RESEARCH METHODOLOGY, DATA ANALYSIS AND INTERPRETATION
5.2 RESEARCH METHODOLOGY
5.2.3 Tests used in the study
158
Questionnaires and rating scales are commonly used to measure qualitative variables.
The nominal scale is a non-quantitative scale of measurement. It uses symbols such as words or numbers to classify the value. The research questionnaire uses this scale for example for gender which is either male or female. Svensson (2001:47)argues that the scaling of responses can vary from the dichotomous alternatives ’yes’ and ’no’ to a mark on a line, as in the visual analogue scale (VAS).
The ordinal scale is a rank-order scale. This scale is used in the present research to distinguish the degree of acceptance by respondents, from strong to weaker.
Some scholars are of the opinion that the median level and the quartiles or, in the case of small samples minimum and maximum (range), are appropriate measurements to describe the distribution of ordinal data. Bar charts point plots of VAS assessments and box and whisker plots are recommended for the graphical display of the distribution of ordinal data (Svensson, 2001:47).
The interval scale used in the present research serves to determine the difference between the equal adjacent points of interval. This study makes use of continuous variables.
Scholars suggest numerical labels as being commonly used for the recordings (Svensson, 2001:47). The study purpose, the properties of study groups and whether assessments or self- or observer-reported are important factors in the choice of research instruments. The structure of the instrument should be described, for example the dimensions of the variable, the number of items and the types of item responses. The joint frequency distribution of paired assessments could be presented in contingency tables or, in the case of VAS assessments, in scatter plots.
159 5.2.3.1 Chi-square goodness-of-fit test
The chi-square goodness-of-fit test is a univariate test used on a categorical variable to test whether any of the response options are selected significantly more or less often than others. Under the null hypothesis, it is assumed that all responses are equally selected. The test is referred to as a single sample chi- square test determining whether frequencies across categories of variables are distributed in relative manner (Tyler R. Harrison, 2002:36). It is also considered as one of the most useful statistics for testing hypotheses when the variables are nominal. The test can give information, either on the significance of any observed differences, or provide details of exactly which categories account for any differences found (McHugh, 2013:143). The model ought to be lower than 0.5.
Accepting the null hypothesis H0 means that the model fixes the data. For a p value greater than 0.05 one accepts the null hypothesis. The alternative hypothesis means that the model does not fix the data (Hansen et al., 2015:89).
Goodness-of-fit procedures are tools in data analysis and used for the detection of model misspecification. They are also employed as formal test statistic corresponding to a particular null hypothesis or might, alternatively, offer a graphical display. Scholars argue that most statistical models are based on some assumptions. Goodness-of-fit procedures are necessary to ensure that conclusions are drawn from the model (Hansen et al., 2015:89).
5.2.3.2 Binomial test
The binomial test determines whether a significant proportion of respondents selects one out of a possible two responses. This test can be extended when data with more than two response options is split into two distinct groups. For categorical questions, the binomial test is used when there are 2 categories and the chi-square goodness-of-fit test when there are 3 or more categories.
5.2.3.3 Independent t-test
The research uses the independent samples t-test which serves to determine if the means of two independent sets of data are significantly different from each other.
160
These two sets of data are mutually exclusive. The null hypothesis for mean differences is presented as: 𝐻𝑜: 𝜇1= 𝜇2, where 𝜇1 and 𝜇2 are the means of the first and second populations respectively. In the case of the null hypotheses being rejected, it means that there is a difference between the means of both samples, which forms the basis for accepting the alternate hypothesis (𝐻1: 𝜇1≠). In making the decision rule, the significant value (p-value) is compared to the alpha level (α) set prior to the test by the researcher. If p < α, the null hypothesis is rejected implying there is a significant difference between the means of both samples. The current study employs the independent sample t-test. It determines whether subordinates who behave in an assertive fashion receive a fairer treatment than those who display low-assertiveness behaviour (Tyler R. Harrison, 2002:54). In this research a one-sample t-test is applied to all Likert scale questions for significant agreement or disagreement, or extent or impact.
5.2.3.4 Descriptive statistics with means and standard deviations
The mean is the arithmetical average. The standard deviation is an approximate indicator of the average distance that the data value is from their mean. For the variance and the standard deviation, the larger the value, the greater the data are spread out, and, the smaller the value the less the data are spread out (Turner, 2011:400,402). Moses (2012:77) defines it as a measure of dispersion used to capture the spread of scores in a distribution of scores.
5.2.3.5 Factor analysis
Factor analysis is a technique that essentially reduces a set of variables to a smaller number of underlying factors and that detects structure in the relationship between variables. It looks for combinations of variables that may represent an underlying latent variable that the researcher has not directly measured but that the variables that have been collected represent (Muijs, 2011:199). It makes sense of a large number of correlations between variables and is an exploratory tool (Robson, 2002:433).
5.2.3.6 Thematic analysis applied in qualitative analysis
Thematic analysis is applied for dealing with the second objective of this research, namely the provision of a theological foundation for the political responsibility of
161
the prophetic voice. (Further study objectives are to propose some theological perspectives to the church and to assess the risks run by the church if it engages in the political prophetic way.
Hennink et al. (2011:16,17) define qualitative research as an approach that allows the researcher to examine peoples’ experience in detail by using a specific set of research methods such as interview, focus group discussion, and observation. The researcher can identify by qualitative research issues from the perspective of the participants in the study and understand their interpretation of, and the perceived meaning they ascribe to, behaviours, events and objects. Qualitative research can also imply studying people in their natural settings to discover how their experiences and behaviours are shaped by their lives’ contexts (Hennink, 2011:9).
Conducting qualitative research is thus helpful in developing an understanding of church leaders’ views on the involvement of the church in political responsibility and of the way in which they experience the insistence on the need for such involvement.
The researcher interprets the meanings given by participants themselves to their views and experiences.
Thematic analysis can be defined as a method to identify, analyse and report on patterns (themes) within data. It minimally organizes and describes data set detail.
Thematic analysis is widely used, but there is no clear agreement about what it exactly entails and how to go about applying it (Braun, 2006:79).
A theme is something captured within important data in relation to the research question and it represents some level of patterned response or meaning within the data set (Braun, 2006:82).
Braun and Clarke distinguish six phases in conducting thematic analysis. The researcher first has to become familiar with the data, after which he or she must generate initial codes, search for themes, review themes, define and name those themes and finally produce the analysis report (Braun, 2006:87).
Scholars emphasize that the codes in such a study are inductive codes and that some concurring of them is applied. The content analysis must be done manually to identify key themes and patterns in the data (Ahmadnezhad et al., 2013:485).
162