sampling, hypothesis, normally distributed data

Question 1

sampling

Answer 1

relatively small number of observations
from which you describe whole population
calculate mean and use confidence interval to determine where mean lies

Question 2

Normal distribution

Answer 2

symmetrical distribution
basis of many statistical tests
if you know mean and SD - can draw every point on a curve

Question 3

null hypothesis

Answer 3

forming it is the first stage of the statistical test
states there is no difference between the groups
relative risk = 1
odds ratio = 0

Question 4

what are confounders

Answer 4

they’re associated with the exposure and the outcome

Question 5

ways to deal with confounding

Answer 5

design stage - randomisation, restriction or matching (in case-control study)
analysis stage - stratification (split analysis by age group for example), standardisation/regression - building a statistical model

Question 6

why would patients be allocated randomly to 2 groups

Answer 6

to reduce the effect of confounders

control unknown confounding variables

Question 7

what does it mean if randomisation is stratified by diabetic status and why would it be done

Answer 7

all people with diabetes are divided equally

so any effect would not be as a result of differing numbers of diabetics

Question 8

confounding variable

Answer 8

a factor that is associated with exposure and outcome of interest
eg age, smoking, socioeconomic deprivation

Question 9

randomisation

Answer 9

ensuring both group have similar proportions of confounding variables

Question 10

regression

Answer 10

control confounding
analysis stage
statistical modelling is used

Question 11

restriction

Answer 11

control confounding

only include people without pre-existing illnesses

Question 12

standardisation

Answer 12

confounding
analysis
used to produce SMR

Question 13

stratification

Answer 13

control confounding
at analysis stage
risks calculated separately for each category of confounding variable

Question 14

what is the 95% confidence interval

Answer 14

an estimated range of values, calculated from a set of sample data which are likely to contain the true population value
a range of values that contain the true risk in 95% of cases

Question 15

what would the confidence interval have to include to accept the null hypothesis of an odds ratio

Answer 15

it would include 1

Question 16

what would the confidence interval include to accept the null hypothesis of a relative risk

Answer 16

0