Designing a study
Introduzido| English | Significado |
|---|---|
| hypothesis/haɪˈpɒθəsɪs/ | hipótese |
| null hypothesis/nʌl haɪˈpɒθəsɪs/ | hipótese nula |
| independent groups/ˌɪndɪˈpendənt ɡruːps/ | an experimental design in which different participants take part in each condition |
| repeated measures/rɪˈpiːtɪd ˈmeʒəz/ | an experimental design in which the same participants take part in every condition |
| random allocation/ˈrændəm ˌæləˈkeɪʃn/ | atribuição aleatória |
| extraneous variable/ekˈstreɪnɪəs ˈveərɪəbl/ | variável extra |
Twelve children and a retracted paper, 1998
- In 1998 The Lancet published a paper by Andrew Wakefield and colleagues about 12 children, suggesting a link between the MMR vaccine and autism. There was no control group, the sample was tiny and selected, and later investigation found serious problems. The journal retracted the paper in 2010, but vaccination rates had already fallen in some countries.
- Large, well-designed studies found no link. A study's design decides what it can show. A clear hypothesis 假设 and careful control of variables come first.
Which design problems did the 1998 MMR paper have?
Dificuldade 2/5 · Moedas 140 + pequeno bônus na primeira conclusão hoje
It was retracted in 2010.
Hypotheses
- A hypothesis is a testable prediction: "Students who sleep 8 hours recall more words than students who sleep 5 hours."
- The null hypothesis 零假设 states no difference or no effect. The data either give evidence against it or do not.
Which is a null hypothesis?
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
No difference or no effect.
Experimental designs
- Independent groups 独立组设计: different people in each condition. Repeated measures 重复测量设计: the same people do every condition.
- Random allocation 随机分配 puts participants into groups by chance, so groups are similar on average.
Match each design to its description.
Dificuldade 2/5 · Moedas 140 + pequeno bônus na primeira conclusão hoje
Each has strengths and weaknesses.
Random allocation makes groups similar on average.
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
Chance spreads differences out.
Controlling variables
- An extraneous variable 无关变量 — time of day, noise, prior knowledge — could affect the result. Keep it constant or balance it across groups.
- A correlation alone cannot show cause: a third factor may explain both variables.

In a memory test, which is an extraneous variable?
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
It could affect the result.
Worked example: does caffeine speed reactions?
- Hypothesis: participants who drink a caffeinated drink have faster reaction times than those who drink an identical-tasting decaffeinated drink.
- Design: 40 volunteers, random allocation to two independent groups; neither participants nor tester know which drink is which (double-blind).
- Controls: same time of day, same reaction-time app, no other caffeine that morning. Compare the mean times.
In a double-blind study, neither participants nor tester know who has which condition.
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
It reduces expectation effects.
Ethics
- Informed consent, the right to withdraw, confidentiality, protection from harm, and a debrief afterwards.
- Deception must be justified and explained at the end.
Which are ethical requirements?
Dificuldade 2/5 · Moedas 140 + pequeno bônus na primeira conclusão hoje
Deception must be explained at the end.
A small, selected sample without a control group cannot show cause. Twelve chosen children could not separate the vaccine from chance or other causes. Compare with a control group and use enough participants.
State a testable hypothesis and a null hypothesis. Choose independent groups or repeated measures, allocate randomly, control extraneous variables, and follow ethical rules. Correlation and small selected samples cannot establish cause.
Design a study testing whether music affects concentration. Give the hypothesis, design, one control and one ethical step.
Hypothesis, design, control, ethics.