Research Methods

Introduction

Research methods are the backbone of psychology. Understanding how data is collected, analysed, and evaluated is essential for every topic on the A-Level specification. This section covers experimental and non-experimental methods, sampling techniques, ethical considerations, descriptive and inferential statistics, correlations, observations, self-report techniques, case studies, and the role of peer review.

Key Concepts

Experimental Methods

Feature	Laboratory	Field	Natural	Quasi
IV manipulated?	Yes	Yes	No	No (pre-existing)
Setting	Controlled	Natural	Natural	Any
Control	High	Low	Low	Variable
Ecological validity	Low	Higher	Higher	Variable
Demand characteristics	High risk	Lower	Low	Variable

Laboratory experiments take place in a controlled environment where the researcher manipulates the independent variable (IV) while controlling extraneous variables. They allow precise measurement and replication but may lack ecological validity.

Field experiments take place in a natural setting where the IV is still manipulated. Participants are often unaware they are in a study, reducing demand characteristics, but extraneous variables are harder to control.

Natural experiments exploit pre-existing differences in the IV (e.g., comparing areas with different crime rates). The researcher cannot manipulate the IV. Useful when manipulation would be unethical or impractical.

Quasi-experiments involve pre-existing groups (e.g., gender, age) where random allocation is impossible. The IV is not manipulated by the researcher.

Experimental Designs

Independent measures (between-groups): Different participants in each condition. No order effects, but participant variables are a confound and more participants are needed.
Repeated measures: Same participants in all conditions. Fewer participants needed and participant variables controlled, but order effects are a problem (mitigated by counterbalancing).
Matched pairs: Participants are paired on relevant variables and each member of the pair is assigned to a different condition. Controls participant variables without order effects, but matching is time-consuming and perfect matching is impossible.

Sampling Methods

Method	Description	Strength	Limitation
Random	Every member of the target population has an equal chance of selection	Unbiased, representative	Difficult to obtain a full list; may still produce an unrepresentative sample
Systematic	Every nth member of the target population is selected	Simple, objective	The list may have a hidden pattern that introduces bias
Stratified	Population divided into subgroups; proportional random sampling from each	Highly representative	Time-consuming; must know population proportions
Opportunity	Selecting whoever is available at the time	Quick, convenient	Biased towards certain types of people (e.g., students)
Volunteer (self-selected)	Participants put themselves forward (e.g., advert)	Willing participants; less ethical concern	Volunteer bias — certain personality types more likely to volunteer

Ethics

The British Psychological Society (BPS) Code of Ethics and Conduct outlines key principles:

Informed consent: Participants must be told the true nature of the study before agreeing to take part. Presumptive, prior general, or retrospective consent can be used when full informed consent is not possible.
Deception: Deliberately misleading participants is acceptable only when it is justified by the scientific value of the study and there is no alternative. Debriefing must occur afterwards.
Right to withdraw: Participants must be told they can leave at any time and have their data removed. This is especially important when payment or course credit is offered.
Confidentiality: Personal data must be protected. Participants should be referred to by number or pseudonym in published reports.
Protection from harm: Participants must not experience physical or psychological harm beyond what they would encounter in daily life.
Debriefing: After the study, participants must be told the true aims, given the right to withdraw their data, and provided with support if distressed.

Descriptive Statistics

Measures of central tendency:

Mean: The arithmetic average. Uses all data points but is distorted by outliers.
Median: The middle value when data is ordered. Not affected by outliers but does not use all data.
Mode: The most frequent value. Can be used with nominal data but may not be representative.

Measures of dispersion:

Range: Highest minus lowest value. Easy to calculate but affected by outliers and ignores the distribution of scores.
Standard deviation: Average distance of each score from the mean. Uses all data and is a precise measure, but can be distorted by extreme values.

Inferential Statistics

The choice of statistical test depends on three factors:

Type of test: Difference (related or unrelated) vs. correlation
Level of data: Nominal, ordinal, or interval/ratio
Design: Related (repeated measures / matched pairs) or unrelated (independent groups)

Test	Purpose	Data Level	Design
Sign test	Difference	Nominal	Related
Wilcoxon	Difference	Ordinal	Related
Mann-Whitney U	Difference	Ordinal	Unrelated
Related t-test	Difference	Interval	Related
Unrelated t-test	Difference	Interval	Unrelated
Chi-squared	Difference / association	Nominal	Unrelated
Spearman’s rho	Correlation	Ordinal	—
Pearson’s r	Correlation	Interval	—

Significance: If the calculated value is equal to or more extreme than the critical value (at p ≤ 0.05), the result is statistically significant and the null hypothesis can be rejected. The 0.05 level (5% probability) is the conventional threshold in psychology, balancing Type I and Type II errors.

Type I error: Falsely rejecting the null hypothesis (a false positive). Type II error: Falsely accepting the null hypothesis (a false negative).

Correlations

Correlations measure the strength and direction of a relationship between two co-variables. They do not establish cause and effect.

Positive correlation: As one variable increases, the other increases.
Negative correlation: As one variable increases, the other decreases.
Zero correlation: No relationship between the variables.

Correlation coefficients range from −1 to +1. The closer to ±1, the stronger the relationship.

Observations

Naturalistic: Conducted in the participant’s own environment with no manipulation. High ecological validity but low control.
Controlled: Some variables are controlled by the researcher. Greater replicability but lower ecological validity.
Covert: Participants do not know they are being observed. Reduces demand characteristics but raises ethical issues (lack of informed consent).
Overt: Participants know they are being observed. More ethical but may alter behaviour (Hawthorne effect).
Participant: The researcher joins the group being studied. Provides an insider perspective but risks loss of objectivity.
Non-participant: The researcher observes from outside. More objective but may miss nuances.

Behavioural categories: Breaking behaviour into discrete, observable units before the observation begins. This improves inter-observer reliability and objectivity.

Time sampling: Recording behaviour at fixed time intervals (e.g., every 30 seconds). Practical for infrequent behaviours but may miss important events between samples.

Event sampling: Recording every instance of a target behaviour. Useful for frequent behaviours but impractical if the behaviour is complex or occurs rapidly.

Self-Report Techniques

Questionnaires:

Open questions: Allow qualitative, detailed responses. Rich data but harder to analyse.
Closed questions: Produce quantitative data (e.g., Likert scales, fixed-choice). Easy to analyse but may oversimplify attitudes.

Strengths: Cost-effective, can reach large samples, standardised. Limitations: Low response rates, social desirability bias, response bias.

Interviews:

Structured: Pre-set questions. Easy to replicate and compare but inflexible.
Semi-structured: Mix of pre-set and follow-up questions. Balance of comparability and depth.
Unstructured: Free-flowing conversation guided by the participant. Rich, detailed data but difficult to replicate and analyse.

Case Studies

An in-depth investigation of a single individual, group, or event. Methods include interviews, observations, and analysis of personal documents.

Strengths: Provide rich, detailed data; useful for studying rare phenomena (e.g., HM, Genie); can challenge existing theories.

Limitations: Cannot be generalised (unique cases); researcher bias; reliance on retrospective data; ethical concerns (privacy, consent).

Peer Review

Peer review is the process by which submitted research is evaluated by other experts in the field before publication.

Single-blind: Reviewers know the author’s identity but the author does not know the reviewer. Most common.
Double-blind: Neither author nor reviewer knows the other’s identity. Reduces bias but harder to arrange.
Open review: Both parties know each other’s identity. Transparent but may inhibit honest criticism.

Strengths: Ensures quality control; prevents flawed research from being published; identifies plagiarism.

Limitations: Anonymity may allow harsh or biased reviews; may suppress innovative findings; slow process; reviewer bias (institutional loyalty, gender bias).

Key Studies

Study	Researcher(s)	Year	Method	Key Findings	Evaluation
On being sane in insane places	Rosenhan	1973	Covert participant observation	Pseudopatients were not detected by staff; labels persisted despite normal behaviour	High ecological validity; ethical concerns (deception); limited generalisability
The halo effect	Thorndike	1920	Correlational study	Ratings of one positive trait correlated with ratings of unrelated traits	Demonstrated cognitive bias; correlational — no causation
Strange Situation	Ainsworth	1978	Controlled observation	Identified three attachment types (secure, insecure-avoidant, insecure-resistant)	High reliability; culturally biased; limited ecological validity

Key Terminology

Term	Definition
Independent variable (IV)	The variable manipulated by the researcher to observe its effect
Dependent variable (DV)	The variable measured by the researcher
Extraneous variable	A variable other than the IV that might affect the DV if not controlled
Confounding variable	An extraneous variable that has systematically varied with the IV
Operationalisation	Defining variables precisely so they can be measured
Hypothesis	A testable prediction about the relationship between variables
Null hypothesis	States there is no significant difference or relationship
Directional hypothesis	Predicts the direction of the difference or correlation
Non-directional hypothesis	Predicts a difference but not its direction
Ecological validity	The extent to which findings generalise to real-life settings
Population validity	The extent to which findings generalise to the wider population
Mundane realism	How closely a study mirrors real-life tasks or situations
Demand characteristics	Cues that allow participants to guess the study’s aims
Investigator effects	When the researcher’s behaviour influences participant responses
Social desirability bias	Tendency to respond in a way that appears favourable
Reliability	The consistency of a measure across time, researchers, or items
Validity	Whether a measure truly reflects what it claims to measure
Nominal data	Categorical data with no order (e.g., eye colour)
Ordinal data	Data that can be ordered but intervals are not equal (e.g., rankings)
Interval data	Continuous data with equal intervals and a meaningful zero may not exist (e.g., temperature)
Meta-analysis	A statistical technique combining results from multiple studies

Evaluation Points

Strengths of Experimental Methods

Control: Laboratory experiments allow precise control of extraneous variables, increasing internal validity and establishing cause and effect.
Replicability: Standardised procedures allow studies to be repeated, testing reliability.
Objectivity: Quantitative data is less open to subjective interpretation.

Limitations of Experimental Methods

Ecological validity: Artificial lab settings may not reflect real-world behaviour.
Demand characteristics: Participants may guess the aim and alter their behaviour.
Ethical constraints: Some variables cannot be manipulated for ethical reasons (e.g., studying the effects of trauma).
Reductionism: Breaking complex behaviour into isolated variables may oversimplify human experience.

Strengths of Non-Experimental Methods

Ecological validity: Naturalistic observations and case studies capture real behaviour.
Rich data: Qualitative methods provide detailed, nuanced insights.
Ethical: Often less intrusive than controlled experiments.

Limitations of Non-Experimental Methods

Causation: Correlations and observations cannot establish cause and effect.
Subjectivity: Qualitative data may be interpreted differently by different researchers.
Generalisability: Case studies involve small, unique samples.

Methodology

Designing a Study

Aim: State the purpose of the research.
Hypothesis: Write a directional, non-directional, or null hypothesis.
Variables: Identify and operationalise the IV and DV.
Design: Choose experimental design (independent measures, repeated measures, matched pairs).
Sampling: Select participants using an appropriate sampling method.
Procedure: Write standardised instructions and a step-by-step protocol.
Ethics: Consider BPS guidelines; obtain ethical approval.
Data analysis: Choose appropriate descriptive and inferential statistics.

Improving Reliability

Standardisation: Use the same procedure, instructions, and materials for every participant.
Test-retest: Administer the same measure twice and correlate the results.
Inter-observer reliability: Have two or more observers record the same behaviour and compare using a correlation coefficient.

Improving Validity

Internal validity: Control extraneous variables, use single-blind or double-blind procedures, standardise instructions.
External validity: Use representative sampling, conduct field experiments, replicate findings across settings.
Ecological validity: Use naturalistic methods or realistic tasks.

Common Pitfalls

Confusing correlation with causation: A significant correlation between two variables does not mean one causes the other. There may be a third (confounding) variable, or the direction of causality may be unclear.
Selecting the wrong statistical test: Always consider the type of data (nominal, ordinal, interval), the design (related, unrelated), and the purpose (difference, correlation) before choosing a test. Using an inappropriate test invalidates the conclusion.
Ignoring ethical guidelines: Failing to obtain informed consent, not debriefing participants, or causing unnecessary distress are serious ethical violations that can invalidate research. Always plan ethics from the outset.

Worked Examples

Example 1: 16-Mark Essay

Question: Discuss the strengths and limitations of using laboratory experiments in psychological research. [16 marks]

Model Answer:

Laboratory experiments are a key research method in psychology, characterised by a controlled environment in which the researcher manipulates the independent variable (IV) while measuring the dependent variable (DV), with all extraneous variables held constant. This method has been used extensively, from Loftus and Palmer’s (1974) study of eyewitness testimony to Bandura’s (1961) Bobo doll experiment.

One major strength of laboratory experiments is their high level of control over extraneous variables. By standardising the environment, instructions, and materials, researchers can isolate the effect of the IV on the DV. This increases internal validity, allowing researchers to establish a cause-and-effect relationship between variables. For example, in Asch’s (1951) line conformity experiments, the standardised line lengths and controlled confederate behaviour ensured that any conformity observed was due to the social pressure of the majority rather than other factors. This level of control is not possible in field or natural experiments.

A further strength is replicability. Because laboratory experiments use standardised procedures, other researchers can repeat the study under identical conditions. This allows findings to be tested for reliability. If the same results are obtained consistently, confidence in the conclusion increases. The replication of Milgram’s obedience studies across cultures and decades (e.g., Slater et al., 2006, virtual re-enactment) demonstrates the reliability and robustness of laboratory findings.

However, laboratory experiments are frequently criticised for their low ecological validity. The artificial setting may not reflect real-life behaviour, meaning findings cannot be generalised beyond the laboratory. Orne (1962) argued that participants in lab experiments are aware they are being studied and may respond to demand characteristics — cues that allow them to guess the study’s aims and behave accordingly. For instance, in Milgram’s original study, participants may have continued administering shocks because they guessed (correctly) that no real harm was being done, rather than because of genuine obedience to authority.

Additionally, laboratory experiments often use unrepresentative samples, in most cases university students, which limits population validity. Henle and Hubble (1978) found that conformity rates varied significantly across cultures, suggesting that findings from Western, educated, industrialised, rich, and democratic (WEIRD) samples may not generalise universally. This is a significant limitation in a discipline that aims to explain human behaviour universally.

In conclusion, laboratory experiments offer high internal validity and replicability, making them invaluable for establishing cause and effect. However, their artificial nature limits ecological validity, and demand characteristics pose a threat to the authenticity of participant responses. Researchers should complement laboratory findings with field and natural experiments to build a more complete understanding of behaviour.

Example 2: 16-Mark Essay

Question: Evaluate the use of correlations as a research method in psychology. Refer to psychological research in your answer. [16 marks]

Model Answer:

A correlation is a statistical technique used to measure the strength and direction of a relationship between two co-variables. Unlike experiments, correlations do not involve manipulating an independent variable; instead, both variables are measured as they occur without manipulation. The correlation coefficient (r) ranges from −1 (perfect negative) to +1 (perfect positive), with 0 indicating no relationship.

One strength of correlations is that they allow researchers to study variables that would be unethical or impractical to manipulate experimentally. For example, investigating the relationship between childhood trauma and adult mental health would be unethical as an experiment, as researchers cannot ethically expose participants to trauma. A correlational design allows this relationship to be studied using pre-existing data, providing valuable insights that can guide further research and intervention.

Correlations are also relatively quick and economical to conduct. They commonly use secondary data that has already been collected, avoiding the need for complex experimental setups. Large datasets, such as those from longitudinal studies or government surveys, can be analysed for correlations without the time and cost of primary data collection. This makes correlations a practical starting point for identifying patterns worthy of more detailed investigation.

However, the most significant limitation of correlations is that they cannot establish causation. Even a strong correlation between two variables does not mean one causes the other. There may be a third, unmeasured variable (a confounding variable) that influences both. For example, a correlation between ice cream sales and drowning incidents does not mean ice cream causes drowning; both are influenced by a third variable — hot weather. This is known as the “third variable problem” and is a fundamental limitation of correlational research.

A related issue is the directionality problem. Even if a causal relationship exists, correlations cannot determine which variable causes which. For instance, a correlation between self-esteem and academic achievement does not reveal whether high self-esteem leads to better grades or whether achieving good grades boosts self-esteem — or both.

Furthermore, correlations can be misleading when presented without appropriate context. A statistically significant correlation may be weak (e.g., r = 0.15) and of limited practical importance, yet it may be reported as though it demonstrates a meaningful relationship. This is particularly problematic in media reporting of psychological research, where correlation is often presented as causation.

In conclusion, correlations are a useful research method for identifying relationships between variables that cannot be studied experimentally. They are efficient and practical. However, the inability to establish cause and effect, the risk of confounding variables, and the potential for misinterpretation mean that correlations should be used as a starting point for research rather than as conclusive evidence. Findings from correlational studies should be followed up with experimental or longitudinal designs to investigate causality.

Summary

Research methods form the foundation of all psychological investigation. Key points to remember:

Experimental methods (laboratory, field, natural, quasi) differ in control, ecological validity, and the ability to manipulate the IV.
Sampling methods affect the representativeness and generalisability of findings; choose the method that best suits the research question and population.
Ethical guidelines (BPS Code of Conduct) must be followed: informed consent, right to withdraw, confidentiality, protection from harm, and debriefing.
Descriptive statistics summarise data (mean, median, mode, range, standard deviation); inferential statistics determine whether results are significant (sign test, chi-squared, Mann-Whitney U, Wilcoxon, t-tests, Spearman’s rho, Pearson’s r).
Correlations measure relationships but cannot establish causation.
Observations, self-reports, and case studies provide complementary data to experiments, each with distinct strengths and limitations.
Peer review ensures quality control in published research but is not without its own biases and limitations.

Mark this page as reviewed