sampling errors

(NL: steekproeffouten)

When researchers investigate a population based on a sample, it is inevitable that errors are made, i.e. results of measurements and analyses within the sample will not correspond to the true value over the entire population.

The way we take a sample can influence the results.

sampling error

We discern between random or systematic sampling errors:

random sampling error

Random sampling errors are only due to chance and cannot be avoided. Every time you take a random sample, you will get a different result within that sample.

systematic sampling error

A systematic sampling error is worse, because it can cause the results to be biased and therefore no longer be representative for the population. Examples of systematic sampling errors:

• In an online survey, only people with internet access are selected. So this sample is no longer random.
• If you conduct a survey in which people participate voluntarily, you have a greater chance that people who are also interested in the subject participate. This can also have an impact on the results.

non-sampling error

After taking the sample, errors can also be made in the measurement and analysis. In this case we speak of non-sampling errors and again we distinguish between random and systematic:

random non-sampling error

For example, a respondent accidentally selects or fills out a wrong answer in a survey.

systematic non-sampling error

Examples:

• a measuring device is not calibrated correctly and systematically gives a slightly higher value than ther actual one
• respondents intentionally underestimate or overestimate reality (e.g. when asked about how much you smoke or drink alcohol on a daily basis)