Key things you should know
01
Start with a good question
Everything else — design, sample, analysis — follows from it.
02
Match design to question
Questions about cause need experiments; questions about experience need qualitative methods.
03
Plan before you collect
Decide measures, sample size and analysis in advance, and write them down.
04
Get ethical approval first
Never recruit or collect data until your institution has approved the study.
05
Protect participants’ data
Collect only what you need, use participant codes and store data securely.
06
Report honestly
Include results that did not support your hypothesis, and state limitations clearly.
07
Read critically
Ask how each study was done before accepting what it found.
08
Keep records
A research diary of decisions makes writing up and defending your methods much easier.
Key research terms
Search the glossary, or browse by topic.
Foundations
- Research question
- The specific question your study sets out to answer. It should be clear, focused and answerable with the data you can collect.Example: Does a short breathing exercise reduce stress before an exam?
- Hypothesis
- A testable prediction about the relationship between variables, usually stated before data are collected.Example: Students who do the exercise will report lower stress than those who do not.
- Null hypothesis (H₀)
- The default assumption that there is no effect or no difference. Statistical tests assess how compatible the data are with it.
- Theory
- An organised explanation of why or how something happens, which generates hypotheses that can be tested.
- Variable
- Anything that can take different values and be measured — for example age, stress score or condition.
- Independent variable (IV)
- The variable the researcher manipulates or groups by, expected to influence another variable.Example: Breathing exercise vs no exercise.
- Dependent variable (DV)
- The outcome that is measured to see whether it changes.Example: Stress score after the task.
- Confounding variable
- A third variable linked to both the IV and DV that could explain the result instead of the IV.Example: If the exercise group also had more sleep, sleep could explain lower stress.
- Operationalisation
- Defining exactly how an abstract idea will be measured.Example: ‘Stress’ measured as the total score on a named, validated questionnaire.
Participants and sampling
- Population
- The whole group you want your conclusions to apply to.
- Sample
- The people who actually take part, drawn from the population.
- Sampling method
- How participants are chosen. Random methods give every member a known chance of selection; non-random methods (convenience, purposive, snowball) are easier but less representative.
- Generalisability
- How far findings from your sample can be applied to the wider population or other settings.
- Inclusion / exclusion criteria
- Rules stating who can and cannot take part, set out in advance.
Study design
- Experiment
- A design in which the researcher manipulates the IV and randomly allocates participants to conditions. The strongest common design for testing cause and effect.
- Quasi-experiment
- Compares groups but without random allocation (for example, existing classes). Causal conclusions are weaker.
- Correlational study
- Measures how variables relate without manipulating them. Shows association, not cause.
- Cross-sectional
- Data collected at one point in time.
- Longitudinal
- The same participants measured repeatedly over time, which helps show change and order of events.
- Randomisation
- Allocating participants to conditions by chance, so groups are similar on average in known and unknown characteristics.
- Control group
- A comparison group that does not receive the intervention (or receives an alternative), used to judge its effect.
- Blinding
- Keeping participants and/or researchers unaware of who is in which condition, to reduce expectation effects.
- Between- vs within-subjects
- Between: different people in each condition. Within (repeated measures): the same people take part in every condition.
- Pilot / feasibility study
- A small-scale study to test procedures, recruitment and measures before a full study. Not designed to test effectiveness.
Measurement quality
- Reliability
- Consistency of a measure. Types include test–retest (stable over time), inter-rater (agreement between observers) and internal consistency (items measure the same thing, often reported as Cronbach’s alpha).
- Validity
- Whether a measure or study actually captures what it claims to.
- Internal validity
- Confidence that the IV caused the change in the DV, rather than something else.
- External validity
- Whether findings apply beyond the study — to other people, places and times.
- Construct validity
- Whether a measure truly reflects the concept it is meant to measure.
- Ecological validity
- How closely the study conditions resemble real life.
- Bias
- Any systematic error that pushes results in one direction — for example selection bias, social desirability bias or researcher bias.
Analysis and statistics
- Descriptive statistics
- Numbers that summarise data, such as mean, median, standard deviation and frequencies.
- Inferential statistics
- Methods that use sample data to draw conclusions about a population, such as t-tests, ANOVA or regression.
- p-value
- The probability of getting results at least as extreme as those observed if the null hypothesis were true. It is not the probability that the hypothesis is true, and it does not tell you how large or important an effect is.
- Statistical significance
- A result is called significant when the p-value falls below a threshold set in advance (often .05). Significant does not mean important, and is never proof of an effect.
- Effect size
- How big a difference or relationship is, for example Cohen’s d or a correlation coefficient. Essential for judging practical importance.
- Confidence interval (CI)
- A range of plausible values for the true effect. A 95% CI is produced by a method that would capture the true value in 95% of repeated samples.
- Statistical power
- The probability a study will detect an effect of a given size if it really exists. Small samples usually have low power.
- Type I and Type II errors
- Type I: concluding there is an effect when there is not (false positive). Type II: missing a real effect (false negative).
- Correlation vs causation
- Two variables moving together does not show that one causes the other; a confounder or reverse direction could explain it.
Qualitative research terms
- Thematic analysis
- A widely used method for identifying, analysing and reporting patterns (themes) in qualitative data.
- Coding
- Labelling segments of data with short descriptions of what they are about, as a step towards themes.
- Theme
- A pattern of shared meaning across the data, organised around a central idea.
- Saturation
- The point at which new data add little new insight. Its use and meaning are debated, so explain how you judged it.
- Reflexivity
- The researcher’s critical reflection on how their own background, assumptions and role shape the research.
- Trustworthiness
- The qualitative counterpart to reliability and validity, often described as credibility, transferability, dependability and confirmability (Lincoln and Guba).
Ethics and data
- Informed consent
- Participants agree to take part after being told, in plain language, what the study involves, its risks and their rights.
- Right to withdraw
- Participants can stop taking part at any time without giving a reason; the information sheet should explain what happens to their data.
- Anonymity
- No one, including the researcher, can link data to an individual.
- Confidentiality
- The researcher knows who took part but protects that information and does not disclose it.
- Pseudonymisation
- Replacing identifying details with codes, with the key kept separately and securely.
- Debrief
- Information given after taking part that explains the study’s aims and offers support or contact details.
- Ethical approval
- Formal permission from an institutional ethics committee, required before recruiting participants or collecting data.
Sources and evidence
- Peer review
- Evaluation of a manuscript by independent experts before publication. It improves quality but does not guarantee a study is correct.
- Primary source
- Original research reported by the people who did it.
- Secondary source
- A summary or discussion of others’ research, such as a textbook or review.
- Hierarchy of evidence
- A ranking of designs by how well they protect against bias for questions of effectiveness — systematic reviews and randomised trials near the top, expert opinion near the bottom. Other question types may need different designs.
- Replication
- Repeating a study to see whether its findings hold. Findings that replicate are more trustworthy.
- Pre-registration
- Publicly recording hypotheses and analysis plans before collecting data, to reduce selective reporting.
What is qualitative research?
Qualitative research explores meanings, experiences and perspectives. It works with words, images and observations rather than numbers, and asks questions such as “what is it like…?”, “how do people make sense of…?” and “why do people…?”.
- Typical methods: semi-structured interviews, focus groups, observation, open-ended surveys, documents and diaries.
- Sampling: usually small, purposive samples chosen because they can speak to the topic, not to be statistically representative.
- Analysis: coding and interpreting data — for example thematic analysis, interpretative phenomenological analysis (IPA), grounded theory or discourse analysis.
- Strengths: depth, rich detail, participants’ own words, ability to discover unexpected issues.
- Limitations: findings are not statistically generalisable, analysis is time-consuming, and interpretation depends on the researcher — which is why reflexivity matters.
What is quantitative research?
Quantitative research measures and counts. It collects numerical data to describe patterns, test hypotheses and estimate relationships or differences, and asks questions such as “how much…?”, “how many…?” and “does X affect Y?”.
- Typical methods: experiments, questionnaires with rating scales, structured observation, behavioural tasks and existing datasets.
- Sampling: larger samples, ideally selected so findings can be generalised; sample size is often planned with a power analysis.
- Analysis: descriptive and inferential statistics — for example t-tests, ANOVA, correlation and regression.
- Strengths: precision, comparability, ability to test cause and effect (in experiments), and replicability.
- Limitations: may miss context and meaning, depends on the quality of its measures, and statistical results are easy to over-interpret.
Qualitative vs quantitative at a glance
| Qualitative | Quantitative | |
|---|---|---|
| Aim | Understand meanings and experiences | Measure, compare and test |
| Typical question | How? Why? What is it like? | How much? How many? Does X affect Y? |
| Data | Words, images, observations | Numbers |
| Sample | Small, purposive | Larger, ideally representative |
| Researcher’s role | Part of the process; reflexivity expected | Aims for distance and standardisation |
| Analysis | Coding, themes, interpretation | Statistics |
| Quality judged by | Trustworthiness (credibility, transferability…) | Reliability and validity |
| Output | Themes, rich description, theory | Statistics, effect sizes, confidence intervals |
Mixed methods
Mixed-methods research combines both approaches in one study — for example, a questionnaire to measure how common something is, followed by interviews to understand why. It is useful when one type of data alone cannot fully answer the question, but it needs careful planning: be clear about why you are combining methods, in what order, and how the findings will be brought together.