Sampling is a process that allows a researcher to scientifically choose who or what is included in an investigation. According to Trochim (2001:41), sampling involves selecting units (e.g. people, organizations) from a population of interest so that one may fairly generalize the results of a study back onto the population from which they were chosen. Sampling is necessary because surveying every person or a whole set of units in a population is often impossible or, at the very least, very costly in terms of time and money. Moreover, since research requires reliable forms of evidence from which to draw robust conclusions, samples help by enabling the detailed examination of a sizeable group or case to take place.
3.5.1. Sampling methods
Sampling can be divided into two types, namely probability or non-probability sampling (Sarantakos, 1997; Neuman, 2006; Flick, 2006; Best and Kahn, 2006). Some scholars also dichotomise them as random or non-random methods (Australian Bureau of Statistics, 2004). The direction taken when sampling or deciding whether to use random or non-random sampling techniques, quantitative or qualitative approaches, is influenced by the research methodology adopted or followed. In a quantitative study, the intention is often to select samples randomly to achieve representative cases. In contrast, qualitative studies often select samples for their usefulness as opposed to their randomness (Best and Kahn, 2006:249). As already mentioned, this study used both quantitative and qualitative methodologies. Non-probability sampling techniques, quota sampling and convenience sampling were used to ensure that certain characteristics of the population were captured without sacrificing representation.
The survey population was categorized according to faculties using quota sampling.
Trochim (2001:57) states that in quota sampling, one selects elements non-randomly according to some fixed quota. Neuman (2006:221) confirms that in quota sampling, the researcher identifies categories into which cases or people will be selected and selects cases to reach a predetermined number in each category. The basic goal of this sampling process, according to Tull and Albaum (1973:38), is to develop a sample that is a replica
of the population of interest. In order to accomplish this, the population is broken down on the basis of key demographic characteristics. A sample is then developed that is proportional to these characteristics, as explained above. Neuman further states that quota sampling ensures that some differences exist in the sample, e.g. age, sex or background.
Literature indicates that undergraduate and postgraduate students often exhibit different information seeking behaviour, and this behaviour also differs according to faculty. This was what prompted the use of quota sampling, with faculties identified as quotas. There are two types of quota sampling, namely proportional and non-proportional. Proportional sampling is when sampling numbers must match the proportions in the population, while non-proportional sampling is less restrictive - one simply has to have enough cases or people to ensure that he/she can talk about even small groups in the population. The target population was selected proportionally at institutional population levels as well as at study levels. Thus, there were more students and staff selected from DUT than from the University of Zululand, and more undergraduate than postgraduate students in the population in line with the population distribution at the two institutions.
For various reasons, it is sometimes difficult to gain access to the individuals one intends to interview or distribute questionnaires to within a limited time frame. Some may be busy, others may lack interest, and others may not be reached for a horde of other reasons. Because of this, convenience sampling was used to select respondents for each quota. Fraenkel and Wallen (2000:112) describe a convenience sample as a group of individuals who are conveniently available for a study. Tull and Albaum (1973:38) state that convenience sampling refers to samples selected not through judgment or probability techniques, but because the elements in a fraction of a population can be reached conveniently. The established quotas guided the researcher in selecting the respondents.
In distributing the questionnaire, the researcher took steps to verify which faculty students or members of staff belonged to. Although many related studies show that departments are an important consideration when investigating behaviour among academics and students, this study did not endeavour to examine behaviour according to the specific and manifold departments; instead, the intention was to have as many
departments and course offerings represented as possible. Convenience sampling also proved handy in selecting academic members of staff, the majority of whom, by nature of their jobs or positions, were busy and had less time on their hands to dedicate to this research. For this reason, academics were selected according to their rank whenever this was possible, although not strictly.
3.5.2. Sample size
Chair et al. (n.d:3) stress that there is no simple rule governing what sample size should be used for all surveys. Fraenkel and Wallen (2000:116) agree that the question remains open as to what constitutes an adequate or sufficient size for a sample. Several things need to be considered in deciding how many cases to take in a sample. Key among them is cost, the budget available, the objectives of the study, and the number of cases needed to detect differences of interest. Analysts usually find that a moderate sample size is sufficient for most needs. Nachmias and Nachmias (1996:194) claim that any increase in the sample will increase the precision of the results. This concurs with Seaberg’s assertions (in Grinnell, 1994:253) that the general rule of thumb is the bigger the sample, the better. There are concerns as well about the level of accuracy expected.
In deciding on the sample size, all the above mentioned issues were considered. Research design also featured significantly, given Neuman’s (2006:149) observation that qualitative research is less concerned with issues of sample size and more with richness, texture and feeling. In other words, qualitative research focuses less on how representative a sample is and more on how it illuminates social life (Neuman, 2006:219).
As noted by Seaberg (in Grinnell, 1994:254), a fairly common problem relating to sample size is the failure to consider the number of categories of the sample which may be required to analyze the data appropriately. In this study, this problem was addressed by disproportionate sampling within the predetermined quotas.
Gay and Aisarian’s (in Leedy and Omrod, 2005:207) guidelines for selecting a sample were used to select the sample size. They state that if a population is beyond a certain point (at least 5,000 units or more), the entire population size is almost irrelevant and a
sample of 400 should be adequate. To reach this figure, the student sample was calculated at a ratio of 1.2 percent of the total population in each institution, while the staff sample was calculated at 4 percent. A few individuals were randomly selected from these samples for interviews.
3.5.3. Sample frame
A sample frame is defined by Neuman (2006:225) as a list of cases in a population or the best approximation of [a given population]. The table below presents the sample frames used in this study.
Table 3.1: Sampling Frame
University of Zululand student population (8613)
Questionnaires Interviews
Faculty Gender Level of
study
Sample size
Gender Level of study
Sample size
M F UG PG M F UG PG -
Arts 13 13 20 6 26 - - - - -
Education 13 14 20 7 27 - - - - -
Science & Agriculture 12 13 20 5 25 - - - - -
Commerce, Administration
& Law
12 13 20 5 25 - - - - -
Total - - - - 103 - - - - 10
Durban University of Technology student population (21316)
Questionnaires Interviews
Faculty Gender Level of
study
Sample size
Gender Level of study
Sample size
M F UG PG M F UG PG -
Accounting & Informatics (Commerce)
32 32 56 8 64 - - - - -
Arts 32 32 56 8 64 - - - - -
Engineering Sciences &
the Built Environment
32 32 56 8 64 - - - - -
Health Sciences 32 32 56 8 64 - - - - -
Total 256 - - - - 15
University of Zululand staff population (281)
Questionnaires Interviews
Faculty Gender Rank/position Sample
size
Rank/position Sample size Sample
chosen by availability
No strict adherence to position or rank
No strict adherence to position or rank
6 members of staff were targeted for interviews
Arts 3
Education 3
Science & Agriculture 3
Commerce,
Administration & Law
2
Total 11 6
Durban University of Technology staff population (595)
Questionnaires Interviews
Faculty Gender Rank/position Sample
size
Rank/position Sample size Sample
chosen by availability
No strict adherence to position or rank
No strict adherence to position or rank Accounting &
Informatics (Commerce)
6 12
members of staff were targeted for
Arts 6
Engineering Sciences &
the Built Environment
6
Health Sciences 6
interviews
Total 24 12
3.5.4. Responses
The responses to the questionnaires and interviews at the University of Zululand and Durban University of Technology are summarized in the table below. Guided by the sample frame, the study sought to sample respondents and interviewees in accordance with the numbers set in the sample frame. To a large extent, this was successfully achieved. In sampling members of staff from both institutions, the most accessible sample was often used because of the failure of certain members of staff to respond to the questionnaires or sit through interviews because of their busy schedules. On the whole, however, the response rate was very high.
Table 3.2: Questionnaire responses by institutional affiliation
University of Zululand Students target
sample
Response Rate
Staff target sample
Response rate
Overall target sample
Overall response rate
103 84 (82 %) 11 9 (82 %) 114 93 (82 %)
Durban University of Technology Students target
sample
Response Rate
Staff target sample
Response rate
Overall target sample
Overall response rate
256 139 (54%) 24 14 (58%) 280 153 (55%)
Table 3.3: Interview responses by institutional affiliation
University of Zululand Students
target sample
Response Rate
Staff target sample
Response rate
Overall target sample
Overall response rate
10 10 (100%) 6 5 (83%) 16 15 (94 %)
Durban University of Technology Students
target sample
Response Rate
Staff target sample
Response rate
Overall target sample
Overall response rate
15 8 (53%) 12 4 (33%) 27 12 (44%)