Probability sampling

Sampling

• Population - entire set of person, objects or events under study

• Source sample - subset of the population in interest

• Sampling - selection of the sample

Probability sampling/random sampling

• Important in quantitative research

• No systematic bias

• Generalise findings to the population of the sample

Simple random sampling

• Each unit in population has equal and independent chance of being selected into sample

• Selected one at a time, independent of one another and without replacement

PROS - Equal chance for everyone, No researcher bias

CONS - Needs large sample, Time-consuming, Hard to list everyone, Can be expensive

Systematic random sampling

• Dividing the sampling frame into intervals

• Randomly selection a starting point then selecting one element from each interval in a systematic way

  • E.g. every fifth person

• Don’t use if there's a possibility of bias in arrangement

PROS - Very representative of population, High precision

CONS - Needs complete, organised lists, Time and effort required

Stratified random sampling

• Strata - population groups that are separate and non-overlapping

• Randomly samples from within each stratum

• Proportionate sampling - sample size is proportionate to the population of the stratum

• Disproportionate sampling - simple size is disproportionate to the population of the stratum, leading to over-sampling of groups

PROS - Ensures all subgroups are included (e.g., age, gender)

CONS - Must know population details to divide into subgroups

Cluster random sampling

• Population is divided into clusters (e.g., areas, schools, hospitals).

• A random sample of clusters is chosen (using simple, systematic, or stratified sampling).

• The whole cluster or a sample from within the cluster is studied.

• Cluster = sampling unit (not the individual).

• Better to choose more clusters and fewer people from each

PROS - Good for very large or spread-out populations, Cost-effective, Easier access within clusters

CONS - More sampling errors than random or stratified methods

Multisatge (pros and cons)

PROS - Reduces cost, Good for large or multi-step studies

CONS - Same limitations as simple random sampling