Designing a study
| English | Chinese | Pinyin |
|---|---|---|
| hypothesis | 假设 | jiǎ shè |
| null hypothesis | 零假设 | líng jiǎ shè |
| independent groups | 组间设计 | zǔ jiān shè jì |
| repeated measures | 组内设计 | zǔ nèi shè jì |
| random allocation | 随机分配 | suí jī fēn pèi |
| extraneous variable | 无关变量 | wú guān biàn liàng |
A study that finds nothing is a valid result
- The commonest failure in a student research project is not a null result. It is adjusting the analysis until something appears.
- State the hypothesis and the null hypothesis 零假设 before collecting anything, and report what you find either way.
- The justification of the method is marked as heavily as the result, which is what makes an honest null result score well.
Your study finds no effect. What should you do?
The method is marked as heavily as the result, which is why an honest null result scores well.
Choosing a design
- Independent groups 组间设计 puts different people in each condition. Simple, but the groups may differ to begin with.
- Repeated measures 组内设计 puts the same people in both. Powerful, but practice or fatigue carries over.
- Independent groups needs random allocation 随机分配; repeated measures needs counterbalancing.
- Choosing between them, and saying why, is a marked decision rather than a detail.
Match each design to the safeguard it needs.
Different people may differ to begin with; the same people carry practice and fatigue across.
Extraneous variables
- An extraneous variable 无关变量 is anything else that could explain your result.
- Control what you can; name the rest in the limitations. An unnamed confound is the one an examiner will find.
- ⚠ "Participants were tested at different times of day" is a real confound and writing it down costs nothing.
You could not control the time of day participants were tested. What should you do?
An unnamed confound is the one an examiner finds. Naming it costs nothing and earns credit.
Name one extraneous variable in a study of music and memory, and say how you would control it.
Example: "Tiredness could affect recall, so all participants would be tested at the same time of day."
Students who sleep more get better grades. Three explanations.
Sleep could improve grades. Or conscientious students both sleep regularly and study well. Or anxiety about grades could reduce sleep, running the arrow backwards.
One correlation, three explanations. Naming the alternatives is the answer; asserting the first is the error the whole methods unit exists to prevent.
More sleep goes with better grades. Which explanations remain possible? Choose all that apply.
Three live explanations, and the correlation alone cannot separate them.
Correlation does not establish cause. The sentence every psychology report needs and the one most often missing — and in this module its absence is marked, not merely noticed.
Do not change your hypothesis after seeing the data and present it as what you predicted. It is the difference between a study and a story, and the write-up shows it: a prediction that fits the result exactly, with no reason given for having made it.