4.4 Data Collection
4.4.2 Sample Selection
A sample is a group within a population that is observed to make inferences about the nature or behaviour of the entire population (Rubin and Babbie, 2009). In qualitative research, it is not necessary to collect data from all the respondents in the target community even if it were possible. Rather, the research objectives and the characteristics of the study population such as size, age, experience and gender should condition the number of respondents to select and how to select them (Hair et al., 2010). Patton (2002) posits that it is advantageous for sampling methods to allow for the identification of a subset of persons with diverse experiences. The following sub-sections explain the possible sample selection methods in qualitative research.
Convenience Sampling
Also known as accidental or haphazard sampling, convenience sampling is when respondents are selected for their availability and willingness to participate in a study (Gravetter and Forzano, 2015). It is typical for studies in health facilities or research involving drug use to rely on convenience sampling. The advantage of convenience sampling is its ease of application and inexpensiveness in attracting respondents (Hair et al., 2015). In spite of this significant advantage, convenience sampling has been adjudged to be a biased method because subsets of the research population could be over or under selected, or outrightly omitted. It is argued that this leads to a high
135 probability that the resulting sample is not representative of the general population (Gravetter and Forzano, 2015).
Purposive Sampling
Also known as the purposeful sampling strategy, researchers engaged in this method intentionally select the respondents that will best contribute to the study or, as Patton (2015) states, selecting information-rich and illuminative cases for qualitative inquiry. Polit and Beck (2014) affirm that this intentional selection is based on the information needs that have emerged from early findings. They add that other strategies exist within purposive sampling to meet the various conceptual needs of qualitative research. These other strategies are maximum variation sampling, extreme (deviant) case sampling, typical case sampling and criterion sampling.
Maximum Variation Sampling
The purpose of maximum variation sampling is to capture the diversity of a phenomenon within a small sample that is to be intensively studied (Rubin and Babbie, 2009). The benefit of maximum variation sampling is that more useful insights are generated by observing phenomena in heterogeneous conditions because they cut across a great deal of variation (Patton, 2015). Problems can arise in the analysis of data if a great deal of heterogeneity is found in a small sample because the inputs are very different from one another. However, Patton (2015) believes that, in the midst of great variations, the common patterns that emerge are particularly valuable for the purpose of capturing the core experiences and shared dimensions of a phenomenon. In the same way, Martella et al. (2013) explain that the purpose of maximum variation
136 sampling is to obtain detailed descriptions of each case or from each respondent and to document the unique and general patterns shared by these respondents.
Extreme (Deviant) Case Sampling
Researchers apply the extreme case sampling technique when there is a conceptual opportunity to answer research questions or meet research objectives from outliers. Martella et al. (2013) discuss this type of sampling as the selection of those cases that are the most outstanding successes or failures related to some topic of interest. Thus the purpose of extreme case sampling is to explore further insights in outlying phenomena.
Typical Case Sampling
Typical case sampling involves selecting participants that provide average or normal responses (Polit and Beck, 2014). Typical cases are straightforward in providing insights into usual occurrences. This sampling technique is useful for completing qualitative profiles of ‘typical’ cases that can be selected through demographic analyses or statistical data that show a normal distribution of attributes that allude to average conditions. The purpose of typical case sampling is to describe the commonalities in cases and a setting and not to make generalised conclusions. However, Patton (2015) cautions that typical case sampling is only an illustrative method and not a definitive process.
137 Criterion Sampling
Users of criterion sampling aim to satisfy a predetermined criterion of importance (Polit and Beck, 2014). Only cases or respondents who meet or surpass the set criterion are included in the sample (Martella et al., 2013). It is argued that during their analysis, researchers applying criterion sampling can explicitly or implicitly compare criterion cases with those that do not satisfy those criteria. In addition, the benefit of criterion sampling is that it facilitates the identification of cases that could be followed-up for further exploration (Patton, 2015).
Quota Sampling
As a sampling technique, quota sampling fills important categories in the larger population with a predetermined number of respondents (Patton, 2015). According to Martella et al. (2013), quota sampling is key to deriving a sample from a heterogeneous population for which researchers have no exhaustive list. The benefit of quota sampling is that subgroups or categories within the population are duly represented (Gravetter and Forzano, 2015). Similarly, it allows for high representation if the population can be subdivided into one or more variables that make up the known proportions of the population and if relationships can be maintained within samples taken from each subdivision. Martella et al. (2013) explain further that quota sampling involves the following three steps:
1. identifying the important categories that are present in the whole population and determining the proportions of the population that fall into the important categories identified
2. determining the sample size and the associated data for each important category
138 3. selecting the members of each category using non-probabilistic
measures like opportunity and volunteer sampling
The benefit of quota sampling is that it allows the researcher to control the composition of the sample with categories that meet set criteria (Gravetter and Forzano, 2015). In addition, Patton (2015) explains that quota sampling ensures that important categories are included in the study whether they are demographic, geographic or theoretical. Quota sampling also facilitates budgeting and logistical calculations because its design specifies the quota for all categories. Furthermore, an overriding advantage of quota sampling is that it can be flexible to allow researchers adjust the size and composition of quotas as the study develops and to make comparisons between different categories (Patton, 2015).