Sample

A “sample” in a program evaluation usually consists of all of the projects sponsored within the program or of the target groups of different projects (e.g. the students who utilise the products developed as part of the projects).

The aim should be to collect data from all these elements (projects, members of target groups). However, if only a selection can be surveyed, , the following is an explanation  of the most important concepts surrounding the samples (Ilmes, 2007):

Population

A population is the entire group of elements to which the conclusions of an investigation should apply (e.g. all of the students on a course, all of the interactions documented using log-files between students and tutors, etc.). When all the elements of a population are investigated, it is called a census. If the population is the students of a course, then a census can normally be easily performed.

Sample

If a census is not possible (e.g. if the population is too large), a selection out of the population („sample“) can be done. When only these selected units are included in the investigation, it is called a partial survey.

Sampling methods

There are different methods for the selection of elements:

Random sampling: In a random sample, the population should be known and defined exactly. Each member should be represented in the population uniquely and thus have an equal or calculable probability of selection.

• Simple random sampling: Each member of the population has an equal chance of being selected. Example: a random number generator is used to make a selection from a list of the students enrolled in a course. The resulting simple random sampling allows the maximum possible degree of representativeness, i. e. the sample reflects the composition of the population.
  
• Complex sampling: Each member of the population does not have an equal chance of selection but a calculable one. One example for this kind of sampling is the stratified selection: Here, the population is divided into different “strata“ which are homogeneous in respect of the variable of interest such as students of medicine, of sociology and information technology, with each member belonging to exactly one stratum. Elements are selected from each stratum by random selection. The resulting stratified sampling has the advantage that elements from “small” strata are adequately represented in the sample.

Non-probability sampling: In non-probability sampling, elements are selected according to certain characteristics. Non-probability sampling techniques cannot be used to derive conclusions about the general population from the sample  by applying statistical models.

• Typical case sampling: In typical case sampling, elements are selected which are considered characteristic of the population.
• Quota sampling: In quota sampling, the distribution of certain characteristics within the sample corresponds exactly to the distribution of these characteristics in the population. This requires that the distribution of these characteristics is known.
  

[1] [2]