Research Methods and Statistics
Psychology · Section 1.2 · 18 study cards
Scientific method, descriptive designs, sampling, correlation, experiments, descriptive and inferential statistics, and research ethics.
Practice this set → Spaced repetition, card by card. No account needed.
Method
The one question that sorts every design
Did the researcher manipulate anything? If yes, it is an experiment and a causal conclusion is available. If no, it is descriptive or correlational, and no matter how large or careful the study, it can only describe or predict. Nothing else in this unit matters as much as getting that first decision right.
Procedure for reading a study description
- Name the design. Look for the manipulation. Words such as assigned, randomly allocated, given or trained signal an experiment. Words such as measured, surveyed, observed, recorded or tracked signal a descriptive or correlational study.
- Name the variables. In an experiment, the independent variable is what the researcher set and it has levels or conditions; the dependent variable is what was measured at the end. In a correlational study there is no independent variable at all, only two measured variables, so do not invent one.
- Check the comparison. Is there a control group, and is it treated identically apart from the independent variable? A missing or non-equivalent control group is the commonest design flaw.
- Check assignment. Were participants placed into groups at random? If groups were pre-existing, such as smokers and non-smokers, the study is quasi-experimental and confounds cannot be ruled out however it is described.
- Hunt the confound. Ask what else differed between the groups. Time of day, experimenter, room, order of tasks, expectations. Name a specific candidate, not just a worry.
- Check the sample. Who was studied and to whom is the conclusion being applied? A convenience sample of students does not license a claim about everyone.
- Check the conclusion against the design. Underline the verb in the researchers' claim. Causes, increases, reduces and improves are causal verbs; if the design was correlational, that verb is the flaw, and you should name the third-variable and directionality problems.
Statistics, at the level exams test
Descriptive statistics summarise the sample you have: centre with the mean, median or mode, spread with the range or standard deviation. Inferential statistics ask whether what you found in the sample is likely to hold in the population. Keep significance and size apart in your head: significance answers is it probably real, effect size answers is it big enough to care about. A result can be one without the other.
Definitions and theorems
Worked example
A researcher reports the following. We recruited 300 adults from a city gym. Each completed a weekly log of hours of exercise and a standard depression inventory. Hours of exercise correlated with depression scores at r = -0.42. We conclude that exercise reduces depression and recommend exercise programmes for the general adult population. Identify the design, name the variables, state one specific third variable, and say what change to the study would license the causal claim.
Name the design. Nothing was manipulated and nobody was assigned to a condition; both exercise and depression were measured as they stood. This is a correlational study.
Name the variables. Because there is no manipulation there is no independent variable. The two measured variables are weekly hours of exercise and score on the depression inventory.
Read the coefficient. r = -0.42 is a moderate negative relationship: more exercise goes with lower depression scores. The sign describes direction only and is not evidence of benefit.
Attack the causal verb. Reduces is a causal claim that a correlational design cannot support. State both reasons. Directionality: depression itself reduces activity and motivation, so the causal arrow may run from depression to exercise. Third variable: a specific candidate is general physical health, which permits more exercise and independently lowers depressive symptoms; chronic pain or employment status would serve equally well.
Attack the sample. Three hundred gym members are a convenience sample of unusually active adults, so generalising to the general adult population is a second, separate flaw.
Say what would fix it. Recruit a sample that is not defined by gym membership and randomly assign participants to a supervised exercise programme or to a control condition matched for time and attention, then measure depression after a fixed period. Random assignment equates the groups at the start, so a difference at the end can be attributed to the exercise.
Report it properly. Give the effect size alongside any significance test, since with 300 participants a small difference can be significant without being clinically meaningful.
Common mistakes
- Reading a negative correlation as a weak one. The sign is direction, the distance from zero is strength. Students routinely rank r = +0.30 above r = -0.70. Cover the sign with a finger before judging strength, then put it back to describe the direction.
- Confusing random sampling with random assignment. Sampling is how people get into the study and governs generalisability; assignment is how they get into the groups and governs causality. An exam item that describes groups formed by chance is testing assignment, whatever the word random tempts you to write.
- Inventing an independent variable in a correlational study. If nothing was manipulated there is no independent variable, only two measured variables. Labelling one of them the independent variable is the first step to writing a causal conclusion the design cannot support.
- Treating a significant result as a big or important one. Significance says probably not chance; it says nothing about magnitude, and large samples make small effects significant. Any question that mentions sample size alongside a significant result is asking about effect size.
- Answering ethics questions with deception is not allowed. Deception is permitted when no alternative exists, risk is minimal and full debriefing follows. The requirements that are non-negotiable are informed consent, the right to withdraw, confidentiality, debriefing, and prior approval by an Institutional Review Board.
Practice it
Reading the method is not the same as being able to recall it under pressure. This set drills 18 cards one at a time and schedules each card separately, so the ones you keep missing come back sooner.