Ch. 2 Part 2 Notes: Experimental Method and Correlational Method
Experimental Method: Overview
Core example question: "Does misery love company?" studied by Stanley Schachter
Context: When people are anxious, do they want to be left alone or be around others?
Schachter’s stance (based on relevant theory): in times of anxiety, people may seek others to help sort out feelings; this reflects the need for affiliation
Structure of the lesson: use the Schachter study to illustrate experimental method, variable types, group design, and control of extraneous variables
The Schachter Question and Hypotheses
Research question: In anxiety, do people desire affiliation (i.e., being with others) more as anxiety increases?
Hypothesis (informal): Increases in anxiety will cause increases in the desire to be with others (need for affiliation)
How hypotheses connect to methods:
The hypothesis motivates the manipulation of a variable (IV) and observation of a consequent change in a second variable (DV)
What is an Experiment? Key definition
Experiment (as a research method):
The investigator manipulates a variable (the independent variable, IV) under carefully controlled conditions
Observes whether changes occur in a second variable (the dependent variable, DV) as a result of the manipulation
Purpose: to detect cause-and-effect relationships (Does IV cause changes in DV?)
In psychology, experiments are a dominant method because of their potential to establish causality
Formal restatement (Schachter example):
IV is the manipulated factor (anxiety level)
DV is the measured outcome (desire to affiliate / need for affiliation)
Variables in the Schachter Study (IV, DV, and the role of extraneous variables)
Independent Variable (IV): the level of anxiety manipulated by the experimenter
In Schachter’s design, anxiety level was varied by deception via a cover story about electric shocks
Actual shocks were not administered; the manipulation was intended to evoke different anxiety levels
Dependent Variable (DV): the desire to affiliate, operationalized as the participants’ preference for waiting with others during a delay
Hypothesized effect: higher anxiety (IV) would lead to greater affiliation (DV)
Explanation of manipulation and measurement sequence:
Participants were told they would undergo a physiological test with possible electricity exposure
One group was told shocks would be very painful (high anxiety); the other group told shocks would be mild/painless (low anxiety)
During a delay before the procedure, participants chose whether to wait alone or with others; this choice served as the DV
Experimental and Control Groups: Design and purpose
Experimental group: participants who received the anxiety-arousing manipulation (high anxiety condition)
Control group: participants who did not receive the anxiety-arousing manipulation (low anxiety condition)
Purpose of the two groups: to compare DV outcomes under different levels of the IV while holding other factors constant
Why two groups? To isolate the effect of the IV on the DV by ensuring that the groups are similar on everything except the IV
Important caveat: in practical research, perfect similarity across all variables is impossible; researchers aim to balance relevant variables that could influence the DV
Extraneous Variables and the Problem of Confounds
Extraneous variables: any variables other than the IV that could influence the DV
Why they matter: if extraneous variables differ between groups, they can confound results and make it unclear whether observed DV changes are due to the IV
Examples from the Schachter context:
Sociability or extroversion could influence the likelihood of affiliating independent of anxiety
Confounding: when an extraneous variable is linked with the IV in a way that makes it difficult to disentangle its effects on the DV
If the high-anxiety group happened to be more sociable on average, observed affiliative behavior could be due to sociability rather than anxiety
The goal: make experimental and control groups alike on relevant extraneous variables to isolate the IV effect
How researchers control extraneous variables: Random assignment
Random assignment (a safeguard):
Each participant has an equal chance of being assigned to any group or condition
Purpose: to produce groups that are similar on extraneous variables by chance, thus reducing systematic differences between groups
Practical example in Schachter’s context: assignment to high-anxiety (experimental) vs low-anxiety (control) should be random to balance potential confounds
Important distinction: random assignment is about group allocation within a study, not about selecting a sample from the population (sampling is a separate issue)
The Logic of Experimental Design (causation through control)
Core logic: if two groups are alike on all aspects that could influence the DV except for the IV, then differences in the DV across groups can be attributed to the IV
Consequence: researchers can isolate the effect of the IV on the DV and make a causal inference about the relationship
In Schachter’s study: the hypothesis that anxiety changes need for affiliation can be tested by comparing DV (desire to wait with others) across high vs low anxiety groups
Step-by-step: Schachter’s Procedure and Results (how the design played out)
Orientation and deception (cover story): participants were told about the study on the physiological effects of electric shocks; the level of threat (painful vs painless) was manipulated to induce different anxiety levels
Delay and measurement of DV: after orientation, participants waited for equipment to be prepared and were asked whether they preferred to wait alone or with others; this choice served as the DV
Experimental and control groups defined by anxiety manipulation:
High anxiety (experimental) group: told shocks would be painful
Low anxiety (control) group: told shocks would be mild/painless
Random assignment caveat: the slide notes emphasize random assignment as a safeguard but Schachter’s description focuses on the intentional manipulation of anxiety levels to create two groups
Results (as shown in the graph):
The X-axis shows the experimental vs control groups (IV manipulation)
The Y-axis shows the DV (percent of participants wanting to wait with others)
Finding: increased anxiety led to a higher desire to affiliate; the high-anxiety group waited with others at roughly twice the rate of the low-anxiety group
Conclusion: the results supported the hypothesis that anxiety increases the need for affiliation
The Experimental vs Control Groups: Summary points
Experimental group: receives the manipulation of interest (anxiety-arousing procedure)
Control group: does not receive the special treatment (neutral condition)
The two groups should be similar on non-IV factors to avoid confounding, enabling attribution of DV differences to the IV
Real-world note: exact sameness across all characteristics is unrealistic; what matters is similarity on variables relevant to the DV
Why random assignment matters (revisited)
Random assignment helps ensure that extraneous variables are distributed similarly across groups, reducing systematic bias
Examples of non-random assignment problems: letting participants choose groups or grouping by friendship networks can introduce confounds (e.g., sociability clusters)
The practical takeaway: random assignment strengthens the internal validity of experiments and supports causal conclusions
Strengths of the Experimental Method (causation and control)
Main strength: can establish cause-and-effect relationships between variables
Why: precise control of the IV and randomization help isolate the IV's effect on the DV while neutralizing extraneous variables
The Schachter study illustrates how an experiment can test a theoretical hypothesis about behavior under different emotional states
No other research method duplicates this level of causal inference for controlled manipulation scenarios
Limitations and Ethical/Practical Considerations of the Experimental Method
One major limitation: not all research questions can be ethically or practically tested with experiments
Example given: the effects of severe nutritional neglect during pregnancy on child health are ethically unacceptable to manipulate directly
When experiments are not feasible due to ethics or practicality, researchers turn to descriptive correlational methods
Transition to the Correlational Method (descriptive approach)
When experiments cannot manipulate the variables, researchers rely on correlational designs
Central claim: correlational methods cannot demonstrate cause-and-effect relationships; they reveal whether a relationship or association exists between variables
The key statistic: the correlation coefficient, denoted as
The correlational approach focuses on predicting one variable from another, not on establishing causation
The goal is to identify links that may prompt further, more controlled investigations
What is a Correlation? Understanding the core idea
A correlation exists when two variables are related to one another
Correlations can be positive or negative, depending on the direction of the relationship, but not necessarily indicating strength or causality
Positive correlation: as X increases, Y tends to increase; as X decreases, Y tends to decrease
Negative correlation: as X increases, Y tends to decrease; as X decreases, Y tends to increase
Examples:
Positive: high school GPA and college GPA (generally, higher HS GPA tends to relate to higher college GPA)
Negative: number of absences and exam scores (more absences tend to relate to lower exam scores)
Important distinction: the terms “positive” and “negative” refer to direction, not to whether the relationship is good or bad
The Correlation Coefficient: Magnitude and Direction
The correlation coefficient r measures the strength and direction of the linear relationship between two variables
Range and interpretation:
Sign indicates direction: positive (+) or negative (−)
Magnitude indicates strength: 0 ≤ |r| ≤ 1
Strength categories (rough guidelines):
|r| near 0: weak or no linear relationship
|r| around 0.3–0.5: moderate relationship
|r| around 0.7–0.9: strong relationship
|r| near 1: very strong relationship
Example comparisons:
r = 0.90 is stronger than r = 0.40
r = −0.75 is stronger in magnitude than r = −0.45
Notation tip: the sign indicates direction; the magnitude indicates strength
Formula (conceptual):
where cov is the covariance and σ denotes standard deviationPractical note: r ranges from −1 to +1; r = 0 indicates no linear relationship
Positive vs Negative Correlations: Examples and implications
Positive example: high school GPA and college GPA (as X ↑, Y ↑ on average)
Negative example: class absences and exam scores (as X ↑, Y ↓ on average)
The direction tells us how the variables move together, not whether one causes the other
Correlation and Prediction: What correlations enable
Prediction as a key utility of correlations: stronger correlations improve predictive accuracy
Example in education: college admission test scores (ACT/SAT) vs college GPA
Typical correlations are moderate (roughly )
These correlations allow admission committees to predict college performance with modest accuracy
Higher correlations would yield stronger predictive power; lower correlations weaken predictive usefulness
Important caveat: even strong correlations do not establish causation
Why Correlation Does Not Imply Causation: The Third Variable Problem
Core problem: a high correlation does not prove that one variable causes the other
Two main issues:
Directionality problem: with a correlation, it is unclear which variable influences the other (X → Y, Y → X, or bidirectional)
Third-variable problem: a third variable Z could cause both X and Y, creating a spurious association
Real-world illustration: social activity (X) and happiness (Y) show a positive correlation, but it is unclear whether social activity causes happiness or vice versa, or if both are influenced by a third variable like extroversion
The third-variable example in notes: extroversion could drive both social activity and happiness, producing a correlation without a direct causal link between X and Y
Takeaway: correlation alone cannot establish causation; experimental manipulation is required to test causal claims
Advantages and Disadvantages of Correlational Research
Advantages:
Allows study of questions that cannot be ethically or practically manipulated
Broadens the scope of phenomena psychologists can examine (e.g., nutrition and child health, urban vs. rural upbringing and values)
Disadvantages:
Cannot conclusively demonstrate causation
Associations can be influenced by third variables; direction of causality remains uncertain without experimental evidence
Summary: correlational research is valuable for exploring relationships and making predictions, but it cannot replace experiments when causal conclusions are required
Summary and Look Ahead
Chapter 2 Part 2 covered:
Experimental method basics, IV/DV concepts, control of extraneous variables, and the role of random assignment
The Schachter (need for affiliation) study as a concrete illustration of manipulation, group design, and DV measurement
Strengths and limitations of experiments (causation vs ethics): why some questions require correlational methods
Introduction to correlational research: what correlations mean, how to interpret r, and why correlation does not imply causation
Third-variable problem and the predictive value of correlations, with real-world examples (SAT/college GPA, absences vs exam scores)
Next topic: correlational method in more detail, including how to measure and interpret correlation strength, prediction, and the limitations in establishing causation
Note: Additional resources and a short video (Schachter’s experiment) were recommended to reinforce understanding
Helpful formulas and definitions (quick reference)
Independent variable (IV) and dependent variable (DV) relation:
Treat X as IV and Y as DV; the experiment tests whether changes in X cause changes in Y
Hypothesis form (conceptual):
If X increases, then Y increases
Prediction relevance: stronger |r| implies better predictive accuracy, but does not prove causation
Key terms to remember:
Extraneous variable: any variable other than IV that could influence DV
Confounding variable: an extraneous variable entangled with IV, complicating causal inference
Random assignment: equal probability of assignment to each group to balance extraneous variables
Third-variable problem: a third variable causing both X and Y, producing a spurious correlation