Research Methods in Psychology

Learning Objectives

• Explain why psychology relies on systematic research methods rather than anecdote or intuition
• Distinguish experimental, correlational, and descriptive research designs
• Identify independent variables, dependent variables, and confounds in a study
• Interpret a correlation coefficient and explain why correlation does not imply causation
• Describe the steps of the psychological research process from question to publication
• Evaluate the strengths and weaknesses of quantitative and qualitative approaches
• Apply ethical principles that govern how psychological research is conducted

Quick Answer

Research methods are the toolkit psychologists use to turn a question about behavior into trustworthy evidence. The experimental method, where a researcher manipulates one variable and randomly assigns participants to conditions, is the only method that can establish cause and effect. Correlational research measures how two naturally occurring variables relate without manipulating anything, which is useful for describing relationships but cannot prove causation. Descriptive methods — surveys, case studies, and naturalistic observation — capture rich detail about behavior as it occurs but offer weaker control. Researchers choose a method based on their goal (causation vs. description), practical feasibility, and ethical constraints, and they typically follow a structured process: forming a question, reviewing existing literature, generating a testable hypothesis, collecting data, analyzing it statistically, and reporting the results for others to evaluate and replicate.

Overview

Anyone can have an opinion about why people behave the way they do. What separates psychology from casual opinion is a shared commitment to testing those opinions systematically, so that a claim survives not because it sounds convincing, but because it holds up against evidence collected in a way that other researchers could repeat and check.

This chapter walks through the main categories of research method, the logic of experimental design, how psychologists analyze the data they collect, and the ethical guardrails that shape what research can and cannot be done. The single most important idea in the whole chapter is the distinction between correlation and causation — a distinction that shows up constantly in both exam questions and everyday misleading headlines.

The Experimental Method

Definition

The experimental method is a research design in which a researcher deliberately manipulates one variable (the independent variable) and measures its effect on another variable (the dependent variable), while controlling other factors, in order to establish cause and effect.

Explanation

The power of the experiment comes from two design features working together: manipulation and random assignment. The researcher assigns participants to conditions (e.g., a treatment group and a control group) purely by chance, which spreads out any pre-existing differences between people (age, personality, mood) roughly evenly across groups. Because the groups start out statistically similar on everything except the variable being manipulated, any difference in outcome can be attributed to that manipulation rather than to who happened to end up in each group.

Example

To test whether a new study app improves exam scores, a researcher randomly assigns half of a class to use the app (the independent variable) and the other half to study as usual, then compares exam scores (the dependent variable) between the two groups.

Real-World Example

Randomized controlled trials, the gold standard for testing new medications and psychological treatments, are a direct application of the experimental method — patients are randomly assigned to receive the real treatment or a placebo, and the difference in outcomes tells researchers whether the treatment itself is responsible for any improvement.

Why It Matters

The experimental method is the only research design that can establish causation with confidence, which makes it indispensable whenever a real decision depends on knowing whether something actually works — from evaluating a new therapy to testing whether a teaching method genuinely improves learning.

Common Misunderstanding

Students often think any study that compares two groups is automatically an "experiment." A comparison only counts as a true experiment if the researcher actively manipulates the independent variable and randomly assigns participants to conditions. Comparing pre-existing groups (e.g., "smokers vs. non-smokers") is not a true experiment, because participants were not randomly assigned to smoke or not smoke — it's a quasi-experimental or correlational design instead.

The Correlational Method

Definition

The correlational method examines the statistical relationship between two variables as they naturally occur, without the researcher manipulating either one.

Explanation

Correlational studies produce a correlation coefficient (r), ranging from -1.0 to +1.0, that describes both the direction and strength of a relationship. A positive correlation means both variables tend to increase together; a negative correlation means one increases as the other decreases. A coefficient near 0 indicates little to no linear relationship. Crucially, a correlation — no matter how strong — cannot tell you which variable causes the other, or whether a third, unmeasured variable is responsible for both.

Example

A researcher finds that hours spent studying and exam scores have a correlation coefficient of r = 0.65, a moderately strong positive relationship, meaning students who study more tend to score higher, on average, though this alone doesn't prove studying causes the higher scores.

Real-World Example

Large-scale public health research often relies on correlational data (since randomly assigning people to smoke, for instance, would be unethical), which is why researchers look for converging evidence — consistent correlations across many different studies, controlling for known confounds — before treating a link as likely causal.

Why It Matters

Correlational research lets psychologists study relationships that cannot ethically or practically be manipulated (such as the link between childhood trauma and adult mental health), and it is often the first step that generates hypotheses later tested through experiments.

Common Misunderstanding

The most common and consequential mistake in all of research methods is treating a correlation as proof of causation. Ice cream sales and drowning deaths rise together every summer, not because ice cream causes drowning, but because a third variable — hot weather — increases both. Any correlational finding should be checked for plausible confounds before assuming a causal relationship, and directionality is also unclear: does A cause B, does B cause A, or does a third variable cause both?

Descriptive Methods

Definition

Descriptive methods — including surveys, case studies, and naturalistic observation — describe behavior as it naturally occurs without manipulating variables or necessarily establishing statistical relationships.

Explanation

Surveys use questionnaires or interviews to gather self-reported data from many people relatively quickly and cheaply, but they rely on honest and accurate self-report, which can be distorted by social desirability bias. Case studies provide extremely detailed, in-depth information about a single person or small group, often used for rare conditions, but their findings cannot be generalized broadly. Naturalistic observation involves watching behavior unfold in its natural environment without any interference, which produces realistic data but sacrifices control over confounding variables.

Example

A researcher interested in coping strategies among survivors of a rare medical condition might use a case study, since only a small number of people are available to study and the depth of individual detail matters more than sample size.

Real-World Example

Jean Piaget's foundational theory of cognitive development in children was built substantially on careful naturalistic observation and case studies of a small number of children (including his own), illustrating how descriptive methods can generate highly influential theories even without large samples or manipulated variables.

Why It Matters

Descriptive methods let researchers study behavior in situations where experiments would be unethical, impractical, or would strip away the natural context that makes the behavior meaningful in the first place — you cannot ethically manipulate whether a child experiences abuse, but you can carefully and ethically observe or interview children who have already experienced it.

Common Misunderstanding

Students sometimes assume descriptive methods are simply a "weaker" or lower-quality version of experiments. They serve a fundamentally different purpose: they answer "what is happening and in what detail," while experiments answer "what caused this to happen." A case study of a rare phenomenon can be far more scientifically valuable than a poorly designed experiment on a common one.

The Research Process

Definition

The research process is the structured sequence of steps psychologists follow to move from an initial question to a published, peer-reviewed conclusion.

Explanation

The process typically runs: (1) identify a research question, (2) conduct a literature review of existing findings, (3) formulate a testable hypothesis, (4) choose an appropriate research design, (5) collect data, (6) analyze the data statistically, (7) draw conclusions about whether the hypothesis was supported, and (8) report results, usually through peer-reviewed publication, so other researchers can evaluate and attempt to replicate the findings. Replication — other researchers independently repeating a study and getting similar results — is what ultimately builds confidence in a finding, since any single study could be a fluke or contain a hidden flaw.

Example

A researcher who notices students seem more anxious before online exams than in-person exams might review existing test-anxiety literature, hypothesize that reduced perceived control increases anxiety, design an experiment manipulating perceived control, collect and analyze anxiety ratings, and publish the results for peer review.

Real-World Example

The "replication crisis" in psychology, where many well-known findings failed to replicate when other labs repeated the original studies, led to major reforms in the field, including pre-registration of hypotheses (stating predictions before data collection) and larger, more rigorous sample sizes.

Why It Matters

Following this structured process, and subjecting it to peer review and replication, is what separates a scientific claim from an untested opinion — it creates a public, checkable trail of evidence rather than asking people to simply trust a researcher's word.

Common Misunderstanding

A common misunderstanding is treating a single published study as settled, permanent truth. Science is cumulative and self-correcting: a single study, even a well-designed one, is one data point. Confidence in a finding grows as independent replications, meta-analyses (statistical combinations of many studies), and converging evidence from different methods accumulate over time.

Statistical Analysis in Psychology

Definition

Statistical analysis is the set of mathematical tools psychologists use to summarize data (descriptive statistics) and to determine whether observed patterns likely reflect a real effect or could be due to chance (inferential statistics).

Explanation

Descriptive statistics — the mean, median, mode, and standard deviation — summarize the basic shape of a data set. Inferential statistics, such as significance testing, help researchers decide whether a difference found between groups (e.g., a treatment group and a control group) is large and consistent enough that it is unlikely to have occurred by chance alone, typically using a threshold like p < .05, meaning less than a 5% probability the result occurred purely by chance if there were truly no effect.

Example

A study comparing two teaching methods might find Method A produced a mean score of 85 and Method B a mean score of 78; inferential statistics would then test whether this 7-point gap is statistically significant or could plausibly be due to random variation between the two groups.

Real-World Example

Meta-analyses, which statistically combine the results of many individual studies on the same question, are increasingly used in psychology (and medicine) to produce more reliable overall conclusions than any single study could offer on its own.

Why It Matters

Statistics let researchers distinguish a genuine, reliable pattern in behavior from random noise, which is essential given that human behavior is naturally variable — without statistical testing, researchers could easily mistake a chance fluctuation for a real psychological effect.

Common Misunderstanding

Many students think statistical significance (p < .05) means a finding is important or large. Statistical significance only indicates that a result is unlikely to be due to chance; it says nothing about the size or practical importance of the effect. A study with a very large sample size can find a statistically significant but practically tiny and meaningless difference between groups.

Key Terms

Term	Definition	Related Concept
Independent variable (IV)	The variable the researcher deliberately manipulates	Experimental method
Dependent variable (DV)	The outcome variable that is measured	Experimental method
Random assignment	Assigning participants to conditions purely by chance	Experimental control, confounds
Confound	An uncontrolled variable that could explain the results instead of the IV	Internal validity
Correlation coefficient (r)	A number from -1 to +1 describing the strength and direction of a relationship	Correlational method
Case study	An in-depth investigation of one person or a small group	Descriptive method
Naturalistic observation	Watching behavior in its real-world setting without interference	Descriptive method
Hypothesis	A specific, testable prediction derived from a theory	Research process
Peer review	Evaluation of research by other experts before publication	Scientific validity
Replication	Repeating a study to see if the same results occur again	Reliability of findings
Statistical significance	The likelihood that a result is not simply due to chance	Inferential statistics, p-value
Informed consent	Ethical requirement that participants understand and agree to a study before it begins	Research ethics

Common Mistakes

Misconception: A correlational study that finds a strong relationship (e.g., r = 0.85) is basically as good as an experiment for proving causation. Why it's wrong: No matter how strong a correlation is, it cannot rule out reverse causation or a third confounding variable. A very strong correlation is more surprising and worth investigating further, but it still doesn't establish which variable, if either, is causing the other. Correct understanding: Only a true experiment, with manipulation of the independent variable and random assignment, can support a causal claim. Strong correlations are a reason to design a follow-up experiment, not a substitute for one.

---

Misconception: A larger sample size always makes a study's conclusions more trustworthy. Why it's wrong: Sample size affects statistical power (the ability to detect a real effect) and reduces the influence of random chance, but it does nothing to fix a flawed design, biased sampling, or a confounded variable. A huge but biased sample can be far less trustworthy than a smaller, well-controlled, randomly selected one. Correct understanding: Trustworthiness depends on design quality (control of confounds, random assignment, representative sampling) at least as much as on sample size; a bigger flawed study just produces a more confidently wrong answer.

---

Misconception: If a finding is published in a peer-reviewed journal, it must be true and doesn't need to be questioned. Why it's wrong: Peer review checks that a study meets basic methodological standards before publication, but it does not guarantee the finding is correct — the replication crisis revealed that many peer-reviewed, published findings failed to replicate when tested again. Correct understanding: Confidence in a finding should grow with independent replication and converging evidence from multiple studies and methods, not from the fact of publication alone.

Comparison and Connections

Feature	Experimental	Correlational	Case Study	Naturalistic Observation
Manipulates a variable	Yes	No	No	No
Can establish causation	Yes	No	No	No
Random assignment	Yes	No	No	No
Typical sample size	Small to moderate	Large	One or a few	Small to moderate
Realism of setting	Often lower (controlled lab)	Can be high	Very high	Very high
Best used for	Testing causal hypotheses	Identifying relationships across a population	Rare or unique phenomena	Behavior in its natural context

Practice Questions

Recall

1. Define independent variable and dependent variable, using an original example. Guidance: IV = the variable the researcher manipulates; DV = the outcome measured. Example: testing whether sleep duration (IV) affects memory test performance (DV).

2. What does a correlation coefficient of r = -0.72 indicate? Guidance: A strong negative relationship — as one variable increases, the other tends to decrease, and the relationship is fairly strong (close to -1), but this does not indicate causation.

Understanding

3. Explain why random assignment is essential for a study to support causal conclusions. Guidance: Random assignment spreads pre-existing differences between participants evenly across groups by chance, so that any difference in outcome can be attributed to the manipulated variable rather than pre-existing group differences.

4. Why can't researchers use a true experiment to study the effects of childhood abuse on adult mental health? Guidance: It would be unethical to manipulate whether a child experiences abuse. Researchers instead rely on correlational and longitudinal designs, which cannot prove causation as directly but can identify strong, consistent associations.

Application

5. A wellness company claims their new app "causes" reduced anxiety because users who use the app report lower anxiety than non-users. Identify the flaw in this claim and describe a better study design. Guidance: This is likely a correlational comparison of pre-existing groups (self-selected users vs. non-users), not a true experiment — confounds like motivation to reduce anxiety could explain both app use and lower anxiety. A better design would randomly assign participants to use the app or not and compare anxiety changes over time.

6. Design a study to test whether a new teaching method improves reading comprehension in third-graders, specifying the IV, DV, and how you would use random assignment. Guidance: IV = teaching method (new vs. traditional); DV = reading comprehension test scores; randomly assign classrooms or students to each teaching method to control for pre-existing differences in ability.

Analysis

7. A news headline states, "Study finds coffee drinkers live longer." Analyze this claim using what you know about correlational research and confounds. Guidance: This is very likely correlational; potential confounds include socioeconomic status, overall health-consciousness, or social habits associated with coffee drinking. Discuss why an experiment manipulating coffee consumption over a lifetime would be impractical, so researchers must rely on careful control of confounds in correlational designs instead.

8. Compare the trade-offs a researcher faces choosing between a large-scale survey and an in-depth case study to study the psychological effects of a natural disaster. Guidance: A survey reaches many affected people, allowing broader generalizations, but sacrifices depth and relies on self-report accuracy. A case study provides very rich, detailed understanding of a few individuals' experiences but cannot be generalized to the wider affected population. The choice depends on whether the research goal is breadth or depth.

FAQ

Why can't we just use experiments for every psychological question? Many important questions in psychology cannot ethically or practically be tested through experiments. You cannot randomly assign people to experience trauma, poverty, or a particular parenting style to see the effect. In these cases, researchers rely on correlational and longitudinal designs, and try to strengthen their conclusions by controlling for known confounds statistically and looking for consistent patterns across many studies.

What exactly does "statistically significant" mean, in plain language? It means the result observed in the study is unlikely to have happened purely by chance, based on a pre-set probability threshold (commonly 5%). It does not mean the effect is large, important, or guaranteed to be real — a statistically significant result could still be a false positive, and a tiny, unimportant effect can still be statistically significant with a large enough sample.

What was the "replication crisis" and why does it matter for how I read psychology research? Starting in the early 2010s, large-scale efforts to repeat well-known psychology studies found that a substantial number of famous findings did not replicate — meaning the original effect wasn't found again under the same conditions. This led the field to adopt reforms like pre-registering hypotheses before collecting data and favoring larger sample sizes. Practically, it means students should treat any single study with appropriate caution and look for whether a finding has been replicated before treating it as established fact.

Is qualitative research (interviews, focus groups) considered less scientific than quantitative research? Not less scientific, just different in purpose and method. Quantitative research (numbers, statistics) is well suited to testing precise hypotheses and measuring effect sizes across many people. Qualitative research is well suited to exploring the meaning, richness, and lived experience behind a phenomenon, especially in early or exploratory stages of research. Many strong research programs use a mixed-methods approach, combining both.

How do researchers handle a study where deception is necessary to get valid results? Deception is only permitted under strict ethical guidelines: it must be scientifically justified (no equally effective non-deceptive method exists), it cannot expose participants to significant risk, and participants must be fully debriefed afterward — told the true nature of the study and given a chance to ask questions or withdraw their data. This is one of the ethical trade-offs explored more fully in the ethics chapter.

Quick Revision

• The experimental method is the only design that can establish causation, through manipulation of an IV and random assignment
• Correlational research measures naturally occurring relationships between variables but cannot establish causation
• A correlation coefficient (r) ranges from -1 to +1, indicating direction and strength, never causation
• Descriptive methods (case studies, surveys, naturalistic observation) capture rich, realistic detail but offer less control over confounds
• Case studies are ideal for rare or unique phenomena; surveys are ideal for large-scale, quick data collection
• The research process runs: question, literature review, hypothesis, design, data collection, analysis, conclusions, publication
• Replication — repeating a study and getting similar results — is what builds real confidence in a finding, not a single publication
• Statistical significance (commonly p < .05) indicates a result is unlikely due to chance, but says nothing about its size or importance
• The replication crisis showed many published findings failed to replicate, leading to reforms like pre-registration
• Confounds are uncontrolled variables that can produce a misleading correlation between two unrelated things (e.g., ice cream sales and drowning)
• Random assignment, not just random sampling, is what allows researchers to rule out pre-existing group differences
• Mixed-methods research combines quantitative and qualitative approaches to capture both breadth and depth

Related Topics

Prerequisites

• Overview of Psychology and the scientific method
• Major Perspectives in Psychology (to understand what different research traditions were trying to test)

Related Topics

• Key Theories and Concepts in Psychology (the theories these methods are used to test)
• Ethics in Psychology (the ethical rules that constrain research design and data collection)

Next Topics

• Ethics in Psychology — detailed treatment of informed consent, deception, and institutional review boards
• Biological Bases of Behavior — how methods like brain imaging extend research into physiology
• Developmental Psychology — longitudinal and cross-sectional designs used to study change over time