Surveys as a research instrument have been one of the key mechanisms through which the social sciences have been able obtain data on people’s lived experiences, while quantitative measures have become mostly associated with survey research (Gorard 2003:90). Furthermore, with the improvement in interviewing methods, the data from surveys were increasingly also more reliable with the effect of making survey research more popular (Almond 2000:9). The development of scoring and scaling techniques enabled survey responses to be categorised “in homogeneous dimensions” (2000:9) which enabled the possibility for more statistical analysis and more complex inferences.
Surveys are regarded as “the best method available to the social researcher who is interested in collecting original data for describing a population too large to observe directly” (Babbie 2013:229). Surveys are a form of data collection by means of standardised questionnaires and can be conducted through various means: interviews, self-administered questionnaires or structured observations (Creswell 2009:146). Elkins and Simeon (2000:31) note that the advantages of interviews or “observing respondents” is the fact that the scope for “fake or […] self-serving and pragmatic answers” is slimmer; however, this is not the case with self-administered surveys, where respondents are more likely to give “short-run responses to transitory contemporary political phenomena”.
Different types of survey designs enable either descriptive, explanatory or exploratory research through means of quantitative measurements for the purposes of measuring trends, attitudes or opinions of a given population12 (Babbie 2013:229). Nachmias and Frankfort-Nachmias (2008:233) state that attitudes are valuable measurements for social research as they provide an account of respondents’ “general inclination”. This is because survey data predominantly “focus on the individual-level causes and consequences” (Levi & Stoker 2000:467). However, groups, consisting of individuals, can also serve as units of analysis (Dale, Wathan & Higgins 2009:520) for the purposes of describing collective characteristics such as race and age (Babbie 2013:99).
12 The population refers to the “aggregate of all cases that conform to some designated set of specifications” (Nachmias
48 The standard method for being able to generalise to a broader social setting demands that surveys be based on representative samples13 of a population (Ioannou 2014:2). Survey research is conducted through probing a sample of a population, enabling the researcher to make generalisations and descriptive findings and other claims about a specific population (Creswell 2009:145; Babbie 2013:262; Allum & Arber 2011:390).
The design of surveys involves the creation of applicable questions to be included in the questionnaire as this specific process also determines the relevance and success of a survey (Gorard 2003:109). Surveys can include open- or close-ended questions, with the latter being more difficult to design, but easier to analyse. Most surveys also include standard background and demographical questions (Gorard 2003:106). This process is then followed by the selection of a sample to be studied and the subsequent data-collection phase through either interviewing or self- administered questionnaires (Babbie 2013:264).
As a research method, surveys have certain strengths and weaknesses (Babbie 2013:262). Responses to the survey “will vary according to the way the question is phrased and who is asking it” (Fukuyama 2001:15). Similarly, when respondents are offered a set of responses in the format of close-ended responses, the particular format may introduce biases as the respondent is forced to select a response from the prescribed options available (Nachmias & Frankfort-Nachmias 2008:233). Responses should be interpreted as being “approximate indicators” of the survey questions framed at the onset (Babbie 2013:263). However, Putnam (2000:138) notes that survey responses “should be interpreted prima facie as accurate accounts of respondent’s social experiences”.
Furthermore, survey research is regarded as weak on validity,14 but strong on reliability15. Despite some factors inhibiting the utilisation of survey data, the World Bank (2011) notes that contemporary researchers are in a favourable position as they have access to “several long- standing surveys […] to compile indexes from a range of approximate items, such as measures of trust in government”. Thus the availability of prior survey data further enables the possibility to make certain descriptive findings about a particular population sample or new research questions (Babbie 2013:262).
13 The sample design should be “as representative as possible of the population from which it is drawn” (Nachmias &
Frankfort-Nachmias 2008:167).
14 Validity is “a measure that accurately reflects the concept it is intended to measure” (Babbie 2013:191).
15 Reliability refers to the “quality of measurement method that suggests that the same data would have been collected
49 3.3.1 Secondary Data Analysis
Secondary analysis involves the study of data collected (mostly through surveys) for the purposes of making statistical findings (Babbie 2013:266). Neuman (2011:374) notes that in the “past two decades many more social scientists have conducted secondary analysis as more data have become available”. Instead of subscribing to the “theory testing” model of social research, secondary analysts incorporate existing survey questions and theory to further corroborate research findings (Allum & Arber 2011:390). Significantly, secondary analysis allows for “meta-analysis in which a researcher brings together a body of past research on a particular topic” (Babbie 2013:266). Secondary analysis enables a study of specific research questions, “while avoiding the enormous expenditure of time and money” necessitated by survey research (Babbie 2013:264). Moreover, it enables the utilisation of primary data, such as national surveys, to study different elements as proxy measures for the purposes of conducting new research questions (Dale, Wathan & Higgins 2009:523, 529; Neuman 2011:374). Babbie (2013:266) emphasises that if the original data had been collected by a reputable institution, conducting secondary research of survey data is of great value and provides additional credibility to the findings made by the secondary researcher. Neuman (2011:337) notes that the issues associated with the unit of analysis and variable attributes are “less of a problem for secondary survey analysis because [the researchers] obtain raw information on each respondent” as most reputable surveys provide all necessary information in technical reports.
According to Kelly and Maxwell (2009:166), the advantage of secondary data analysis is that it allows for statistical methods and quantitative analysis to extract information from primary data. The applicable statistical techniques employed by the researcher will be dependent on the research question under investigation as well as the type of variables used (Allum & Arber 2011:391). However, secondary data analysis also has certain disadvantages such as disagreement regarding the validity of the original data gathered and the suitability of questions posed in the original questionnaire to serve as valid measurements for the purposes of secondary analysis (Neuman 2011:374; Babbie 2013:266).
3.3.2 Descriptive Analysis
A central tenet of the social research process is to observe and describe social characteristics in a consecutive manner by means of descriptive observations (Babbie 2013:27, 91). Descriptive analysis is a valuable mechanism to avoid over-generalisation and swift conclusions based on only a few observed items. Moreover, descriptive research provides the researcher with an account of, amongst other things, “behaviour, which often leads to hypotheses” (White & Arzi 2005:139).
50 According to White and Arzi (2005:143), the outcome of descriptive research is “insights, rather than conclusions”. The main purpose of descriptive analysis is to delineate what is being observed so that general insights can be drawn from the observations made (Babbie 2013:21).
In this study the main rationale for describing the different variables is to analyse and track the development of social cohesion and reconciliation, so as to gain insights into conflict transformation in South Africa. Moreover, this study is predominantly based on inductive reasoning and adopts a more reflective approach (Hay 2011:153) as it “moves from the particular to the general, from a set of specific observations to the discovery of a pattern that represents some degree of order among all the given events” (Babbie 2013:21). According to Fox and Bayat (2007:106), induction implies “that the researcher collects the data and then extrapolates from that to achieve insights into human behaviour”. Therefore, when approached from an inductive point of view, theory is the outcome of the research (Bryman & Bell 2011:13).
3.3.3 Similar Studies: The SCORE index
The Social Cohesion and Reconciliation (SCORE) index and its framework was developed as a tool to measure two indicators of peace: reconciliation and social cohesion (Ioannou 2015:2). The index, based in Cyprus, was created in 2013 under the auspices of the United Nations Development Programme (UNDP), the Centre for Sustainable Peace and Democratic Development (SEED) and the Action for Cooperation and Trust (ACT). Up until 2014 the SCORE index was conducted only in Cyprus, Bosnia and Nepal (Ioannou 2014:5). The SCORE index measures social cohesion, in terms of “the quality of coexistence between individuals within their own group and the institutions that surround them”, and reconciliation, through the on-going effort to establish peace between conflicting groups (UNDP 2014). As the index works with the hypothesis that social cohesion affects and predicts reconciliation, it is therefore understood that improved social cohesion will lead to a greater degree of reconciliation (UNDP 2014).16
Given that both social cohesion and reconciliation represent multifaceted and abstract constructs, the SCORE methodology analyses social cohesion and reconciliation in terms of multi-component constructs. These constructs in turn consist of various other sub-components that are of a lesser degree of complexity to track (Ioannou 2014:2). These sub-components are then directly observed through various indicators or attitudes and measured through questionnaires (Ioannou 2014:3). This process is made possible as attitudes can be described by their “content (what the attitude is about), their direction (positive, neutral etc.) and their intensity” (Nachmias & Frankfort-Nachmias
51 2008:233). These indicators measure trust in institutions for social cohesion and the level of social distance between groups for the purposes of measuring reconciliation (Ioannou 2014:3); they include people’s opinions on whether they would be in favour of having certain types of relations with an opposing group such as close relatives by marriage, friends, neighbours and as co-workers (Ioannou 2014:7).
Social capital or trust within a society is severely affected by inter-group anxiety and stereotypes as people with higher inter-group anxiety avoid having contact with members of other groups and hence are more likely to retain stereotypes of the opposing group (Ioannou 2015:20). Conversely, the quantity and intensity of horizontal interactions or contact among individuals in a community would facilitate the emergence of a decrease in inter-group anxiety, reducing stereotypes and giving rise to reconciliation and more desirable norms such as trust.
The SCORE index underpins the need for studying variables related to both social cohesion and reconciliation in order to gather more insights regarding conflict transformation. However, it should be noted that this study will not attempt to duplicate the SCORE index, but will draw on aspects of the SCORE index while offering a unique focus on South Africa and limited to the availability of indicators.