Semester 1

Question 1

Does a sample state exactly what the overall population is like?

Answer 1

No, it only states what the sample is like

Question 2

What is the difference between statistics and probability?

Answer 2

statistics uses samples to infer information about a population
Probability uses the information about the population to make predictions- often about a sample

Question 3

Samples from a population vary so how can we find the actual mean value for the population?

Answer 3

Take the means of the averages from the samples to get an overall mean. This will give us a normal distribution of means.

Question 4

How can you find out how precise an overall mean is? Define this method of analysis

Answer 4

use a 95% confidence interval
This is the range that will contain the samples mean 95% of the time, so the lower the confidence interval is the greater the precision

Question 5

Define precision and accurary

Answer 5

Accuracy:How close a measured value is to the true value
Precision: Ability to do the same measurement over and over and get same result

Question 6

Define bias

Answer 6

When one result is favoured over another

Question 7

What are the two types of bias?

Answer 7

Selection bias and information bias

Question 8

Describe the two types of selection bias

Answer 8

• External validity: leads to errors in generalisability because the sample is not representative of the general population
• Internal validity: leads to errors in comparability because the groups being compared are not from the same population (comparing GP pts without a disease with hospital pts with a disease- should compare GP pts with/without disease)

Question 9

What is information bias? State somethings that may cause it

Answer 9

Measurement error leading to a certain result favoured.

• differential recall error (case- control)
• differential observer/ interviewer error
• Differential measurement error
• Differential misclassification error

Question 10

What is confounding

Answer 10

A separate characteristic of a population that could influence the outcome making comparison between populations difficult.
Eg nurses exp more burnout but is this bc they’re nurses or bc they’re often young females, which also causes more burnout

Question 11

How do you account for confounding variables?

Answer 11

Standardise the data

Question 12

What is difference between direct and indirect standardisation?

Answer 12

Direct= standardised against each other
Indirect= standardised against an acceptable value

Question 13

What is the difference between descriptive epidemeology and analytical epidemeology?

Answer 13

Descriptive describes a population using surveys- can be misleading 100% who drink water die
Analytical epidemeology compares groups to find risk factors

Question 14

How is a cross sectional survey done?

Answer 14

Take a group of people and find out prevalence of the disease

Question 15

What is difference between prevalence and incidence?

Answer 15

prevalence is number/ proportion of people in a population living with a certain disease- how widespread disease is
Incidence is the rate of new cases- risk of getting disease

Question 16

What is a cohort study? How is it done? When is it usually used?

Answer 16

Prospective: take a group of people, expose some and don’t expose others then wait and count outcome events in patient years
Retrospective: take group of people where some were exposed and others weren’t in the past and then count outcomes in pt years
Good for rare exposures

Question 17

What is the difference between odds ratio, risk ratio and rate?

Answer 17

Odds ratio is the proportion that experienced the final outcome, irrespective of time
Risk ratio: proportion that experienced outcome at end of a certain time period
Rate: incidence of outcome during time period (it could be that only 2/10 died in the time period (low risk) but the rate of death was 1/10 per week (high rate ratio))

Question 18

What is a case control study? Why would it be favourable of cohort? What are limitations?

Answer 18

Find people with the disease/outcome and see what proportion were exposed
Quicker and good for rare outcomes, but can only get odds & risk ratio