Abstract
In this article, I explore the utility of effectively maintained inequality theory in examining educational inequality in South Africa at the end of the apartheid era. As an obviously unequal country, South Africa provides an excellent opportunity to test the claim that even with large quantitative differences in achievement, qualitative differences will matter. Using data from the early 1990s, I find that there were extensive quantitative differences in secondary school transitions across respondents in different racial categories. The minority White population was consistently able to achieve both more and better education. At the same time, though, qualitative distinctions mattered. For the majority of the population, particularly Africans, the quality of education attained varied across parental background. These outcomes are important not only for examining the veracity of effectively maintained inequality, both in terms of racial and class differences but also because they illustrate how educational differences have served to perpetuate inequality over time in a society that no longer allows for the explicit denial of opportunity by race.
Introduction
South Africa consistently ranks as one of the most unequal countries in the world, with many arguing that the country represents both a developed and developing country. The GINI coefficient, a measurement of income inequality, is, at .65, the highest currently measured (The World Bank, 2011). Life expectancy, health status, housing, and other key social outcomes also vary widely (Statistics South Africa, 2011). Education is no exception: While the overall average level of education in South Africa is quite high compared with other countries on the continent, particularly regarding the proportion of students who make it to secondary education (UNESCO, 2011), the distribution within the country is extremely uneven (Fiske & Ladd, 2004; Lu & Treiman, 2011). What is crucially important about educational inequality is its central role in determining most other forms of social inequality, both at any particular moment and over time. Differences in educational achievement change slowly. Most education is achieved early in life, and cohort replacement processes needed to change its distribution across a population are slow. This is true even during periods of rapid social transformation that change the distribution of political and economic opportunity, something that happened with the democratic transition in South Africa during the 1990s. Consequently, educational differences created from the mid-20th century onward have great importance for understanding continuing inequalities in contemporary South Africa.
In this article, I examine educational inequality in South Africa within the framework of EMI. In some ways, South Africa would seem an odd choice for examining this theory of educational attainment. Much of the inequality observed there can be attributed to the racist policies that began in the colonial era and were then fully realized under apartheid. This system of laws, only fully abolished in 1994, was quite explicit in enforcing different educational opportunities for residents depending on their official racial classification. The resultant disparity in years of educational attainment would be disputed by no one. As an obviously unequal country, however, it provides a central case to examine the possibility that despite large quantitative differences in achievement, qualitative differences still matter. As Lucas (2001, p. 1652) writes, “it is possible that even when quantitative differences are common, qualitative differences are also important; if so, I posit that the socioeconomically advantaged will use their socioeconomic advantages to secure both quantitatively and qualitatively better outcomes.” This is a generally untestable corollary of the theory, as most industrialized countries have such a small range of quantitative differences in educational achievement. South Africa provides a venue for examining this hypothesis. In addition, this country allows an examination of how EMI can be extended to additional qualitative distinctions beyond class background, in this case racial differences, and how these interact with class.
In this article, I specifically examine educational inequality at the end of the apartheid era. Using data from the early 1990s, I show that that while there were extensive quantitative differences across respondents in different racial categories, qualitative differences still existed. Race did matter: The minority White population was consistently able to achieve both more and better education. For the majority of the population, particularly Africans, though, quality of education varied across parental background. These outcomes are important not only for examining the veracity of EMI, both in terms of racial and class differences but also because they have helped perpetuate inequality over time in a society that no longer allows for the explicit denial of opportunity by race.
The Setting
South Africa is by far the most developed country in Africa, with an economy that dwarfs its neighbors—four and a half times larger than Angola, the next largest economy in southern Africa, and 50 times larger than that of Zimbabwe (The World Bank, 2011). It is also marked by great inequalities (Terreblanche, 2002). Two unique historical developments that shaped the nature of this inequality, and which are important for understanding the development of educational inequality, are the comparatively long history of the country as a mining and industrial power in the region, and the nature of race relations since the colonial era.
The primary cause behind South Africa’s position as the leading industrial economy in Africa is its early development as a source of precious minerals for the world economy. Colonial South Africa was initially settled by the Dutch as a farming outpost, meant to supply the trading ships making their way around the Cape en route to Asia. Dutch migrants settled on the southern tip of the continent to provide a stable source of food, and as a result, the Cape Colony grew into a significant agricultural economy with a small, but proportionally large, population of people of European descent (Magubane, 1979; Thompson, 1990). With the takeover of the colony by the British in the early 19th century, both the Cape and other newer settlements on the coast grew as more English settlers moved to the developing colony. In turn, the inland areas became more colonized as the descendants of the Dutch settlers moved into the interior to separate themselves from the new English authorities (Omer-Cooper, 1994).
The international fortunes of the economy changed greatly with the discovery of gold and diamond deposits at the end of the 19th century. These deposits would in the long run result in great wealth, yet in the short run required high levels of capital investment. Some of these funds were obtained to enable consolidation, for example, the use of European capital by Rhodes to consolidate diamond mining in the 1880s (Omer-Cooper, 1994; Pampallis, 1991). Much of it was for machinery and physical transformation of the landscape, though, as the particular geography of the diamond and gold deposits required heavy industrial investment before they could be profitably extracted (Omer-Cooper, 1994; Thompson, 1990). For this reason, the country has for the past 100 years been continuously located within important global economic and political systems of trade that otherwise bypassed much of the continent during that time (Lewis, 1990; Magubane, 1979).
A second consequence of the development of large-scale mining was the creation of an industrial labor market in South Africa. The mines not only required a large amount of capital to be profitable but also a great deal of labor. This labor would eventually come from Africans in the broader region, after they had been brought under colonial control through a series of wars against local political groups (Omer-Cooper, 1994). These wars dispossessed residents of their previous economy, forcing them to become dependent on wages from the mines (Browett, 1982; Lipton, 1985). They were additionally denied access to land for independent farming through legislative acts that granted such land to settlers of European descent (Terreblanche, 2002). Finally, the wealth generated by the mining sector led to the development of other industries that in turn generated additional labor demands. The rise in manufacturing shifted even more employment away from agriculture, so that by 1948 the majority of labor in South Africa was employed in industrial jobs (Browett, 1982). The demands of industry shaped the labor market throughout the apartheid era, which in turn affected the educational system, as discussed below.
The second unique aspect of South Africa that shapes economic inequality is the specific nature of race relations that have developed over its history. This complex system of race relations relates to its history as a settler society (Glaser, 2001). Europeans moved to the country for permanent residence as farmers, encouraged to do so by their home countries (Magubane, 1979). These settlers gradually increased their control over the region through wars and migration, establishing more widespread settlements. Consequently, there has always been, compared with the rest of Africa, a relatively large portion of residents of European ancestry. At the turn of the 20th century, this was approximately 26% of the total population, in 1948, approximately 21%, and at independence in 1994, 14% (Ross, 1999; Thompson, 1990). There has also been a distinct group of people with both European and African ancestry, who formed a population located socially in between the colonial authorities and the African groups. In addition, due to the country’s location within the British colonial system, there was immigration to the country from South Asia. This resulted in a substantial proportion of residents of Indian ancestry. The development of this four race 1 system under the colonial authorities was then linked to differential rights and opportunities for groups.
There were many ways that one’s racial identification shaped economic opportunities for South Africans across the course of the 20th century. One clear early example of this was the creation of differential rights of access to land, which was codified under the British authorities in the 1913 Land Act (Thompson, 1990). This act limited the rights of those of African descent to own land, solidifying White control of the best farmland in the country. Other examples include the Mines and Works Act and the Native Labour Regulation Act, both passed in 1911, that reserved certain jobs in the mines and railways for Whites (Omer-Cooper, 1994). The system of racial inequality became most developed, however, following the Nationalist Party victory in 1948. Over the subsequent decade this party codified a system of laws, known as apartheid, which classified all residents into one of four racial categories and systematically created differential systems of rights and opportunities for each. This system of legislation effectively linked racial classification and inequality for decades. Apartheid was greatly weakened during the 1980s, and officially dismantled in 1994 (Price, 1991; Ross, 1999). This system of laws, though, continues to greatly affect economic inequality since they shaped the early achievement of much of the current population. This is true not only economically but also in terms of their education.
The Development of Education in South Africa
South Africa has the most developed educational system on the continent. The country of over 50 million has nearly universal primary education, with approximately four million students attending 6,304 secondary schools, a Gross Enrollment Ratio of 91%, and over 800,000 students in one of the country’s 23 universities (South Africa Department of Basic Education, 2010a, 2013). These numbers are much higher than those found in any other country in the region. While today nearly universal and standardized, the educational system developed unevenly and sporadically in South Africa, in tandem with the changing demands of an industrializing society. Over the course of its development many politicians explicitly listed the labor market demands of an increasingly manufacturing-based economy as the reason for requiring at least a basic education for all residents (Fiske & Ladd, 2004). Conceptions of race and ethnicity held by those with political power also mattered greatly. Prior to the Second World War education consisted of a mixture of different systems, with three main distinct pieces. The first was federally supported public schools, mostly developed by the English colonial state after the formation of the Union of South Africa in 1902 (Christie, 1991; Omer-Cooper, 1994). The government imported teachers from abroad, and created a structure and curriculum to match the English system of education. These schools were primarily set up to educate Whites, with the additional purpose of further Anglicizing the culture of this new member of the British Empire (Christie, 1991; Rakometsi, 2008).
The second piece of the educational system was set up in reaction to the establishment of the English system of public schools. Many Whites in South Africa did not want a predominantly English curriculum with English as the medium of instruction (Christie, 1991). This group, mostly Afrikaaners but also other Whites of non-English descent, set up a parallel system of schools based on principles known as Christian National Education. These schools, primarily found in the provinces that were the former Boer states incorporated into the Union following the end of the English–Boer war in 1902, retained Afrikaans as the medium of instruction (Omer-Cooper, 1994). Their curriculum also varied from the federal public schools in ways that Afrikaaners felt allowed them to retain their own sense of history and identity (Fiske & Ladd, 2004).
These two systems primarily served to educate Whites, with the South African state providing little education for non-Whites. A third system of education, for Africans and other non-Whites, consisted primarily of schools established and run by Christian (usually foreign) missionaries (Christie, 1991; Hyslop, 1993; Martineau, 1997; Thompson, 1990). These included not only primary and secondary schools but also some colleges. The most preeminent among these, the South African Native College at Fort Hare, was founded in 1916 (Fiske & Ladd, 2004; Pampallis, 1991).
Soon after World War II, the entire educational system changed greatly with the imposition of apartheid. White schools were combined into one system, and federal control extended over them. The central piece of legislation that changed education for the majority South Africans was the Bantu Education Act of 1953 (Davis, 1972; Fiske & Ladd, 2004; Rakometsi, 2008). This act established extended federal control to education for Africans, forcing Christian organized schools to give up control or close (Christie, 1991; Davis, 1972; Hyslop, 1993; Kallaway, 1984; Thomas, 1996). This was extended in the 1960s to include the educational institutions of all non-White students, through the Coloured Education Act and Indian Education Act (Rakometsi, 2008; Ross, 1999; Wieder, 2001). This did initially lead to an increase in overall attendance, because even though education was not compulsory at first for African Blacks, there were so few attending even a year or two of primary school under the previous patchwork system of independent schools that enrollment could only rise (Hyslop, 1993; Louw, van der Berg, & Yu, 2006; Motala, Dieltiens, & Sayed, 2009; Thompson, 1990). The end result, though, was the expansion of already large inequalities in educational attainment by race.
The creation of separate schools for students of different racial classifications was accompanied by very unequal regulations, curriculum, and funding (Case & Yogo, 1999; Smith, 2011). There were different mandatory levels of education, with much lower requirements for non-Whites during this period (Motala et al., 2009). White schools emphasized more academic subjects, while schools for Africans had much lower academic expectations, and emphasized more “practical” subjects that prepared their pupils for blue-collar work futures (Fiske & Ladd, 2004, Hyslop, 1993; Kallaway, 1984; Maharaj, Kaufman, & Richter, 2000; Nkomo, 1990; Rakometsi, 2008). Schools teaching non-Whites received a fraction of the government expenditures that were given to White schools. In 1946, the government was paying more than 20 times per capita for White education as for Blacks (Maharaj et al., 2000; Thompson, 1990). This ratio did decline over time, although it never approached parity: By 1975, this had changed to a 15 to 1 ratio, and dropped to 4 to 1 by 1989 (Christie, 1991; Thomas, 1996). Consequently, African schools were much more crowded than those attended by Whites, and the physical plant significantly worse (Case & Yogo, 1999; Davis, 1972; Mncwabe, 1993). Finally, on average, the teachers in these schools were much less qualified than those found in White schools (Fiske & Ladd, 2004; Ross, 1999; Wieder, 2001).
Given these differences in educational opportunity across race, it comes as no surprise that previous studies of education have documented extreme inequalities in the amount of education attained by South Africans (Anderson, Case, & Lam, 2010; Lu & Treiman, 2011; Sibanda, 2005). Whites consistently show higher average levels of education, followed by Asians, Coloureds, and then Africans (Anderson et al., 2001; Louw et al., 2006; Thomas, 1996). Though differences between groups did decline greatly over the course of the 20th century, at independence, there were still vast interracial differences (Fiske & Ladd, 2004). The 1996 census, the first taken of the country as a whole, showed that the median educational level for Africans was only some primary school, with nearly one quarter having no education at all. In contrast, the median level for Whites was a complete high school education, with nearly one quarter having some postsecondary education (Statistics South Africa, 1999). Even though political change happened relatively quickly in the 1990s, changes in the distribution of education could not hope to keep pace because such a large proportion of those currently living in South Africa went to school during the apartheid era (or even before). The inequalities created then continue to affect the population even today, 20 years after the transition to democracy.
Another interesting finding regarding the increase in educational inequality during apartheid is that it was exacerbated due to rising levels of education among the most educated non-Whites, rather than a more general rise in educational attainment for one population group versus another (Louw et al., 2006; Nattrass & Ardington, 1990; Thomas, 1996). This inequality was tied to the goal of “separate development.” A central feature of the apartheid ideology, this was the idea that all four racial groups should develop as separate and “complete” societies (Thompson, 1990). Consequently, separate spaces and services were created, ranging from the establishment of different schools all the way up to the formation of separate political states. This policy had the consequence of providing occupational opportunities for a small portion of the non-White population, as separate institutions required a minimum corps of well-educated workers from each racial group (Crankshaw, 1997). Establishing Black hospitals required Black doctors and nurses, for example. The creation of separate homelands, which were supposed to function as separate countries, led to additional political, occupational, and eventually, educational opportunities for Africans (Dreyer, 1989; Rakometsi, 2008). Such jobs were a small proportion of overall employment for non-Whites, especially in rural areas; however, they did shape opportunity. Consequently, apartheid was not a straightforward denial of education for non-Whites, but instead a system that led to some Blacks achieving a high level of education, with the vast majority having little opportunity for learning.
The overall picture for educational inequality thus remains somewhat mixed. There are great differences by race, but also the potential for different levels of inequality within each racial group. Few studies have examined how quality of education might also vary as well as quantity. One exception is Case and Deaton (1999), who found that large pupil/teacher ratios, more commonly found in Black schools, greatly lowered math scores. In addition, the role of gender, age, and parental education is unclear. Previous studies have found smaller than expected differences by gender (Anderson et al., 2001; Thomas, 1996). Looking at age, while most studies of industrial societies find that younger generations are better educated, this relationship might be partially reversed in South Africa given the makeup of the educational system before and after the implementation of apartheid laws. Consequently, there might be an overall lower relationship between parent and child education than normal, given the radical change in the educational system after 1948.
Study Design and Data
To examine inequality in educational attainment, I focus on major transitions during students educational careers. The South African educational system loosely resembled that found in many former English colonies. It consists of 6 years of primary education, followed by 3 years of lower secondary, 3 years of higher secondary school, then university (Behr, 1988; Maharaj et al., 2000). Higher secondary school was not compulsory for anyone in the period I examine, and secondary schools are for the most part unified, with no parallel vocational system of schooling. The final 3 years of school culminated with a matriculation exams, or National Senior Certificate, that students must pass to be certified to attend university. In 2009, approximately 61% of graduating students passed this exam, which is roughly the same proportion as passed in 1994, although the intervening years had a great deal of variation around this number (South Africa Department of Basic Education, 2010a, 2010b). There are thus three transitions crucial for each student’s educational career: transition into lower secondary school, transition into higher secondary school, and transition into university. This last transition is relatively rare during the time period examined, particularly for non-Whites, and so will not be included in this analysis.
In comparison with many studies of educational transitions, these might seem like relatively low-level transitions. However, in the context of the South African labor market, these are crucial. Educational achievement there is still lower than found in most European countries, even though it is higher than other African societies. As shown in Table 1, the average number of years of education for the population as a whole is only 7 years. These numbers greatly differ by race. For example, while nearly all Whites had completed secondary school, with on average having 11 years of schooling, Africans had on average only 5 years. This was still useful for work, though, as achieving even a lower secondary education gave one a set of qualifications that helped get a better job (Treiman, McKeever, & Fodor, 1996).
Mean Years of Educational Attainment by Race.
Note: Standard deviation in parentheses.
Source: Adapted from Survey of Socioeconomic Opportunity and Achievement.
The data for the analysis come from a 1991 household survey conducted in South Africa, the Survey of Socioeconomic Opportunity and Achievement (Treiman, Moeno, & Schlemmer, 2001).The survey involved face-to-face interviews with a stratified random sample of individuals living in South Africa and the nominally independent “states” of Tanskei, Venda, Bophuthatswana, and Ciskei (the TVBC states). The TVBC states were not officially part of South Africa until they were reincorporated with the new constitution and multiracial elections of 1994; however, they were never truly independent countries. The instrument was conducted in English, Afrikaans, Zulu, and Sotho, and had a response rate of over 90%. This survey asked respondents about sources of income, job history, and other standard demographic questions. A stratified random sample of 9,086 respondents was collected. The data are weighted to be representative of the entire adult population of South Africa, including the TVBC states. Only those aged 20 years and older are included in the analysis.
All information on respondent’s education comes from a retrospective educational history that details each period of schooling. In the analysis, I examine the transition into lower secondary school, equivalent to attending seventh grade, and the transition into upper secondary education, equivalent to attending 10th grade. To test for inequality on qualitative dimensions of education I distinguish those who have completed either of these transitions by whether they studied more complex subjects during their lower or upper primary education, in this case math and science. Three divisions are created to compare those who studied math, those who additionally studied science, and those who studied neither. The dependent variables for each respondent thus have four possible values: did not attend that level of education, attended, attended and studied math, and attended and studied both math and science.
The independent variables used in the analysis are meant to capture the main influences on educational opportunity in South Africa. These include race, gender, age, urban location, and parental background. As detailed above, one’s racial classification greatly determined opportunities for schooling in South Africa. In this analysis, race is measured as a set of dummy variables distinguishing Africans, Asians, and Coloureds from Whites. Gender also greatly shaped educational opportunity in 20th-century South Africa, with women consistently having fewer opportunities than men (Martineau, 1997; Sibanda, 2005). In the analysis, gender is dummy coded as female. Educational opportunities were also more available in urban areas, so I include whether the respondent lived in an urban area at age ten 2 . Age has a potentially more complicated relationship to educational inequality. While educational opportunity increased over time in general, certain protests against the apartheid regime might have lowered educational attainment for some particular subgroups in the 1970s and 1980s (Fiske & Ladd, 2004; Maharaj et al., 2000; Rakometsi, 2008). Two dummy variables are added into the analysis to capture this potential effect. Those who were of age to start elementary school before the advent of the Bantu education system, and those who were of age to start elementary school after the time of the Soweto riot, which marked the beginning of the anti-education aspect of the anti-apartheid movement. These categories are related to age, however, not strongly enough to cause problems with multicollinearity.
Parental background is measured with two variables. The first is parental education, coded as the highest level of education attained by either parent. The second is father’s occupation, coded using a seven-category CASMIN or Erikson–Goldthorpe–Portocarero (EGP) coding scheme (Erikson & Goldthorpe, 1992). This is a condensed version of the standard class scheme used in comparative mobility research, one adapted for the less advanced industrial setting of South Africa. These classes are the service class (I + II, large proprietors, higher professionals, lower professionals, and higher and lower managers), routine nonmanual workers (III), self-employed with or without employees (IVa + IVb), farmers (IVc), skilled manual workers and manual supervisors (V, VI), semiskilled and unskilled manual workers (VIIa), and unskilled agricultural workers (VIIb). The use of a seven-category scheme does entail some loss of information; however, this coding scheme is necessary given the sample size and the nature of the labor market in this period. While a six-category scheme might have been preferable given the small sample sizes for the agricultural categories, in South Africa it is crucial to differentiate small farmers and farm laborers. This is true not only because of the standard of living enjoyed by holders of these two different occupations but also due to the way this distinction has been tied to race since the 1913 Land Act restricted agricultural land ownership to Whites throughout most of South Africa.
There are two modeling frameworks to assess EMI. The first, using order probit or ordered logit regression, is not appropriate for these data because preliminary analyses showed that the parallel regression assumption is not met. For this reason, I instead utilize a multinomal logit model framework. Although not as parsimonious, this model better fits these data. Examining the IIA assumption using different tests does reveal some conflict in determining whether this is the appropriate test. The Hausmann test shows mixed results, but the Small–Hsiao test consistently demonstrates that this assumption is not invalid. Thus, while not perfect, this method is preferable for examining these educational transitions.
Findings
Looking first at Table 2, the multinomial regression models show strong relationships between race and educational achievement. Non-Whites are much less likely to have made the transition into lower secondary school. Africans in particular are the least likely in the data to have gone to secondary schools that included studies in mathematics and/or science. The differences between those designated as non-White in South Africa are rather small, however, compared with the difference between any one of these groups and Whites. There is also a strong gender effect. While women are somewhat less likely to have started lower primary school than men, the big difference is that they are much less likely to have taken both mathematics and science class while there. Those in urban areas are much more likely to have attained any of these levels of schooling, and older respondents are less likely to have done so. The dummy variables testing for potential nonlinear time period effects were not statistically significant, once age is controlled for.
Mulitnomial Logistic Regression of Entry Into Lower Secondary Education.
Note. SE = standard error; Adapted from Survey of Socioeconomic Opportunity and Achievement.
There are positive effects of parental education across the board. Respondents whose parents had more education were more likely to obtain more, and better, educational levels. There are also substantial differences across father’s occupation. Compared with those who had fathers in the top two occupational groups, all respondents are less likely to have attended secondary school. The children of farmers, either small farm owners or agricultural workers, are by far the least likely to have attended any lower secondary school. Those with fathers who were working class or lower middle class were also less likely to have attended lower secondary school, while those whose fathers were working class less likely to have taken mathematics and/or science than children of lower middle-class fathers.
Another way of examining the importance of parental background on educational attainment, and one that more directly addresses the claims of EMI, is to look at the predicted probability for achieving these levels of education for respondents whose parents had different occupations. To do this, I set the values for the other variables as urban, male, of average age for those who attended lower secondary school (in this case, age 37). I compare Whites and Africans, excluding other non-White groups to simplify the graphs. I also examine two levels of parental background: those with parents who attended only through primary school and those who completed an upper secondary education. The resulting predicted probabilities are reported in Figure 1.
Predicted probabilities for lower secondary school attainment.
The data clearly show that parental background is crucially related to educational achievement for Africans in South Africa. The most common outcome for those whose parents had low education and lower occupational attainment is not attending lower secondary school, whereas respondents whose parents had less education but held better jobs are most likely to attend a lower secondary school. For those whose parents had more education, there is a smaller, but still pronounced difference between parents of different occupational levels. Regardless of parent’s job, these respondents are more likely to attend and also take math and science courses compared with other possible outcomes. Interestingly, there is little difference across occupation for Whites. The pattern of making the transition into lower secondary school is nearly identical across White respondents, no matter what the level of parental education or occupation.
Moving on to upper secondary school, the models shown in Table 3 again show that non-Whites are much less likely to have made the transition into this level of school. These respondents are again least likely to attend more academically rigorous upper secondary schools. Overall, Coloured respondents are much less likely to attend upper secondary school, across nearly all types of schooling, than all other groups, and Asians are more likely to attend rigorous schooling than other non-Whites. The data again show that women are less likely to make this educational transition, as are those in nonurban areas. The effect of age is again negative, with older respondents less likely to have this level of schooling, and again time period is not statistically significant.
Mulitnomial Logistic Regression of Entry Into Upper Secondary Education.
Note. SE = standard error; Restricted to those completing lower secondary school. Adapted from Survey of Socioeconomic Opportunity and Achievement.
Looking at parental background, again respondents whose parents were more educated are more likely to attend any type of secondary school, as are those whose parents were in the upper class. Children of farmers are the least likely to have attended higher secondary school, and respondents whose fathers were working class were also less likely to have taken mathematics and/or science than children of lower middle class fathers.
The predicted probabilities for different theoretical persons, displayed in Figure 2, show a similar story to that regarding middle school. As with Figure 1, these are reported for urban males, this time with an age of 33 years, though, as this is the average age for those who have attended upper secondary school. Again, the data show that parental occupation makes a difference on the type of education received for African men. Among those whose fathers attained only a lower level of education, not attending higher secondary school was the most common outcome, although it nearly as likely for those whose fathers had higher occupational standing to attend upper secondary school and take both math and science courses as not attend higher secondary school. Among the children of more highly educated fathers, African men with fathers who held working-class jobs were most likely to not attend secondary school at all, whereas those whose fathers held the top occupations were most likely to attend and take math and science courses. As was the case for middle school, for Whites there is little evidence that parental background makes much difference: The most common outcome for all Whites is to attend a better secondary school regardless of their family of origin.
Predicted probabilities for upper secondary school attainment.
Conclusion
The data on South Africa show support for the theory of EMI with regard to Africans. Not only is the respondent’s level of educational achievement related to the occupation of their parents, but also the quality of that education. This holds even after controlling for age, gender, and growing up in an urban area. Other research has shown that the level of education of caregivers and economic resources of the home are major predictors of educational success in South Africa (Anderson et al., 2001; Liddell & Rae, 2001; Sibanda, 2005; Thomas, 1996; Townsend, Madhavan, Tollman, Garenne, & Kahn, 2002). These findings additionally show the importance of these factors for determining how good that educational level is for the majority of South Africans.
The story for White respondents is quite different. For both of the levels of educational analyzed, there is very little variability for Whites in this era. The predominant outcome is attending the best secondary schools—no matter what educational or occupational background their parents had. This shows that apartheid not only led to the well-documented unequal outcomes in the quantity of education but additionally the quality. It also shows that in grossly unequal societies, where a small minority holds a considerable advantage in every arena, it is possible to reduce variation in schooling opportunities at earlier levels to eliminate much of the inequality within that one particular minority. This is not to say that important differences in educational attainment did not exist among Whites in South Africa. Instead, what is likely the case is that differences in quality could not be picked up by the crude measures in these data. Alternatively, they might only become visible in postsecondary education, which was not examined in this study.
By showing that inequality generally follows the pattern predicted by EMI, this study further supports the idea that qualitative differences matter even when quantitative differences are pervasive. In South Africa, this is important for better understanding both the perpetuation of racial inequality after the political and social transformation of 1994, but also for fully understanding educational (and subsequent social) inequality among Africans. While rights and opportunities have been made more equal, inequalities that existed within the educational system when most residents were in school have created a legacy of inequality within the labor market that cannot be quickly changed. It also affects the delivery of education today, since the teachers and culture of learning still remain from this era (Fiske & Ladd, 2004; Smith, 2011).
Footnotes
Acknowledgements
I am indebted to the other members of the research group, especially Samuel R. Lucas and Delma Bryne, for comments on earlier drafts. The meetings were supported by a grant from the EQUALSOC International Network of Excellence of the European Union 7th Framework Programme.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received the following financial support for the research, authorship, and/or publication of this article: The author received no direct funds for this research. However, this research was presented and discussed at a group meeting which was funded by the EU, as noted in the acknowledgment. Those funding sources did provide some funds for researchers from Europe.