Difficulty: Introductory | Prerequisites: Part 1 of these notes
Source: Social Psychology: The Science of Everyday Life, 3e, Greenberg et al. | Chapter 1
Tags: scientific method, theory, hypothesis, correlational method, experimental method, quasi-experiment, longitudinal design, field research, independent variable, dependent variable, correlation coefficient, random assignment
This section covers how social psychologists do their work: the cycle of theory and research, the major research designs (correlational, experimental, quasi-experimental, longitudinal, field), and the key vocabulary of variables, hypotheses, and measurement. If Part 1 was the "what" and "why" of social psychology, this is the "how." Understanding these methods is essential because exam questions will ask you to identify which design a given study uses and what conclusions it can support.
Social psychologists build theories to explain social behaviour, derive testable hypotheses from those theories, and then test them through research. The main designs are correlational (measuring associations), experimental (manipulating a variable to test causation), quasi-experimental (comparing naturally occurring groups), longitudinal (tracking people over time), and field research (observing behaviour in real-world settings). Only true experiments with random assignment can establish causation. Correlation does not imply causation, because of the reverse causality and third variable problems.
Theory
An explanation for how and why variables are related to each other. A good theory organises facts, generates hypotheses, guides research, raises new questions, and has practical applications.
Hypothesis
An if-then statement that specifies what the relationships between variables should look like if a theory is correct. In simple terms, it is a testable prediction derived from a theory.
Fact (in science)
The content of research observations that have been repeatedly found by many researchers. Facts are the raw material; theories explain them.
The cycle of theory and research
Theory produces hypotheses, hypotheses are tested through research, research outcomes feed back into revised theory. The correct sequence is: Theory, then Hypothesis, then Research Outcome, then Revised Theory.
Correlational method
A research approach in which two or more variables are measured and compared to determine the extent to which they are associated. No variables are manipulated.
Correlation coefficient
A positive or negative value (ranging from -1 to +1) that shows the strength and direction of the association between two variables. A negative correlation means higher scores on one variable are associated with lower scores on the other.
Experimental method
A study in which a researcher takes active control of an independent variable, manipulates it, and observes its effect on a dependent variable. Random assignment is essential.
Independent variable (IV)
The variable the researcher manipulates or assigns participants to. When you randomly assign participants to different conditions, you are creating an independent variable.
Dependent variable (DV)
The variable that is measured. In simple terms, it is what you think will be affected by the independent variable.
Random assignment
A procedure through which participants are assigned to experimental conditions in such a way that they have an equal chance of being in any condition. This controls for individual differences and the third variable problem.
Quasi-experiment
A study in which naturally occurring groups of participants are compared on a dependent variable without random assignment (e.g. comparing Native Americans and White people on environmental attitudes).
Longitudinal design
A research approach that follows the same group of people over an extended period, measuring variables at multiple time points. Best suited for tracking developmental change.
Field research
Research conducted in real-world settings rather than the laboratory. Its chief weakness is that researchers lose control over variables.
Interaction
A pattern of results in which the effect of one independent variable on a dependent variable depends on the level of another independent variable. The stereotype threat finding (race priming hurts Black students' performance but not White students') is a classic interaction.
Meta-analysis
A statistical technique that analyses a large number of previously conducted studies to understand the overall pattern of findings on a given topic.
Power analysis
Used to determine the appropriate sample size needed for a study to detect an effect if one exists.
Parsimony (parsimonious theory)
A theory that explains a large number of diverse observations with relatively few and simple principles. In simple terms, simpler is better, all else being equal.
Theory produces hypotheses. Hypotheses are tested through research. Research outcomes lead to revised theory.
Correct sequence: Theory → Hypothesis → Research Outcome → Revised Theory.
Hypotheses are derived from theories, in order to test the accuracy of the latter.
A good theory should: (1) organise disparate facts, (2) be logically consistent, (3) provide direction for research, (4) generate new questions, and (5) have practical applications.
A theory does NOT need to show how correlation explains causation.
Even if current methods cannot fully test a theory's hypotheses, the theory may still ultimately prove accurate.
Two or more variables are measured and compared. No manipulation occurs.
Produces a correlation coefficient (r), ranging from -1.0 to +1.0.
A positive r means both variables move in the same direction; a negative r means they move in opposite directions; r near 0 means little or no relationship.
If r = +1.0 for two psychological variables, it probably means they are measuring the same underlying construct.
Best suited for questions about whether two things are related (e.g. is self-esteem related to healthy eating?) and for studying personality or dispositional variables.
Limitation: correlation does not imply causation, because of two problems.
Reverse causality: when a correlation is found, it may be impossible to determine which variable causes the other. A correlation between self-esteem and academic performance does not tell you which came first.
Third variable problem: a correlation between two variables may be driven by their mutual association with another, unmeasured variable. Swimming pools and heatstroke deaths are correlated because both are driven by high temperatures (the third variable).
The researcher manipulates the independent variable, randomly assigns participants to conditions, and measures the dependent variable.
The experimental method can make researchers more confident about likely causal order.
Three conditions for a true experiment: (1) the IV is manipulated, (2) participants are randomly assigned, (3) everything else is held constant.
Experiments help establish causation because: (a) the researcher controls temporal sequence, (b) participants are randomly assigned, (c) everything else is held constant. Note: "greater external validity" is NOT a reason experiments establish causality.
Experimental methods are preferable for determining causal relationships; correlational methods are preferable for studying personality or dispositional variables.
Naturally occurring groups are compared without random assignment.
Example: comparing Native Americans and White people on environmental attitudes. You cannot randomly assign someone to an ethnicity.
Weaker causal inference than a true experiment because of the lack of random assignment.
The same group is tracked over an extended period.
Best suited for studying developmental change (e.g. how religious beliefs develop from adolescence through early adulthood, or whether childhood video-game play predicts adult aggression).
Research conducted in natural, real-world settings.
Example: a researcher counting how many people look both ways before crossing a street.
Advantage: captures social behaviour as it naturally occurs.
Chief weakness: the researchers lose control over the variables.
Best suited for questions about naturally occurring behaviour where laboratory control is not feasible (e.g. do people with expensive cars stop less at stop signs?).
"Correlation means there is no causal relationship." Wrong. Correlation may indicate causation, but further research (typically an experiment) is needed to make that claim.
"If a hypothesis is disproven, the whole theory must be discarded." Wrong. One failed hypothesis leads to theory revision, not necessarily rejection. A theory should generate multiple hypotheses.
"A quasi-experiment is the same as a true experiment." Wrong. A quasi-experiment lacks random assignment, which limits causal inference.
"Longitudinal studies can establish causation." Not on their own. They show temporal order (which helps), but without random assignment, third variables remain a problem.
Be able to identify the research design from a scenario description. The exam will describe a study and ask you to name it (correlational, experimental, quasi-experimental, longitudinal, field).
Know the correct sequence of the theory-research cycle: Theory → Hypothesis → Research Outcome → Revised Theory.
"Correlation does not imply causation" will be tested from multiple angles. Know the reverse causality problem and the third variable problem by name and be able to apply them to novel examples.
Know what random assignment does and does not solve. It addresses the third variable problem and pre-existing group differences. It does NOT solve the causal sequence problem (the experimenter controls that by manipulating the IV first).
Understand the difference between an independent variable and a dependent variable, and be able to identify each in a described study.
Know the five characteristics of a good theory and that "showing how correlation explains causation" is NOT one of them.
True or False: The correlational method allows researchers to draw causal conclusions. Answer: False. Only the experimental method supports causal conclusions.
Fill in the blank: The correct sequence of the theory-research cycle is Theory → _____ → Research Outcome → Revised Theory. Answer: Hypothesis.
True or False: In a quasi-experiment, participants are randomly assigned to conditions. Answer: False. Quasi-experiments compare naturally occurring groups without random assignment.
Fill in the blank: A correlation coefficient of -0.75 indicates a _____ (strong/weak), _____ (positive/negative) relationship. Answer: strong, negative.
True or False: Field research gives researchers strong control over variables. Answer: False. The chief weakness of field research is loss of control over variables.
Q: Researchers follow 200 people from ages 9 to 25, measuring violent video-game play and aggression each year. What type of study is this?
A: A longitudinal design.
Q: A researcher randomly assigns people to read about a natural disaster or a control topic, then measures their travel intentions. What type of study is this?
A: An experimental design.
Q: A researcher finds a positive correlation between self-esteem and academic performance. What can she conclude about causation?
A: Nothing. It is impossible to conclude anything about the causal relationship from correlational data alone.
Q: What is the difference between a correlational hypothesis and an experimental/causal hypothesis?
A: A correlational hypothesis predicts an association between variables ("the more neurotic a person is, the more negative emotions they will experience"). An experimental hypothesis predicts that manipulating one variable will cause a change in another.
Q: Taylor analyses a large number of previously conducted studies on musical and mathematical ability. What is this called?
A: Meta-analysis.
Q: Which research method is best for determining whether one variable has a causal influence on another?
A: An experiment.
The research methods covered here are the tools you will see applied in every subsequent chapter. When you read about a study on conformity, aggression, or attraction, you will be expected to identify the design and evaluate its conclusions accordingly. The distinction between correlation and causation is one of the most frequently tested concepts across all of introductory psychology, not just social psychology. Meta-analysis will come up again when the course discusses replication and the reliability of findings across the field.
Scientific method psychology, theory vs hypothesis, correlational method examples, correlation coefficient definition, experimental method social psychology, independent variable vs dependent variable, random assignment definition, quasi-experiment examples, longitudinal design psychology, field research social psychology, reverse causality problem, third variable problem, meta-analysis definition, power analysis sample size, parsimony theory, interaction effect psychology, Greenberg social psychology 3e Chapter 1