Sampling Methods & Size Calculations Flashcards
What are the types of probability sampling?
Simple random
Systematic random
Stratified random
Cluster random
Advantages to simple random probability sampling
Easy
Large samples likely to be representative
No info needed prior to sampling
Each member of the population has equal chance of being selected
Disadvantages to simple random probability sampling
Larger samples often needed to be representative
Comparisons between small subgroups may be difficult
Describe systematic random probability sampling
A random sample, with a fixed periodic interval, is selected from a larger population.
Starting point is chosen at random.
When can systematic random probability sampling be especially beneficial?
When don’t have access to complete list of population in advance.
Advantages of systematic random probability sampling
Simple + more straight forward than simple random sampling
Low risk of error
Avoids risk of clustering groups
A numbered list of people isn’t required
Disadvantages of systematic random probability sampling
Not all individuals have equal chance of selection
Assumes size of population can be determined
Small risk that researcher could influence the order of the list in their favour
When is stratified random sampling used
When we want to guarantee that a certain characteristic of our sample is included in the sample itself
i.e age or gender
What does stratified random sampling ensure
A proportional number of people from a subpopulation
Advantages of stratified random sampling
Fewer people needed to achieve the same representativeness compared to simple random sample.
Can customise sample to be comparable to wider population.
Can make inferences about small subgroups.
Disadvantages to stratified random sampling
Info is req before for particular stratification variables
Becomes difficult if theres too many stratification variables
What would cluster random sampling be appropriate for?
School based intervention
List the non-probability sampling methods
Convenience
Purposive
Quota
Convenience sampling
Ind. selected for ease
No equal probability of being selected
Advantages to convenience sampling
Quick + easy compared to all probability sampling
When is convenience sampling mostly used?
For pilot studies
Disadvantages to convenience sampling
Can’t estimate sampling error
Exclusion bias i.e gender
Can’t generalise from the sample to the population
Purposive sampling
Ind are selected because they posses a particular characteristic of interest or they represent a specific group.
No equal probability of being selected
What is purposive sampling mainly restricted to?
Qualitative studies
Advantages of purposive sampling
Guarantees characteristics of interest will be selected.
Good for qualitative studies
Disadvantages of purposive sampling
Can’t estimate sampling error
Exclusion bias, biased responses
Can’t generalise from the sample to the population
Quota sampling
Sampling continues until the quota for a specific characteristic is achieved, after which no more people with that characteristic will be selected.
What is quota sampling often used for?
Social surveys
Advantages to quota sampling
Ensures the sample have some characteristics the same as the population.
Common in large social surveys.
Disadvantages to quota sampling
Can’t estimate sampling error
Exclusion bias
Can’t generalise from sample to population
What may it mean if a sample size is too small?
Underpowered
What may a sample size that is too big mean for the study?
Becomes overpowered
Ethical issues relative to sample sizes being too big
Unethical due to causing unnecessary inconvenience
Wastes time + money
Extremely unethical if exposing more people than required to pain/risk of injury.
Ethical issues relative to sample sizes being too small
Could fail to identify a true meaningful effect as being statistically sig. different from ‘no effect’.
Unethical due to waste of resources, peoples time and could cause misleading conclusions.
What does the sample size calculation derive from?
Desired power - Should not be less than 80%
Desired significance level - 0.05
Estimate of the true effect size
What comes under the estimate of the true effect size
Means/SD = Cohens D
Proportions = Rel. risk or Odds Ratio
When are cohens D or rel. risk or odds ratio used
Cohens D = Data on continuous scale
Rel. risk or odds ratio = Used for dichotomous outcome variables (i.e yes or no).
Power
Probability that, for a given effect size, your study will find a statistically significant effect.