Sampling techniques

We want to make some inferences about some population, based on data available from a sample.

Sampling errors refer to uncertainties that arise due to extrapolation, and generally cannot be avoided. Bias refers to flaws in the data collection or analysis that can be avoided. Typically, bias means the population that the sample is representative of, is different from the target population that we want to learn more about.

Sampling techniques

technique description advantages disadvantages
simple random treat population as one group easy unreliable if the population is varied in composition
convenience take nearest or a cluster of values useful if population is uniform generally unreliable
systematic take every Nth{N\text{th}} value sample over entire population less able to take fewer or more samples, susceptible to missing or overemphasizing patterns, tedious
quota convenience sampling for groups somewhat representative of the population if groups are well-chosen sometimes unreliable
stratified simple random sampling for groups representative of the population if groups are well-chosen more technical in the set up

IA moderation mostly uses stratified sample, where works are chosen to cover the whole spread of marks.