Black-White Differences in Achievement: The Importance of Wealth Amy J. Orr Linfield College In this article, the author argues that wealth, which is an indicator of both financial and human capital, can affect academic achievement, as well as help totoexplain the gap in black-white test Delivered by Ingenta User Unknown scores. Analyses reveal that wealth affects achievement through its effect on the amount of IP: 12.217.228.119 Date: 2004..12..08..18..05.. cultural capital to which a child is exposed. Because blacks have substantially less wealth than do whites, wealth can help to explain a portion of the racial achievement gap. The implications of the findings are discussed. S tudies have generally found that black students have fewer educational opportunities than do white students and are at a disadvantage compared to whites on several educational outcomes. Of particular importance is academic achievement, often measured by performance on standardized test scores. Black adolescents consistently have lower scores on standardized tests than do white adolescents. Although research has shown that the gap in black-white test scores has narrowed over the past 30 years, it is still substantial, and at the current rate of change, it may take approximately 50 years to close the gap (Hedges and Nowell 1998). Because of the importance of test scores in determining college attendance and selectivity, wages, and job placement (Kane 1998), the factors that lead to inequalities in test scores must be uncovered so that they may be alleviated. For many years, researchers have been searching for explanations as to why blacks, on average, have lower achievement than do whites. One major finding has been that much of the differences can be attributed to black-white differences in family background characteristics, such as socioeconomic status (SES) and family size or composition. Wealth, defined as the value of all assets of a particular family or individual minus any debts owed, has not received sufficient attention. Blacks have less wealth than do whites, regardless of income, education, and occupation (Blau and Graham 1990; Eller 1994; Merida 1995; Oliver and Shapiro 1989, 1995). For example, Merida (1995) found that middle-class blacks have only 15 percent of the wealth held by middle-class whites (when class is measured with the traditional three factors). While researchers have paid extensive attention to SES as a determinant of achievement, wealth is often not included in conceptualizations of this factor. Rather, income, educational attainment, and occupational prestige have been viewed as sufficient indicators of economic well-being and social status. It has only been recently that wealth has been viewed as an important component of SES and as sufficiently different from income to demand our attention when discussing important educational outcomes. The most recent research conducted in this area was by Conley (1999, 2001) and Phillips, Brooks-Gunn et al. (1998). Conley found significant effects of wealth on several academic outcomes, including educational attainment (high school graduation and college completion), the probability of dropping out, the risk of expulsion, and the risk of being held back a grade. Conley did not consider academic achievement, although he alluded to its possible connection to wealth in 2001, when he posed the question: “Can an effect of wealth be detected in students’ aspirations, educa- Sociology of Education 2003, Vol. 76 (October): 281–304 281 282 Orr tional expectations, and even performance?” used to create educational opportunities. For (p. 70). Phillips, Brooks-Gunn et al. (1998) example, assets, in the form of savings began to answer this question by including accounts, certificates of deposit (CDs), indiwealth in models of academic achievement vidual retirement accounts (IRAs), stocks, and among children aged five to six. Their failure bonds, produce a flow of income through to find an effect of wealth may have been the interest and dividends. Having investments in result of the young age of their sample. the form of businesses, farms, or real estate Perhaps wealth does have an effect on can also produce a profit that can be used to achievement, but at a later age. I explore this expand educational opportunities. possibility via analyses of theDelivered academicby Ingenta Other to types of assets can either help to achievement of an older sample of children. I minimize expenses coming out of a family’s User Unknown argue that wealth is not only a determinant of earned income or can serve as a stockpile of IP: 12.217.228.119 achievement, in general, but can help to resources that are available when needed Date: 2004..12..08..18..05.. explain the gap in black-white test scores. (Oliver and Shapiro 1995). For example, although few may view home ownership or the ownership of a car as a way to expand educational opportunities, owning these THEORY types of assets can provide financial advanThe importance of wealth may be best under- tages that nonownership does not. Home stood by using broad notions of capital, such owners are provided with tax deductions, as those used by Bourdieu (1986). According exclusions, and deferrals. While home owners to Bourdieu, the forms of capital are convert- may pay mortgages, part of this expense ible, although they differ in their liquidity, or taken from earned income is returned to the ease by which they can be converted. them through special tax breaks that are not Some, like economic capital, are more liquid available to renters. The money that is than others and can be more easily converted returned to home owners, or not paid to the to a different form (in this case, social or cul- government, can be used to expand educatural capital). As a form of economic capital, tional opportunities. Once a mortgage is paid then, wealth can be used as both a direct off, the money once used to pay it can be financial resource and can be converted into diverted into academic resources. Renters other types of capital, either cultural or social. continue to pay this monthly expense, which Wealth, as economic capital, can help parents takes away finances that could be used to pay to purchase many important educational for the education of their children. In addition to its direct impact on a famiresources, such as books, computers, private schooling, and various status symbols. Wealth ly’s economic well-being by producing can also be cashed in or converted to give the income and/or minimizing expenses that next generation some degree of cultural, as come out of a family’s earned income, wealth well as social, capital (for example, by allow- can benefit a family by providing a sense of ing parents to spend more time with their economic security (Oliver & Shapiro 1995). children). The presence of various forms of Parents may be more willing to spend earned capital can contribute positively to a child’s income on items that they do not deem to be necessities if they have financial resources achievement. that they can fall back on in times of need. Many educational resources can be viewed as Wealth as Economic (Financial) “wants,” rather than “needs.” Families who Capital feel financially secure, because of wealth, may Wealth can contribute to the financial capital be more likely to invest in these items. Although wealth can affect academic available to a family. Assets can produce income and/or can minimize expenses that achievement in many of the same ways that may normally come out of a family’s earned income does, it may actually intensify these income. Certain types of assets can translate effects. Two families with similar incomes but directly into added income, which can be different amounts of wealth may be unable to Black-White Differences in Achievement 283 provide the same opportunities for their chil- has consequences for the quality of public dren. While middle-income families provide schools. Wealth is associated with higher more opportunities for their children than do property taxes, which fund local school dislow-income families, middle-income families tricts. A larger tax base generally translates with a greater amount of wealth can provide into a higher per-pupil expenditure. Children more opportunities for their children than can who attend wealthier public schools can middle-income families with a lesser amount of expect a lower student-to-teacher ratio; highwealth. Children from families with little wealth er-quality teachers; better facilities; and the may have a decreased chance of attending col- presence of educational resources, such as Delivered Ingenta toand up-to-date books (see, e.g., lege, less access to high-quality schools, fewerby computers User Unknown educational resources in their homes, fewer Education Week 1998). IP: 12.217.228.119 opportunities to accumulate cultural capital, and limited access to items that may Date: increase 2004..12..08..18..05.. Educational Resources Wealth can detersocial status. mine the amount of educational resources available to a child in the home. Greater College Expectations Wealth can affect the wealth can lead to greater educational academic achievement of a child through its resources; for example, families who have effects on the child’s motivation, aspirations, wealth may be better able to provide books, and expectations. For example, home owners computers, or tutors for their children. With have an advantage over non-home owners in such academic materials at home, children’s that they have the ability to take out a second learning time can be increased. For instance, mortgage and have increased creditworthi- children can spend time outside school readness. A second mortgage can directly trans- ing or working on a computer, thereby gainlate into extra finances for educational expen- ing knowledge that they may not have ditures. Increased creditworthiness can learned in school (Downey 1995; Jones 1984; improve the likelihood of being approved for Milne et al. 1986). an educational loan. Parents may be more likely to encourage a child to plan for college Cultural Capital Wealth, as economic capiif they are certain that the finances will be tal, can be cashed in to obtain cultural capital available either from a second mortgage or (Bourdieu 1986), defined as “instruments for an educational loan, if not from income. the appropriation of symbolic wealth socially Parents’ encouragement to attend college designated as worthy of being sought and can affect a child’s motivations and aspira- possessed” (Bourdieu 1977:488). DiMaggio tions. Previous research has shown that both (1982) showed that cultural capital, meamotivations and aspirations affect effort, sured by attendance at symphony concerts which, in turn, affects academic achievement and arts performances or exhibitions and a (Natriello and McDill 1986).1 cultivated self-image, has a positive effect on academic achievement. Whether a child has Quality of Schooling The lack of additional access to these types of opportunities can financial resources may limit the ability to depend on the amount of wealth available to send children to private schools. Research has a family; children from families with little shown that attendance at a private school wealth may have less access to cultural capican have a positive effect on a child’s achieve- tal. ment (Bryk, Lee, and Holland 1993; Hoffer 1997–98). The cost of private schooling has Social Acceptance and Self-esteem Several been estimated to be over $4,000 per year, scholars have recognized the importance of which can be more than a quarter of the wealth in contributing to the degree of an income of a moderate-income family individual’s social acceptance. For example, (Coleman, Schiller, and Schneider 1993; Warner (1949/1994:195) noted that wealth Tracy 2001). Having additional income from could be put to use to acquire “symbols that assets can help to absorb this cost. can ultimately be translated into social accepAs Conley (1999) pointed out, wealth also tance by those who have sufficient money to 284 Orr Particularly for adolescents, the social status aspire to higher levels than they presently conferred by having the best house in the enjoy.” These symbols may include “the neighborhood may instill confidence that ‘right’ house, the ‘right’ neighborhood,’ could translate to success in an academic set[and] the ‘right’ furniture” (Warner ting. Conversely, living in conditions that are 1949/1994:195). Weber (1946:188) also disinferior to those of one’s peers may cause anxcussed the connection between wealth and iety and a lack of confidence in one’s family social acceptance, stating that “only the resiand oneself. Since underlying conceptions of dent of a certain street (‘the street’) is considpoverty often rely on a notion of relativity, housing quality may be the most visible way ered as belonging to ‘society’, is qualified for Delivered to poverty (or affluence) manifests that relative social intercourse, and is visited and invited.”by Ingenta User Unknown itself in the lives of adolescents and with Although these authors’ comments were respect to their primary group affiliations. (p. IP: 12.217.228.119 directed to the social acceptance of adults, 65) Date: there is reason to believe that wealth can2004..12..08..18..05.. also contribute to the social acceptance of children. Human Capital and Wealth Previous research has shown that an “adolescent society” exists within schools Human capital theory stresses the importance (Coleman 1961; Eitzen 1975; Goldberg and of human characteristics as resources that are Chandler 1989; Schultz 1990) and that with- available to expand an individual’s opportuniin this society, there is a status system based ties. Certain individuals have advantages over on various criteria. Children place a fairly high others because they have attributes that are value on participation in athletics and other valuable in attaining particular goals. Human extracurricular activities and personal popu- capital theory could help to explain the effect larity. Popularity is often based on the types of wealth on academic achievement, as well of material possessions one has and the as on the gap in black-white test scores. The clothes that one wears (Schultz 1990). For characteristics of parents with wealth may differ in important ways from those of parents example, Ferguson (1999) reported that in a with little or no wealth. particular community, 82 percent of the One way in which wealthy parents may young black men, 65 percent of the young differ from less-wealthy parents is that they white men, 81 percent of the young black may be more oriented toward and concerned women, and 80 percent of the young white about the future. This concern may lead them women stated that having “cool clothes” to start planning for and investing assets for makes a student popular. The ability to have the future early because they know that their these types of “status symbols” can be affectpresent behavior may greatly affect their ed by wealth. If children or their families do future status. However, financial security may not have discretionary income at their dispos- not be the only goal of “future-oriented” indial, children may be unable to keep up with viduals. It is likely that an orientation toward the latest fashions or trends, which can be the future affects other areas of their lives as costly. Having these items may increase their well, such as a concern about the education popularity or social status, which has been of their children. shown to have a positive effect on children’s Parents with a strong sense of the impact self-esteem (Walker and Greene 1986). of their present behavior on future outcomes Research has shown that positive self-esteem may be more likely to invest in their children’s can lead to higher academic achievement education than may those who do not have (Liu, Kaplan, and Risser 1992; Sterbin and this sense or have lost hope in the Rakow 1996). present–future connection.2 Knowing how Wealth may also affect self-esteem in important education is to several later outanother way. As Conley (1999:65) pointed comes, future-oriented parents may be more out, housing (an integral part of wealth) likely to make monetary investments in edu“serves as an outward sign of social status.” cation, as well as in relationships and experiAs he explained: ences that may help their children to achieve Black-White Differences in Achievement 285 better. For example, parents with this of the arts, music, and literature, such as takattribute may be more likely to be concerned ing them to museums or art performances. about the quality of their children’s school or “Status-seeking” parents may not only the educational resources available to the provide their child with knowledge of or children at home. They may also be more experience with the arts, music, and literalikely to spend time participating in educa- ture, but may reproduce this characteristic in tional activities that can benefit their children their children, instilling in them an interest in academically, such as reading books to them, these areas. By increasing the cultural capital helping them with their homework, or get- available to and possessed by their children, Delivered Ingenta to may be benefiting their chilting involved at the school. This type ofby these parents User Unknown parental support, commonly referred to as dren academically. As I discussed earlier, IP:positively 12.217.228.119 social capital (Coleman 1988), can teachers tend to have higher expectations of affect children’s achievement. Date: 2004..12..08..18..05.. children who have cultural capital. These The link between having parents who are higher expectations can have a positive affect oriented toward the future and high academ- on a child’s achievement. In light of these ic achievement may not be due solely to the arguments, I posit the following hypothesis: parents’ actions. Such parents can serve as Hypothesis 1: After parental education, occurole models for their children, possibly pation, and income are controlled, parental increasing the children’s chances of becomwealth will have a positive effect on a child’s ing future oriented. An orientation toward achievement. the future could lead to increased motivation to do well in school, improving the likelihood of academic success. Wealth and Racial Differences in Parents with wealth may also differ from those with little or no wealth in their concern Achievement for social status. Owning a home or other Since blacks have less wealth than do whites, types of assets is a way to improve one’s pres- they may be at an educational disadvantage tige. Those who invest in assets may be more even if their incomes are equal to those of concerned about their social standing, inde- their white counterparts. Coming from famipendent of their concern with financial welllies with lower levels of wealth, black children being. Veblen (1934:34) discussed the conmay have less access to financial, human, culcern for status as a motivation for the accutural, and social capital. With lower levels of mulation of wealth when he stated, “Among financial capital, they may have less access to the motives which lead men to accumulate high-quality schools, to educational resources wealth, the primacy, both in scope and intenin their homes, to items and experiences that sity, therefore, continues to belong to this could increase their cultural capital, to colmotive of pecuniary emulation.” He argued lege, and to status symbols. In addition, they that individuals who accumulate assets are may have lower motivation, aspirations, and likely to be concerned about the views of othself-esteem. Limited access to the human capers and to use the ownership of assets as a way to “gain the esteem and envy of [their] ital that is available in wealthy families can lead to lower levels of cultural and social capfellowmen” (p. 32). There are several ways in which an individ- ital and hence to lower achievement. ual can improve his or her social status. Therefore, because black and white families Investing in assets is one, and having cultural differ substantially in the amount of wealth capital is another. Individuals who seek status they possess, wealth may help explain blackmay be motivated to increase their cultural white differentials in achievement. capital, as well as to invest in the cultural capHypothesis 1, Corollary 1: After controlling for ital of their children (see, e.g., Bourdieu parental education, occupation, and income, 1986). These parents may be especially likely increases in the gap in wealth between blacks to provide their children with experiences and whites will lead to increases in the gap that will increase their children’s knowledge between blacks’ and whites’ achievement. 286 Orr Impact of Wealth on Older Children’s Achievement children progress through school. For example, Phillips, Crouse, and Ralph (1998) found that from Grade 1 to Grade 12, the gap in Wealth may affect the achievement of older mathematics scores widened by 0.18 stanchildren differently from that of younger chil- dard deviations. Although the gap in reading dren for several reasons. First, older children scores remained fairly constant, the gap in are more aware of costs and constraints. vocabulary scores widened by approximately Aspirations and plans to attend college may 0.23 standard deviations.3 These findings ebb as children begin to realize the costs of imply that a black student who enters school Delivered Ingenta attending college and their parents’ capacityby with the to same test scores as a comparable User Unknown to pay for it. white student will be likely to have lower test Second, wealth may have IP: a stronger 12.217.228.119 scores than the white student at the time he impact on adolescents than Date: it does2004..12..08..18..05.. on or she leaves the 12th grade. younger children because adolescents identiThe arguments presented in the previous fy more strongly with their peers and less with section help to explain why the achievement their parents (Steinberg and Silverberg 1986). gap between black and white students may During early adolescence, children have high widen during the later years of schooling. levels of conformity with their peers (Berndt Because of their lower levels of wealth, black 1979; Costanzo 1970) and are sensitive about parents may be less likely to finance their chilhow they are viewed by others (Elkind 1967; dren’s college education directly or absorb Okun and Sasfy 1977). Therefore, fitting in the cost of college loans. This argument was with peers can play an important role in how supported by Hauser and Anderson (1991), adolescents feel about themselves. Popularity who argued that a shift from grants to college at this point in their lives can often be based loans in the 1980s may have been responsible on the type of car that they drive and the for the decline in the number of black seniors clothes that they wear (Ferguson 1999; who aspired to attend college. If black adoSchultz 1990), both of which can be affected lescents are aware that their parents cannot by wealth. As with younger children, if ado- contribute financially to their education, they lescents or their families do not have discre- may have low expectations and may theretionary income at their disposal, they may fore not be motivated to do well. In addition, have less access to items that can increase the inability to keep up with other students in their popularity. Since they place more the way of cars and clothes, as well as living emphasis on conforming to their peers than in “inferior” housing, can have a negative do younger children, they may be more effect on the self-esteem or self-confidence of affected by wealth. Without wealth, having a older black children and, in turn, could lead “nice” car (or any car, for that matter) may to lower academic achievement (Liu et al. not be feasible. If discretionary income is not 1992; Sterbin and Rakow 1996) and conavailable, adolescents may also be unable to tribute to the widening achievement gap afford the latest fashions. Having these items between black and white students as they may increase their popularity or social status, grow older. which has been shown to have a positive Hypothesis 2, Corollary 1: The effect of wealth effect on adolescents’ self-esteem (Walker and on racial differences in achievement will Greene 1986). increase as children grow older. Hypothesis 2: The effect of wealth on achievement will increase as a child grows older. Widening of the Achievement Gap in Later Grades Research has shown that the gap in test scores between blacks and whites widens as The Importance of Types of Assets Not all assets may contribute equally to a child’s academic achievement. While wealth, in general, is easily convertible, some assets are more readily converted into ready cash than are others. It is these assets that may be Black-White Differences in Achievement 287 the most useful in contributing to educational opportunities while a child is attending school. Assets that produce income, such as CDs, stocks, bonds, and savings accounts, can be used for a family’s immediate wellbeing, whereas assets like homes and vehicles are less likely to be converted into cash unless it is absolutely necessary to do so (see Oliver and Shapiro 1995).4 from school surveys, transcripts of interviews, and interviewers’ observations of the children’s home environments. This data set was chosen for the analyses because of the appropriateness of the measures it contains. While educational data sets, such as the National Educational Longitudinal Survey of 1988 and High School and Beyond, contain a wide range of information about a Delivered by child’s Ingenta to education, they typically do not conHypothesis 3: Wealth in the form of incomeUser Unknown tain measures of family wealth. Data sets that producing assets will have a greater effect on IP: 12.217.228.119 achievement than will wealth in the form of do contain data on wealth, such as the Panel Study of Income Dynamics (PSID) and the non-income-producing assets. Date: 2004..12..08..18..05.. Survey of Income and Program Participation, Blacks are less likely to invest in income- have historically not collected adequate eduproducing assets than are whites (Oliver and cational data. 5 The NLSY79 Mothers and Shapiro 1995). Hence, black families may Children data set contains data not only on have fewer assets that can be converted easi- wealth, but on education.6 ly into cash for educational expenditures. The sample contained school-aged chilTherefore, black-white differences in academ- dren (aged 5–14) who were given the ic achievement may be a result not only of Mathematics portion of the Peabody the overall value of assets, but of the types of Individual Achievement Test (PIAT) in 1996. assets owned. The children selected were black and white (Caucasian); Hispanics and Asians, as well as Hypothesis 3, Corollary 1: The greater the black-white differences in the amount of other ethnic groups, were excluded from the wealth held in income-producing assets, the sample. Children for whom there were missgreater the black-white differences in achieve- ing data on one or more of the variables of interest were also excluded. See the appendix ment. for weighted descriptive statistics for the limited sample (the sample produced after listwise deletion of missing data) and the full sample (prior to listwise deletion); the two METHOD samples did not differ significantly. Data and Sample The data used in the analyses came from a subset of the data set of the National Longitudinal Survey of Youth (NLSY79) entitled “NLSY79 Mothers and Children.” This subset includes approximately 3,000 women who were aged 14–21 in 1979 and had become mothers since that time. These women were interviewed annually from 1979 through 2002. In addition to information about the women, the data set contains information about their spouses and children. The data on the children, which have been collected biannually since 1986, were derived from interviews with the mothers; interviews with the children; and various socioemotional, cognitive, and physiological assessments. These data also contain information obtained Variables Dependent Variable The dependent variable in the analyses was academic achievement, measured by standardized scores on the Mathematics subscale of the PIAT in 1996.7 The PIAT, which consists of three components (Mathematics, Reading Comprehension, and Reading Recognition), is a widely used brief assessment of achievement and ability (Baker et al. 1993) that has been administered to a large number of children over a wide range of ages. Since the PIAT allows for comparability across ages, it is attractive as a measure of academic achievement. As with the other components of the PIAT, the Mathematics Assessment has been used extensively as a measure of a child’s achieve- 288 Orr ment (see, e.g., Baharudin and Luster 1998; these commodities. The second variable was Hao 1995; Kaestner 1997; Luster and created by taking the value of non-incomeMcAdoo 1994; Rodgers, Rowe, and May producing assets (the home and the vehicle) 1994). The Mathematics subscale is consid- and subtracting any debt owed on these posered to be a highly reliable and valid assess- sessions. Again, the two wealth variables ment, having a test-retest reliability that reflect a five-year average (1990–94). ranges from 0.73 for 5th graders to 0.84 for SES was measured by three variables: fam12th graders with an across-grade median of ily income (a continuous variable ranging 0.74 (see Baker et al. 1993; Dunn and from $0 to $100,000 or more), parental eduDelivered Ingentaand to parental occupation. The final Markwardt 1970). The assessment has a highby cation, Unknown cross-year reliability and a highUser concurrent income variable used in the analyses was IP: 12.217.228.119 correlation with the full PIAT composite score average family income over a five-year period (Baker et al. 1993). Date: 2004..12..08..18..05.. (1990–94). Parental education (a continuous The PIAT Mathematics Assessment “mea- variable ranging from 0 to 20, with 0 being sures a child’s attainment in mathematics as “no education” and 20 being “8 years of coltaught in mainstream education” (Baker et al. lege or more”) was derived by taking the 1993). The assessment consists of 84 increas- average of the mother’s education and the ingly difficult multiple-choice items. Skills mother’s spouse’s education. Parental occumeasured range from recognizing numerals pation, the average of the mother’s score on to measuring concepts of advanced geome- the Duncan Index and the mother’s spouse’s try and trigonometry. score, ranged from 1 to 97. In the analyses, the family occupation variable reflects an Independent Variables The primary inde- average of Duncan scores from 1990 to 1994. pendent variables in the analyses were wealth Race was coded 1 if the child was black and 0 (or net worth), SES, and race. Wealth vari- if the child was white. ables were created by adding the value of all household assets and subtracting household Mediating Variables The mediating varidebt. Assets included (1) the market value of ables in the analyses were cultural capital, residential property; (2) the total amount of educational resources in the home, social money assets, such as stocks, bonds, CDS, capital, child self-esteem, and school qualiIRAs, Keogh accounts, tax-deferred plans, and ty.10 Cultural capital was measured by a comsavings accounts; (3) the market value of posite variable comprised of 2–5 individual vehicles; (4) the market value of farm, busi- variables, depending on the age of the ness, and other property; (5) the market value child.11 For every child, the individual variof estates, trusts, and inheritances; and (6) ables were recoded to dichotomous variables the value of other assets over $500. Debts and then averaged. For children aged 5, the included (1) the amount of mortgages and variables averaged were “the child has gone back taxes owed on residential property; (2) on an outing with parents at least once a the amount of other debt on property; (3) the month” and “the child has been taken to a amount of debt on farm, business, and other museum at least once or twice in the past property; (4) the amount of money owed on year.” For children aged 6–9, the variables vehicles; and (5) the total amount of other averaged were “there is a musical instrument debts over $500. The wealth variables were that the child can use at home,” “the child created for five years (1990–94).8 The final gets special lessons or activities,” “the child wealth variable used in the analyses was the has been taken to a museum at least once or average of the family’s net worth over this twice during the past year,” and “the child period.9 has been taken to a performance during the Wealth was divided into two variables to past year.” Finally, for children aged 10 or test Hypothesis 3. The first variable measured older, the variables averaged were “the child the net worth of income-producing assets. It has been taken to a museum at least once or was created by taking the value of all income- twice during the past year,” “the child has producing assets minus any debts owed on gone on an outing with the parent(s) in the Black-White Differences in Achievement 289 past month,” “the child receives special vidual items. The greater the child’s score, the lessons or activities,” “the child has been greater the child’s general self-worth is said to taken to the theater at least once or twice be. The original score received on this assessduring the past year,” and “there is a musical ment was recoded to a range of 0 to 1, so instrument that the child can use at home.” that all cases, regardless of age, had the same In all cases, the range of this variable is 0 to 1. range on the general self-esteem variable. Educational resources were also measured Quality of education was measured by by a composite variable. This variable was school sector and school location. School secconstructed in the same manner as cultural tor was coded 1 if the school was public and Delivered to capital. For children aged 5, the variablesby 0Ingenta if the school was private. School location User Unknown averaged for the educational-resources com- was coded 1 if the school was in an urban 12.217.228.119 posite were “the child has 10 or IP: more books area and 0 otherwise.12 at home,” “the family receives two Date: or 2004..12..08..18..05.. more magazines regularly,” and “the child has a Control Variables. Other important variables record or tape player.” For children aged 6–9, that were included in the analyses were the the variables averaged were “the child has 10 child’s age, mother’s age, mother’s score on or more books at home” and “the family the Armed Forces Qualification Test (AFQT), receives a daily newspaper.” Finally, for chil- the child’s gender, number of the child’s sibdren aged 10 or older, the variables averaged lings, and the marital status of the parents. were “the child has 10 or more books at The child’s age and mother’s age were conhome,” “the family receives a daily newspa- tinuous variables measured in years. Gender per,” and “the child has a computer at was coded 1 if the child was female and 0 if home.” In all cases, the range of this variable the child was male. The number of siblings was 0 to 1. ranged from 0 to 9. Marital status was coded The measurement of social capital differed 1 if the mother was married and 0 otherwise. according to the children’s ages. For a child In the analysis, the variable reflects the numaged 5–9, social capital was measured by ber of years the mother had been married how often the mother read to the child, over a five-year period (1990–94). coded 1 if the mother read to the child at least three times a week and 0 if less often. Procedures For a child aged 10 or older, social capital was measured by whether or not the parent or Ordinary least-squares (OLS) regression was parents had helped the child with homework used to test the hypotheses. Since the cases were not independent (in many cases, there in the past week, coded 1 if yes and 0 if no. For children younger than age 8, self- were multiple children per household), the esteem was measured by maternal responses tests were based on a heteroscedasticity-conto statements that dealt with whether or not sistent covariance matrix. Specifically, HCO the child felt worthless or inferior and was used (see White, 1980). Several models whether or not the child was liked by other were estimated; each model is explained in children—coded 1 if the child did not feel detail in the following section. When I comworthless or inferior and 1 if the child was pared the models, I considered changes in liked by other children, 0 otherwise. These the sizes of the regression coefficients and scores were averaged to create a self-esteem examined the statistical significance of the score that ranged from 0 to 1. If a child was coefficients. It is important to note that the natural log aged 8 or older, the self-esteem score was the child’s raw score for self-worth on the Self- of income and the natural log of wealth were Perception Profile for Children, which mea- used when the variables were included in the sures a child’s sense of general self-worth and analyses. It is widely believed that the relaself-competence. For this study, the child’s tionship between wealth and achievement score on the general self-worth subscale was outcomes, as well as that between income used. This subscale includes six items, scored and achievement, is nonlinear (Conley 1999; 1 to 4; the raw score is the sum of the indi- Henretta and Campbell 1978; Mayer 1997). 290 Orr For example, when considering the purchase Descriptive statistics are also presented for of educational resources, a $1,000 increase in blacks and whites separately. The two groups wealth may mean more to a family who has a differed in several areas. The black children net worth of only $3,000 than to a family scored an average of 8.09 points lower on the who has a net worth of $100,000. For the PIAT Mathematics Assessment than did the family with a lower net worth, an extra white children (a standard deviation gap of $1,000 is likely to have more of an impact on 0.62). Over a 5-year period, the black mothwhether money is put toward the education ers were married for approximately 2 years, of a child than it would for a family with a while the white mothers were married for Delivered Ingenta to 4 years. The black mothers high net worth. Therefore, once a certainby approximately User Unknown level of wealth is reached, the effect of wealth were only slightly younger than the white on achievement is likely to plateau. IP:1312.217.228.119 mothers. The black children had fewer educaSince some individuals had Date: a negative 2004..12..08..18..05.. tional resources in their homes and lower levvalue on the net worth variable, it was neces- els of cultural capital. The white children were sary to make adjustments to these numbers less likely than the black children to live in so that the natural log of wealth could be cal- urban areas and to attend public schools. The culated. All net worth values that were less black children and white children were close than or equal to 0 were recoded 1, so that the to equal in age, number of siblings, levels of natural log could be ascertained. This proce- self-esteem, and levels of social capital. The dure is similar to that used by Henretta and white parents tended to be employed in Campbell (1978). Approximately 11 percent more prestigious occupations and, on averof the sample was affected. 14 age, to have completed more years of education. The black mothers scored an average of 530 points on the AFQT, while the white mothers score an average of 758 points (a difRESULTS AND DISCUSSION ference of 1.17 standard deviations). Over a 5-year period, the black mothers and white Descriptive Analyses mothers spent approximately the same Table 1 presents the weighted descriptive sta- amount of time out of the workforce. As tistics for the sample. The sample was 15 per- expected, the black families and white famicent black and 48 percent female. The aver- lies differed widely in the amount of wealth age score on the PIAT Mathematics they had and in the amount of family income Assessment was 104.29. The average child earned over a 5-year period. The black famihad two to three siblings, and, on average, lies had only 19 percent of the wealth held by the mothers had been married 3.81 out of the the white families and only 56 percent of the previous 5 years. The average age of the chil- income (see Table 2). dren was 10–11 years; the average age of the mother was 35. In addition, 74 percent of the children lived in urban areas, and 87 percent Wealth, Achievement, and the attended public schools. The average child Racial Test Score Gap came from a home with relatively few educational resources and little social capital, but The results of the regression of achievement had fairly high self-esteem and access to a fair on race, SES, and wealth are presented in degree of cultural capital. The average family Table 3. Model 1 contains only the race varihad an income of $45,703 over a 5-year peri- able. With no other factors considered, blacks od and an average net worth of $77,103 over score 8.34 points, or .6 standard deviations, a 5-year period (see Table 2). The parents had below whites on the PIAT Mathematics an average of approximately 13 years of edu- Assessment, a statistically significant differcation and an average occupational prestige ence.15 The control variables and the traditional variscore of 30.16 over a 5-year period. On average, the mothers had been out of the work- ables used to measure SES are added in Model force for 2 out of the previous 5 years. 2. The larger the family size and the older the Black-White Differences in Achievement 291 Table 1. The Variables in the Analyses: NLSY79 Mothers and Childrena Total Sample Variables Dependent Variable PIAT math score Whites Blacks Mean SD Mean SD Mean SD 104.29 12.97 105.48 12.51 97.39 13.44 0.52 2.69 0.50 1.16 2.14 10.37 34.57 529.95 2.22 2.49 2.32 165.97 10.58 2.47 Control Variables Delivered by Ingenta to 0.50 Female 0.48 0.50 0.48 User Unknown Number of siblings 2.50 1.03 2.47 1.01 IP: 12.217.228.119 Number of years mother Date: 2004..12..08..18..05.. married (out of 5) 3.81 1.91 4.10 1.69 Age of the child 10.07 2.46 10.02 2.45 Mother’s age 35.21 2.26 35.32 2.23 Mother’s AFQT 724.11 193.80 757.66 177.90 Grandparents’ education 11.51 2.43 11.67 2.39 Independent Variables Black Parental education Parental occupation (5 years) Years mother unemployed Log family income (5 years) Log wealth (net worth) (5 years) Log net worth of incomeproducing assets (5 years) Log net worth of non-income-producing assets (5 years) Mediating Variables Educational resources Cultural capital Urban Public Self-esteem Social capital aDescriptive 0.15 13.12 0.35 2.03 NA 13.17 NA 2.07 NA 12.76 NA 1.75 30.16 13.07 30.90 13.03 25.89 12.50 2.11 1.81 2.10 1.82 2.15 1.74 10.40 0.89 10.49 0.84 9.86 0.97 9.58 2.70 9.95 2.42 7.41 3.17 7.54 3.27 8.02 2.99 4.78 3.47 9.07 2.69 9.51 2.24 6.54 3.56 0.34 0.70 0.74 0.87 0.87 0.25 0.41 0.29 0.44 0.34 0.21 0.43 0.35 0.72 0.72 0.86 0.88 0.25 0.41 0.28 0.45 0.35 0.20 0.43 0.26 0.58 0.87 0.93 0.85 0.25 0.36 0.32 0.34 0.26 0.23 0.43 statistics are weighted by the child weights. mother, the lower the achievement of the child. Children whose mothers’ score high on the AFQT have higher achievement themselves. As the number of years a mother is married increases, a child’s mathematics achievement increases. Mathematics achievement is higher for children whose parents have more years of schooling and work in more prestigious occupations. Income, on the other hand, is not significant after the mother’s AFQT score and parental occupation are controlled. As expected, the effect of race dramatically decreases in size from Model 1 to Model 2 (dropping 6.52 points). However, the racial gap in achievement 292 Orr Table 2. Black-White Mean Differences in Income and Wealth: NLSY79 Mothers and Children, Five-Year Average (N = 2,098) Measure Total Sample White Black Wealth (net worth) Income $77,102.89 $45,702.71 $87,545.28 $48,857.00 $16,686.49 $27,452.97 Note: The values presented for all net worth variables are based on the "true" values of Delivered by values. IngentaDescriptive to the variables, which include negative net worth statistics are weighted by the child weights. User Unknown IP: 12.217.228.119 Date: 2004..12..08..18..05.. traditional socioeco- deprivation. The remains significant with nomic controls in place. Model 3 adds wealth (log of total net worth) to the equation. The results indicate that wealth affects mathematics achievement, even after SES (parental education, occupation, and income) are controlled. Children who come from families with little or no wealth score lower on the PIAT Mathematics Assessment than do those from wealthier families, regardless of the parents’ income, education, and occupation. Compared to other indicators of SES, wealth has the largest standardized effect on a child’s scores (the standardized coefficients for wealth, parental education, and parental occupation are, respectively, 0.077, 0.057, and 0.058). The addition of wealth in this model (controlling for the variables listed earlier) decreases the coefficient for race by 15 percent. Wealth, therefore, also explains a portion of the black-white achievement gap. As the gap in wealth between children increases, so will the gap in test scores.16 The only other coefficient in the equation to undergo significant change is that for marital status. When wealth is in the model, the coefficient for marital status is insignificant. This finding supports the findings of previous studies that found financial capital to be a primary explanation for why children from single-parent households do poorly in school (see, e.g., Downey 1994; Milne et al. 1986; Myers et al. 1987). However, most studies have relied only on measures of income; when income is controlled, marital status has often remained significant. Various theories have been used to explain this result, such as the allocation of resources or interpersonal results presented here indicate that wealth may be the missing factor in the explanation of the low achievement of children from single-parent families. Mediating Variables Model 4 (Table 4) includes the mediating variables educational resources, cultural capital, self-esteem, and social capital, as well as the variables for school sector and location. The greater the exposure to cultural capital, the greater the child’s mathematics achievement. A child’s self-esteem does not appear to affect performance on the PIAT Mathematics Assessment. Attending a public school or a school in an urban area has no significant effects. Although social capital is negatively associated with mathematics achievement, the effect is dependent on the child’s age. Supplemental analyses indicated that children aged 10 or older whose parents have helped them with homework score lower on the test than do those whose parents have not, perhaps because children who require some type of assistance are having difficulty in school and are predisposed to score poorly on the mathematics test. In contrast, social capital does not have a negative effect on achievement for children aged 5–9 whose parents read to them at least three times a week. When the mediating variables are included in the equation, the effect of wealth shrinks by 15 percent. Therefore, the effect of wealth on academic achievement works, in part, through the effect of wealth on cultural capital and social capital. To explore this finding Black-White Differences in Achievement 293 Table 3. The Regression of PIAT Mathematics Achievement on Race, Socioeconomic Status, and Wealth (unstandardized coefficients)a Variables Model 1 Constant 105.37*** (0.34) Controls Female Number of siblings Years mother married Delivered by Ingenta to — User Unknown — IP: 12.217.228.119 Date: 2004..12..08..18..05.. — Age of the child — Age of the mother — Mother’s AFQT — Grandparents’ education — Independent Variables Black 94.11*** (5.29) -0.22 (0.51) -0.59* (0.25) 0.35* (0.15) -0.13 (0.10) -0.32** (0.13) 0.02*** (0.002) 0.18 (0.12) -0.25 (0.51) -0.57* (0.25) 0.21 (0.16) -0.14 (0.10) -0.35** (0.13) 0.02*** (0.002) 0.17 (0.12) -1.54* (0.71) 0.39* (0.18) 0.06* (0.03) 0.21 (0.16) -0.35 (0.32) 0.34** (0.12) 2,098 0.242 Parental occupation (5-year average) — Years mother unemployed — Log family income (5-year average) — Log wealth (net worth) (5-year average) — 2,098 0.091 2,098 0.239 N R-squared Model 3 91.74*** (5.22) -1.82* (0.71) 0.42* (0.18) 0.06** (0.03) 0.19 (0.16) -0.03 (0.32) — Parental education -8.34*** (0.58) — Model 2 Source: NLSY79 Mothers and Children. aStandard errors are in parentheses. Standard errors are corrected for the nonindependence of siblings. * p < .05, ** p < .01, *** p < .001. further, I examined the correlations between wealth and the mediating variables. Wealth is positively and significantly correlated with cultural capital but is not significantly correlated with social capital. The persistent effect of wealth may be due to the unavailability of a measure of a child’s educational expectations, the omitted schoolquality variables, and/or the lack of more sophisticated measures of parental orientation toward “status seeking” and toward the future. Each of these elements was believed to play a role in the relationship between wealth and achievement. 294 Orr Table 4. The Regression of PIAT Mathematics Achievement on Race, Socioeconomic Status, Wealth, and Mediating Variables (unstandardized coefficients)a Variables Model 4 Constant 93.60*** (5.50) Controls Female Number of siblings Years mother married -0.29 (0.51) Delivered by Ingenta to User Unknown -0.67** IP: 12.217.228.119(0.25) 0.21 Date: 2004..12..08..18..05.. (0.16) Age of the child Age of the mother Mother’s AFQT Grandparents’ education Independent Variables Black Parental education Parental occupation (5-year average) Years mother unemployed Family income (5-year average) Log wealth (net worth) (5-year average) Mediating Variables Educational resources Cultural capital Urban school Public school Social capital Self-esteem N R-squared -0.14 (0.11) -0.34** (0.13) 0.02*** (0.002) 0.12 (0.12) -1.43* (0.72) 0.35 (0.18) 0.06* (0.03) 0.22 (0.16) -0.43 (0.32) 0.29* (0.12) -0.19 (0.77) 2.47* (0.99) 0.72 (0.58) -0.90 (0.85) -2.43*** (0.68) 2.34 (1.33) 2,098 0.253 Source: NLSY79 Mothers and Children. Standard errors are in parentheses. Standard errors are corrected for the nonindependence of siblings. * p < .05. ** p < .01, ***p < .001. Black-White Differences in Achievement 295 On one additional note, the only traditional socioeconomic variable that remains significant is parents’ occupation. The effect of parents’ education is not statistically significant once the mediating variables are taken into account. Although the effect of race is reduced, it remains significant, meaning that some of the influence of race does not work through these additional factors. 17 Effects of Age and Assets a variety of factors that could contribute to inequalities in educational opportunities, including the characteristics of the school, the student body, and teachers; the school facilities and curriculum; the attitudes of students; and family background characteristics. In discussing the implications of the results of the study, Coleman et al. made it clear that family background stood above all others as a Delivered by contributor Ingenta to to educational inequalities. User Unknown According to Coleman (1990:119): Composition of IP: 12.217.228.119 Taking all of [the] results together, one impliDate: 2004..12..08..18..05.. cation stands out above all: That schools bring The results of the regression of achievement on race, SES, wealth, and the interaction variable for wealth and age are presented in Table 5. The coefficient for the interaction between age and wealth is not significant. Therefore, wealth does not appear to discriminate significantly among different age groups.18 Table 6 presents the results of the regression of mathematics achievement on race, SES, and the variables for the amount of net worth held in income-producing and nonincome-producing assets. The net worth of income-producing assets has a positive and significant effect on achievement, whereas the net worth of non-income-producing assets is statistically insignificant. Since black children tend to come from families that hold a smaller amount of net worth in income-producing assets, part of the racial difference in test scores may be attributed to the composition of assets.19 CONCLUSION More than three decades ago, sociologist James S. Coleman was charged by the U.S. Department of Education with studying educational inequality between black and white children. Specifically, he was to lead a team of researchers in a project that would examine “the lack of availability of equal educational opportunities for individuals by reason of race, color, religion, or national origin in public educational institutions at all levels in the United States” (see Mosteller and Moynihan 1972:4–5). In Equality of Educational Opportunity, Coleman et al. (1966) discussed little influence to bear on a child’s achievement that is independent of his background and general social context; and that this very lack of an independent effect means that the inequalities imposed on children by their home, neighborhood, and peer environment are carried along to become the inequalities with which they confront adult life at the end of school. Over the past 37 years, Coleman et al.’s (1966) interpretation has held up. Family background remains a strong predictor of a child’s achievement (see, e.g., Crane 1996; Downey 1994; Jones 1984; Lee 1993; Milne et al. 1986; Myers et al. 1987; Teachman 1987; Vanfossen, Jones, and Spade 1987) and contributes to racial differences in test scores (see, e.g., Hedges and Nowell 1998; Phillips, Brooks-Gunn et al. 1998). Children who come from disadvantaged families tend to have lower achievement than do those who come from more advantaged families. This disadvantage does not tend to disappear as a child progresses through school (Phillips, Crouse, and Ralph 1998). Although research has demonstrated that an important link between family background and achievement still exists, current studies may actually underestimate these effects. Without considering a family’s wealth, researchers may be excluding an important factor that contributes to a child’s achievement. The results of this study indicate that wealth has a positive effect on achievement, even after a family’s SES is held constant. Wealth also explains a portion of the blackwhite differences in achievement. The results also indicate that the impact of wealth does not differ significantly on the basis of chil- 296 Orr Table 5. The Regression of PIAT Mathematics Achievement on Race, Socioeconomic Status, Wealth, and Interaction Variables (unstandardized coefficients)a Variables Model 5 Constant 96.99*** (6.04) Controls Female Number of siblings Years mother married Delivered by Ingenta to -0.24 User Unknown (0.51) IP: 12.217.228.119-0.58* (0.25) Date: 2004..12..08..18..05.. Age of the child Age of the mother Mother’s AFQT Grandparents’ education Independent Variables Black Parental education Parental occupation (5-year average) Years mother unemployed Log family income (5-year average) Log wealth (net worth) (5-year average) Interaction Variable Age*Wealth N R-squared 0.20 (0.16) -0.42 (0.33) -0.35** (0.13) 0.02*** (0.002) 0.17 (0.12) -1.55* (0.71) 0.40* (0.18) 0.06* (0.03) 0.22 (0.16) -0.34 (0.32) -0.01 (0.38) 0.03 (0.03) 2,098 0.243 Source: NLSY79 Mothers and Children. Standard errors are in parentheses. Standard errors are corrected for the nonindependence of siblings. * p < .05, ** p < .01, *** p < .001. Black-White Differences in Achievement 297 Table 6. The Regression of PIAT Mathematics Achievement on Race, Socioeconomic Status, Wealth, and Asset Composition (unstandardized coefficients)a Variables Model 6 Constant Controls Female Number of siblings Years mother married 97.20*** (5.34) Delivered by Ingenta to -0.23 User Unknown (0.51) IP: 12.217.228.119-0.49 (0.25) Date: 2004..12..08..18..05.. 0.17 (0.16) -0.13 (0.10) -0.34** (0.13) 0.02*** (0.002) 0.16 (0.12) Age of the child Age of the mother Mother’s AFQT Grandparents’ education Independent Variables Black Parental education Parental occupation (5-year average) Years mother unemployed Family income (5-year average) Log net worth of income-producing assets Log net worth of non-income-producing assets N R-squared -1.42* (0.72) 0.34 (0.18) 0.05* (0.03) 0.24 (0.16) -0.57 (0.32) 0.42*** (0.11) 0.05 (0.12) 2,098 0.245 * p < .05, **p < .01, *** p < .001. Source: NLSY79 Mothers and Children Standard errors are in parentheses. Standard errors are corrected for the nonindependence of siblings. dren’s ages and hence cannot help to explain the widening racial gap in test scores as children progress through school. The effect of wealth on achievement is explained mainly by the effect of wealth on the amount of cultural capital to which a child is exposed. This finding supports Bourdieu’s (1986) notion of capital: Economic capital (wealth) can be converted into other forms of capital (in this case, cultural capital) 298 Orr to reproduce status. The analyses also support tem is characterized as being open—allowing Bourdieu’s (1986) concept of liquidity: The individuals to escape the levels at which they type of economic capital that tends to be were born—there are limits to mobility, some cashed in is the type that is the most liquid (in of which are not based on personal strengths this case, income-producing assets). In these or deficiencies. analyses, the amount of wealth held in I have shown that parental wealth is a conincome-producing assets has a significant straint on the educational resources and effect on achievement, whereas the net worth opportunities available to a child. Since eduof non-income-producing assets does not cation is a primary avenue by which individuDelivered Ingenta play an important role in a child’s achieve-by als attain tostatus (Blau and Duncan 1967), ment. User Unknown parental wealth can influence where a child The final finding reported in this is will be placed in the stratification system later IP: article 12.217.228.119 that after class factors are controlled, race has in life. Because parental net worth is linked to Date: 2004..12..08..18..05.. a significant negative effect on achievement. educational outcomes, children from families Black children tend to score lower on stan- with little or no wealth are likely to remain at dardized achievement tests than do white a lower level in the stratification hierarchy children, even after parental income, educa- than are those from wealthier families, theretion, occupation, and wealth are taken into by reproducing, to some degree, the status account. This effect cannot be accounted for ranking of their parents. by other differences in family characteristics, The fact that blacks, on average, have subsuch as family size and composition, a finding stantially different levels of wealth than do that is consistent with that of other studies whites also demonstrates the limits of the that have dealt with the gap in black-white achievement ideology in this country and test scores (Jencks and Phillips 1998). calls into question the idea that the United States has an open stratification system. The accumulation of wealth among the black IMPLICATIONS OF THE FINDINGS population is limited by forces that are unrelated to achievement (see, e.g., Oliver and A common belief held by many Americans is Shapiro 1995). Blacks face discrimination in that if you want to be successful, all you have several markets, which limits the amount of to do is try. Success, according to many, can wealth black families possess. Although these be attained by any who want it badly practices are not formally sanctioned, they enough. This perspective depicts the United are effective in limiting the life chances of States as a meritocracy in which success is black Americans. The lack of wealth among certain individubased on achievement, rather than on ascription, and in which opportunities to achieve als and segments of the population has are open equally to all. The stratification sys- important consequences for these individuals tem that exists is open, and any barriers to and groups, as well as for the society as a success are thought to be personal, rather whole. When an entire group has limited access to wealth, it has decreased access to than social (MacLeod 1987). The findings of this study, in addition to the opportunities, resources, and power that several previous studies that have shown a wealth confers on its owners. These limitasignificant relationship between family back- tions can serve to disenfranchise an entire ground and achievement, contradict the segment of the population, making it difficult notion that opportunities to achieve are for them to participate fully in societal instituequally open to all individuals. The results tions and to exercise influence in such areas indicate that ascriptive characteristics, such as as politics and the economy. This situation parental wealth, contribute to whether or not also has negative consequences for society in a person is successful. Despite the emergence that valuable resources are not used and the of egalitarian ideologies, individuals in likelihood of cross-cultural misunderstandings American society do not have equal opportu- is increased (Walsh 2000). nities. While the American stratification sysMy findings demonstrate the importance Black-White Differences in Achievement 299 of a multidimensional approach to the deter6. For a discussion of the differences in the minants and consequences of stratification. measurement of wealth in various U.S. surEconomic well-being and social status can be veys, see Juster and Kuester (1991). attained in several ways. While income, edu7. With the exception of the child’s age cation, and occupation—the most widely and mother’s age, the variables were taken used indicators of class position—can explain from surveys and assessments administered a great deal about access to resources and prior to 1996. opportunities, they are not exhaustive. 8. Since the NLSY data were not collected In the debates over the relative influence of in 1995, when the variables were averaged Delivered Ingenta class and race, it is often argued that the num-by over five to years, they were averaged from ber of blacks in the middle class hasUser grown. Unknown This 1990 to 1994. claim has been used to demonstrate race IP: that 12.217.228.119 9. Missing data are a major concern with has lost its importance as a determinant life regard to information on assets. Fortunately, Date: of 2004..12..08..18..05.. chances for black Americans. However, when the response rate in the NLSY79 has historithe indicators of middle-class status include cally been high for the questions on owneronly income, education, and occupation, these ship of assets (approximately 99 percent for successes are overestimated. While blacks have each). With regard to the value of assets and come closer to parity with whites in income, debts, the response rates have also been education, and occupation, the substantial high. In 1992, for example, the response racial differences in wealth continue to affect rates for the value of assets ranged from 94 educational and social opportunities. percent to 99 percent, and the response rates for questions about debts ranged from 97 percent to 98 percent. For a further discussion, see Center for Human Resource NOTES Research (1997). 10. While the child’s expectations were 1. This argument is supported by the results of Conley’s (1999, 2001) research, discussed as a possible mediator, the data set which found that wealth is a strong predictor does not include information on expectations for children younger than age 10. Since of college attendance. 2. Minorities, for example, may have had adding this variable would have restricted the experiences that have demonstrated that the sample too greatly, the impact of expectaconnection between present efforts and tions was not explored. 11. The questions asked of the mothers future rewards is not valid and may have lost hope in the idea that what they do in the pre- and children varied with the children’s ages. Individual components used to construct the sent can affect their future outcomes. 3. For simplification purposes, Phillips, mediating variables were chosen primarily Crouse, and Ralph (1998:235–36) stated: “A from those used to construct the NLSY HOME 1.0 standard deviation gap on an IQ test is 15 Inventory Scale. While the HOME Scale is points. A 1.0 standard deviation gap on the available for use in the analyses, I believe it is math or verbal SAT is 100 points. Thus, a important to distinguish among the various black-white gap of 0.80 standard deviations is types of resources and capital available to equivalent to a 12 point gap on an IQ test or children. Therefore, new variables were created using the individual components. Some an 80 point gap on the verbal SAT.” 4. This argument is also supported by the additional components were included on the results of Conley’s (1999) research, which basis of a review of past research. 12. Because of the lack of better informafound that liquid (income-producing) assets have a stronger effect on academic outcomes tion about each child’s school, school quality than do illiquid (non-income-producing) was measured by the chosen variables. School survey data were collected beginning assets. 5. The PSID began to collect data on chil- in 1994–95, but only for a subset of the children, called the Child Development dren. Supplement, only in 1997. 13. The use of the natural log of income 300 Orr and wealth in statistical models is supported debt is not a statistically significant predictor by past research. For example, Mayer (1997) of a child’s test scores. used the natural log of income when consid15. The calculation for the standard deviaering the effect of income on several out- tion gap was based on the unweighted stancomes, including a child’s score on the PIAT dard deviation for the overall PIAT Mathematics mathematics and reading assessments. After score (13.93). When weighted descriptive starunning goodness-of-fit statistics for several tistics were used in the calculation, the gap is functional forms of income, Mayer concluded 0.62 standard deviation units. that “the goodness-of-fit statistics are nearly 16. The models were also analyzed using Ingenta the same for all specifications, butDelivered a nonlinearby scores ontothe PIAT Reading Comprehension User Unknown form usually provides a better fit than the lin- and the PIAT Reading Recognition subscales as IP: 12.217.228.119 ear form.” The natural log of wealth has also the dependent variables. The results were simibeen used in statistical modelsDate: that 2004..12..08..18..05.. have lar to those found with the PIAT Mathematics explored the effect of net worth on various subscale. outcomes (see, e.g., Conley 1999; Henretta 17. Preliminary analyses indicated that the and Campbell 1978). “effects” do not differ by race. 14. It may be argued that this procedure is 18. The possible interaction between wealth problematic, since the value of a family’s debt and age was also explored by running individmay affect a child’s achievement—the higher ual regressions for different age groups (ages the debt, the less likely a family would be to 5–9 and 10–14). Wealth was significant in both invest in educational resources or opportuni- regressions. ties for a child. Separate regressions were run 19. This result is similar to the results of other to ascertain whether the value of a family’s studies found that income-producing (liquid) debt is a significant determinant of achieve- assets are the most important type of assets ment. The results indicated that the value of (Conley 1999, 2001). Black-White Differences in Achievement 301 APPENDIX Comparison of the Limited Sample and Full Samplea Limited Sample (N = 2,098) Variables Dependent Variables PIAT mathematics score Control Variables Female Number of siblings Number of years mother married (out of 5) Age of the child Mother’s age Mother’s AFQT Grandparents’ education Mean Mean Delivered by Ingenta to 104.29 12.97 103.24 User Unknown IP: 12.217.228.119 Date:0.48 2004..12..08..18..05.. 0.50 0.49 Independent Variables Black Parental education Parental occupation (5 years) Years mother unemployed Log family income (5 years) Log wealth (net worth) (5 years) Log net worth of incomeproducing assets (5 years) Log net worth of non-incomeproducing assets (5 years) Mediating Variables Educational resources Cultural capital Urban Public Self-esteem Social capital SD Full Sampleb SD N 13.84 3,680 2.50 1.03 2.52 0.50 1.08 3,680 3,650 3.81 10.07 35.21 724.11 11.51 1.91 2.46 2.26 193.80 2.43 3.70 9.65 35.15 698.70 11.24 1.99 2.72 2.31 208.00 2.83 3,662 3,680 3,679 3,557 3,563 0.15 13.12 30.16 2.11 10.40 9.58 0.35 2.03 13.07 1.81 0.89 2.70 0.16 12.99 29.11 2.26 10.34 9.29 0.37 2.17 14.19 1.86 0.96 2.96 2,897 3,680 3,680 3,680 3,607 3,623 7.54 3.27 7.13 3.56 3,650 9.07 2.69 8.76 3.02 3,664 0.34 0.70 0.74 0.87 0.87 0.25 0.41 0.29 0.44 0.34 0.21 0.43 0.38 0.68 0.77 0.86 0.86 0.30 0.42 0.31 0.42 0.35 0.23 0.46 3,653 3,616 3,668 3,499 3,323 3,608 Source:NLSY79 Mothers and Children. aDescriptive statistics are weighted by the child weights. bThe sample includes all children with a valid PIAT mathematics score in 1996. REFERENCES Baharudin, Rozumah, and Tom Luster. 1998. “Factors Related to the Quality of the Home Environment and Children’s Achievement.” Journal of Family Issues 19:375–403. Baker, Paula C., Canada K. Keck, Frank L. Mott, and Stephen V. Quinlan. 1993. NLSY Child Handbook: A Guide to the 1986-1990 National Longitudinal Survey of Youth Child Data. Columbus: Center for Human Resource Research, Ohio State University. Berndt, Thomas. 1979. “Developmental Changes in Conformity to Peers and Parents.” Developmental Psychology 15:608–16. 302 Orr Blau, Peter M., and Otis Dudley Duncan. 1967. The DiMaggio, Paul. 1982. “Cultural Capital and American Occupational Structure. New York: School Success: The Impact of Status Culture John Wiley & Sons. Participation on the Grades of U.S. High Blau, Francine D., and John W. Graham. 1990. School Students.” American Sociological Review “Black-White Differences in Wealth and Asset 47:189–201. Composition.” Quarterly Journal of Economics Downey, Douglas. 1994. “The School Performance 105:321–39. of Children from Single-Mother and SingleBourdieu, Pierre. 1977. “Cultural Reproduction Father Families: Economic or Interpersonal and Social Reproduction.” Pp. 487–511 in Deprivation?” Journal of Family Issues Power and Ideology in Education, edited by 15:129–47. Delivered Ingenta to Jerome Karabel and A. H. Halsey. New York:by —-. 1995. “When Bigger Is Not Better: Family Size, User Unknown Parental Resources, and Children’s Educational Oxford University Press. —-. 1986. “The Forms of Capital.” Pp. in Performance.” American Sociological Review IP:241–58 12.217.228.119 Handbook of Theory and Research the 60:746–61. Date:for2004..12..08..18..05.. Sociology of Education, edited by John G. Dunn, Lloyd M., and Frederick C. Markwardt, Jr. Richardson. New York: Greenwood Press. 1970. Peabody Individual Achievement Test Bryk, Anthony S., Valerie E. Lee, and Peter B. Manual. Circle Pines, MN: American Guidance Holland. 1993. Catholic Schools and the Service. Common Good. Cambridge, MA: Harvard Education Week. 1998. Quality Counts ‘98. University Press. Washington, DC: Editorial Projects in Center for Human Resource Research, Ohio State Education. University. 1997. The National Longitudinal Eitzen, D. Stanley. 1975. “Athletics in the Status Surveys NLSY79Users’ Guide. Columbus, OH: System of Male Adolescents: A Replication of Author. Coleman’s ‘The Adolescent Society.’” Coleman, James S. 1961. The Adolescent Society: Adolescence 10(38):267–76. The Social Life of the Teenager and Its Impact on Elkind, David. 1967. “Egocentrism in Adolescence.” Education. New York: Free Press. Child Development 38: 1025-1034. —-. 1988. “Social Capital in the Creation of Human Eller, T. J. 1994. Household Wealth and Asset Capital.” American Journal of Sociology Ownership: 1991: Current Population Reports 94:95–120. (Series P-70, No. 34). Washington, DC: U.S. Government Printing Office. —-. 1990. Equality and Achievement in Education. Boulder, CO: Westview Press. Ferguson, Ronald F. 1999, November. “Racial TestColeman, James S., Ernest Q. Campbell, Carol J. Score Trends 1971–1996, Popular Youth Hobson, James McPartland, Alexander M. Culture and Community Academic Mood, Frederic D. Weinfeld, and Robert L. York. Standards.” Paper presented at the annual 1966. Equality of Educational Opportunity. conference of the Institute for Educational Washington, DC: U.S. Government Printing Initiatives, Notre Dame, Indiana, November Office. 1999. Coleman, James S., Kathryn Schiller, and Barbara Goldberg, Alan, and Timothy Chandler. 1989. Schneider. 1993. “Parent Choice and “The Role of Athletics: The Social World of Inequality.” Pp. 147–82 in Parents, Their High School Adolescents.” Youth and Society 21:238–50. Children, and Schools, edited by Barbara Schneider and James S. Coleman. Boulder, Hao, Lingxin. 1995. “Poverty, Public Assistance, CO: Westview Press. and Children in Intact and Single-Mother Conley, Dalton. 1999. Being Black, Living in the Red: Families.” Journal of Family and Economic Issues Race, Wealth, and Social Policy in America. 16(2–3):181–205. Berkeley: University of California Press. Hauser, Robert, and Douglas Anderson. 1991. —-. 2001. “Capital for College: Parental Assets and “Post-High School Plans and Aspirations of Postsecondary Schooling.” Sociology of Black and White High School Seniors: Education 72:59–72. 1976–86.” Sociology of Education 64:263–77. Costanzo, Philip. 1970. “Conformity Development Hedges, Larry V., and Amy Nowell. 1998. “Blackas a Function of Self-Blame.” Journal of White Test Score Convergence Since 1965.” Personality and Social Psychology 14:366–74. Pp. 149–81 in The Black-White Test Score Gap, Crane, Jonathan. 1996. “Effects of Home edited by Christopher Jencks and Meredith Environment, SES, and Maternal Test Scores Phillips. Washington, DC: Brookings Institute on Mathematics Achievement.” Journal of Press. Educational Research 89:305–14. Henretta, John C., and Richard T. Campbell. 1978. Black-White Differences in Achievement 303 “Net Worth as an Aspect of Status.” American Myers, David, Ann Milne, Keith Baker, and Alan Journal of Sociology 83:1204–23. Ginsburg. 1987. “Student Discipline and High Hoffer, Thomas B. 1997–98. “Social Background and School Performance.” Sociology of Education Achievement in Public and Catholic High 60:18–33. Schools.” Social Psychology of Education 2:7–23. Natriello, Gary, and Edward McDill. 1986. Jencks, Christopher, and Phillips, Meredith, eds. “Performance Standards, Student Effort on 1998. The Black-White Test Score Gap. Homework, and Academic Achievement.” Washington, DC: Brookings Institution Press. Sociology of Education 59:18–31. Jones, Lyle. 1984. “White-Black Achievement Okun, Morris, and Joseph H. Sasfy. 1977. Differences.” American Psychologist 39(11): “Adolescence, the Self-Concept, and Formal Delivered by Ingenta to 1207–13. Operations.” Adolescence 12:373–79. User Unknown Juster, F. Thomas, and Kathleen A. Kuester. 1991. Oliver, Melvin, and Thomas Shapiro. 1989. “Race “Differences in the Measurement Wealth, and Wealth.” Review of Black Political Economy IP:of12.217.228.119 Wealth Inequality, and Wealth Date: Composition 17:79–91. 2004..12..08..18..05.. Obtained from Alternative U.S. Wealth —-. 1995. Black Wealth/White Wealth: A New Surveys.” Review of Income & Wealth 37(1):33Perspective on Racial Inequality. New York: 62. Routledge. Kaestner, Robert. 1997. “Are Brothers Really Phillips, Meredith, Jeanne Brooks-Gunn, Greg J. Better? Sibling Sex Composition and Duncan, Pamela Klebanov, and Jonathan Educational Achievement Revisited.” Journal of Crane. 1998. “Family Background, Parenting Human Resources 32:250–84. Practices, and the Black-White Test Score Kane, Thomas J. 1998. “Racial and Ethnic Gap.” Pp. 103–45 in The Black-White Test Score Preferences in College Admissions.” Pp. Gap, edited by Christopher Jencks and 431–56 in The Black-White Test Score Gap, Meredith Phillips. Washington, DC: Brookings edited by Christopher Jencks and Meredith Institute Press. Phillips. Washington, DC: Brookings Institute Phillips, Meredith, James Crouse, and John Ralph. Press. 1998. “Does the Black-White Test Score Gap Lee, Seh-Ahn. 1993. “Family Structure Effects on Widen after Children Enter School?” Pp. Student Outcomes.” Pp. 43–75 in Parents, 229–72 in The Black-White Test Score Gap, Their Children, and Schools, edited by Barbara edited by Christopher Jencks and Meredith Phillips. Washington, DC: Brookings Institute Schneider and James S. Coleman. Boulder, CO: Westview Press. Press. Liu, Xiaoru, Howard Kaplan, and Will Risser. 1992. Rodgers, Joseph Lee, David C. Rowe, and Kim May. “Decomposing the Reciprocal Relationships 1994. “DF Analysis of NLSY IQ/Achievement between Academic Achievement and General Data: Nonshared Environmental Influences.” Self-Esteem.” Youth and Society 24:123–48. Intelligence 19:157–77. Luster, Tom, and Harriette Pipes McAdoo. 1994. Schultz, Clarence. 1990. “The Adolescent Society “Factors Related to the Achievement and Revisited.” Free Inquiry in Creative Sociology Adjustment of Young African American 18(1):3–12. Children.” Child Development 65:1080–94. Steinberg, Lawrence, and Susan Silverberg. 1986. MacLeod, Jay. 1987. Ain’t No Makin’ It: Leveled “The Vicissitudes of Autonomy in Early Aspirations in a Low-Income Neighborhood. Adolescence.” Child Development 57: 841–51. Boulder, CO: Westview Press. Sterbin, Allan, and Ernest Rakow. 1996, November. Mayer, Susan E. 1997. What Money Can’t Buy: “Self-Esteem, Locus of Control, and Student Family Income and Children’s Life Chances. Achievement.” Paper presented at the annual Cambridge, MA: Harvard University Press. meeting of the Mid-South Educational Merida, Kevin. 1995, October 9. “Worry, Research Association, Tuscaloosa, Alabama. Frustration Build for Many in the Black Middle Teachman, Jay. 1987. “Family Background, Class.” Washington Post, pp. A1, A22–A23. Educational Resources, and Educational Milne, Ann, David Myers, Alvin Rosenthal, and Alan Attainment.” American Sociological Review Ginsburg. 1986. “Single Parents, Working 52:548–57. Mothers, and the Educational Achievement of Tracy, Mary E. 2001. Mission and Money: A CHS School Children.” Sociology of Education Report on Finance, Advancement, and 59:125–39. Governance. Washington, DC: National Mosteller, Frederick, and Daniel P. Moynihan. Catholic Education Association. 1972. On Equality of Educational Opportunity. Vanfossen, Beth, James Jones, and Joan Spade. New York: Random House. 1987. “Curriculum Tracking and Status 304 Orr Maintenance.” Sociology of Education In Social Stratification: Class, Race, and Gender 60:104–22. in Sociological Perspective, edited by David B. Veblen, Thorstein. 1934. The Theory of the Leisure Grusky. Boulder, CO: Westview Press. (Original Class. New York: Random House. work published 1949) Walker, Lynn, and John Greene. 1986. “The Social Weber, Max. 1946. From Max Weber: Essays in Context of Adolescent Self-Esteem.” Journal of Sociology Hans Gerth and C. Wright Mills (eds. Youth and Adolescence 15:315–22. and trans.). New York: Oxford University Press. Walsh, Mary Williams. 2000, March 25. “Net White, Halbert. 1980. “A Heteroskedastic-Consistent Worth Shrinking Despite Boom Times.” Los Covariance Matrix Estimator and a Direct Test of Angeles Times, p. A16. to Warner, W. Lloyd. 1994. “Social ClassDelivered in America.”by Ingenta Heteroskedasticity.” Econometrica, 48:817–38. User Unknown IP: 12.217.228.119 Amy J. Orr, Ph.D., is Assistant Professor of Sociology, Department of Sociology and Anthropology, Date: Linfield College, McMinnville, Oregon. Her2004..12..08..18..05.. main fields of interest are education, race and ethnicity, gender, and social policy. She is currently conducting further research on wealth and achievement, retention of minority students, and college writing. The author thanks Maureen Hallinan, Richard Williams, David Hachen, Felicia LeClere, and Warren Kubitschek for their comments and suggestions and John Bauer for statistical support. Address correspondence to Amy J. Orr, Department of Sociology and Anthropology, Linfield College, 900 SE Baker Street, McMinnville, OR 97128; e-mail: aorr@linfield.edu.