What you'll learn

• What makes a subject “scientific”, including objectivity, falsifiability, control and replication.
• Why some psychological research looks very scientific, while other research is harder to fit into a scientific model.
• How research methods, statistics and graphs support psychology’s claim to be a science.
• How to use AO1, AO2 and AO3 when writing essays on this debate.

The big question: can psychology be a science?

Psychology studies the mind and behaviour. That immediately creates a challenge: behaviour can be observed, but thoughts, feelings and intentions often have to be inferred from what people say or do.

Natural sciences such as biology and chemistry often study things that can be measured directly. Psychology sometimes does this too, for example measuring brain activity, reaction times or test scores. But it also studies experiences like memory, obedience, prejudice, attachment and mental illness, which are more difficult to measure objectively.

What makes research scientific?

Empirical evidence

Psychology often uses empirical evidence. For example, Loftus and Palmer (1974) did not simply guess that leading questions affect memory; they measured participants’ speed estimates after different verb conditions such as “hit” or “smashed”.

Objectivity

Psychology improves objectivity through standardised procedures, operationalised variables and statistical analysis. For example, Bandura et al. (1961) used behaviour categories to record children’s aggression towards the Bobo doll.

However, complete objectivity is difficult because psychologists make choices about what to measure, how to interpret behaviour and which theories to use.

Operationalisation

This is crucial in psychology because many concepts, such as obedience, anxiety or memory, are not directly visible.

Control and standardisation

A controlled procedure keeps important conditions the same for all participants, so the researcher can be more confident that changes in behaviour are caused by the independent variable.

In Milgram (1963), the setting, verbal prods and shock generator were standardised. This made the study easier to replicate and helped create strong control. But it also raised ethical issues because participants were deceived and experienced stress.

Replication

Bocchiaro et al. (2012) can be compared with Milgram because both investigated obedience and disobedience, but in different contexts. Replication does not always mean copying a study exactly; it can also mean testing whether a general effect appears in a new situation.

Falsifiability

A scientific claim must make clear predictions. For example, “leading questions influence memory estimates” can be tested. A vague claim such as “people remember things strangely because memory is mysterious” is not very scientific because it does not clearly say what evidence would count against it.

The scientific method in psychology

Psychologists often move through a cycle: theory, hypothesis, operationalisation, data collection, analysis, conclusion and replication.

The diagram also shows why statistics matter. Scientific psychology does not just describe behaviour; it asks whether patterns are strong enough to be unlikely due to chance.

Research methods and statistics as part of science

Levels of measurement

• Nominal data: categories, such as “obeyed” or “disobeyed”.
• Ordinal data: ordered scores, such as ranks from most aggressive to least aggressive.
• Interval data: equal units, such as scores on a standardised test where the distance between scores is meaningful.

Descriptive statistics

Descriptive statistics summarise data.

• Mode: the most frequent score or category.
• Median: the middle score when data are ordered.
• Mean: the arithmetic average.
• Range: highest score minus lowest score.
• Variance: how spread out scores are around the mean.
• Standard deviation: a measure of spread around the mean.
• Ratios, percentages and fractions: useful for comparing proportions, such as the percentage of participants who obeyed.

Graphs and diagrams

Different graphs suit different data:

• Bar chart: compares separate categories.
• Histogram: shows frequency of continuous data grouped into intervals.
• Line graph: shows change over time or across ordered conditions.
• Pie chart: shows proportions of a whole.
• Scatter diagram: shows the relationship between two co-variables in a correlation.

Inferential statistics

Psychologists usually compare an observed value from a statistical test with a critical value from a statistical table. The table depends on the test used, sample size, whether the hypothesis is one-tailed or two-tailed, and the chosen significance level.

A common significance level is p = 0.05. This means the researcher accepts a 5% probability of finding a significant result by chance if the null hypothesis is actually true.

Key symbols you may see include:

• = means equal to.
• < means less than.
• > means greater than.
• << means much less than.
• >> means much greater than.
• ∞ means infinity.
• ~ can mean approximately or distributed as, depending on context.

Choosing a statistical test

For OCR, you should know the named non-parametric tests and when they are used:

• Mann-Whitney U: difference test, independent groups, ordinal data or non-parametric conditions.
• Wilcoxon Signed Ranks: difference test, repeated measures or matched pairs, ordinal data or non-parametric conditions.
• Chi-square: association between nominal categories.
• Binomial Sign test: difference from chance or between related nominal outcomes.
• Spearman’s Rho: correlation/association using ranked or ordinal data.

A parametric test is suitable when the data are interval level, approximately normally distributed, and the groups have similar variance. Parametric tests are often more powerful, but only when their assumptions are met.

Type I and Type II errors

A very strict significance level reduces the risk of a Type I error but can increase the risk of a Type II error.

Positions within the debate

Position 1: psychology can be scientific

This position argues that psychology can use controlled experiments, objective measurement, standardised procedures and statistical testing.

Sperry (1968), for example, used a highly controlled split-brain procedure to investigate hemispheric lateralisation. Visual information was presented to one hemisphere at a time, making the study precise and replicable. This supports psychology’s scientific status, especially within the biological area.

Loftus and Palmer (1974) also supports this position because it used a controlled laboratory experiment, manipulated the verb in a question, and measured speed estimates quantitatively.

Position 2: psychology is not fully scientific

This position argues that people are too complex, conscious and socially influenced to be studied in exactly the same way as objects in natural science.

For example, Milgram’s participants may have responded to demand characteristics, the authority of Yale University, or the unusual laboratory setting. This means the study may not perfectly reflect everyday obedience.

Similarly, Bandura et al. (1961) measured aggression in a controlled setting, but hitting a Bobo doll may not be the same as real-life aggression.

Position 3: psychology should use multiple methods

A balanced view is that psychology is partly scientific but should not rely only on laboratory experiments. Case studies, interviews, observations and qualitative data can be useful when studying complex experience.

For example, a purely quantitative study might tell us how many people obeyed, but qualitative data may help explain why they obeyed.

Applying the debate to core studies

Milgram (1963)

Milgram’s obedience study used standardised prods, a controlled setting and quantitative obedience rates. This supports psychology as scientific. However, it involved deception, lack of fully informed consent and potential psychological harm, so it raises BPS ethical concerns around respect, responsibility and integrity.

Bocchiaro et al. (2012)

Bocchiaro et al. studied obedience, disobedience and whistleblowing in a more contemporary context. It used standardised instructions and quantitative comparisons between predicted and actual behaviour. This shows how scientific methods can test socially relevant behaviour.

Loftus and Palmer (1974)

This study is a strong example of experimental control. The independent variable was the verb used in the critical question, and the dependent variable was the speed estimate. However, artificial video clips reduce ecological validity.

Bandura et al. (1961)

Bandura’s Bobo doll study used observation categories and controlled conditions, supporting objectivity and reliability. Ethically, though, exposing children to aggressive models raises concerns about protection from harm.

Similarities and differences with other debates

Reductionism versus holism

Science often encourages reductionism, where complex behaviour is broken into measurable parts. This helps control and measurement, but may ignore the whole person and their context.

Nature versus nurture

Scientific psychology can study both biological influences and environmental influences. For example, Sperry (1968) supports biological explanations, while Bandura (1961) supports learning from the environment.

Freewill versus determinism

Scientific explanations often look for causes of behaviour, which can seem deterministic. For example, social learning theory explains aggression through observation and reinforcement. But human choice and meaning can be harder to capture scientifically.

Usefulness and ethics

Scientific methods can make psychology useful because findings can inform therapy, education, policing and policy. But scientific value does not remove ethical responsibility. Research must still follow BPS principles: respect, competence, responsibility and integrity.

Validity, reliability, ethnocentrism and sampling bias

Scientific research should be reliable and valid, but psychology often uses restricted samples. Milgram used American male volunteers, and many classic studies used Western participants. This creates possible ethnocentrism and sampling bias, limiting generalisability.

AO1, AO2 and AO3 for essays

AO1: describe

Define science and explain principles such as objectivity, operationalisation, falsifiability, control, replication and hypothesis testing.

AO2: apply

Use details from a source or study. For example, if a scenario describes a researcher using standardised instructions and a statistical test, link this to objectivity and scientific control.

AO3: evaluate

Judge strengths and limitations. You might discuss internal validity, ecological validity, reliability, ethics, reductionism, usefulness, and whether the study can be replicated.