Welman and Kruger (2005) refer to sampling as a subset of a population. The reason why we should sample the population is that the size of population may in some cases make it impractical and uneconomical to involve all the members of the population. We distinguish between probability sampling and non-probability sampling.
Probability sampling refers to the possibility that any member of the population has an equal chance to be sampled. Probability sampling consists of the following examples: simple random, stratified random, systematic and cluster sampling. (Welman & Kruger, 2005 and Leedy & Ormrod, 2015) Simple random sampling applies when each member of the population has an equal chance to be sampled. Stratified random sampling can be opted for when the population consists of different sub populations, for example, different management positions in an organisation. Systematic sampling refers to when one takes a number, for example every 10 on a list of 100 of the entire population. The sample will be then every tenth member on the list. Cluster sampling can be applied when we have a large population. One should first draw a sample (cluster) of the entire population.
The non-probability sampling implies that all members in a population has the chance to be included in this sample. Non-probability sampling takes one of the following formats namely: accidental or incidental sampling, purposive sampling, quota sampling and snowball sampling. An accidental sampling method can be employed when some members of the population are readily available and nearby while a purposive sampling can be opted for when the researcher based on experience deliberately chooses the sample that is representative of the entire population. Quota sampling, in contrast, refers to a sampling technique where the same percentage is sampled for each stratum of the population. If the researcher approaches a few members of the entire population, then the few members approached are then used as informants to identify other members of the population and this is known as snowball sampling.
This study employed the purposive non-probability sampling to sample the council and executive management. The entire population of the middle level and lower management of the academic cadre was targeted. The aim with the sampling technique for the council was to get the views of the chairpersons and the SRC representatives on the Councils. The chairperson’s main responsibility is to ensure that these public institutions of higher learning comply with the acts that govern these institutions. The SRCs representative represents the voice of the students, the most important stakeholders at these institutions of higher learning. It was impractical and uneconomical to target the entire population of the SRC and Councils of UNAM and NUST. This corresponds with Welman and Kruger (2005).
The sample for the executive management was the Vice-Chancellor as CEO, because amongst other things they are overall responsible, on a day-to-day basis, for academic affairs and research assisted by their Pro Vice-Chancellors. These Pro Vice-Chancellors are the Pro Vice-Chancellor: Innovation and Research and Development (UNAM) and the Pro Vice-Chancellor: Academic Affairs (UNAM) and the Deputy Vice-Chancellor: Academic Affairs and Research (NUST). The purpose of this targeted sample is twofold: the Vice-Chancellors are the custodians of academic affairs and research, two of the fundamental responsibilities of their institutions. The Pro/Deputy Vice-Chancellors are the heads of research and academic affairs at their institutions. Their perception of leadership for organisational transformation can filter through to all levels in their institutions.
The entire population of the academic Deans and HODs at the two public institutions were sampled. The reason to get their perceptions on leadership as the middle and lower level management of the academic cadre was twofold. The first reason was to eliminate individual biases, in the responses of the Vice-Chancellor’s, chairpersons of councils and SRC representatives on councils, which seems to be a challenge from self-perception assessment and own performance. This self-perception corresponds with Ellaad, (2003); Walfish, McAllister, O’Donell ; Lambert, (2012). Self-perception refers to the high rating that someone may give to their own performance (Ellaad, 2003). The second reason is that the Deans and HODs may have different perceptions, based on their experience in their departments and faculties. Table 4.2 on page 176 gives a breakdown of the sample size for this study.
Table 4.2: Sampling Size for this Study
POPULATION GROUP UNAM NUST
Council (Chairperson and SRC Representative) 2 2
Executive Management: Vice-Chancellor 1 1
Executive Management: Pro Vice-Chancellor Innovation, Research and Development for UNAM and Pro Vice-Chancellor: Academic Affairs and The Pro Vice-Chancellor for Academic and Research of NUST 2 1
Deans of Faculties (excluding the Dean of Students) 8 6
Heads of Departments 69 20
Total 82 30
Grand Total= 112 (82+30)
4.6 VALIDITY AND RELIABILITY
Validity in general means that the methodology employed will produce accurate, significant and reliable results to properly address the research problem (Leedy ; Ormrod, 2015). Creswell (2014) concurs and adds that validity requires that the results obtained via the collection instruments are accurate, meaningful and reliable to address the research problem. In qualitative research the researcher needs to adhere to three types of validity, namely, internal validity, external validity and construct validity (Leedy ; Ormrod, 2015). Internal validity raises the issue if the design and the data will allow the researcher to draw truthful assumptions about cause and effect and other relationships within the data. To ensure internal validity the researcher must take all actions to avoid the possibility that different conclusions may be drawn from the data.
In order to ensure internal validity for this study the researcher opted to collect data through two instruments to strive for methodological triangulation. The data were collected through in depth semi-structured interview schedules and a semi-structured survey. External validity refers to the extent that conclusions can apply universally to other similar research problems. Strategies to ensure external validity (Leedy ; Ormrod, 2015) are: to conduct research in a real life setting, ensure a representative sample from the population and to replicate the study to a different population. To uphold validity for the qualitative component of this study, studied the only two public institutions of higher learning. The researcher sampled the entire sub population of the middle and lower management of the academic cadre to ensure external validity. The Councils’ chairpersons represent the external stakeholders (government, society and the industry) and the Vice-Chancellors and Pro Vice-Chancellors represent the academic cadre at the executive level of management. The Deans of the respective faculties represent middle level management of the academic cadre while the HODs represent the lower level of management of the academic cadre.
With quantitative research three methods of validity are differentiated. These three types are content validity, predictive/concurrent validity and construct validity (Creswell, 2014). Content validity refers to the suitability of the content of an instrument. This means, do the questions accurately measure and assess what the researcher wants to know. To uphold content analysis for this study, the questions set for the instrument should provide answers to the research questions and ultimately contribute to the development of a normative leadership model to guide organisational transformation at public institutions of higher learning in Namibia. Predictive/concurrent validity determine whether results link with other results. The third type of validity, namely construct validity, refers to whether the items measure theoretical concepts. Construct validity becomes lately the overriding objective in validity, and it has concentrated on whether the results serve a beneficial purpose and have positive impact when the results are used in practice (Hubly & Zumbo, 1996). Creswell (2014) defines reliability in quantitative research as the degree to which identical or similar results may be acquired if another similar study will be conducted, using the same methodologies on a comparable sample. Reliability refers to the trustworthiness of the data (Creswell, 2014).
Gibbs (2007) claims that qualitative validity refers to when the researcher validates for the accuracy of the findings by employing certain procedures while qualitative reliability, in contrast, checks if the researcher’s approach is consistent across different researchers and different studies. This means that the instrument should yield the same results if another instrument is used for the particular individual. To further enhance the validity the following strategies were employed (O’Cathain, 2010). This study opted to use detailed descriptions and presentation of the data collected to give the readers the option to draw their own conclusion from the data presented. The researcher is an employee at NUST and a registered student at UNAM for his PhD. The researcher also did his first degree at UNAM. Lastly the researcher consulted with experts in the field of this study on the conclusions drawn to ensure that the researcher made appropriate interpretations and valid conclusions. To uphold reliability for this study, detailed transcribing of the semi-structured interview was conducted by a professional. The transcribed interviews were validated by the researcher to the interviews recorded were transcribed verbatim. A statistician was appointed to manage the analysis of the hard copy survey questionnaires from the Pro Vice-Chancellors and the Deans as well as the electronics survey questionnaires for the HODs, via SurveyMonkey.
To achieve measurement of validity and content validity of the survey questionnaires the researcher made certain that all the questions set were aimed to get the most appropriate responses to answer the research questions. This study employed the following strategies to eliminate response bias and interviewer bias. In order to eliminate response bias during the interviews the questions set and probing were done in such a way that the interviewees did not feel they should respond negatively about their organisations. The researcher abstained from making comments, facial expression or voice tones that may lead the respondents to react negatively towards the questions asked during the semi-structured interview.
Bias refers to any condition or influence that may manipulate the data. Leedy and Ormrod (2015) identify four types of biases in descriptive research. These biases are sampling bias, instrumentation bias, response bias and research bias. Sampling bias refers to a sufficient and representative number of participants from the population who is targeted to address the research problem (Leedy & Ormrod, 2015). To avoid sampling bias all the members of the academic cadre were sampled. Instrumentation bias is when questions are structured in such a way to lead the responses in a set direction (Leedy & Ormrod, 2015).
The neutral questions set for the instruments were guided by the research questions. Response bias relates to the beliefs and values of the respondents that may impact the responses. These responses may be based on what the respondents believe is true or they believe that the researcher wants to hear (Leedy & Ormrod, 2015). To reduce response bias the respondents for the survey questionnaire had only a specific time frame to complete the survey questionnaire and the study opted for the semi-structured face to face interviews to avoid allowing participants time to think and prepare their responses in advance as can be done with a structured interview. Research bias raises the concern about the expectations, values and general beliefs of the researcher that may cause certain conclusions to be drawn or only certain variables to be studied (Leedy & Ormrod, 2015). To reduce or eliminate this bias the researcher at all times, from the proposal stage throughout this study, adhered to ethical principles to uphold objectivity.