Model

The dependent variable is constructed using the answer to the question ‘Were you, or anyone in your household, threatened or attacked because of your ethnic group, your tribe or your nationality during May/June 2008?’15 A probit model was chosen to deal with the binary character of the dependent variable. The following equation is estimated:

where

where X{ is a vector of characteristics for the household i and Wj control for characteristics of the ward j. The analysis is clustered by enumeration area to allow correlation of covariance within locations. When ward level characteristics are introduced as controls, the standard errors are corrected for a potential Moulton bias (1990) while clustering at the ward level.

Table 12.2 reports the key descriptive statistics and the differences between subsamples.

Demographic and migration history variables

The size of the household describes the number of persons, including children, sharing the same shelter and budget. Single migrants, especially young men, can potentially be perceived as outsiders in the sense that they come to work and do not struggle for their families - at least present with them - like most in the community.

A dummy variable is used whether the respondent was born in South Africa. For an in-depth analysis of the characteristics of foreigners in relation to victimization, the data can be broken down by country of origin. Foreigners are expected to face a higher chance of threat.

A binary variable is used to capture whether the household is composed of both South Africans and foreigners. On the one hand, it can be assumed that foreigners living with South Africans would be assimilated to insiders. For instance, it is assumed that these foreigners would have easier access to community life through their partners. On the other hand, foreigners who have South African partners can be perceived as competitors. In the aftermath of the violence, interviews with South Africans in the zones affected by violence revealed that foreigners were blamed for stealing women. In this case, mixed households would be more at risk than non-mixed households. A priori, the effect is unknown.

A binary variable captures whether the respondent speaks IsiZulu. This stands as a proxy for the household. Speaking IsiZulu, the dominant African street language, is expected to decrease the probability of being attacked or threatened for being an outsider. Potentially speaking the language facilitates integration and reduces recognizability.

The time since arrival in South Africa/in the location provides information on the level of integration. Long-term migrants should be more integrated in their host communities than new arrivals that have less time to settle and be involved in community activities. The former should therefore face a lower risk of being attacked. If the hypothesis were verified, it would corroborate the hypothesis that the recent influx of migrants could have been a major cause of the May 2008 attacks. Yet the length of time since arrival, particularly in Alexandra, could be correlated to the level of poverty of the household. The poorer the household, the longer it takes to afford moving to a safer neighbourhood (Richards et al. 2007). Meanwhile, we control for the different aspects of poverty. Both durations are de facto highly correlated. Therefore the time in the current location is preferred for the analysis. The impact of the time spent in South Africa is also considered for foreigners.

A binary variable is used to identify rural migrants. Under apartheid rules, rural migrants were perceived as ‘outsiders’ in Alexandra (Nieftagodien 2008).

Table 12.2 Descriptive statistics: differences between sub-samples

Total

Foreign vs native

Alexandra vs inner-city

Household size

3.99

  • -1.60
  • (0.22)***
  • 5.25
  • (0.21)***

Mixed household

0.23

  • 0.18
  • (0.02)***

Speak IsiZulu

0.72

  • 0.05
  • (0.02)**
  • 0.04
  • (0.02)**

Length in current location

10.49

  • 2.64
  • (0.18)***
  • -2.28
  • (0.18)***

Length in South Africa/Gauteng

11.69

  • 3.03
  • (0.16)***
  • -1.64
  • (0.17)***

Rural background

0.46

  • 0.15
  • (0.02)***
  • -0.29
  • (0.02)***

Secondary school education

0.63

  • -0.02
  • (0.02)
  • 0.24
  • (0.02)***

Wealth index

50.48

  • -9.50
  • (1.35)***

28.09 (1 17)***

Relative poverty dummy

0.35

  • 0.04
  • (0.02)**
  • -0.31
  • (0.02)***

Criminal record

0.03

  • 0.03
  • (0.01)***
  • -0.02
  • (0.01)*

Source: Author’s calculations.

Notes: Standard errors reported. *, **, ***: differences significant at 1%, 5%, and 10%.

The regime purposely created preferences towards residents. Migrants from rural areas in the country were consistently stigmatized into an out-group. Moreover, rural migrants are expected to have more difficulties in adapting to a new urban environment than migrants coming from urban centres, whether they originate from cities in South Africa, or elsewhere in Africa.

The data also captures the main purposes of migration, with economic reasons, educational reasons, escaping conflict or political oppression, and familial reunion being the most often cited. We can expect that different types of migrants face different degrees of vulnerability and victimization.

 
Source
< Prev   CONTENTS   Source   Next >