Chance II

Question 1

What are p-values?

Answer 1

p-values are the probably of getting study estimate when there is really no association, just because of sampling error (chance). If probability is really low, unlikely the estimate is due to sampling error

Question 2

What is hypothesis testing?

Answer 2

Parameter can indicate whether there is true association, or no association
Study generates an estimate and we can find if there is an associate or not
If our study finds an association and there isn’t one, we got it wrong - this is what a p-value tells us, the probability of finding an association when there truly isn’t one

Question 3

What is the null hypothesis and the alternative hypotheses

Answer 3

Null = no association
Really is no association in population, parameter = null value

Alternative = parameter does not equal null
Really is an association in the population

Question 4

How do we interpret p-values?

Answer 4

p<0.05 = reject null hypothesis and accept alternative hypothesis hence association is statistically significant. Chance is an unlikely explanation of the study finding

p>0.05= fail to reject null hypothesis and reject alternative hypothesis hence association is not statistically significant. The study finding is consistent with chance as an explanation

Question 5

What are type II errors?

Answer 5

Type II error is where you find that there is no association in your study however in the null hypothesis there is an association
Type II errors incorrectly fail to reject null hypothesis when we should have. Typically due to having too few people in the study (wider CI). Bigger sample size = narrower CI = more likely to get small p. Smaller sample size = wider CI = less likely to get small p

Question 6

Relationship between CI and p-value

Answer 6

You can see whether a p-value is greater or less than 0.05 with a 95% confidence interval
If the 95% CI crosses/includes null value than p>0.05 and results not statistically significant
If the 95% does not cross/include null value than p<0.05 and results are statistically significant

Question 7

Limitations of p-values

Answer 7

Statistical significance threshold is arbitrary and artificial - at threshold 5% will still find a statistically significant association when there really isn’t one 1/20 times
Just gives evidence about consistency with the null hypothesis and doesn’t say anything about precision
Statistical significance is not clinical significance - doesn’t say anything about whether the results are valid, useful or correct