220.127.116.11 Generalising from samples The analysis of survey data undertaken so far has not attempted to make generalisations to wider populations from which the samples were taken. It is important to be aware that the data is for a sample, not a population, and analysis so far is about the sample.
One cannot simply transfer results for a sample onto the population from which it is drawn. Even representative samples are not identical replicas of the population but are liable to variation due to sampling error. The bigger the sampling error the less accurate your sample will be in estimating values for the population.
If the sample is not a representative sample (see Section 8.3.9 then it will be biased as well.
So making statements about wider populations must be done with care. It is important to be aware of both bias and sampling error when generalising about populations from samples.
The other thing that to be aware of at this stage of the analysis is that although a relationship between two variables might have been identified, it is not possible to say how strong that relationship is on the basis of the analysis so far.
Measuring the degree or extent of a relationship or an association can be done statistically (see Section 18.104.22.168).