|
Chapter 2
METHODOLOGY AND SAMPLE DEMOGRAPHICS
Ted Leggett
The data is derived from two surveys, one conducted in May 2002 throughout the Johannesburg Central and Hillbrow police station areas (n= 1,100), and a follow-up survey conducted in July 2002 that was limited to ten residential hotels in Hillbrow. The initial survey form included questions about criminal victimisation, opinions on safety and police performance, and questions geared to evaluate community cohesion. The follow-up survey omitted some questions and included an additional section.
Those who said that they had been victims of certain crimes were asked a more detailed schedule of questions on the specifics of this incident. If the individual respondent had been the victim of hijacking, assault, or robbery, the detailed interview schedule for those crimes was administered. If the respondent’s household had been the victim of vehicular theft or burglary, or if anyone in the household had been murdered, then the appropriate questionnaire was used. This resulted in detailed accounts of more than 300 recent robberies, and over 100 burglaries and assaults, with factual and opinion information on criminal justice response to the incident. All this information is laid out in the chapters that follow.
Picking a sampling frame for these police station areas is something of a challenge. The most recent Census data stems from 1996, and no one would dispute that the area has changed radically since then. Both areas were projected to experience a population decrease of about 3% per annum from the 1996 estimates. If this projected decrease did occur, this would place the present population levels at about 91,000 for Hillbrow and about 21,000 for Johannesburg Central, for a total population of 112,000 for inner Johannesburg.
If these figures are indeed accurate, our initial survey of 1,100 respondents represents a 1% sample of the total population of the area. The final sample in the initial survey was 326 respondents in Johannesburg Central (30%) and 774 in Hillbrow (70%). The Census-derived population ratios are about 20% in Johannesburg Central and 80% in Hillbrow. This would suggest an over-sampling of Johannesburg Central, assuming its resident population is really a fifth the size of that in Hillbrow.
But there are many reasons to believe the official estimates of population size and composition are not reliable, which would make using these figures to construct a sampling frame problematic. These issues are discussed below.
Immigration and ethnicity
One does not need to spend much time in inner Johannesburg to realise that a significant portion of the population is foreign-born. Tours with the police, conducted as part of the time use study, reveal that a lot of these people are in South Africa illegally, although many have some form of ‘documentation’, much of it expired, fraudulently altered, or otherwise suspect. These people often live in buildings that are either not zoned for residential occupancy or which have been officially closed down. For this and other reasons, they are unlikely to have participated in the official headcounts. In addition, many are likely to have arrived in the area after the 1996 Census.
These hypotheses were born out in the survey data. In the main survey, nearly a quarter of the respondents identified themselves as foreign born, and 68% said they had moved to their present residence since the 1996 Census (although it was not clear how many moved there from within the policing area). In the follow-up residential hotel survey, an even greater share was found to be foreign born (38%). Eighty-nine percent of these people said they had arrived in this country in the last five years, after the 1996 Census.
According to the Census data, the 1996 ethnic breakdown of the two station areas is shown in Figures 2.1 and 2.2. Based on what we know about immigrant populations from other research5, today’s inner Johannesburg is likely to be more black, younger, and more male than the Census projections would indicate.
Figure 2.1: Ethnic breakdown, Johannesburg Central
Figure 2.2: Ethnic breakdown, Hillbrow
Indeed, the survey data revealed a population that conformed to all of these suppositions. The sample was 57% male. Most of the respondents were young adults: 83% were between the ages of 21 and 40 years. In terms of ethnicity and nationality, the sample in the initial survey was 92% black, 3% coloured, just over 2% Indian, and 3% white. In Johannesburg Central, the breakdown was 90% black, 5% coloured, 5% Indian, and no whites were interviewed. In Hillbrow, the breakdown was 93% black, 2% coloured, 1% Indian, and 4% white. The residential hotel sample was almost entirely black.
In the initial survey, 26% of the respondents were Zulu-speaking South Africans, 13% Sotho-speakers, and 10% Xhosa-speakers. Comparing language groups between the black South Africans to the national ratios determined in the 1996 Census shows that inner Johannesburg represents a kind of a microcosm of the overall black population, except for an overrepresentation of Sotho and Venda speakers (Figure 2.3).
Figure 2.3: Black South Africans by language group
Nearly a quarter of the sample in the initial survey were nationals of other African countries (271 individuals). A full 6% of the sample was Zimbabwean – 70 respondents. In addition, 26 Nigerians, 26 Basotho, 25 Mozambicans, ten Congolese, seven Ghanaians, and six Somalis were polled. Figure 2.4 shows the breakdown between South Africans and nationals from other parts of the continent in the initial survey.
Figure 2.4: Nationality and ethnicity
The residential hotel poll (n=201), as expected based on the author’s earlier qualitative work, had a much higher share of foreign nationals (38%), most of who were Nigerian (45 respondents). In fact, 35% of all male residential hotel respondents were Nigerian and in some hotels, over 80% of all male respondents were Nigerian. Figure 2.5 shows the ethnic breakdown of the residential hotel sample. The vast majority of these Nigerian hotel tenants (92%) classified themselves as Ibo, with most claiming their home area as Lagos (54%) or somewhere in the southeast portion of the country, including Port Harcourt and Calabar (32%). The significance of this finding will be discussed in Chapter 5.
Figure 2.5: Residential hotel residents by nationality and ethnicity
Geographic sampling
The inadequacy of the Census data leaves us with little choice but to base our sampling on something other than population demographics. Given the large size of the sample relative to the official population estimates (over 1%), the decision was taken to proceed geographically. Since this survey is about crime and policing, it seemed logical to select the sample from within the delineated police patrol areas, and this was done in Johannesburg Central. These areas were selected by police in large part to distribute workload evenly, and so each should represent an equivalent slice of the overall crime pie. For Hillbrow, where it was felt a more intensive sample was needed in an area of higher population density, smaller geographic units were used: the police Crime Analysis System data blocks.
Within these areas, a simple random sampling of households was done, and the interview conducted with the head of the household when available, or whoever answered the door if not, provided that person was able to speak on behalf of the household and was over the age of 18. All the questions were pre-coded and this information was transcribed into the SPSS data analysis software package.
The residential hotel poll was taken from hotels known to the author based on earlier research to be hot spots for the sale of drugs and sex. Clearly, this is not an unbiased sample, and this information should be treated as a set of case studies. The author was surprised to learn that many of these hotels had changed character in recent years, and no longer fit the model suggested by earlier qualitative work.6
Methodological limitations
Because the survey was spread evenly over police patrol areas in Johannesburg Central, and the patrol areas are smaller where crime is more likely to be reported, low crime areas were sampled more sparsely than high crime areas in terms of questionnaires per square kilometre. This might result in a slightly exaggerated crime picture. But it is also true that the smaller patrol areas corresponded to areas of higher population density, so this effect may not be significant.
It might be expected that our fieldworkers would encounter many of the same problems that Census workers would encounter in terms of locating and interviewing people who do not want to be recognised as residents. To determine national origin, respondents were simply asked “where do you come from?” The surveyors were all experienced black South African fieldworkers, many of whom had worked in Hillbrow before, and were encouraged to press the matter if the person did not appear to be local in origin (for example, if they showed an apparent lack of fluency in local languages). The high number of those who admitted to being foreign nationals suggests that this problem was largely overcome, but there remains the possibility that the immigrant population is even higher than a quarter of the total populace. If this is the case, the immigrant perspective was under-sampled.
Despite the general willingness of migrants to talk, a high level of refusals was encountered in the residential hotel survey, and fieldworkers were barred from entering some of the most notorious hotels. Safety considerations also prompted the avoidance of certain hotels by fieldworkers. This would suggest an under-sampling of some of the most crime-prone areas.
Because the survey was conducted in daylight hours, only those at home during the day were polled. This common problem often results in what is sometimes termed a ‘housewife survey’, in which only those involved in the care of the home during business hours are contacted. However, given the employment status of the respondents (Figures 2.6 and 2.7), this might be less of a problem in inner Johannesburg than in other areas. Many are apparently involved in work that does not preclude their being at home during the day.
Figure 2.6: Employment status, South African citizens
Figure 2.7: Employment status, non-South African citizens
Indeed, only 11% of the locals and 9% of the foreign nationals described themselves as housewives. Just under a third said they were employed full time in the formal sector, meaning that they paid taxes on their earnings, even if self-employed, and 58% of the respondents identified themselves as the head of the household.
Only 18% of those polled described themselves as unemployed, with just over a quarter being full time formal employees, another quarter being occasional or informal sector workers, and another quarter being students, housewives, or pensioners. This profile varied for non-South Africans, with greater participation in the informal sector and in casual, part-time work, and less in the formal sector, as students, and housewives. In the residential hotels, the employment profile was 44% informal self-employment and 7% formally employed full time, with only 10% describing themselves as unemployed. Among the Nigerian hotel residents, 74% described themselves as informally self-employed.
Levels of unemployment were about the same between locals and foreigners, which argues against the commonly held view that foreigners are here to sponge off the South African society. Questions about income were not asked, because this information is typically unreliable in areas where a large portion of the populace earns their living in informal or casual markets.
|
|
|
