ch4 - 2

Question 1

Probability sampling

Answer 1

Each element of the population has a known chance of being selected as a subject

Question 2

Non-probability sampling

Answer 2

The elements of the population do not have a known chance of being selected as a subject

Question 3

Probability sampling types

Answer 3

• Simple random sampling
• Systematic sampling
• Stratified sampling
• Cluster sampling

Question 4

Non-probability sampling types

Answer 4

• Convenience sampling
• Quota sampling
• Judgement sampling
• Snowball sampling

Question 5

Simple random sampling

Answer 5

Each population element has an equal chance of being chosen

(+) offers highest generalisability (-) costly? Good sampling frame needed

Question 6

Systematic sampling

Answer 6

Select random starting point and then pick every n^th element

(+) simplicity (adds a degree of system or process)
(-) Low generalisability if there happens to be a systematic difference between every n^th observation

Question 7

Stratified sampling

Answer 7

Divide the population in meaningful (homogenous) groups, then apply SRS within each group

(+) All groups are adequately sampled, allowing for group comparisons
(-) More time consuming
(-) Requires homogenous subgroups

Question 8

4 criteria for stratified sampling to be possible

Answer 8

1. Elements within strata or subgroups must be homogenous
2. Elements must differ or be heterogenous between strata or sub groups/subpopulation different between groups
3. The stratification variables must be related to the characteristic of interest
4. The number of strata usually varies between 2 and 6. Beyond 6 strata, any gain in precision is more than offset by the increased costs

Question 9

Cluster sampling

Answer 9

Divide the population in heterogenous groups, randomly select a number of groups and select each member within these groups
(+) Geographic clusters can be created (-) subsets of naturally occurring clusters are typically more homogenous than heterogenous

Question 10

Convenience sampling

Answer 10

Select subjects who are conveniently available

(+) convenient (inexpensive and fast) (-) lower generalisability

Question 11

Quota sampling

Answer 11

Fixed quota for each subgroup

(+) when minority participation is critical (-) lower generalisability

Question 12

Judgement sampling

Answer 12

Select subjects based on their knowledge/professional judgement

(+) Convenient (inexpensive and fast) when a limited # of people has the info you need (-) lower generalisability

Question 13

Snowball sampling

Answer 13

“Do you know people who…” obtaining referrals from referrals from referrals

(+) for rare characteristics (“experts”) (-) first participants strongly influence the sample

Question 14

Determining the sample size, rules of thumb

Answer 14

• sample size ≥ 75, <500
• multivariate research : ≥ 10x parameters to be estimated
• subsample (e.g. male/female): ≥ 30 per subsample

General rule: Larger sample size = lower sampling error* provided the appropriate sample design is used