Sampling method affects who can appear in the data. Simple random sampling gives every member an equal chance; systematic sampling follows a fixed interval after a start; stratified sampling represents defined subgroups. Clustered, quota, convenience and judgement methods make different practical trade-offs and may introduce bias.
A distribution should be described as a whole. Consider typical value or centre, spread, overall shape, clusters, gaps and unusual values rather than choosing one striking observation.
Primary data are collected for the current investigation. Secondary data already exist and must be checked for source, definitions, date, coverage and sampling method before conclusions are reported.
Large samples can reduce random fluctuation but cannot automatically repair a systematically biased selection method. Reliability depends on both sample size and sample design.
When working with reconciliation data concerning Aboriginal and Torres Strait Islander Peoples and non-Indigenous Australians, use reputable sources, retain context and avoid turning sensitive social data into decontextualised percentages.