Biostats

Question 1

What are biostatistical methods used for?

Answer 1

They are designed to quantify risks and health outcomes in the population and to establish evidence for associations.

Biostats provides the foundation for evidence based PH

Question 2

What is statistical bias?

Answer 2

A systematic error in the estimation of a quantity of interest that causes the expected value of a statistic to differ from the true population value.

Question 3

What is sample selection bias?

Answer 3

When individuals of the sample aren’t representative of the population.

Question 4

What is a random sample and what are the benefits and limitations?

Answer 4

Every individual in the population has an equal chance of being selected.

Benefits –> representative
Limitation –> not helpful to study rare conditions
Challenge –> getting the data (logistics)

Question 5

What is population surveillance, what is it used for and what is one example of it?

Answer 5

Systematic collection of data in the population to examine risk behaviors and health trends

Descriptive statistics are computed to summarize the prevalence of risk factors and outcomes

Example: NHANES –> health and nutritional data from interviews and physical exams. yearly cross-sectional capture of the US population. Uses a probability sampling design instead of random sample to get data for specific groups.

Question 6

How do you account for sampling variability?

Answer 6

Low sampling variability –> repeated sampling from population would yield similar answers

• confidence intervals
• margins of error
• error bars

Question 7

How do you establish an association and what does the strength of the evidence depend on?

Answer 7

As the exposure changes the outcome tends ro change with it

• study design
• accuracy of measurements
• strength of association
• accounting for confounding

Question 8

What is statistical power? why is it useful?

Answer 8

The probability that your proposed study will give a statistically significant result.

• Larger sample size = higher power
• Important to calculate before doing a study to avoid wasting resources
• Helps determine if you should have found something

Question 9

What is the relationship between measurement error and the strength of evidence?

Answer 9

• measurement error can introduce bias into the study
• more accurate measurements = stronger evidence
• there is a tradeoff between accuracy and cost when measuring exposure to outcomes

Question 10

What has to happen for something to be statistically significant? and what does it depend on?

Answer 10

The change we observe would have to be more extreme than what we would expect to see by chance.

depends on: sample size, magnitude of the chanage and consistency

Question 11

What is a p value?

Answer 11

P values give the probability of getting the observed value or something more extreme under the assumption that there is no actual effect.

Smaller p - values provide stronger evidence

Question 12

What is a confounder and what are examples of real confounders that have affected real data?

Answer 12

The effect of a hidden variables (s) that distorts the relationship between exposure and outcome.

• Smoking in the association between coffee and cancer risk
• Maternal infection in the association between acetaminophen and childhood asthma
• Healthy use effect in the association between hormone replacement therapy and reduce Heart risk

Question 13

How can you eliminate confounding variables from affecting you data?

Answer 13

• RCTs are the gold standard to avoid confounding

Question 14

Difference between statistical and practical significance

Answer 14

Statistical - evidence of a non-zero effect, not necessarily a large effect so it might not have practical significance.

Question 15

What is a meta-analysis and why is it useful?

Answer 15

When you combine evidence from different studies.

replication of results is important to build strong evidence.

Question 16

What is the target population?

Answer 16

Group you ideally want to generalize results to.

Question 17

What is the source population?

Answer 17

Specific individuals from which a representative sample will be drawn - they are at risk of being a case

Question 18

Study population?

Answer 18

group you can get access to, realistically generalize results too

Question 19

sample

Answer 19

group you actually ends up in your study analysis