Designing a study
Introduced| English | Meaning |
|---|---|
| hypothesis/haɪˈpɒθəsɪs/ | A proposed explanation or prediction that can be tested using evidence. |
| null hypothesis/nʌl haɪˈpɒθəsɪs/ | The default claim about a population parameter that a statistical test assesses against an alternative. |
| independent groups/ˌɪndɪˈpendənt ɡruːps/ | A design in which each person takes part in only one condition. |
| repeated measures/rɪˈpiːtɪd ˈmeʒəz/ | A design in which the same people complete every condition. |
| random allocation/ˈrændəm ˌæləˈkeɪʃn/ | random allocation |
| extraneous variable/ekˈstreɪnɪəs ˈveərɪəbl/ | extraneous variable |
A correlation leaves several explanations open
- Students who report more sleep also report higher grades. Sleep may matter, but study habits or worry could also contribute to the association.
- State the hypothesis and null hypothesis 零假设 before the planned analysis. A null result should be reported with the method and uncertainty, not hidden.
Your study finds no effect. What should you do?
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Report the method, findings and uncertainty honestly. No detected effect does not prove the true effect is exactly zero.
Choose the design for the question
- Independent groups 组间设计 use different people; repeated measures 组内设计 use the same people across conditions. Consider feasibility, individual differences and carryover.
- Random allocation 随机分配 helps balance groups. Counterbalancing helps with order effects. Neither guarantees removal of all differences or makes an unsuitable method suitable.
Match each design to the safeguard it needs.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
Different people may differ to begin with; the same people carry practice and fatigue across.
Separate an extra influence from a confound
- An extraneous variable 无关变量 may influence the outcome. It becomes a confound when it varies systematically with the condition and offers an alternative explanation.
- Different testing times are a possible concern; check whether time is linked to condition. Record controls and remaining limitations instead of silently excluding inconvenient data.
You could not control the time of day participants were tested. What should you do?
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Report testing-time variation and whether it was linked to condition. An extra variable is a confound when it varies systematically with the condition.
Name one extraneous variable in a study of music and memory, and say how you would control it.
Difficulty 3/5 · 200 coins + a small bonus on first completion today
Example: test conditions at comparable times and record sleep-related limitations. A common time alone does not guarantee identical tiredness.
Plan the report before seeing the answer
- Define the measure, analysis and ethical safeguards. Exploratory changes can be useful, but label them as exploratory and explain why they were made.
- No detected effect does not prove the true effect is exactly zero. Consider sample size, measurement and uncertainty rather than promising a score for a particular result.
More sleep goes with better grades. Which explanations remain possible? Choose all that apply.
Difficulty 2/5 · 140 coins + a small bonus on first completion today
Three live explanations, and the correlation alone cannot separate them.
Music is always tested in the morning and silence in the evening. Why is this a concern?
Difficulty 1/5 · 100 coins + a small bonus on first completion today
Testing time supplies an alternative explanation. The concern does not establish that the result is false or which factor actually caused it.
Check a transfer design
- If music participants are tested in the morning and silence participants in the evening, time and music are entangled. A time-of-day difference can compete with the music explanation.
- A fairer allocation and comparable procedure address that problem. Correlation alone cannot separate direction of effect and third-variable accounts.
A fairer allocation and comparable procedure address that problem. Correlation alone cannot separate direction of effect and third-variable accounts.