Observation method
Observation is not casual looking. It is selecting what to watch, recording it (notes, video, checklists) and analysing it.
- Naturalistic vs controlled: in a real setting (a playground) or in a lab where conditions are arranged.
- Non-participant vs participant: the observer stays outside, or becomes part of the group.
Strengths: shows behaviour as it really happens. Limits: takes time; people may change behaviour when watched; the observer's own views can bias what is recorded.
Experimental method
An experiment tests a cause-and-effect relation.
- Independent variable (IV): what the researcher changes (noise). Dependent variable (DV): what is measured (reading time).
- Experimental group gets the IV; control group does not. Groups are made similar by random assignment.
- Extraneous variables (other things that could affect the DV) are controlled by elimination (remove them), constancy of conditions (keep them the same for all) and counterbalancing (change the order of tasks across participants).
- Lab experiments give tight control; field experiments happen in real settings; quasi-experiments use groups that already exist (e.g. flood-hit and non-flood villages) because the IV cannot be assigned.
Correlational research
We measure two variables and find the correlation coefficient (r), from −1.0 to +1.0.
- Positive: both rise together (study hours and marks).
- Negative: one rises as the other falls (late-night screen time and sleep).
- Zero: no relation (shoe size and marks).
The closer r is to ±1, the stronger the link. Correlation helps prediction, but it does not prove cause — a third factor may affect both.
Survey research
Surveys collect views, attitudes, habits and facts from many people.
- Personal interviews: structured (fixed questions in fixed order) or unstructured (flexible talk). Can be one-to-one or in groups.
- Questionnaire: written questions, open-ended or closed (yes/no, multiple choice); by post, in person or online.
- Telephone survey: quick, but people may hang up.
Limits: people may give socially desirable answers or not remember correctly.
Psychological testing
A psychological test is a standardised, objective tool to measure qualities like intelligence, aptitude, personality or interest.
- Reliability: gives consistent scores (test-retest, split-half).
- Validity: measures what it claims to measure.
- Norms: average scores of a large similar group, used to compare a person's score.
- Types: by language — verbal, non-verbal, performance; by administration — individual or group; by difficulty — speed tests and power tests.
Case study
A case study is an in-depth study of one person, family, group or event. It uses interviews, observation, tests, school and medical records, and diaries. It is common in clinical psychology and in studying rare events.
Strength: rich, detailed picture. Limit: findings about one case cannot easily be generalised to everyone, and memory of past events may be faulty.
Try it
Measure your hours of phone use and hours of sleep for 7 days. Plot them. Is the correlation positive, negative or zero? Then try the free-play step in 3D.
Key formulas and definitions
- Experiment: IV (changed) → DV (measured); experimental group vs control group
- Control of extraneous variables: elimination, constancy of conditions, counterbalancing
- Correlation coefficient r: −1.0 ≤ r ≤ +1.0 (positive, negative, zero)
- Good test = standardised + reliable + valid + norms
- Correlation ≠ causation
Worked examples
1. A researcher gives one group 10 minutes of yoga daily and the other group nothing for a month, then measures stress. Name the IV, DV, experimental and control groups.
IV: yoga. DV: stress score. Experimental group: the yoga group. Control group: the no-yoga group.
2. Classify each correlation: (a) temperature and ice-cream sales, (b) absences and marks, (c) height and liking for music.
(a) Positive, (b) negative, (c) zero.
3. Ice-cream sales and drowning cases rise together. Does ice cream cause drowning?
No. Both rise in hot weather (a third variable). Correlation does not show cause.
4. A reading test gives Aman 48 today and 22 next week without any change in him. What is wrong with the test?
It is not reliable: a reliable test gives consistent scores.
5. Which method: studying a child who grew up alone in a forest?
Case study: a rare, single case studied in depth using many sources.
6. Why is a village hit by flood compared with one not hit a quasi-experiment?
The researcher cannot assign the flood (IV) randomly; the groups already exist, so it is only "like" an experiment.
Common mistakes
- Thinking correlation proves cause. Only a well-controlled experiment shows cause.
- Mixing up experimental and control groups. The control group does not get the IV.
- Saying a test is valid just because it is reliable. A test can be consistent but measure the wrong thing.
- Believing observation is easy. It needs careful selection, recording and control of observer bias.