Hypotheses, Variables and Experimental Designs
Predicting Behaviour Scientifically
Every psychology investigation starts with a precise, testable prediction called a hypothesis. A vague idea such as “music affects concentration” is not enough — a hypothesis must say exactly what is being changed and exactly what will be measured. Getting this right at the planning stage is what allows a study to be tested, criticised and repeated by other researchers.
Null and Alternative Hypotheses
Every experiment or correlational study has two hypotheses. The null hypothesis states there is no difference and no relationship — any pattern in the data happened by chance. For example: “There will be no significant difference in the number of errors made by participants tested in a noisy room compared with a silent room; any difference is due to chance.”
The alternative hypothesis is what the researcher actually expects to find. Depending on the study, it can predict a difference between conditions (e.g. “participants tested in a noisy room will make significantly more errors than those in a silent room”) or a correlation between two variables that are simply measured, not manipulated (e.g. “there is a positive correlation between hours of sleep and exam score”). If the results are unlikely to be due to chance, the null hypothesis is rejected in favour of the alternative.
The null hypothesis always predicts no difference, no relationship and no pattern — any result is due to chance. The alternative hypothesis predicts either a difference between conditions or a correlation between co-variables.
Variables: What You Change and What You Measure
In an experiment, the independent variable (IV) is the thing the researcher deliberately changes — for example, whether participants revise using flashcards or by rereading notes. The dependent variable (DV) is the outcome the researcher measures — for example, the number of words correctly recalled from a list.
Not every study manipulates an IV. In a correlational study, the researcher simply measures two co-variables — for example, hours of sleep and exam score — and checks whether they rise and fall together, without changing either one directly.
Extraneous variables are anything else that could affect the DV. In the flashcard study, if one group happened to have better baseline memory, that could explain a difference in recall instead of the revision method — so it must be controlled.
Controlling Extraneous Variables: Standardisation
One way to control extraneous variables is standardisation. This means every participant experiences exactly the same procedure, instructions, materials and environment, except for the condition of the IV they are in. In a study on noise and concentration, this might mean giving every participant the same puzzle, the same time limit and the same instructions, so the only thing that differs between groups is the noise level.
Standardisation controls extraneous variables by keeping every part of the procedure identical across conditions, so any difference in the DV can be confidently attributed to the IV rather than to something else.
Experimental Designs: Repeated Measures and Independent Measures
A repeated measures design uses the same participants in every condition. Each person completes the task under condition A and then condition B. This controls for participant variables (differences between people, such as memory ability) because each person acts as their own comparison. A weakness is order effects — performing the task twice may cause practice or fatigue, which can be reduced by counterbalancing.
An independent measures design uses different participants in each condition — one group experiences condition A, a separate group experiences condition B. This avoids order effects entirely, since no one repeats the task. Its weakness is participant variables: because the groups contain different people, differences between conditions could be caused by individual differences rather than the IV.
Counterbalancing is one way to manage order effects in repeated measures — half the participants complete condition A first, the other half complete condition B first, so any practice or fatigue effect balances out across the whole sample.
There is no single ‘correct’ design. A researcher chooses repeated measures when they want to remove participant variables and have fewer people available; they choose independent measures when order effects would be a serious problem, such as a task that cannot be attempted twice without practice changing performance.
Putting It Together for Designing an Investigation
These planning decisions — hypothesis, IV, DV, extraneous variables, standardisation and design — are exactly what you will be asked to work through when designing an investigation. Being able to state each one clearly, in the right order, is the foundation for every other research methods skill on this course.