Difficulty: Introductory | Prerequisites: Introduction to Social Psychology notes
Before studying the content of social psychology, you need to understand how the evidence is gathered. This section covers the scientific method as applied to social behaviour: how researchers design studies, measure variables, rule out alternative explanations and draw valid conclusions. It also addresses common pitfalls in everyday reasoning (like hindsight bias) that motivate the need for rigorous methods in the first place.
Social psychologists use descriptive, correlational and experimental methods to study behaviour. Only experiments (with random assignment and manipulation of an independent variable) can establish causation. Understanding confounds, moderators, reliability and validity is essential for evaluating any study's claims.
Hindsight bias
The tendency to believe, after learning an outcome, that one would have predicted it all along. In simple terms, it is the "I knew it all along" effect. It makes research findings feel obvious even when they were not predictable beforehand.
Scientific method
A systematic process of observation, hypothesis formation, experimentation and analysis used to test ideas about the world. Think of it as a structured way of asking questions and testing answers so that personal biases are minimised.
Hypothesis
A specific, testable prediction about the relationship between two or more variables. In simple terms, it is an educated guess that can be proven wrong.
Research question
A broader, open-ended inquiry that guides a study but does not make a specific prediction. Think of it as the question a hypothesis is trying to answer.
Operationalisation
The process of defining an abstract concept in terms of specific, measurable procedures or indicators. In simple terms, it is how you turn something vague (like "aggression") into something you can count or measure (like the number of times a participant presses a button to deliver a loud noise).
Independent variable (IV)
The variable that the researcher manipulates or varies in an experiment. Think of it as the "cause" side of the cause-and-effect relationship the experiment is testing.
Dependent variable (DV)
The variable that is measured to see if it is affected by changes in the independent variable. In simple terms, it is the outcome, the thing you are watching to see if it changes.
Confound
An extraneous variable that varies systematically with the independent variable, making it impossible to determine which one caused the observed effect. Think of it as a hidden "third thing" that muddies the results. If participants in your treatment group are all tested in the morning and the control group in the evening, time of day is a confound.
Moderator
A variable that affects the strength or direction of the relationship between two other variables. In simple terms, a moderator answers the question "for whom?" or "under what conditions?" For example, gender might moderate the relationship between stress and coping style, meaning the relationship looks different for men and women.
Reliability
The degree to which a measure produces consistent, repeatable results. Think of it as: if you step on a scale three times in a row, does it give you the same number each time?
Internal validity
The extent to which a study can establish a causal relationship between the independent and dependent variables, free from confounds. In simple terms, it is confidence that the IV, and not something else, caused the change in the DV.
External validity
The extent to which the findings of a study can be generalised to other settings, populations and times. Think of it as: do these lab results hold up in the real world?
Implicit Association Test (IAT)
A computer-based measure that assesses the strength of automatic associations between concepts (e.g., race and valence) by measuring reaction times. In simple terms, it tests biases you might not be aware of by seeing how quickly you pair certain ideas together.
Descriptive methods observe and record behaviour without manipulating anything. They answer "what is happening?" but cannot explain why.
Naturalistic observation: Watching behaviour in its natural setting without interference. Strength: high ecological validity. Limitation: no control over extraneous variables.
Archival research: Analysing existing records (newspapers, databases, public documents). Strength: can study phenomena across long time periods. Limitation: the data were collected for another purpose and may not measure exactly what you need.
Survey research: Asking people to report their own attitudes, beliefs or behaviours via questionnaires or interviews. Strength: can reach large samples efficiently. Limitation: self-report bias, including social desirability (people reporting what sounds good rather than what is true).
Measures the relationship between two variables without manipulating either one
Produces a correlation coefficient (ranging from -1.0 to +1.0) that indicates the strength and direction of the relationship
Critical limitation: correlation does not imply causation. A correlation between ice cream sales and crime rates does not mean ice cream causes crime; both may be driven by a third variable (hot weather)
Useful when experiments would be unethical or impractical (e.g., you cannot randomly assign people to smoke for 20 years)
The only method that can establish a cause-and-effect relationship
Requires:
Manipulation of an independent variable (the researcher changes something)
Random assignment of participants to conditions (so that pre-existing differences are distributed evenly)
Measurement of the dependent variable
Control of extraneous variables to rule out confounds
If the groups differ on the DV, and the only systematic difference between groups was the IV, then the IV caused the effect
A confound is any variable, other than the IV, that differs between conditions and could explain the results
Example: if the experimental group is tested in a quiet room and the control group in a noisy room, room noise is a confound
Random assignment is the primary tool for eliminating confounds
A moderator changes the relationship between two variables
Recognising a moderator in a study: look for an interaction effect, where the effect of the IV on the DV depends on the level of a third variable
Example: a study finds that persuasion increases with argument quality, but only for people who are paying close attention. Attention level is the moderator
Note: You will not be tested on mediators
Be able to read a bar graph and describe what pattern of results it shows
Be able to work in reverse: given a written description of results, sketch or identify the correct bar graph
Key things to look for: which condition is higher, whether there is an interaction (do the bars show different patterns across groups?), and what the axes represent
Self-report scales: Questionnaires asking people to rate their attitudes or feelings (e.g., Likert scales)
Physiological measures: Heart rate, skin conductance, cortisol levels, as indices of stress, arousal or emotion
The IAT: Measures implicit (automatic) associations by comparing reaction times when pairing concepts
These techniques differ in what they can capture: self-report taps conscious attitudes; physiological measures and the IAT can access processes people may not be aware of or willing to report
Understanding research methods is not just an academic exercise. Being able to tell whether a study supports a causal claim or merely a correlational one is critical for evaluating health advice, policy arguments and media headlines. If a news article says "eating chocolate linked to weight loss," knowing the difference between correlation and causation prevents you from being misled.
"If a study has a large sample, it must be valid." A large sample improves statistical power but does not fix problems with internal validity. A confounded experiment with 10,000 participants is still confounded.
"Correlational studies are useless because they cannot prove causation." Correlational research is valuable for identifying relationships, studying variables that cannot be ethically manipulated and generating hypotheses for experimental testing.
"A moderator is the same as a confound." A confound is an unwanted variable that threatens internal validity. A moderator is a variable the researcher is interested in, one that reveals for whom or under what conditions an effect holds.
"Reliability and validity are the same thing." A measure can be reliable (consistent) without being valid (measuring what it claims to). A broken clock that always reads 3:00 is perfectly reliable but not valid.
Know the definition of a confound and be able to identify one in a study scenario
Know the definition of a moderator and be able to recognise one in a study description
Be able to distinguish between descriptive, correlational and experimental research, and know which can establish causation
Be comfortable interpreting bar graphs and translating written results into a graph pattern
Understand the difference between reliability, internal validity and external validity
Mediators will not be tested
True or false: Correlational research can establish that one variable causes another. (False: only experimental research can establish causation.)
Fill in the blank: The variable the researcher manipulates is the ____, and the variable measured is the ____. (Independent variable; dependent variable.)
True or false: Random assignment helps eliminate confounds. (True.)
Fill in the blank: A measure that produces consistent results is said to have high ____. (Reliability.)
True or false: A moderator is the same as a confound. (False: a moderator is a variable of interest that changes the strength or direction of a relationship.)
Q: A researcher finds that people who exercise more report lower levels of anxiety. Can the researcher conclude that exercise reduces anxiety? Why or why not?
A: No. This is a correlational finding. It could be that exercise reduces anxiety, that lower anxiety leads to more exercise, or that a third variable (such as overall health or free time) drives both. Only an experiment with random assignment could establish causation.
Q: In a study, participants in the experimental group receive a new therapy in a bright, comfortable room, while control participants receive no therapy in a dimly lit, cramped room. Identify the confound.
A: Room environment is a confound. It varies systematically with the IV (therapy vs. no therapy), so any difference in outcomes could be due to the therapy, the room or both.
Q: A study finds that a motivational speech improves exam performance, but only among students who already had moderate self-esteem. What role does self-esteem play in this study?
A: Self-esteem is a moderator. It changes the relationship between the IV (motivational speech) and the DV (exam performance): the speech works for moderate-self-esteem students but not for others.
Q: What is hindsight bias, and why does it make scientific research necessary?
A: Hindsight bias is the tendency to believe, after learning an outcome, that you would have predicted it. It makes scientific research necessary because it gives people false confidence in their intuitions, leading them to believe that studying behaviour systematically is unnecessary when it is not.
Research methods underpin every other section. The studies discussed in the Introduction (Milgram, Stanford Prison Experiment, Bystander Effect) are all examples of experimental research. The IAT returns in the Social Cognition and Social Attribution sections. Graph interpretation skills are needed throughout.
research methods social psychology, hindsight bias, I knew it all along, scientific method, hypothesis vs research question, operationalisation, independent variable, dependent variable, IV DV, descriptive research, naturalistic observation, archival research, survey research, correlational research, correlation does not equal causation, experimental research, random assignment, confound, confounding variable, moderator, interaction effect, reliability, internal validity, external validity, generalisability, IAT, implicit association test, self-report, physiological measures, bar graph interpretation