeaway
  • Scientists emphasize falsifiability, replication, and critical evaluation to separate robust findings from spurious results.

Cultural and societal context of psychology research

• Western-centric origins of much classic psychology
  • Much foundational work came from white, relatively narrow populations (e.g., college students in the Northeast U.S.).
  • Contemporary psychology increasingly addresses cultural validity and diversifies samples.
• Cross-cultural implications
  • Differences in family structure and social roles (e.g., elder care in Western vs. Eastern cultures) influence mental health and meaning-making.
  • Culture can shape how psychological processes manifest and how they should be measured.
• Implications for interpreting research
  • When reading studies, consider population, cultural context, and whether findings generalize.
  • Be mindful of cultural biases in interpretation and the need for replication in diverse samples.

The scientific method and the role of skepticism in science

• The Vienna Circle and scientific rigor
  • Historical development of a formal approach: theory -> hypotheses -> experiments -> data-driven conclusions.
  • Emphasizes the community aspect: peer review, replication, and cumulative knowledge.
• The danger of post-truth and misinterpretation of science
  • A single flawed paper can gain a “stamp of science” and influence public discourse if not scrutinized.
  • The vaccine-autism controversy is cited as a cautionary example of how poor science can mislead public health decisions.
• How to assess scientific claims
  • Look for peer-reviewed sources and replication of results.
  • Evaluate internal validity (confounds, manipulation checks) and external validity (generalizability).
  • Consider alternative explanations and whether the authors ruled them out.
  • Be wary of sensational headlines that oversimplify correlations as causations.

SONA and practical notes for students

• SONA as a tool for experimental participation
  • Students sign up for experiments; early sign-up can fill slots quickly.
  • Some platform features (e.g., direct email connections between students) may not function perfectly yet due to deployment issues.
  • The instructor plans to follow up with announcements once technical issues are resolved.
• What to expect when participating
  • There is typically a process of consent, paperwork, and then the experimental task.
  • Some studies may involve seemingly odd setups (as in the slide-knockover example) to manipulate psychological states.
• What to take away for exams
  • You should be able to distinguish correlation from causation and explain why experiments are required to establish causality.
  • Understand the roles of IV and DV, control conditions, random assignment, and confounds.
  • Be able to describe operational definitions, manipulation checks, internal/external validity, and replication.
  • Recognize how cultural validity can affect generalizability and interpretation of results.
  • Be prepared to discuss peer review and the ethical responsibilities of scientists in publishing and communicating results.

Quick recap of key takeaways

• Correlation does not equal causation; causal inference requires controlled manipulation and ruling out third variables.
• Experiments use random assignment to create equivalent groups, manipulate the IV, and measure the DV to test hypotheses.
• Operational definitions specify exactly how variables are manipulated and measured; manipulation checks help confirm the manipulation worked.
• Internal validity focuses on ruling out confounds; external validity concerns generalizing findings beyond the lab.
• Replication and cultural validity strengthen theories and their applicability.
• Peer review is a crucial gatekeeping process to ensure scientific quality, but it is not infallible; critical thinking remains essential when interpreting findings.
• The scientific method provides a rigorous framework for answering questions, testing theories, and refining our understanding of human behavior.

r = rac{ ext{cov}(X,Y)}{\sigmaX \sigmaY}

• This is the standard formula for the Pearson correlation coefficient, reflecting the strength and direction of a linear relationship between two variables $X$ and $Y$.