E-module 1 - Principles of research design and hypothesis testing

Question 1

What are the principles behind generation of research questions?

Answer 1

Asks a question that can be answered by testing of a hypothesis

• requires a question mark
• think about what needs answered and whether it can be answered when writing

Question 2

Definition of a hypothesis?

Answer 2

A predictive statement that be tested in order to answer a related research question

Question 3

What are the 2 types of hypothesis?

Answer 3

Null - “there will be no change in etc.”

Alternate - “there will be a specific and reproducible change/directional change in etc.”

Question 4

What are the 2 outcomes of statistical tests regarding hypotheses?

Answer 4

Can only REJECT or FAIL TO REJECT (no accepting)

Question 5

Definition of a P-value?

Answer 5

The likelihood that the observed difference was observed by chance (between 0 and 1)

Question 6

What is the P-value compared to and what conclusions are evaluated from this?

Answer 6

Compared to the SIGNIFICANCE LEVEL
Can reject or fail to reject hypothesis based on which side of the significance level the p-value lies
- e.g. p = 0.00123 where significance level is p = 0.002, can reject hypothesis as statistically significant

Question 7

Definition of an ‘a-value’?

Answer 7

Probability of rejecting a null hypothesis when it is true

- this is the significance level e.g. a=0.05 is the same as 5% possibility of rejecting null hypo when it is true

Question 8

Suggestions of
p<0.001
p<0.05
p>0.05
regarding evidence against hypothesis, outcome of hypothesis, and resulting significance

Answer 8

p<0.001 = very strong evidence against hypothesis, reject H0, highly significant difference

p<0.05 = strong evidence against hypothesis, reject H0, significant difference

p>0.05 = weak evidence against hypothesis, fail to reject H0, no significant difference

Question 9

What happens if p=0.05 or v close to it and what should you do for this?

Answer 9

p=0.05 or being very close is a marginal result

- p-values should always be reported, particularly in these cases in order for readers to make their own conclusions

Question 10

What are the boundaries for rejecting or failing to reject H0?

Answer 10

p < a - reject H0

p > a - fail to reject H0

Question 11

Can you prove a null/alternate hypothesis?

Answer 11

No

- can only fail to disprove it (fail to reject)

Question 12

Is statistically significant the same as clinically significant?

Answer 12

No
- no steadfast reason why but you’d imagine that statistically significant is only the same when the significance level for clinically significant is also the measure of statistics significance
e.g. statistically sig = a = 0.05 where clinical sig = a = 0.001, NOT the same
BUT where statistical sig = 0.0005 and clinical sig = 0.0005, ARE the same

Question 13

Using incidence of depression in males and females as the research topic, generate a research question, null hypothesis, and alternate hypothesis.

Answer 13

Research question - Is there a difference between the incidence of depression in males and females?

Null hypothesis - There is no difference between the incidence of depression in males and females

Alternate hypothesis - There is a difference/ increase/ decrease in the incidence of depression in males compared with females