E-module 1 - Hypothesis testing

Question 1

What is a research question?

Answer 1

• provides the frame for the entire research project
• Everything read, experimented and discussed should relate back to it.

Research questions may be quite broad and non-specific:
“Does exercise prevent osteoporotic fractures?”

Or they may be narrow and specific:
“Can walking 4 miles a day prevent osteoporotic fractures in post-menopausal women over the age of 55?”

Question 2

What is a hypothesis?

Answer 2

A testable statement that predicts a new finding before the answer is known.
2 types:
- Null hypothesis
- Alternative hypothesis (may be several)
These predictions are stated at the outset of a piece of work, and once data have been collected, carefully chosen statistical tests are able to test the validity of the hypothesis.

Question 3

How can the research question be evaluated?

Answer 3

Hypothesis testing often in combination with significance testing using statistics

Question 4

What is a null hypothesis?

Answer 4

The null hypothesis (H0) states that there is no dependent relationship between two variables.

• most important hypothesis to test, as being able to reject it demonstrates the existence of a relationship.
• not often referred to in scientific papers as authors and readers understand its existence as tacit (i.e. doesn’t need to be specifically stated)

Question 5

What is an alternative hypothesis?

Answer 5

The alternative hypothesis predicts a specific and reproducible relationship between variables.
- may predict the direction of the finding, useful if results of an experiment can only go in one direction.

Even with directional hypothesis, when analysing data, check in both directions in case an effect is missed

Question 6

How is the null hypothesis rejected?

Answer 6

Statistical tests are used to test validity of hypothesis.

• determine whether reject or fail to reject a hypothesis
• hypothesis that cannot be rejected is the most likely but not necessarily correct

Question 7

What is a p-value?

Answer 7

P = probability that the observed difference was observed by chance.
- number between 0 and 1 that calculates the probability your observed results are real i.e. a study based on a small group reflects what would be observed in the entire population

p-value is interpreted – and used to reject or fail to reject the hypothesis – by comparing the calculated probability to the significance level (alpha).

Question 8

What is the alpha-value?

Answer 8

Alpha is the probability of rejecting a null hypothesis when it is true.

For example, a significance level of 0.05 indicates a 5% risk of concluding that a difference exists when there is no actual difference i.e. rejecting the hypothesis when it should not be rejected.

Question 9

How does the size of the p-value determine the validity of the hypothesis?

• very small p-value
• small p-value
• large p-value
• p-value close to the cut-off

Answer 9

• very small: (p ≤ 0.001) indicates very strong evidence against the hypothesis, so reject the hypothesis.
• small (p ≤ 0.05) indicates strong evidence against the hypothesis, so you reject the hypothesis.
• large (p > 0.05) indicates weak evidence against the null hypothesis, so you fail to reject the hypothesis.
• p-values very close to the cut-off (0.05) are considered to be marginal (could go either way).