Topic 9/10 - Hypothesis testing

Question 1

Why do we need hypothesis testing?

Answer 1

It is to be able to successfully weigh up evidence to make evidence based decisions.

Question 2

What are the 3 main steps of hypothesis testing?

Answer 2

1) Set up research question

2) Weigh up evidence

3) Explain conclusion

Question 3

What is the acronyms for the hypothesis testing process?

Answer 3

H
A
T
P
C

Question 4

What are the components of the ‘set up research question’ step?

Answer 4

H: Hypothesis (i.e. use of H0 (null) vs H1 (alternate)

Question 5

What are the components of the ‘Weigh up evidence’ step?

Answer 5

A: Assumptions (tell us if our hypothesis is valid)

T: Test Statistic

P: p-value

Question 6

What are the components of the ‘Explain conclusion’ step?

Answer 6

C: Conclusion

Question 7

What is the null hypothesis?

Answer 7

It assumes that the difference between the observed value (data) and expected value (EV) is due to chance alone

Question 8

What is the alternate/alternative hypothesis?

Answer 8

Assumes that the difference between observed value (data) and expected value (EV) isn’t due to chance alone

Question 9

Should we use a box model for null hypothesis?

Answer 9

Yes, use a box model for null hypothesis, i.e. if null hypothesis is H0 = 0.8, assume

Question 10

Main difference between null and alternate hypothesis?

Answer 10

Null is the default, status quo, and that theres nothing new (contains = >=, <=) whereas alternate is that what we think could happen, and is used to prove the research question (contains /= , >, <)

Question 11

What is the importance of assumptions step?

Answer 11

A conclusion isnt transparent if assumptions arent stated

A conclusion is potentially invalid if assumptions arent justified

FOr example, important assumptions are independence and that everyone has equal chances

Question 12

What is the test statistic?

Answer 12

It is a measure of the difference between what is observed in the data and what is expected from the null hypothesis

If null hypothesis is true, then test statistic is the standard unit corresponding to the observed value

Question 13

What is the equation for test statistic?

Answer 13

(Observed value - expected value) / standard error

Question 14

What is the p value?

Answer 14

P value is the observed significance level.

It is a way of weighing up whether the sample is consistent with the H0. The p value is the chance of observing a test statistic if H0 is true. It says nothing about the alternate hypothesis

Question 15

What does a big p value mean

Answer 15

Not statisticall significant

Data consistent with null hypothesis

We can retain the null hypothesis

Question 16

What does a small p value mean (by convention, small is p <0.05)

Answer 16

Statistically significant

Evidence against null hypothesis

We can reject the null hypothesis

Question 17

WHat are some common mistakes with the p value?

Answer 17

P value is not the chance that the null hypothesis is true

A large p value doesnt mean that H0 is true

The use of 0.05 isnt mandatory, the significance of 0.05 is a convention. Some people may use 0.01 and say results are highly significant

Question 18

What should be mentioned in the conclusion part?

Answer 18

Statistical conclusion: We conclude that the data is consistent with the null hypothesis

Scientific conclusion: therefore the new treatment does/doesnt have an effect rate significantly higher/lower or equal to ____

Question 19

What are the 2 types of hypotheses?

Answer 19

1 sided

2 sided

Question 20

What are 1 sided hypotheses?

Answer 20

Specific direction of alternate hypothesis

I.e. H1 : p > 0.8

Question 21

What are 2 sided hypotheses?

Answer 21

Doesnt specify direction of alternate hypothesis , i.e. H1 : p =/ 0.8

In this case we can wusually double the p value as it has two sides : P > 0.8, P <-0.8