"While the use of questionnaires and schedules (quantitative or positivistic research techniques) are mostly discouraged, and in many instances regarded as inappropriate, by symbolic interactionists for discovering how people define the situations they are in and how they construct their actions accordingly, this does however, not imply that such techniques cannot be used at all in research that is largely qualitative in nature. Indeed, it has been suggested that such ‘positivistic’ techniques can and should be used in combination with, for example, participant observation to a greater extent as is presently the case in sociological research, in order to reach a greater level of understanding of the particular issues at hand."14

Research strategy

For the greater part, this exploratory study of the social world of former combatants and demobilised personnel in South Africa was conducted within a quantitative methodological framework supplemented by a qualitative approach. These two well-known and recognised approaches to research differ vastly from each other. In the ensuing discussion, the key differences between the two approaches are outlined.15

The quantitative paradigm is based on positivism, which takes scientific explanation to be nomothetic (based on universal laws). Its aims are to measure the social world objectively, to test hypotheses and to predict and control human behaviour. In contrast, the qualitative paradigm stems from an anti-positivistic, interpretive approach, is idiographic and thus holistic in nature, with the main aim to understand social life and the meaning that people attach to everyday life.

Quantitative researchers use a deductive form of reasoning. In contrast, qualitative researchers use an inductive form of reasoning.16 Quantitative research takes universal propositions and generalisations as a point of departure, whereas qualitative research aims to understand phenomena within a particular context. In terms of epistemology, quantitative researchers see themselves as detached from, not as part of, the object that they study. Researchers can therefore be objective — they do not influence the study object and are not influenced by it. In contrast, qualitative researchers are subjective because they interact with the subject (object of investigation).

With regard to methodology, the quantitative paradigm emulates the physical sciences in that questions or hypotheses are stated and subjected to empirical testing to verify them. In contrast, qualitative methodology is dialectical and interpretive. During the process of interaction between the researcher and the subject, the subject’s world is discovered and interpreted by means of qualitative methods.17

Qualitative researchers are concerned with understanding rather than explanation; with naturalistic observation rather than controlled measurement; with the subjective exploration of reality from the perspective of an insider as opposed to the outsider perspective that is predominant in the quantitative paradigm.

Before concluding this discussion of quantitative and qualitative paradigms, four points have to be emphasised:

Although these paradigms undoubtedly guide social science researchers, the distinctions between the two paradigms are not as clear-cut in practice as the preceding discussion implies.



There is a trend at present to bring the ‘Cinderella’ of science into research circles. Qualitative interviewing is generally used to verify knowledge obtained by means of quantitative data collection, or for a preliminary exploration before undertaking more structured survey research.

The views of many scholars (for example, in the field of programme evaluation that used to be the stronghold of the quantitative paradigm) on quantitative and qualitative research styles have changed dramatically in recent years. In an excellent overview of the ‘paradigm debate’ over the past four decades, it is clear that the long and heated conventional-alternative paradigm debate has abated and that equal attention and weight should be given to these two paradigms.18

It appears that local scholars have different views on the paradigm debate in South Africa. A review of methodological and theoretical developments in South African social science concludes that, until recently, there has been no real antagonism, similar to that in American universities, between proponents of qualitative and quantitative approaches in the country.19 Referring to local works, the review concluded that these contributions "have been conciliatory and constructive in tone."20

Unit of analysis

The exploratory study‘s unit of analysis can be defined as the following: a study undertaken during the first quarter of 2001 (a specific point in time) of key characteristics (gender, age, race, employment status and other demographic features), interests (receiving further training, career chosen if given the opportunity, assistance to find a job, and others), opinions or perceptions (regarding a number of matters such as the extent of training received before leaving the SANDF, whether the SANDF did enough to assist after leaving it), feelings (towards the Citizen Force, the commandos, and the Service Corps of the SANDF), and actions (deciding to resign or demobilise, study further after leaving the SANDF, seeking employment), of a group of South African former combatants and demobilised soldiers.

Selection of respondents from the population of interest

In social research, the focal group from which scholars wish to learn something is known as the ‘population. This concept is used in the statistical rather than geographical sense. While the population of interest — also called the ‘universe’ — include all elements or cases of human beings, collectives of individuals or groups, organisations, or social artefacts of a particular study, the population of interest in social research is normally composed of individuals.

When the population of interest is small, a researcher can administer a data collection tool, for example, a questionnaire, to its entire selection of cases or elements. However, universes in quantitative research, such as survey research, are usually relatively large and geographically distributed. In cases where it is difficult, if not impossible, to solicit information from all people, a sample of the universe is selected and information is solicited from this group.

The main problem with sampling is the danger that a particular study’s sample may not be representative of the population as a whole:

"Representativeness is the major issue in sampling, and the reason for its importance is simple: you want to be able to make inferences about the population as a whole based on what you find to be true of the sample. If the sampling error is large — if different samples yield vastly different results — then your conclusions about the population are likely to be incorrect."21

The question therefore is how researchers can ensure that a sample is representative:

"The answer offered by all methodologists is that random sampling is the only technique available that will ensure an optimal chance of drawing a sample that is representative of the population from which it was drawn. This then leads us to consider the two kinds of sampling available to researchers: probability sampling, which is based on randomisation, and non-probability sampling, which does not implement randomisation."22

While contentious, quantitative researchers avoid haphazard, accidental, convenient, quota-based non-probability types of sampling. There are special situations where some form of non-probability sampling such as purposive, judgmental or snowball sampling is the only avenue open to a researcher who needs to obtain knowledge of a difficult-to-reach specialised population.

Snowball sampling is often used in social research where research has to be conducted into relatively unknown phenomena and where it is extremely difficult, if not impossible, to identify cases:

"Snowball sampling (also called network, chain referral, or reputational sampling) is a method for identifying and sampling (or selecting) the cases in a network. It is based on an analogy to a snowball, which begins small but becomes larger as it is rolled on wet snow and picks up additional snow. Snowball sampling is a multistage technique. It begins with one or a few people or cases and spreads out on the basis of links to the initial cases."23

Since the universe of former combatants in South Africa dating back to 1994 cannot be established, a random sample of representative cases could not be drawn for inclusion in the present study. Cases had to be selected non-randomly. Non-probability sampling was used to ensure that typical characteristics of former combatants were included in the sample.

Quota sampling was first used to establish the following three sets of categories and areas from which former combatants were selected:

key segments of demobilised SANDF personnel — the different military musterings from both statutory and non-statutory forces: MK, APLA and AZANLA;



key socio-demographic characteristics: age, gender, race and language; and



geographic areas: the South African provinces, including both rural and urban areas.

Secondly, purposive or judgmental sampling was utilised to obtain subjective information obtained veterans’ organisations, the South African Service Corps, and a number of community organisations to identify potential respondents from the target group.

Finally, snowballing was also used. Additional respondents were identified by asking former combatants, who were interviewed, to name other former combatants from their friendship networks. They were also asked to assist the researchers in explaining the objectives and nature of the study to their friends. It was also made clear to the respondents that participating in the research would not harm them but would benefit all South African former combatants in the long run.

Sample

With the assistance of veterans’ organisations, the Service Corps, former combatants, interested family members and community organisations in the different provinces, 307 former combatants were finally included in the study. The sample included:

154 MK combatants;

50 APLA combatants;

15 (MK) SDU combatants;

1 AZANLA combatant;

1 Inkatha VIP member; (total: 221);

43 SADF/SANDF members; and

7 members of formal TBVC24 forces (total: 51).

Against the background of a total of approximately 40 000 demobilised SANDF personnel (between 1994 and 2001), the sample is reasonably large. Although the sample was not drawn with the aim to guarantee representivity (and thus generalisation of the study findings), former soldiers who were interviewed did not differ too much, or too little, from one another to sketch a conceivable profile of them.

With regard to the three selection criteria, it should be noted that key military forces were covered in the sample. Except for race, the variety in socio-demographic characteristics was large enough to reflect a fair profile. Unfortunately, the racial segmentation in South Africa did not allow the snowball technique to cover all four racial categories. White, Asian and coloured race categories were underrepresented. The geographic areas were fairly well represented. Respondents came from all provinces and represented former soldiers from both rural and urban areas.

While the majority of respondents were from the non-statutory forces (see working definition in the appended questionnaire), members from the statutory forces (SADF/SANDF and TBVC defence forces) were also included. Thus, the sample reflected both elements of an exclusive definition of demobilisation and rationalisation and the South African context-bound use of ‘demobilisation’ following the demise of the apartheid state and corollary structures. In a few cases (by far the minority), people who were ‘rationalised’ through dismissal or voluntary severance (for example, because they chose not to move from their home base after the TBVC defence forces were integrated) were included.

Although the research results obtained from this sample cannot be generalised to all South African former combatants, the researchers were convinced that sufficient respondents were recruited from the most typical categories of the former combatant population. The sample was sufficient to explore and describe the social world of local former combatants and to construct a profile of demobilised SANDF staff. This profile can be used to make relevant recommendations, related to society and policy. Minimal requirements for drawing a representative sample of demobilised soldiers include:

the establishment of a generally agreed upon definition of demobilised SANDF personnel;



knowledge of the biographical and demographic characteristics of this population;



access to full and accurate information for research purposes (the exact numbers of staff demobilised or rationalised);



their ranking;



time and place(s) of service;



time and place of severance; and



method of severance.

Future studies in this regard should have such information available before a representative sample can be drawn that may lead to conclusions which can be generalised. It is the belief that this exploratory study could be a first important step towards such future research that will benefit former soldiers.

Data collection

As already mentioned, two well-known and recognised approaches to research were used in the study. While the social survey, a quantitative method, was primarily employed to explore and describe the social world of South African former combatants, unstructured or qualitative methods such as open-ended questions included in the survey schedule, as well as fieldnotes and reports written by fieldworkers were also utilised to gather information, albeit to a limited extent.

The social survey

For the purpose of this study, the following definition of the social survey was used:

"The survey may be defined as a methodological technique that requires the systematic collection of data from populations or samples through the use of the interview or the self-administered questionnaire. The investigator approaches a sample of persons who have been exposed to a set of events or experiences and interviews them with respect to these experiences … [A] group of persons are observed at one point in time and questioned about their behaviours, attitudes, and beliefs with respect to a series of issues."25

The following two types of survey were used:

Individual interviews: Fieldworkers visited respondents at the places were they were found (residences, work or other places). They recorded respondents’ answers to previously constructed questions in a schedule.



Completion of individual questionnaires in a group context: Former combatants and SADF staff were gathered in groups at specific venues where they completed questionnaires in the presence of fieldworkers, who explained certain questions and, when necessary, assisted those respondents with difficulties.

Aspects of former combatants’ social situation (fields of interests, attitudes, perceptions, actions, background, demographic information, and others) were carefully studied and converted into specific questions which formed part of the survey questionnaire. Since the goal in survey research is that every respondent should interpret both the questions and answers similarly, special attention was given to both the wording of the questions and their possible responses, as well as their placement in the schedule. Finally, in formulating the questions, the two types of surveys were borne in mind.

Three strategies that proved to be very valuable in the design of the survey questionnaire, were:

discussions with informants, including personnel of the SANDF, members of veterans’ organisations, former combatants and other people who were knowledgeable of about combatants;



reviews of previous research undertaken on the topic; and



a reality check and pre-testing. The initial draft of the interview schedule was tested by conducting a focus group with a few of the former combatants who were willing to discuss the items or concepts the researchers planned to include in the study. This focus group pointed out a number of ambiguities, hidden assumptions, or conceptual complexities that the researchers overlooked. After these revisions were made, and the interviewers were trained, a pilot test of the interview schedule was undertaken.

Qualitative methods

In any discussion of qualitative data analysis, it is important that there is a clear understanding of the meaning of qualitative data and how the collection and storage of details about the social world should be done. It is also important to know exactly how to store the data before it is analysed. In lieu of this, some important aspects of qualitative analysis are outlined below.

The term ‘data; refers to:

"the rough materials researchers collect from the world they are studying; they are the particulars that form the basis of analysis. Data include materials the people doing the study actively record, such as interview transcripts and participant observation field notes. Data include what others have created and the researcher finds, such as diaries, photographs, official documents, and newspaper articles. Data are both the evidence and the clues. Gathered carefully, they serve as the stubborn facts that save the writing you will do from unfounded speculation. Data ground you to the empirical world and, systematically and rigorously collected, link qualitative research to other forms of science. Data involve the particulars you need to think soundly and deeply about the aspects of life you will explore … Some qualitative studies rely exclusively on one type of data, interview transcripts, for example, but most use a variety of data sources."26

For the purposes of this study, qualitative data included:

rough material gathered by fieldworkers from particular aspects of the world of former combatants;



material actively recorded by fieldworkers themselves;



any information gathered during the course of the study that was not expressed in numbers;



any human creation and product of former combatants including, for example, words, letters,



drawings, photographs and household garbage;



escriptions of situations, events, people, interactions and observed behaviour;



direct quotations and excerpts from self-accounts by former combatants about their experiences, attitudes, beliefs and thoughts;



entire passages from fieldnotes; and



meaning former combatants attached to the world in their own words.

Four qualitative methods for data collection were employed in the study:

participant observation;

fieldnotes;

fieldworker reports; and

open-ended questions included in the quantitative schedule.

Participant observation

Participant observation refers to a social process during which:

the fieldworker or researcher interacts with some of the people (former combatants or soldiers) before, during and after the completion of the schedule;



in the milieu of the latter; and



the fieldworker observes relevant social behaviour during these exchanges and systematically notes them down.

Collecting data from research participants in their natural settings was especially helpful in obtaining some understanding of the social situations in which former combatants and soldiers lived and worked. This first-hand information also facilitated the interpretation of the qualitative data gathered by means of the schedule.

Fieldnotes

The mainstay of qualitative research is the written account of what researchers (fieldworkers) hear, see, experience and think in the course of collecting and reflecting on the data in their studies. While memorising observations during the current study was often unavoidable, fieldworkers were requested to make notes as far as possible. The compilation of fieldnotes is not as straightforward as it appears, and does not merely represent summaries of events, but rather detailed reproductions of what occurred. Following recent developments in the field of qualitative research, fieldworkers were asked to look out for the following: What activities occurred? Where did the activities occur? What are the views and/or theories of the author/research participant?

More particularly, fieldworkers had to try and cover the following areas when compiling their notes:

respondents’ physical appearance, dress, mannerisms and style of acting (they had to attempt to concentrate on particular aspects of research participants that might set them apart from other participants or inform them about their affiliations);



respondents’ dialogue and conversations which fieldworkers had to paraphrase and summarise;



a description of the physical setting where the schedule was completed;



the particular events that the subjects were involved in, in what manner they were involved in the activities, as well the nature of the events; and



the fieldworker’s own behaviour during the interaction with the respondent (since qualitative fieldworkers or researchers are viewed as the instruments of data collection, it is crucial to take stock of their own behaviour, assumptions and anything else that might affect the data that is collected and analysed).

Data analysis

The following definition of qualitative data analysis portrays the key features of the process of making sense of the data in the current study:

"Data analysis is the process of systematically searching and arranging the interview transcripts, field-notes, and other materials that you accumulate to increase your own understanding of them and to enable you to present what you have discovered to others. Analysis involves working with data, organising them, breaking them into manageable units, synthesising them, searching for patterns, discovering what is important and what is to be learned, and deciding what you will tell others. For most projects, the end products of research are dissertations, books, papers, presentations, reports or, in the case of applied research, plans for action. Data analysis moves you from the rambling pages of description to those products."29

Selection and training of fieldworkers

Fieldworkers or interviewers are the ‘foot soldiers’ in any research project, whether consisting of structured scheduled interviews or qualitative interviews. Consequently, it is crucial that they are carefully selected and properly trained. Careful attention to the selection and training of interviewers is usually time well-spent.

A number of social scientists identified two important phases in interviewer training: ‘general’ and ‘study-specific’. These phases can be described as follows:

"The first phase consists of a general orientation to the requirements and routines of standardised interviewing. This can be done by means of lectures, demonstrations, and videos, and might include, for example, information about reading questions as worded, the meaning of different instructions on interview schedules, and demonstrations of nondirective probing … The second phase of training is specific to the study being planned. It involves thoroughly familiarising interviewers with the purposes of the study, the specific questions to be asked, the layout of the interview schedule, and the nature of the respondents to be interviewed. And, most important, interviewers should actually conduct some supervised practice interviews during this phrase of training."30

In this study, particular attention was paid to the general and project or study-specific training. This included practice interviews under the supervision of the Social Sciences Consultancy (SSC).

While some social scientists may prefer other data collection methods, virtually all researchers, at some time during their careers, will use quantitative or qualitative interviews to collect data. There are a number of important principles, skills and steps in conducting these two types of interviews. While all these factors may be taken into account during the execution of the interview, there can be no doubt whatsoever that the role of the interviewer in the entire process is of the utmost importance. Fieldworkers should therefore be carefully selected and well-trained if the collected data is to be valid, reliable and credible.

Promising young researchers were approached to assist with collecting information from former combatants. A three-day training course was presented on various dimensions of social research, including the ‘general’ and ‘study-specific’ phases of interviewer training, and other aspects of particular interest to the present study. Through this general and specific training, 11 fieldworkers (some with previous experience and others with no previous fieldwork training) were equipped and empowered to conduct fieldwork — a crucial stage in a project of this nature.

Data analysis

Inferential methods — the instruments used in this study to make sense of the former combatant phenomenon — are in accordance with the positivistic model. Suffice it to state here:

This model of science is founded on the assumption that social phenomena can be unambiguously defined and delineated. Moreover, this is linked to assertions that particular features of such realities can be categorised ‘objectively’ by researchers and can be measured by the application of numbers to such categories, as well as the number of cases within each of the types.



The related belief is that phenomena are amenable to statistical analysis. This is clearly reflected in the way quantitative researchers formulate and test explanatory and predictive models of causation. Quantitative researchers assume that aspects of the social world can be simplified and represented by such models.

The key steps and decisions taken during the inferential process of the study are discussed below.

Computing data

After fieldworkers had completed the schedules, these were checked to establish whether they were properly completed. Thereafter, six people captured the data on computer using of SPSS, a statistical programme. The data was entered as soon as the completed schedules were received. This enabled researchers to control the capturing process continuously.

As already mentioned, the schedule contained precoded, as well as open-ended questions. Recoding and coding for numeric and non-numeric data were done after the raw data was entered into the computer. The answers to open-ended questions were entered into the datafile verbatim. Non-numeric data, or string variables, included, among others:

type of training or education received;

aspired careers;

demographic data, such as residential area; and

reasons for unemployment.

After having reached agreement on clustering, the researchers clustered or categorised the answers to open-ended questions into fewer, and therefore more manageable, categories. Recoding and newly computed variables were then created.

The capturing process pointed to some adjustments that had to be made during the completion of the schedules. It became clear that questions relating to employment and non-employment by the SANDF had to be dealt with more cautiously, since, for example, some respondents indicated that they were (briefly) employed, but never wore a uniform (the so-called ‘demobilised’ category or people on the CPR). Furthermore, some respondents who completed the schedule themselves did not always complete the entire schedule. Of course, this is a common shortcoming in self-administered surveys, and its effect can often be reduced, for example, by rephrasing and reordering the items or questions. This was not possible in this study, since readjusting the schedule would have led to answers becoming incomparable.

During the cleansing of the complete data set, it appeared that the schedule could have been improved in several respects. For instance, questions on training experiences within non-statutory and statutory forces caused some confusion because the answers to these questions showed questionable overlaps. Completion of the schedule was not always done correctly, which also resulted in inconsistencies in answers given to different questions. While some of these inconsistencies (for example, employed respondents giving reasons for unemployment) could not have been solved, others could have been corrected by building controls into the schedule. Inconsistencies in answers to the questions pointed to the need in future studies to rephrase a number of questions and for fieldworkers to take more trouble to ensure that respondents feel comfortable with both the schedule and the interview situation.

It must be pointed out that, since the former combatant phenomenon is very sensitive, more intense training of the interviewers is required. Fieldworkers have to be sensitive to the following aspects during the interview: recruiting of former SANDF personnel to assist with the research; the manner in which potential respondents are introduced to the interview; the place where the interview is to be conducted; and guiding the respondents through the questions. As is clear from the section dealing with the fieldwork, the current study offers some very useful insights into how interviewers should deal with the specific problems and circumstances in which former combatants find themselves and how to ensure a ‘safe’ interview environment.

Computing the former combatant phenomenon: Numbers (frequency counts) and percentages

An important consideration in quantitative research is to obtain the incidence, or number of times the phenomenon they want to analyse occurs in their study area, over a given period of time. While quantitative researchers can utilise official sources like data gathered by a government department, or ask a sample of people about their knowledge of and/or perceptions of an issue, these still have to be represented in some quantitative value. In the study, both numbers and percentages of responses were calculated.

Numbers are generally regarded as inappropriate and misleading, since only a minimum of simplistic scientific analyses can be made with frequency counts:

"The essential problem with comparing the absolute frequencies of an observed trait or status is that such counts ignore the total number of cases in each group (e.g. males vs. females; whites vs. nonwhites) eligible to occupy the status or exhibit the particular trait. When the total number of cases in the comparison groups is equal, then the frequencies can be compared directly. When the comparison group differs in size, however, we need to take explicitly into account the size difference."31

Percentaging provides a general measure that takes proportions into account:

"The percentage is … a … way of expressing the frequency occurrence of a status or trait per 100 cases of subjects. The percentage is computed by dividing the frequency of delinquency by the number of subjects and then multiplying by 100 (percentage = frequency of delinquents ÷ number of subjects X 100)."32

People find percentaging more appealing because it provides a sense of proportion or an implicit comparison with other known proportions.

In concluding the discussion of inferential analysis, the following cautionary words are in order.

It is important to realise that the study focused only on two-variable relationships. Since social phenomena are complex and particular phenomena are related to various other social phenomena, it is important in social analyses that a third (possibly even more) relationship between the studied phenomena and other social phenomena should be examined. Using more than two variables in social research requires that the researcher use sophisticated techniques such as multivariate or factor analysis. This type of analysis — to examine the relationship between particular dimensions of the former combatant phenomenon and various other variables at the same time — was not used in this study. The main reason was the fact that the study was meant to be exploratory. It was therefore not possible study to:

gain an estimate of the substantive significance of a combined set of explanatory variables;



attribute weights to each explanatory variable, which is proportionate to the importance of any variable in explaining variations in some aspect of former combatant behaviour level, while holding constant the effects of all other variables; and



construct causal models of the former combatant phenomenon:



it is common to posit and test a composite model which specifies the collective and respective contributions of a number of explanatory variables at one and the same time, rather than examine a number of separate two-variable hypotheses. This is done by the use of techniques of multivariate analysis (including the calculation of multiple correlations) which permit the examination of the strength of association of several variables at one and the same time."33

Furthermore, the data analysis is underpinned by the belief that quantitative data (survey research) enables the drawing of inferences about the population or sample surveyed. However, generalising the study’s research results to the larger group of demobilised SANDF personnel, as has already been pointed out, is not possible, since the sample on which they are based is not representative of all demobilised SANDF personnel in South Africa. Such representivity was impossible as information on demobilised soldiers is limited and seemingly not accurate or reliable. This is partly due to a lack of capacity and resources to build such a datafile, the speed at which the demobilisation process coincided with the transformation and rationalisation processes, and because information on SANDF personnel is classified. In any discussion of this study’s findings, the analyses can therefore at best be regarded as tentative and subject to further study.

Data quality

Due to the fact that social research investigates human beings, it is bound to contain errors. It therefore goes without saying that the main goal of the social scientist is to minimise observer effects or nuisance variables as far as possible.

The following excerpts from fieldreports describe how one nuisance variable, attached to former combatants’ perceptions and their circumstances, posed a major challenge to fieldworkers:

"I have done more than 20 face-to-face interviews. I conducted most of the interviews at the respondents’ homes. The interviewees were mostly in favour that the interviews were conducted at their homes. There were a few who didn’t feel comfortable with me knowing where they lived and who gave me a venue of their choice and I would just go wherever they said. Not all these military people have settled down. There are still those who are unsettled and believe that the war is not over yet. They mostly belonged to MK and APLA. They are still militant and claim that there are still incidents where MK and APLA members are being abducted and killed mysteriously at their homes at night. Others are imprisoned unconstitutionally. One respondent gave an example of his colleague with whom he served with during exile. Because he’s an MK man, he was arrested on charges of assault and because of his background he’s been awaiting trial for the past 12 months. He claimed his colleague told him whilst visiting him in detention that one of the prison officials said that the government is not comfortable with them running around loose. Therefore, it’s a relief for the high commissions when they are locked up so that an eye can be kept on them. He went on to say that ‘our government is aware of all these incidents but is just keeping a blank (blind) eye as though they know nothing’. I observed that this particular respondent was near breaking point. I found him very depressed and looking as though he had been told that the world is coming to an end. Life has beaten the worst out of him. He looked as though he didn’t care about life anymore. When I was about to leave he started crying. I tried to comfort him and told him that he was not alone as such studies as those I’m conducting are ways and means to see how they can be helped. He then told me of how he has lost his two children and wife because he could no longer support them after he had been retrenched (demobilised). He thanked me for listening to him for he needed to talk to someone. I left feeling sorry for him"34

"The fieldwork was a touching experience of a mistake that deserve not be repeated, as it was more harmful than good. Some respondents were very emotional and couldn’t go ahead with the interviews. I had to understand that and tried to comfort some were I was able; it was really difficult to see a man crying his heart out. My experience and comprehension of the situation were that these people were failing dismally to restore the dignity they once had; there was no hope or light at the end of the tunnel that they will ever make it in the future. Lucky for those who were able to go back to school and study further. It has earned them the right to a new employment."35

Another set of observer effects involved logistic arrangements that had to be made with the team of fieldworkers, various members of veterans’ organisations and SANDF officials and others who assisted with contacting former combatants and/or prospective respondents from different geographic areas themselves. This complicated process often brought unforeseen and unexpected situations. The following accounts by fieldworkers describe these difficulties:

"However, the only weakness was the choice of the fieldworkers. Some of them are currently employed on a full-time basis and they were not readily available to conduct such interviews. Alternative arrangements had to be made from time to time because of this unavailability. It is suggested that future fieldworkers should be drawn from the pool of unemployed people who could become available when their service is required, and that they should be capacitated in field research."36

"[M]ost ex-militants were reluctant and uncertain of the motive behind the whole exercise … those who left the military because of reasons known to them, felt that they are being traced and targeted, and hesitated to participate."37

"Some venues were small for the envisaged numbers expected. This is due largely to the fact that we accepted venues offered to us. On the positive side, it seemed to make it easier for some participants to discuss their responses … Fieldworkers were at times given short notice to attend interview meetings. This might have been due to the time it took to make arrangements despite having warned/spoken against this ... AAC needs to improve on the office communication and co-ordination in order to afford fieldworkers information and messages in good time."38

Conclusion

While attempts were made to minimise nuisance variables in the study, and it is believed that these attempts were successful enough for the research findings to be regarded as valid, it must be emphasised that, because of the exploratory nature of the study, these results should at best be regarded as tentative, even though they are valuable and insightful.

Because of the current scope of the study, there are undoubtedly many aspects of the ‘world’ of South African former combatants that still need to be unravelled. It is hoped that future researchers will undertake large-scale studies of the circumstances of these former military personnel. Hopefully, research of this nature will aid in unravelling and finding solutions to the difficulties that this group experiences in their day-to-day living because of the various social changes that came about with the newly established democracy.