Purposeful (or purposive) sampling
The sampling strategies of the qualitative researcher are guided by principles of ethics and the opportunity of gaining access to people whom they can observe and interview in-depth, and from whom they can obtain rich data. The selection of participants (settings, or units of time) is criterion-based, that is, certain criteria are applied, and the sample is chosen accordingly. Sampling units are selected for a specific purpose on which the researcher decides, therefore the term ‘purposive’ or ‘purposeful’ sampling is used. For instance, the researcher chooses a sample on the basis of group membership, on the basis of the experiences that participants have had or the type of treatment and care that they were given.
The group is specified in advance. Some researchers use the term ‘criterion-based’ sampling (Schensul et al., 1999; Endacott and Botti, 2004), because most sampling strategies, even random or theoretical sampling, are highly purposive. However, purposeful or purposive is the term used by most qualitative researchers, and it is based on the judgement of the researcher. Purposeful sampling can also include the site or setting of the research.
At the start of the research, researchers must ask two questions: what to sample and how to sample. People generally form the main sampling units. The appropriate informant is chosen by the researcher or may be self-selected. Sometimes researchers can easily identify individuals or groups with special knowledge of a topic, occasionally they advertise or ask for informants who have insight into a particular situation or are experts in an area of knowledge. These voluntary participants selected for the research are often those that are most articulate because the researchers find it easier to communicate with them and elicit rich data, but this might lead to a neglect of certain individuals that are powerless or inarticulate and who should be included; indeed they might be very important as their voices are often marginalised.
Individuals are sampled for the information they can provide about a specific phenomenon, be it a condition, such as an illness, a treatment (for instance a particular medicine, manipulation, counselling), a type of care, professional decision-making, etc. They could be nurses who have cared for people undergoing treatment, patients who have had day surgery or midwifery students who are interviewed about their clinical experience and so on. Identification of a particular population provides boundaries between those who are included in the study and those who stay outside it (inclusion and exclusion criteria). The members of the sample share certain characteristics. The sample is thus chosen on the basis of personal knowledge of the person selected about the phenomenon under study.
Useful informants would be people who have had experiences about which the researcher wants to gain information. For example, individuals who have diabetes might share experiences and the meanings that these have for them with the health researcher.
Informants with special knowledge or experience might consist of newcomers, people who are changing status, or those who have been in the setting for a long time. Individuals who are willing to talk about their experience and perceptions are often those persons who have a special approach to their work. Some have power or status; others are naïve, frustrated, hostile or attention seeking, although researchers must remember that the latter are not always the best informants because they may have a mainly negative perception of the organisation or institution under discussion – ‘an axe to grind’. Ethically it is important that the persons in the sample are not jeopardised by ‘confessing’ to their practices (unless illegal) and uncovering their thoughts.
As in all research, the researcher needs to clarify the rationale for inclusion and exclusion of particular people or other sampling units.
Example of purposive sampling
Bisson et al. (2009) give an account of their qualitative research which aimed to gain the view of a variety of sufferers of Huntington’s disease on decision-making on which they develop a care pathway for future decisions and powers of attorney. They used purposive sampling in order to gain a range of perspectives from various individuals such as sufferers, people with the gene, carers and clinicians in the field. They also included a lawyer, medical ethicists and advisors from the Huntington’s Disease Association. This sample was chosen to gain a full range of perspectives from individuals. Theoretical sampling was also used to collect the views of both males and females, old and young participants.
Sampling types
There are various forms of sampling. We shall discuss only the most often used and important types. An overview of a whole range can be found in Patton (2002) and Kuzel (1999), although many sampling types overlap. The commonest methods are as follows:
- Homogeneous sampling
- Heterogeneous sampling
- Total population sampling
- Chain referral sampling (snowball sampling)
- Convenience or opportunistic sampling
- Maximum variation sampling
- Theoretical sampling
Homogeneous sampling
This involves individuals who belong to the same subculture or have similar characteristics. Nurses often use homogeneous sample units when they wish to observe or interview a particular group, for instance specialist nurses. Midwives may wish to examine the perspectives of community midwives on their role in the community. In these examples, a homogeneous group is being studied. The sample can be homogeneous with respect to a certain variable only – for instance, specific occupation, length of experience, type of experience, age or gender. The important variable would be established before the sampling starts.
Example of homogeneous sampling
Examples of homogeneous sampling would consist, for instance, of a group of adolescent schoolgirls between the ages of 13 and 15 who are being interviewed about a topic that is of importance to them, or a number of orthopaedic surgeons who have used a Taylor Spatial Frame. For the purpose of the specific studies, they would be homogeneous samples.
Heterogeneous sampling
A heterogeneous sample contains individuals or groups of individuals who differ from each other in a major aspect. For instance, nurses may wish to explore the perceptions of nurses, social workers and doctors who care for patients with HIV. The three groups form a heterogeneous sample. Heterogeneous sampling is also called maximum variation sampling (Patton, 2002) because it involves a search for individuals with widely differing experiences and for variations in settings.
Example for heterogeneous sampling
Researchers might wish to explore the perspective of people with a chronic illness and the ways they choose strategies for managing their condition. The heterogeneous sample might comprise males and females across a broad range of ages with different jobs and from a variety of different backgrounds. This sample would be chosen to maximise contrasts between the participants.
The sample might consist of people from a naturally occurring population – such as members of a local carers’ group, a specific ward, a community of patients. Some sampling is based on early findings with a group and cannot be determined prior to the study. For instance, a midwife could sample women who have just given birth to their first child and find that it would be interesting to select older and younger primiparae because they might have different ideas about childbirth. Sometimes married couples are chosen as samples or people who live together. Occasionally the sample consists of focus groups, for instance self-help groups, or groups with similar conditions or experiences.
Total population sampling
A sample is called a total population sample when all participants selected come from a particular group; it is used infrequently in qualitative research. For instance, all the nurses with specific knowledge or a skill, such as those with training and experience in counselling, might be interviewed because the researcher focuses on this skill, and there might be few available with the particular expertise. There are some diseases where those who suffer from them are very small in number, and the researcher might interview all of these. All midwives in one midwifery unit might be observed, because the specific setting in which they work or the special techniques they adopt are seen as important. Not many qualitative studies carry out total population sampling.
Chain referral or snowball sampling
A variation of purposive sampling is chain referral or snowball sampling (the former is a term originally coined by Biernacki and Waldorf (1981)). A previously chosen informant is asked to identify other potential participants with knowledge of a particular area or topic, and these in turn nominate other individuals for the research. Researchers use snowball sampling in studies where they cannot identify useful informants, where informants are not easily accessible or where anonymity is desirable, for instance in studies about drug addiction or alcohol use. Penrod et al. (2003) suggest that chain referral sampling is useful in situations where people are vulnerable and when they are not easily accessible: this might include groups who are labelled negatively by society (for instance, those that suffer from sexually transmitted diseases), those with whom researchers discuss sensitive topics (such as sexual behaviour) or those individuals who fear being exposed or criminalised (i.e. substance users).
Example of chain referral sampling
A sample of Lesbian couples were interviewed by Spidsberg (2007) about maternity care. After initial recruitment through sending leaflets, snowballing took place through recruitment by women already interviewed through word of mouth information to friends who then volunteered to be interviewed.
Convenience or opportunistic sampling
The terms convenience or opportunistic sampling are self-explanatory. The researcher uses opportunities to ask people who might be useful for the study and easy to access. To some extent, of course, most sampling is opportunistic and arranged for the convenience of the researcher. Researchers usually adopt this sampling strategy when recruiting people is difficult, though this is not the best way of sampling.
The researcher chooses individuals whose ideas or experiences will help achieve the aim of the research; occasionally variations in the sample have no specific influence on the phenomenon to be explored, and in this case a convenience sample can be selected.
Example of convenience sampling
Rodham et al. (2006) studied risk behaviours in adolescence. As a convenience sample for easy accessibility in their locality, they selected four schools from in the Bath area and asked for volunteers over 16 of the school population to participate in the study.
Another convenience sample might consist of all midwives who work in a particular hospital because the researcher has easy access to them and they fit other criteria specified for the research.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
