What is the population
All the individual items which of are interest in the given situation
What is a census
A survey that collects data from the whole population
Adv and Disadv of a census
Adv:
- Accurate, unbiased representation of whole population
Disadv:
- Time-consuming and expensive
- Large volumes of data to process
- Impractical if testing involves destruction
What is a sampling unit
Individual members of the population that can be sampled
What is a sampling frame
A full list of all the sampling units, giving a unique number (identification) to each unit
Name the sampling methods
Simple Random
Systematic
Stratified
Opportunity
Quota
Self-selected
Cluster
When to carry out Simple random sampling
When each item in the population has an equal chance of being selected
Full list of the population
Each selection is independent
How to carry out Simple Random sampling
Give a unique identification number to each item in the population
Use a random number generator to randomly select the numbers and match it to the item from the sample
The items selected will make up the sampling frame
Adv and Disadv of Simple Random sampling
Adv:
- Unbiased
Disadv:
- Inconvenient over a large area
What does Systematic sampling do?
Chooses every nth member from a population to be sampled
How to carry out Systematic sampling
Give a unique identification number to each item in the population
Calculate a regular interval: n = Population size / Sample size, starting at a random item between 1 and n
Continue to sample every nth item from the starting point
Adv and Disadv of Systematic sampling
Adv:
- Unbiased
- Quick
- Suitable for large samples
Disadv:
- Could coincide with a pattern (eg every 10th product is faulty and you sample every 10th)
- Sampling frame coule be biased
How does Stratified sampling work
Uses the same proportion of each category in the sample as there is in the population
How to carry out Stratified sampling
Divide the population into categories
Calculate the total population
Calculate the number of items needed in each category:
Size of category sample = (number in category / population size) x sample size
Then, select the sample from each category at random (after we know how many items needed for each category)
Adv and Disadv of Stratified sampling
Adv:
- Can be a good representation if there are disjoint categories (no overlap)
- Useful when results may vary based on category
Disadv:
- Expensive
What is Quota sampling
It involves dividing the population into categories and the sampler has a ‘quota’ to meet - a required number of responses
How to carry out Quota sampling
Divide the population into categories
Give each category a ‘quota’ (number of members to sample)
Collect data until quotas are met in all categories (without random sampling)
Adv and Disadv of Quota sampling
Adv:
- Non-response it not an issue as the quota must eventually be met
- A full list of the population isn’t required
Disadv:
- Easily biased by interviewer as they choose who to sample (could exclude some of the population)
What is Opportunity Sampling / Convenience sampling
When the sample is chosen from a selection of the population that is convenient for the sampler
(eg taking the first ‘n’ items found, or the first ‘n’ people that walk past an interviewer)
Adv and Disadv of Opportunity / Convenience sampling
Adv:
- Quick and easy
Disadv:
- Easily very biased as it isn’t random
- Might not be very representative
What is Self-selected sampling
When individuals in the sample have chosen to be included
(eg respondents to a survey posted publicly on the internet)
What is Cluster sampling
A sampling technique where the population is divided into clusters (smaller groups), then random clusters are selected to form a sample
How to carry out Cluster sampling
Divide the population into clusters: should cover whole population
Randomly select a certain number of clusters to use in the sample (based on required sample size)
Use all members in selected clusters or randomly sample within each cluster to form sample
Adv and Disadv of Cluster sampling
Adv:
- Quick and cheap
- Can use random sampling within the cluster method
Disadv:
- Less representative of entire population
- Can’t always naturally separate clusters
