Social Research Glossary   A B   Citation reference: Harvey, L., 2012-18, Social Research Glossary, Quality Research International, http://www.qualityresearchinternational.com/socialresearch/ This is a dynamic glossary and the author would welcome any e-mail suggestions for additions or amendments. Page updated 24 January, 2018 , © Lee Harvey 2012–2018. A fast-paced novel of conjecture and surprises

_________________________________________________________________

Sampling

core definition

Sampling is the process of selecting a number of items or individuals from a wider group or population (via the sampling frame).

explanatory context

Introduction

Sampling is widely used in the social sciences as it is rarely feasible to survey or enquire about entire populations, however narrowly defined it may be.

Sampling is essential in order to make social scientific analysis manageable. It is also much more efficient. A small sample can provide relatively 'accurate' data on large populations. Fifteen hundred people in an opinion poll can give a good indicator of how 20 million people at a general election will vote.

Selecting samples raises problems of analysis for researchers because the selection process may be such that the sample is not representative of the population, which can lead to bias. In all cases, there will be sampling error when samples are used. Statistical analysis can take account of sampling error but cannot account for bias.

Sampling frame

The sampling frame is the listing of the items that make up the population from which the sample is to be drawn. The existence of a sampling frame is necessary for most random sampling (except area sampling).

If a sampling frame is incomplete or out of date (and most are) then bias can result.

The absence of a sampling frame is probably the main (but not only) reason for using non-random sampling techniques.

Random samples

Introduction and Definition

Random samples are those where the probability of any member of the population being selected for a sample is known. This is sometimes referred to as scientific sampling.

In practice, the application of random sampling, tends to be on the basis os an equal probability for each member of the population.

Random sampling is therefore systematic and quite different from haphazard sampling.

There are a number of different ways of drawing a random sample.

Simple sandom samples

Simple random sampling is eqivalent to puting all the names of the subject group in a container and then drawing out the required number for the sample. This is usually done without replacment to avoid the same person appearing in the sample more than once.

In practice, simple random sampling is done by selecting from a list using random numbers. Computers are often used for this task.

Systematic random samples

Systematic random samples are like simple random samples but instead of the sample being selected from a list by using randomly generated numbers a random start point is selected and then every nth item on the list is included in the sample. E.g. if the sample contains 1000 names and a sample of 100 is required, then a random start point is selected by generating a random number between 1 and 1000 and every 10th item is taken from the list.

There is a possibility that this process might give an unrepresentative sample if the list is ordered in a particular way. For example, if the list was an address list in house number order then every tenth house might generate only even numbers which would (in Britain) lead to the houses selected usully being all on one side of a street, and it is possible that this might generate a sample with an unrepresentative housing class.

Cluster samples

In cluster sampling the population is viewed as being divided into groups (e.g. schoolchildren into school classes, bank workers into bank branches etc.) and these groups are then sampled. This method may often save time and money but at the expense of statistical precision.

In single-stage cluster sampling the process stops wih the selction of the whole group.

In multi-stage cluster sampling further sub-samples may be selected from within the whole groups, e.g. if we wanted to select a national sample of doctors we might (a) sample area health boards (b) sample practices within these boards.

Multi-stage random samples

Multi-stage random sampling is a means to select a sample covering a large and dispersed population without having to contact people in scattered locations. This is important from a resource point of view if, for example, a national sample is being selected for interview.

Multi-stage random sampling works by selecting in stages. For, example a sample of 1000 voters in Britain might be chosen by first selecting 20 constituencies at random from the list of all Britaish constituencies and then choosing 2 wards at random within each constituency, and then selecting twenty five voters from the electoral register of each ward. The sample will be selected at random (i.e. the probability of selection for each individual would be knowable in theory) but the location of the interviewees would be more concentrated than if a sample of 1000 people had been selected by simple random sampling techniques from the electoral register for Great Britain.

In practice, multi-stage random sampling tends to bring in an element of bias as it is rare that the different probabilities of selection for individuals are taken into account, usually it is assumed that each member has an equal chance of being in the sample. In the above example this would not be the case if, for example, a constituency of 100000 voters and one of 50000 were selected at stage one. The probability of any individual from the first constituency being in the sample would be half that of a voter in the second constituency.

Stratified random samples

Stratified random sampling occurs when the population is ordered into strata, on some basis, prior to selection using one of the random techniques. The criteria used to select the strata depends on the purpose of the investigation. If a gender comparison is being undertaken, the population under investigation would be split into male and female and the sample selected so that the population proportions are reflected in the sample.

For example, if the population consisted of doctors and the sampling frame showed that twice as many doctors were males than females then if a sample of three hundred was required, a hundred females would be selected from the female doctor list and two hundred males from the male doctor list.

Alternatively, (and less satisfactorily) equal numbers might be selected from each list in the above example but the males given a weighting twice that of females. While this overcomes the different probabilities of selection of males and females it reduces the potential variablity of the male sample.

Stratified random sampling is often combined with multi-stage sampling in practice. For example, when selecting a sample of voters, opinion pollsters rank order British constituencies according to the percentage Conservative vote at the last general election (thus stratifying the population on the basis of political opinion at the constituency level) and then use systematic random sampling to choose the constituencies and then random sampling to select wards and voters.

Area samples

Area sampling is an attempt at random sampling when a sampling frame is not available. A geographic area is selected for the study (sometimes at random, but more often the study is directed to a particular area) and the dwellings in it are selected in some random fashion, such as calling at every nth house in each street in the area. The residents may then make up the sample or there may be a further screening if the research is only interested in certain categories of respondents (for example, people over 75 years of age).

Non-Random Samples

Introduction

A non-random (or non-probability) sample is any sample where it is not possible to say that all the members of the relevant population had a known probability of being selected at the outset.

Snowball samples, quota samples, volunteer samples are all types of non-random samples.

Non-random samples are also sometimes known as purposive samples. Non-random sampling, such as convenience sampling, is therefore almost always biased.

One possible exception, in practice, is the use of quota sampling. This method does not give every member of the population an equal (or known) chance of being in the sample but it attempts representativeness by selecting respondents, from those available, in proportion to a predetermined quota that reflects data already known about the population (usually demographic data such as age, gender etc.).

Convenience samples

Convenience sampling is the crudest form of sampling, in which anyone who is convenient is included in the sample. Samples of this sort might be used in the pre-pilot stage of a research project but are of little use otherwise.

Quota samples

A quota sample is one where the population is divided into sub-groups (e.g. social classes) and the interviewers have to find a quota of people from each sub-group so that the sub-groups are in the same proportion in the sample as they are in the population.

Sometimes quotas are decided arbitrarily without a full knowledge of the population demographic data (e.g. a survey comparing male and female, old and young may decide on a quota frame which specifies equal numbers of males and females above and below a specified age limit, such as 40).

Snowball samples

Snowball samples occur when the researcher makes contact with a suitable subject and is then directed to, or makes contact with other members in a network of contacts in a kind of 'chain letter' style.

Volunteer samples

A volunteer sample is one that is recruited using volunteers. They are self-selecting, being a subgroup of a population who are prepared to be involved in the research. Volunteer samples are often recruited through advertisements.

Purposive sampling

Theoretical sampling

analytical review

Mann and Richards (undated) describe 'sample' as follows

Since no investigation can have everything as its subject, the researcher needs to select. In hypothetico-deductive research, the selection of a representative sample is regarded as particularly important, whereas qualitative researchers recognise a range of possibilities. Here, more effort is expended on understanding the ‘sample’ which is the subject of study than in establishing that it is in some way ‘representative’.

Elwell's Glossary of Sociology (undated) defines sampling as:

Taking a small representative part of a population for purposes of drawing inferences from the analysis of the sample characteristics to the population as a whole.

associated issues

related areas

sampling error

statistics

Researching the Real World Section 8.3.9

Sources

Elwell's Glossary of Sociology, undated, available at http://campus.murraystate.edu/academic/faculty/frank.elwell/prob3/glossary/socgloss.htm, page not available 20 December 2016.
Mann, S. and Richards, K, undated, Research Methods: Introduction to Qualitative Research
, available at http://www2.warwick.ac.uk/fac/soc/al/degrees/ma/core/research_methodology/ma_introduction_to_qualitative_research_sm__kr.pdf, accessed 24 June 2013, page not available 20 December 2016.

copyright Lee Harvey 2012–2018

A NOVEL Who bombed a Birmingham mosque?
Top

A B