What a hypothesis is

1. A hypothesis is a precise, testable prediction of what a study will find.
2. It is written before any data is collected, working from the aim of the study.
3. The aim is the general purpose, such as finding out whether background music affects recall.
4. The hypothesis turns that aim into a prediction specific enough to be shown wrong.
5. Both variables are operationalised, which means saying exactly how each one is set up or measured.
6. An operationalised prediction reads "people recall fewer words from a 20-word list while pop music plays than in silence".
7. Every study carries two hypotheses: the alternative and the null.

The alternative hypothesis

1. The alternative hypothesis predicts that a real pattern will show up in the data.
2. It claims the difference or the relationship is produced by the variables being studied.
3. It is also called the research hypothesis or the experimental hypothesis.
4. When the results show the predicted pattern, the alternative hypothesis is accepted and the null hypothesis is rejected.

The null hypothesis

1. The null hypothesis states that there will be no difference between the conditions, or no relationship between the co-variables.
2. Any small gap left in the numbers is put down to chance rather than to the independent variable.
3. When the results show no clear pattern, the null hypothesis is accepted and the alternative hypothesis is rejected.

Predicting differences, correlations and no patterns

1. Predicting a difference is used when two conditions or two groups are compared.
   1. The prediction names both conditions and says which one scores higher on the measured outcome.
   2. One example is "pupils who revise with a quiz score higher on a 20-word spelling test than pupils who reread their notes".
   3. Difference hypotheses belong with experiments, where an independent variable is manipulated and a dependent variable is measured.
2. Predicting a correlation is used when two co-variables are measured in the same people.
   1. The prediction names both co-variables and which way the link runs.
   2. A positive prediction reads "the more hours of sleep a pupil has, the higher their score on a 20-mark maths test".
   3. A negative prediction reads "the more hours a pupil spends on a phone after midnight, the lower their score on a 20-mark maths test".
   4. A correlation hypothesis predicts a link between the two co-variables, and does not claim that one produces the other.
3. Predicting no pattern is the null hypothesis written for the same study.
   1. For a difference study it reads "there is no difference in spelling scores between pupils who revise with a quiz and pupils who reread their notes".
   2. For a correlation study it reads "there is no relationship between hours of sleep and score on a 20-mark maths test".

Directional and non-directional predictions

1. A directional prediction states which way the result will go, such as which condition scores higher or whether a correlation is positive or negative.
2. A non-directional prediction states that the two conditions will differ on the measured outcome.
3. Both are forms of the alternative hypothesis, so each one is paired with the same null hypothesis.

Strengths and weaknesses of a precise hypothesis

1. A hypothesis that operationalises both variables tells another researcher exactly what to repeat, which makes a check on reliability possible.
2. A fixed measure, such as words recalled from a set 20-word list, means two researchers marking the same participant reach the same score.
3. Naming the measure forces the question of whether it captures the behaviour being studied, which is a question about validity.
4. "Measures memory" could mean recognition, free recall or how long something is held, so a result based on it can be read three ways.
5. Because the prediction is written down before the data comes in, a result in the opposite direction counts against it.
6. One weakness is that the prediction narrows what is looked at, so an unexpected pattern in the data is easy to walk past.
7. A weakness of a correlation hypothesis is that a strong link can still be produced by a third variable that was never measured.
8. A weakness of a difference hypothesis is that it predicts the outcome without saying why the two conditions differ.

Writing a hypothesis for a given study

1. Name the two things the investigation compares or links.
2. Decide whether the study looks for a difference between conditions or a relationship between co-variables.
3. Write the alternative hypothesis as one full sentence, naming both variables and how each one is measured.
4. Turn that same sentence into the null by replacing the predicted pattern with "no difference" or "no relationship".
5. Match the rest of the plan to it, so the people tested, the task set and the score recorded all fit the prediction.
6. When the study is already described, work its hypothesis out from the aim given in that description.

Turning this into marks

1. When both hypotheses are asked for, write the alternative and the null as two separate full sentences rather than one sentence with "not" added.
2. Name the measure, such as words recalled from a 20-word list, because "affects memory" does not secure the knowledge mark.
3. State which condition scores higher when the prediction is directional, because a sentence that only says the groups differ answers a different question.
4. The mark most often lost is writing the null as the opposite prediction instead of as no difference or no relationship.