Sample

Non-response rate

Even when the aim is to complete a census or if a random selection is made,  the sample can be biased because, for example, some of the selected people might not fill in the questionnaire, or they might refuse to take part in an interview, or they might not be available, etc. The non-response is often systematic, e.g. the people who do not participate in a survey about a course may be the ones who did not like the course. In such a way, the results might be given a more positive bias. Thus, it is necessary to examine who the non-responders are and to what extent the non-response rate influences the results. If necessary, a further survey should be conducted.

Sample size

In determining the sample size (i.e. the number of elements in the sample) you should consider the following:

• The probability with which the value found in the sample corresponds to that of the population increases as the sample size increases. 
  
• The more diverse the distribution of the variable of interest  within the population, the larger the sample required.
  
• Reliable judgments of sample size can only be made if a hypothesis-testing survey with a predetermined effect size is planned.

Further information about sample can be found at Wottawa & Thierau (1998, p. 114-143), Rossi, Lipsey & Freeman (2004, p. 31-65), Weiss (1998, p. 87-96), Sapsford & Jupp (1996), Gabler & Hoffmeyer-Zlotnik (1997), Bortz & Döring (2003).

[1] [2]