Caroline Schaerer and Andrea Baranzini
of individuals with high education attainment at the district and hectare levels. As expected, we observe a greater variability of the distribution of foreign population and of high educated individuals when measured at the hectare level as compared with the district. In Figure 2, we also notice that foreign population can reach a concentration of 40 to 45% of the total population especially in the districts near the city centres. Of course, given the smaller size of the hectare, this proportion can even reach more than 60% in some hectares. See also OCSTAT (2005) for a detailed description of the relative concentration of foreigners by origin in the canton of Geneva.
Comparing Figure 2 with Figure 3, it appears at first glance an inverse relation- ship between the presence of foreigners and the level of educational attainment: the higher the share of foreigners, the lower the educational attainment level. Indeed, the correlation between the share of foreigners and the share of individuals with high education level at the district level amounts to -0.91 per cent in Zurich (-0.73 per cent in Geneva) and to -0.46 per cent at the hectare level (-0.22 per cent in Geneva). Such a path will be discussed in more detail in Section 3 below.
In order to characterise in more detail the distribution of the Geneva and Zurich population, in the next section we measure several segregation indices.
Measuring segregation: where are the people living?
The literature on residential segregation is considerable, particularly on the basis of race and ethnicity (see the influential work by Duncan and Duncan, 1955; Massey and Denton, 1988). This literature has developed measures of residential segregation by considering five different dimensions, i. e. evenness, exposure, concentration, centralization and clustering. For each of the dimensions, there exists a vast choice of segregation indices, from the single group indices, which refer to the segregation of one group with respect to the population as a whole; the two-group indices, which measure the segregation between two specified groups, to the more recent multi-group indices (see Reardon and Firebaugh, 2002). Recently, thanks to the development of the geographic information systems (GIS) technology, segregation indices have been extended in order to better account for the spatial distribution of the different groups within a city (see Reardon and O’Sullivan, 2004; Wong, 2003; Omer and Benenson, 2002). There is however little agreement on which measure should be best used in a specific context. Therefore, Massey and Denton (1988) recommend the adoption of different indices of segregation in order to account for the different facets of segregation. In this paper, we concentrate on the two most discussed and used dimensions of segregation, i. e. evenness and exposure, and we calculate aspatial indices. How-
© Seismo Verlag Zürich