Topic 12: sampling methods + selection bias Flashcards
(29 cards)
Define random error
- Error by chance = inherent in sampling process
Define systematic error
- Bias
- Via man-made action when conducted study
Describe random errors in epidemiological studies
- Random error decreases = increase in sample size
- 95% Cl narrows = increase in sample size
- Down to 0 = total population included
Describe systematic error in epidemiological studies
- Systemic bias not influneced by sample size
Define selection bias
- Systematic error resulting from = participants in study not representative of source population
- Result = biased sample = biased estimate
Give the sampling methods
1) Non-probability sampling
- Convenience
2) Probability sampling
- Simple random
- Stratified random
- Cluster
- Multi-stage
3) Systematic sampling
- Simple systematic
- Prop quota sampling
Define probability sampling
- Sample selected via probabilistic methods
- Random selection = strong statistical inferences about whole group
Define non-probability sampling
- Samples selected by convenience
- Non-random selection based on convenience + other criteria = easy to collect data
Define systematic sampling
- Sample selected via simple systematic rule
Describe simple random sampling
- Most straightforward
- All individuals in sampling frame = same probability of being selected independantly of others
- Use = quantitive research
- Sample size = large = chosen indiduals representative of source population
- Tools = random number generator etc
Give the advantages/disadvantages of random sampling
ADVANTAGES:
- Ensures representative sample
- Less costly/time-consuming than other methods
- Ideal for quantitive study + testing hypothesis
DISADVANTAGES:
- If sampling ferame too large + population geographically diverese = impractical to perform random
- Time consuming + costly = if large sample
Describe stratified random sampling
- Same as simple BUT within specific strata
- Size of sample = proportional to specifc stratum size in population
- Strata = e.g. district/gender/age division
Give the advantages/disadvantages of stratified sampling
ADVANTAGES:
- Precise conclusions = every subgroup represented
- Allows comparison between subgroups
DISADVANTAGES:
- More time-consuming than simple random
- Can’t be used if members cannot be classified into a subgroup confidently
- Higher complexity = errors
Describe cluster sampling
- Based on the structure of natural clusters/groups in population
- E.g. schools/city/hospitals
- Random sample of clusters → selecting all individuals in selected clusters
- Sample frame = all clusters
- Either include every individual in clusters OR if large clusters = sample within
Give the advantages/disadvantages of cluster sampling
ADVANTAGES:
- Reduce cost + time when population spread over large area
DISADVANTAGES:
- Differences between clusters = errors
- Cannot guarantee sample cluster = representative of population
- Representativeness compromised = few clusters selected + clusters too specific + contain too few individuals
Describe multi-stage sampling
- Uses hierarchial structure of natural clusters
- After random cluster selection → random selection of individuals within cluster
STAGES:
1) Random selection of large clusters = schools
2) Random selection of smaller cluster within = class
3) Random selection of individuals within
Give the advantages/disadvantages of multi-stage sampling
ADVANTAGES:
- Improve sample representiveness if sample geographically diverse/small
- Less costly/time consuming
DISADVANTAGES:
- Representativeness compromised = if too few clusters selected + too specific + few individuals
Describe simple systematic sampling
- Selected based on systematic rule
- E.g. people with name starting with a letter + born on a specific month + every nth person
Give the advantages/disadvantages of simple systematic sampling
ADVANTAGES:
- Acceptable + convenient alternative to random sampling
- Faster + cheaper
DISADVANTAGES:
- Representativeness compromised = selection not random
Describe proportional quota sampling
- Similar to stratified sampling = sample selected based on existing strata
- But systematic sampling within strata not random
- Sampling will continue until % + sample quota met
Give the advantages/disadvantages of pro quota sampling
ADVANTAGES:
- Acceptable + convenient alternative for random sampling
- Ensures original population structure compared to simple systematic
DISADVANTAGES:
- Representative compromised = selection not random
Give non-random sampling methods
1) Convenince sampling = most accessible
2) Voluntary response sampling = volunteer themselves for study
3) Purposive sampling = select individuals with characteristics wanted in study
4) Snowball sample = participants recruit others
Describe convenience sampling
- Most used non-probability
- Selection non-random = based on convenince
- E.g. ask people around you to take a survey just because they’re available
Give the advantages/disadvantages of convenience sampling
ADVANTAGES:
- Cheap + fast + convenient
DISADVANTAGES:
- Representiveness compromised = selection fully non-random
