Statistical sampling Flashcards
What is random (simple) sampling
A sample where each individual has a equal chance of being chosen
How would you form a random sample from a population of n individuals
- Give all individuals a unique integer from 1 to n
- Generate a random number (with a calculator)
- Continue until the sample size is reached (discard duplicates)
- Chose the people assigned the generated numbers
What are the benefits of random sampling
- Equal chance of selection, so reduces bias
- Requires less population knowledge to be completed
- Simplest form of sampling
What are the potential problems with random sampling
- Data is not necessarily representative of the population
- Time consuming to allocated numbers and then choose
- Ineffective with smaller populations
What is systematic sampling
A sample that selects individuals at fixed periodic intervals
How would you form a systematic sample from a population
Choose every nth individual, depending on the size of the population and of the desired sample size.
What are the benefits of systematic sampling
- Simple and convenient to use
- Creates a sample of members evenly distributed in the sample, so reduces bias
- Works effectively for large populations
What are the potential problems with systematic sampling
- Requires the knowledge of an approximate population total
- Interacts with periodic traits (ie. systematic errors are either ignored or exaggerated)
- It creates a fractional chance of selection, which means that most individuals are systematically missed, with no chance of selection
What is stratified sampling
A sample with subgroups (strata) in the same proportion as the whole population
How would you form a stratified sample from a population
- Calculate the proportion of the population made up by the subgroups (number in subgroup / total population)
- Multiply this proportion by the desired sample size
- Use random sampling to select this number of individuals from each group.
What are the benefits of stratified sampling
- Representative of the whole population
- Reduces bias
- Using random sampling to fill each category if simple to do
What are the potential problems with stratified sampling
- Requires detailed knowledge about individuals in the population to calculate the proportions
- The random sampling evolved is time consuming
- Individuals overlapping categories may cause issues
What is cluster sampling
A sample made up of a pre-existing sub group (cluster) that is representative of the whole population.
How would you form a cluster sample
Choose a cluster of individuals that are representative of the whole population to use as a sample.
(eg. people on the same flight at an airport - as long as the sample doesn’t need to be representative of destinations)
What are the benefits of cluster sampling
- Requires minimal knowledge about the total population size
- Requires minimal effort to construct a sample
- Can be used on very large populations to save time