Topic 10: sampling + random error

Question 1

Define sample

Answer 1

• Selected subset of source population
• Representative of source population

Question 2

Define source population

Answer 2

• Group of all individuals = we’re interested to assess parameters
• Can be general population or sub-population e.g. everyone with condition in country

Question 3

What is the purpose of a sample?

Answer 3

• Study something we can’t study in whole population due to practical restrictions e.g. financials/time

Question 4

Define sampling

Answer 4

• Process of selecting number of individuals from all individuals in source population

Question 5

Define sampling frame

Answer 5

• List with all individuals in population = used for sampling

Question 6

Define sampling units

Answer 6

• Individuals potentially to be selected

Question 7

Describe sampling in descriptive research

Answer 7

• Investigate prevalence/incidence of condition in population
• Sample accuracy important = represents specific source population

Question 8

Describe sampling in analytic research

Answer 8

• Investigate association between exposure + outcome
• Can be general with source population depending on research

Question 9

Describe sampling when investigating biological effects

Answer 9

• E.g. effect of smoke on risk of cancer
• Can be more general with source population = no need to restrict specific region

Question 10

Describe sampling when investigating social/cultural effects

Answer 10

• E.g. effect of social class on risk of heart disease
• Need to restrict source population to specific region from where sample is derived

Question 11

Define an estimate

Answer 11

• When measuring the sample of a population = determine proportion of characteristic = estimate
• Sampling error inherent
• Measure of quantity in sample = represents true quantity in source population

Question 12

Describe statistical inference

Answer 12

• Sample estimate used to draw inferences about population from the sample
• Use statistics to determine degree of uncertainty in estimate of interest

Question 13

Describe the parameter

Answer 13

• Measure of quality/association in population of interest e.g.
  > Mean age
  > Prevalence of obesity
  > Mean difference in bp between M/F
  > Odds ratio for association between smoking + cancer
• Sample is used to make an estimate about something true for the whole group = quantify corresponding to population parameter
• Thereform mean of sample size ≈ mean of population parameter

Question 14

Define sampling variation

Answer 14

• Variation between different sample estimates from same source population

Question 15

Define sampling error

Answer 15

• Difference between sample estimate + actual population parameter when measuring in sample rather than source population
• Due to chance = random error
• Sample size plays role on magnitude of error

Question 16

Describe sampling distribution

Answer 16

• We take repeated samples with sample size = n → e.g. n=3 take many samples of 3 values from the population
• Calculate mean of each sample
• Plot means on histogram = will cluster around true mean = estimate population parameter

Question 17

Describe standard error

Answer 17

• Uncertainty of how well sample estimate represents population parameter
• Estimates SD of sample distribution = average error that can occur with sample size = n

Question 18

How to calculate SE?

Answer 18

SE = S/√n
- S = sample SD
- n = sample size

Question 19

Define confidence intervals

Answer 19

• Range within which we are confident with degree of uncertainty = true population parameters lie

Question 20

Describe how the 95% confidence interval calculated

Answer 20

LOWER CI:
- Sample estimate - 1.96 x SE
UPPER CI:
- Sample estimate + 1.96 x SE
> 95% confident population parameter within interval sample estimate +/- 1.96 SE