Sampling

SAMPLE

A sample is a subset of a POPULATION

“A population is a group of potential participants to whom you want to generalize the results of a study. And generalizability is the name of the game; only when the results can be generalized from a sample to a population do the results of the research have meaning beyond the limited setting in which they were originally obtained.” (Salkind, 1996, pp. 85-86)

Generalisability is typically assessed by looking at the external validity of the study (please refer to Lecture 7). It is important to keep in mind though that the way in which you select the sample (the sampling strategy or type of sampling that you use) will affect the generalisability of the study!

TYPES OF SAMPLING (SAMPLING STRATEGIES)

NON-PROBABILITY SAMPLING

-      selecting a sample from a population in a way which is NOT RANDOM i.e. not every element in the population has an equal, non-zero probability (chance) of being selected

-      Advantage: convenient and economical

-      Disadvantage: No way to estimate the probability of each element being included in the sample, and no guarantee that each element has some chance of being included

-      E.g. interviewing the 1^st 30 students in the Matrix on a Monday – those with no classes = no chance.

Types of non-probability sampling strategies:

Convenience samples

-      Availability and willingness to respond are the selection criteria for the sample

-      Includes volunteer sampling

-      Includes snowball sampling:

Ø Appropriate when members of a special population are difficult to locate

Ø Ask members of target population to provide information to locate other members of the same population they happen to know

Quota samples

-     The researcher first identifies categories of people e.g. male/female and then decides how many people to include in the sample from each of these categories.

-     Advantage: the researcher can ensure that population differences are accounted for

-     Disadvantage: once categories are fixed, choice of persons to fill these categories is still haphazard, thus misrepresentation and researcher bias in choice of sample may occur.

Purposive samples

-     The researcher handpicks the elements to be included in the sample on the basis of expert judgement

-     Sample consists of those who have certain desired characteristics or who are likely to provide useful information for the study being done

-     E.g. – market research, voter trend samples etc…

PROBABILITY SAMPLING

-       selecting a sample from a population by means of RANDOM sampling

-      RANDOM SELECTION / RANDOM SAMPLING: A selection procedure in which every element in the population has a known non-zero probability of being chosen for the sample i.e. selecting a sample from the whole population in such a way that the characteristics of each of the units of the sample approximates the characteristics of the total population.

-     Advantage: reduces bias

-     Disadvantage: Not always practical, not necessarily time and cost efficient

-     E.g. a list of all students registered is obtained from the dean’s office and the participants’ names are picked randomly*

Types of Probability Sampling Strategies

Simple random sampling

-      Basic technique, used if relatively homogenous population

-      Strategy in which each possible sample of a specified size in a defined population has an equal chance of being chosen.

-      Normal procedure: a sampling frame is established (a list of elements in the population from which the sample is drawn). Each element in the population (sampling frame) is numbered, and the required sample size is decided. A table of random numbers is then used to determine each element in the sample.

-      Seldom used in practice – labourious and inefficient process

           

Systematic sampling

-     Used in preference to simple random sampling if a sampling frame (list of all elements in population) is available, and the population is relatively homogenous in character

-     The sample size required is divided into the size of the sampling frame, to yield a value k. A table of random numbers is then used to select only the first element in the sample, thereafter every k^th element is then included.

Stratified random sampling

-     Used when a sampling frame is available, but the population does not appear to be relatively homogenous.

-     Instead the sample is considered in terms of sub-populations or strata (which are relatively homogenous). Each stratum is defined, and a separate sampling frame for each is constructed. Random samples are then drawn from each stratum.

-     There are two commonly used methods for determining the number of subjects selected from each stratum: drawing equally-sized samples from each or draw samples on a proportional basis (i.e. representative of the proportion of the entire population that the stratum represents).

Cluster sampling

-     If a sampling frame is difficult or impossible to develop, often an aggregate or cluster of elements is used as a sampling unit in which each cluster stands an equal chance of being included in the sample i.e. cluster sampling

-     Problematic as it may introduce particular biases into the research process, e.g. not all the clusters may be equivalent, which affects the validity of the study.

·       SAMPLING WITH REPLACEMENT

·       SAMPLING WITHOUT REPLACEMENT