• Keine Ergebnisse gefunden

THE DISTRIBUTION OF ETHNIC GROUPS

As indicated previously, there is positive correlation between education and the size of the migration flows from the respective areas. Therefore, if education is a determinant

of migratory behavior, then the variation in educational achievement among provinces may explain the relative importance of Central and Nyanza provinces as sources of migra- tion and, conversely, the relative unimportance of Eastern and Coast provinces. On the other hand, there may be conditions within these provinces that determine both the educa- tional level and the propensity t o migrate.

In order t o check on such a common determinant of education and migration, a comparison was made of the distribution of major ethnic groups among provinces in the survey sample and the 1969 census (see Table 5.4). We assumed that the ratio ofmales aged 15 t o 5 0 to the total African population is the same within each province; the distribution of male Africans in each ethnic group within each province was considered t o be the expected value. The deviation of observed migrants in each ethnic group within each province from the corresponding expected values is statistically significant. The dominant deviations from expected values, in order o f importance (with an indication whether the observed value is above o r below the expected value) are as follows: Kikuyu, Central Province - above; other tribes, Rift Valley Province - below; Kisii and Kipsigi, Rift Valley Province - below; Kisii and Kipsigi, Nyanza Province - below; Luo, Nyanza Province - above; Embu and Meru, Eastern Province - below; other tribes, Eastern Province -below;

Kamba, Eastern Province - above; Luhya, Western Province - above. Therefore there is a distinct tendency for the Kikuyu and, t o a lesser extent, the Luo, Kamba, and Luhya t o have an above-average propensity t o migrate. Conversely, other tribes, the Kisii, Kipsigi, Embu, and Meru have a low propensity t o migrate.

These results are not fully consistent with those of Segal who seeks toexplain differ- ences in the extent of rural-to-urban migration among ethnic groups on the basis of their respective traditional cultural characteristics. According t o Segal (1970, pp. 107-108):

The major theoretical framework utilized in formulating this study's major hypothesis of ethnically based urban migratory differential was scale theory as explicated by Godfrey and Monica Wilson. The major aspect of the scale theory involved was the prediction that largescale traits will be associated with high urban migratory rates and that small-scale traits will be associated with lower ones. Assuming that m y identification of specific traits as large o r small in scale is accurate, this general prediction is borne out.

On the basis of lus analysis of the 1962 census data the Luo were seen t o have a higher propensity t o migrate than the Kikuyu.* It is possible that the ranking of the Luo above the Kikuyu in 1962, but the reverse in 1969, reflects the proximity of the 1962 census t o the emergency. Kikuyu had been forced out of some of the major urban centers, especially Nairobi, and Luo were brought in t o replace them. The 3 years between the lift- ing of the emergency regulations and the census may have been t o o short a period for a natural adjustment in the spatial allocation of people t o have occurred.

*The analysis of Segal is open to question because of errors in measurement. If as many Luo were urbanized as he claimed, more than 100 percent of the population in the major urban centers in Kenya would be Luo. This error need not totally invalidate his thesis because differences in rural and urban sex ratios among ethnic groups, rather than the number of migrants, is used to measure migra- tion in his analysis.

TABLE 5.4 The percentage distribution of ethnic groups in each province as reported in the migration survey and the 1969 census (males only).

Province Klkuyu Embu and Meru Kamba Luhya Luo Kisii and Kipsigi Coast tribes Other tribes Total Nyanza

Sample Census Western

Sample Census Rift Valley

Sample Census Central

Sample Census Eastern

Sample Census Coast

Sample Census Total

Sample 36.4 2.2 15.3 17.5 18.6 1.5 7.5 1.0 100

Census 20.4 6.3 11.1 13.5 14.3 11.1 7.1 16.2 1 0 0

SOURCE: Central Bureau of Statistics (1970, Table 11).

In our regression analysis, as reported in the previous chapter, five ethnic groupings

- Central Bantu, Western Bantu, Coastal Bantu, Nilotic, and other tribes - were entered as dummy variables in the polytomous logistic model of migration. In the analysis, Coastal Bantu was omitted so the coefficients for the other four indicate whether their propensities t o migrate differ from that of the Coastal Bantu. The results were not reported in Table 4.1 for three reasons: only the coefficient for other tribes was statistically significant, R' for each changed by one percentage point only, and several of the other explanatory vari- ables were then found t o be no longer significant. As expected, other tribes had a signifi- cantly below-average propensity t o migrate. For the ethnic groups that show considerable rural-urban migration there are not significant differences between them in migration propensities, given the effect of the other explanatory variables.

In conclusion then, variation between ethnic groups in their propensity t o migrate cannot be denied. Our analysis merely questions whether these differences are a function of cultural influences independent of the economic realities of each ethnic group. The regression coefficients for the ethnic groups that dominate the migration process typically are not significant, while other tribes (those not listed specifically) stand out as a distinct exception in our economic analysis; this indicates that Segal's thesis may be a better explanation for why some groups do not migrate than for why other groups d o migrate.