Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Effects of Family Structure on the Earnings Attainment Process: Differences by Gender Author(s): Mary Ann Powell and Toby

L. Parcel Source: Journal of Marriage and Family, Vol. 59, No. 2 (May, 1997), pp. 419-433 Published by: National Council on Family RelationsNational Council on Family Relations Stable URL: http://www.jstor.org/stable/353480 Accessed: 17/11/2010 05:50
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ncfr. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

National Council on Family Relations is collaborating with JSTOR to digitize, preserve and extend access to Journal of Marriage and Family.

http://www.jstor.org

MARY ANN POWELL AND TOBY L. PARCEL

The Ohio State University

Effects of Family Structureon the Earnings AttainmentProcess: Differences by Gender

This study compares how being raised in an original, two-parent family and being raised in other family structures affects educational achievement, occupational status, and earnings attainment for a national sample of 30- to 59-year-old women and men. Data are derived from the 1989 Panel Study of Income Dynamics. Findings suggest that family structure has different effects by gender. Although both men and women from original, two-parent families earn more, on average, than those from other family structures, for women, this effect occurs through educational attainment. For men, the association between family structure and attainment is explained by other family background variables, including smaller family size, being Catholic, higher levels of parental education, and being White. Men who are raised by both natural parents are not advantaged educationally, compared with those who grow up in other types offamily structures. A cohort analysis for men that compares baby boomers with prebaby boomers, however, suggests contradictory effects of family structure that deserve more exploration.

of Department Sociology, 300 BrickerHall, The Ohio State University, 190 N. Oval Mall, Columbus, OH 43210 1 (powell.130@osu.edu; parcel. @osu.edu). Key Words:divorce, earnings attainment, family structure, gender,two-parentfamilies.

How does family structure in childhood affect an individual's subsequent socioeconomic wellbeing? We know that the absence of the father lowers cognitive test scores for young children (Mott, 1993) and that being raised by a single parent may interfere with high school graduation (Coleman, 1988; McLanahan, 1985; Shaw, 1982) and with the availability of funding for higher education (Steelman & Powell, 1991). McLanahan and Sandefur (1994) argue that children raised by only one biological parent are deprived of economic and social resources, which negatively affects their future success. Boys who grow up in original, two-parent families have a greater chance of achieving a higher occupational status than those raised in families headed by women (Duncan & Duncan, 1969), and boys who are raised by their mothers have a greater chance of ending up in occupations with lower status than those their fathers held (Biblarz & Raftery, 1993). Amato and Keith (1991) indicate that children who experienced parental divorce are more likely than those raised in original, two-parent families to exhibit psychological, behavioral, social, and academic problems. A great deal of evidence suggests that being raised without both natural parents involves costs for both children and the larger society. (See McLanahan & Booth, 1991, and Seltzer, 1994, for comprehensive reviews and Demo, 1992, for some counter arguments.) These studies do not tell us, however, whether divorce affects children when they mature to

Journal of Marriage and the Family 59 (May 1997): 419-433

419

420 adulthood. In addition, prior research on the effects of family structure on socioeconomic outcomes frequently has failed to consider effects by gender by considering men only (Biblarz & Raftery, 1993; Duncan & Duncan, 1969; Greenberg & Wolf, 1982; Krein, 1986) or has been limited by confining attention to outcomes in young adulthood (Hill, Augustoniak, & Ponza, 1987). This article extends previous work by providing a clear causal picture of the effects of childhood family structure on adult socioeconomic outcomes for both men and women. We differentiate two broad categories of famistructure: those respondents who were raised ly by both naturalparents most of the time until they were 16 years old (original, two-parent family status) and those who report that they were not. We provide a rationale for using these broad categories and expand on this definition. We study (a) how growing up with both natural parents affects educational attainment, occupational status, and earnings, compared with growing up in other family structures, and (b) whether these effects differ by gender.
THEORETICAL GUIDANCE

Journal of Marriage and the Family Duncan, 1988; Holden & Smock, 1991; Peterson, 1996; Weiss, 1984), but also time may be lacking for activities, such as supervising children's homework (Astone & McLanahan, 1991). These conditions also may apply to never-married, single-parent families. Explanations of resource dilution do not offer insight into possible gender differences in the allocation of limited resources when families are disrupted or when there is only a single parent. Socialization explanations emphasize the importance of role modeling whereby parents foster or deter attainment in their children through their own examples (Hill et al., 1987; McLanahan & Bumpass, 1988). Because disruption often results in children being raised by their mothers, the lack of a male role model could affect boys negatively by removing an example of how men achieve occupational success. The lack of this model may serve as a disincentive for boys to achieve. Through role modeling, girls may learn that women are able to manage a family alone, which might lead to early or single childbearing. Socialization models also predict that daughters could respond to the lack of a male figure in their lives by marrying earlier to fill a void (McLanahan & Bumpass, 1988). Early or single childbearing or early marriages can mean that girls' education suffers. Additionally, the socialization models predict that disruption may weaken parents' social control (Mueller & Pope, 1977). Although the mechanisms for detrimental effects differ in human capital and key status attainment explanations, each theory predicts negative implications for subsequent achievement. Effects of Family Structure on Educational Attainment We know that the economic disadvantages experienced by single-parent families can result in lower rates of high school graduation and diminished parental funding of higher education. Additionally, Astone and McLanahan (1991) found that children in stepfamilies and with single parents receive less parental encouragement and attention regarding educational activities and have a higher likelihood of leaving high school than those in original, two-parent families. They found that if socioeconomic status was controlled, there were equally negative effects for single-parent families and stepfamilies. McLanahan and Bumpass' (1988) study of adult outcomes of women indicates that those who spend some time in one-parent families are

Human capital and status attainment perspectives provide useful bases for studying the effects of family structure on adult socioeconomic outcomes. Becker and Tomes (1986) propose that parents influence the adult earnings of their children by "expenditures on their skills, health, learning, motivation, 'credentials,' and many other characteristics" (p. S5). This theory posits that the investments people make in education and training will be reflected in returns that can be measured in earnings (Becker, 1975). Investments may be made on several levels, including societal, parental, and individual levels. Family disruption can diminish the investments that parents make in their children (Astone & McLanahan, 1991; Steelman & Powell, 1991). Two central explanations that suggest how family structure might have an impact on adult outcomes are extensions of status attainment theories. Explanations of resource dilution lead us to anticipate that family disruption will have negative consequences because disruption spreads thin whatever resources the family has (Blake, 1989; Coleman, 1988). Families with less money, more children, or only one parent in residence may have fewer resources for support. In disrupted families, not only is there normally a drop in income (Bianchi & McArthur, 1991; Hoffman &

Family Structure, Gender, and Earnings Attainment more likely to marry and have children early, to get pregnant before marriage, and to divorce. Much of the effect of marital disruption on teenage motherhood is explained by its effect on education. Women whose families were disrupted during their childhood are more likely to enter marriages earlier; in these marriages both partners have limited education (Mueller & Pope, 1977). McLanahan and Sandefur (1994) found that in White families, young men and women are more likely to drop out of high school when they come from disrupted families, and women are much less likely to complete college than men. They indicate that these findings are not consistent across all four surveys they study, however, and must be viewed as tentative. The effects are not as strong in Black and Hispanic families, and they report no sex differences. Short-term, detrimental effects also may have long-term consequences. For example, Mott (1993) found that father absence predicts lower cognitive test scores and poorer emotional adjustment for 5- to 9-year-old children. Lower cognitive scores could lead to poor performance in school with implications for future achievement. A number of people have studied how original, two-parent family status during childhood affects educational attainment. There are indications that both men and women raised without both natural parents may experience some disadvantages. Because education is an important part of our model and because earlier findings have been tentative, we include education to confirm and clarify prior work. Effects of Family Structure on Occupational Status In an early study, Duncan and Duncan (1969) found that for men, growing up in an original, two-parent family increases the probability of pursuing an occupation that has relatively high socioeconomic status. More recently, Biblarz and Raftery (1993) found that family disruption affects occupational mobility directly because men from disrupted backgrounds are more likely to enter low-status occupations. According to Biblarz and Raftery, the intergenerationaltransmission of status and occupation takes place, in part, through role modeling, socialization, and sharing values between fathers and sons. When families break up, the son's access to his father is diminished. Biblarz and Raftery imply that this lack of contact results in the failure of sons to model the occupational roles of their fathers. Coming from a dis-

421 rupted family both directly lowers men's occupational destinations, compared with their father's occupations, and weakens the association between the father's occupation and his son's occupational destination, resulting in greater universalism. Comparable research on women is nonexistent, a gap this article will fill. Effects of Family Structure on Earnings Attainment Studies of how family structureaffects adult earnings attainment have been limited. In research on a sample of 23-year-old men from the Panel Study of Income Dynamics (PSID), Greenberg and Wolf (1982) found that parental marital disruption caused reductions in the mother's home time and family income, which resulted in lower earnings for men in young adulthood. Effects were strongest for Whites, but also were present for Blacks. Because only young men were studied, this research may not give us a very good indication of future earnings potential, nor does it address whether these effects are the same for women. Using the Survey of Mature Women and the Survey of Young Men to match mothers and sons from the National Longitudinal Survey of Labor Market Experience (NLS), Krein (1986) found that for a group of men who were 28-38 years old in 1980, the earnings of those raised by a single parent were negatively affected only indirectlythrough lower educational achievement. Those who lived with one parent during preschool were most likely to achieve less education by age 28. Krein cautions that additional negative effects may show up later because effects were measured soon after the completion of education. In a comprehensive study using the Panel Study of Income Dynamics, Hill et al. (1987) explored several aspects of attainment for a group of 24-year-old men and women, including wages for young men and income-to-needs ratios for young men and women. They found that wages for White men were reduced because of net income deprivation in the family of origin caused by family disruption. Net income deprivation refers to a situation in which there is a net loss in family income that cannot be made up by the compensating efforts of mother and children. The lower income constrains childhood developmental activities and ultimately reduces the respondents' adult attainments. Additionally, the family income-to-needs ratios for men and women indicated a reduction in the ability to meet family

422
needs only for White and Black sons because of net income deprivation. There was no significant reduction on this dimension for the women in the study. Although this study included men and women, it looked only at income-to-needs ratios (not earnings) for women, and, again, it studied a very limited age range of adults early in their working years. Alternative Explanations To establish the impact of family structureduring childhood on education, occupational status, and earnings, we consider alternative explanations for the anticipated associations. Race is related to both earnings and to original, two-parent family status. Blacks have higher rates of single parenting than Whites (Bumpass, 1984). Religion may have an impact on family status because of Roman Catholic prohibitions against divorce. Smaller families may be better able to provide educational opportunities for children (Blake, 1989; Downey, 1995). Married women work fewer hours and earn less than unmarried women (Peterson, R., 1989), and married men earn more than unmarried men. These facts suggest that marital status is a possible alternative explanation. Finally, following human capital arguments, years of work experience and hours worked positively affect earnings (e.g., Parcel, 1979; Parcel & Mueller, 1983), and thus are sensible controls. Regarding the potential for statistical interaction, we know that status attainment processes vary by gender (Parcel & Mueller, 1983). Women obtain weaker returns to investments such as experience or to characteristics such as occupational status. Not only do adult socioeconomic outcomes vary by gender, but Morgan, Lye, and Condran (1988) report that families with male children are less likely to experience marital disruption than families with only female children. Krein and Beller (1988) report that the negative effects of living in a single-parent family are greater for boys than for girls. It is sensible, therefore, to consider whether the effects of family structure on several adult outcomes will also vary by gender. Hypotheses The causal chain for studying earnings links family background characteristics to educational attainment, which leads to occupational status that helps determine variations in earnings. Changes

Journal of Marriage and the Family in family background can have an impact on earnings, either directly or indirectly through their effects on education or occupational status. These basic effects should be in the same direction for men and women with one exception. Marriage should have a negative effect on earnings for women and a positive effect for men. We expect educational achievement, occupational status, and earnings attainment for children from original, two-parent families to be greater than for children from other family structures. We believe that the negative effects of disrupted family status on future earnings will differ for men and women. For women, the effects will operate through lower educational achievement. For men the effects will operate through occupational outcomes. METHODS Sample and Measurement We use data from the Panel Study of Income Dynamics, a longitudinal survey that began in 1968. It consists of a representative sample of individuals and the families in which they reside in the U.S. The study is conducted by the Survey Research Center (SRC) at the University of Michigan and consists of two independent samples with different sampling frames. The first, known as the SRC sample, included nearly 3,000 families from the master sampling frame and was a cross-sectional, national sample using a stratified multistage sample of the civilian, noninstitutional population of the U.S. (Hill, 1992). The second sample, from the Survey of Economic Opportunity and known as the SEO sample, sampled standard metropolitan statistical areas and also included nonstandard metropolitan statistical areas in the South. It involved unequal selection probabilities (Hill, 1992). This sample included about 1,900 families with low income in 1968 who were interviewed by the Census Bureau for the Office of Economic Opportunity (Inter-university Consortium for Political and Social Research, 1993). Together, both samples included a total of 4,802 families in 1968. The members of the original study have been followed through the years as new families are formed when members leave their families of origin, resulting in over 7,000 families in the 1989 interviews. Each year since the study began, data on income, family structure, and socioeconomic background have been collected. If a married or per-

Family Structure, Gender, and Earnings Attainment manently cohabiting couple is in the survey, the man is normally designated the head of the family. Single women with or without children also are designated family heads. Surveys are conducted annually with the head of the family when possible, although in couples the wife is interviewed in some instances. In this study, we use a subsample of respondents (n = 3,523) from the 1989 panel (Wave XXII) who were family heads or wives in that year, who were aged 30-59, were from the civilian work force, and who reported an occupation and had earnings greater than zero in 1988. The combined PSID sample is a probability sample but with unequal selection probabilities, making weighting necessary in order to represent the population of the U.S. PSID weighting variables compensate for both unequal selection probabilities and sample attrition (Hill, 1992). Because the original families were selected in 1968, a recent increase in Latino immigration is not reflected in the primary sampling units, and that population is not adequately represented in this sample. We use only sample members from the original families in the study. Some were heads and wives in 1968, and others were children. Additionally, although limited data are available on all other family members, the data for heads and wives are much richer and include detailed reports on family background, occupation, and earnings that are not available for other family members. Limiting the study to heads and wives eliminates 156 other family members who are primarily adult children still living with a parent or parents. We have limited the study to sample members who were 30-59 years old in 1989 because our interest is in people in their primary working years. By age 30, many individuals have completed most of their education and have had time to establish their own households or marry, thus becoming heads or wives. Placing an upper age limit of 59 on the sample helps assure that most respondents have not yet retired from the labor force. We look primarily at those with earnings greater than zero for two reasons. Methodologically, we eliminate zero-earners based on an analysis by Hauser (1980), who demonstrates that including values of zero on the log of earnings gives a distribution an extreme, negative skew. Substantively, we are interested in individual earnings of employed men and women. Previous work has examined total family income, which

423
confounds marital status with an individual's own economic fate, especially for women. Although data indicate that women frequently recover from the economic losses of divorce by remarrying (Duncan & Hoffman, 1985), there is some evidence to suggest that they are less likely to remarry if they are economically independent (Ambert, 1983). In order to understand what (if any) biases this decision may introduce, we include a Heckman correction to determine how family structure affects the hazard of being excluded from the sample because of nonearner status ( Berk, 1983; Heckman, 1980). The results are reported separately near the end of this article. Variables Original, two-parent family status. Because our respondents were born between 1930 and 1959, detailed longitudinal data on their family histories are unavailable. We must rely on their adult recollections of their childhood. As Corcoran, Gordon, Laren, and Solon (1991) note, retrospective accounting can be inaccurate and thus injects some error into the measurement of variables. This affects our measure of family status. Respondents were asked, "Were you living with both of your natural parents most of the time until you were 16?" Although this measure lacks detail about duration or the ages at which children lived with a single parent, there is some evidence that this may not matter. McLanahan and Sandefur (1994) found that the timing and duration of single parenting were unimportant for understanding high school dropout rates or teen birth rates. We believe our measure succeeds in capturing the essence of the adult's perception of the family structure in which he or she was raised. We believe people know if their two natural parents were together during their childhoods, and we think this measure validly taps childhood family structures. Adult children who were not raised in original, two-parent families may have a variety of experiences, including being raised by single, nevermarried mothers or by one parent after divorce or after the death of the other parent. Or they may have been raised in stepfamilies or have grown up with people other than their parents, such as foster parents or grandparents. Although several studies have indicated that it may be important to look at the family structure of the disrupted family in which the child is raised (Acock & Kiecolt, 1989; Biblarz & Raftery, 1993; Demo & Acock,

424

Journal of Marriage and the Family


TABLE 1. MEANS AND STANDARD DEVIATIONS OF ORIGINAL, TWO-PARENT FAMILIES AND OTHER FAMILY STRUCTURES FOR WOMEN AND MEN

Women Original,TwoParentFamilies M Family background Numberof siblings Catholicab Parents'educationbd Whiteabc Age Education Total yearsabc Dummies in years
0-1la

Men OtherFamily Structures M (SD) (2.95) (.40) (1.50) (.45) (8.43) (2.23) (.37) (.50) (.43) (.27) (.27) (.47) (18.57) (8.83) (678.85) (13.05) (1.03) (10.37) OriginalTwoParentFamilies M 3.50 .27 3.75 .93 41.69 13.66 .11 .33 .23 .18 .16 .85 41.82 20.99 2249.13 43.25 (SD) (2.64) (.44) (1.52) (.26) (8.13) (2.45) (.31) (.47) (.42) (.39) (.36) (.35) (21.29) (8.90) (645.35) (12.41) OtherFamily Structures M 3.50 .18 3.52 .80 42.00 13.38 .14 .36 .21 .16 .13 .83 40.92 21.62 2194.38 42.40 10.12 32.48 (SD) (2.95) (.38) (1.28) (.40) (7.97) (2.49) (.34) (.48) (.41) (.36) (.34) (.38) (21.16) (8.81) (604.54) (11.63)

(SD) (2.80) (.44) (1.56) (.33) (8.05) (2.32) (.30) (.49) (.42) (.33) (.34) (.44) (18.97) (8.18) (672.43) (12.93)

3.58 .27 3.70 .87 41.73 13.34 .10 .41 .23 .12 .14 .74 42.86 14.37 1727.69 33.19

3.62 .20 3.63 .72 42.46 12.75 .17 .43 .25 .08 .08 .68 38.58 15.42 1782.02 34.27 9.34 16.03

12C 13-15 16ac 17+ac Marriedabc Occupationalstatusa Years workedc Hours workedin 1988C Hours workedper weekc Earnings Logged earningsabc Dollars (in thousands)ac

9.45 (.97) 17.91 (13.52) (n = 1355)

(n = 339)

10.21 (.79) 36.18 (44.80) (n = 1497)

(.80) (33.79) (n = 332)

aStatistically significantdifferencebetween women from original,two-parentfamilies and other family structures. bStatistically significantdifferencebetween men from original, two-parentfamilies and otherfamily structures. cStatisticallysignificantdifferencebetween men and women. dParents'education is categorized from 1 to 8. 1 = 0-5 years of schooling, 4 = completion of high school, 8 = postbaccalaureate work.

1991; Kobrin & Waite, 1984; Wojtkiewicz, 1992), a recent, detailed analysis of high school graduation by Wojtkiewicz (1993) concludes that concern about family type overcomplicates the issue because major effects are observed with a dichotomous indicator. In this study, we classify all men and women who lived with both natural parents most of the time until age 16 as from original, two-parent families. Those who state that they did not live with both natural parents until age 16 are classified as from a perceived disrupted family. Other family background variables. Sex is coded 1 if the respondent is a female; males are coded 0. Religion is coded 1 if Roman Catholic, 0 otherwise. Race is coded 1 if White, 0 otherwise. Married respondents are coded 1, 0 otherwise. The number of siblings is the actual number of siblings during childhood as reported by the respondent. Parents' education is used to control for

family socioeconomic background. This is calculated as the mother's education plus the father's education, divided by two. Although parental education may not capture all aspects of the socioeconomic status of the childhood family, generally it is highly correlated with occupational status and takes advantage of available information. Education and occupational status. We use total years of education as a dependent variable. As an independent variable, we create categories to represent credential completion, with categories for 0-11 years, 12 years (the comparison group), 13-15 years, 16 years, and 17 or more years. Dummy variables allow us to examine relative differences in the importance of levels of education for men and women on occupational status and earnings. Occupational status is based on Stevens and Featherman's (1981) MSE12 revised index of Duncan's original, 1950, (SEI) index of occupational socioeconomic status.

Family Structure, Gender, and Earnings Attainment Earnings. Earnings are logged to reduce the impact of outliers and to allow interpretationof relative differences, rather than absolute amounts (Peterson, T., 1989). Peterson reviews the myriad of functional forms that sociologists have used to estimate earnings equations, including log linear forms. He mathematically demonstrates that controlling for annual hours worked on earnings is equivalent to setting an hourly wage, which confounds the exploration of differences in earnings between subgroups. We agree with his arguments, and we control for hours worked by setting the coefficient to 1 for that variable. Peterson also notes that earnings inequality is generated both by differences in hourly wages and by differences in hours worked. When logged earnings is the dependent variable and we control for hours worked, once we fix hours worked, all variation is due to the variation in the hourly wage rate. This allows the dependent variable-earnings, to be interpreted in terms of the hourly wage, and the independent variables then are interpreted in terms of the hourly wage, a convenient interpretation. If earnings is the dependent variable and if we simply control for hours worked, it implies that the average hourly wage rate depends on the numbers of hours worked when it actually depends on other factors, such as education and experience. (See T. Peterson, 1989, for the derivation of equations.) This variable, adult earnings, is based on Parcel (1979), and includes "earnings when employed by others, income from selfemployment, and other income derived from respondents' own labors" (p. 267). The number of years worked since age 18 includes full-time work that the respondent reported.

425 Missing values. We use mean substitution for missing values. The percentage of missing values ranged from 0 for number of siblings, marital status, occupational status, hours worked, and earnings to 7.5% for parent education. Most variables were missing values in fewer than 1% of cases. Analytic Strategy After presenting basic descriptive data, the central analytic strategy is a series of multiple regression analyses where variables are entered in the order determined by theory. Based on this fully recursive model, we perform separate analyses for men and women. We then consider interactions and investigate possible differences between earners and nonearners in a hazard analysis. DESCRIPTIVE RESULTS Means and Standard Deviations We present means, standard deviations, and notations of differences of means tests in Table 1. Both men and women who were raised in original, two-parent families complete more total years of education and have higher logged earnings than those from other family structures. Some of the differences are modest. Additionally, a comparison of means for women and men indicates that men achieve more years of education at higher levels than women, and they have higher earnings. Men and women are similar in levels of occupational status. Bivariate correlations between the respective dependent variables and predictors are in the expected directions and generally statistically significant (data available from the first author).

TABLE 2. YEARS OF EDUCATION REGRESSED ON FAMILY STRUCTURE (EQUATION 1) AND FAMILY BACKGROUND (EQUATION 2) FOR WOMEN AND MEN

Women Equation1 IndependentVariables Original,two-parentfamily Constant R2 Adjusted R2 b (SE) .59** (.14) 12.75 .01 .01 Equation2 Equation1

Men Equation2

P
.10

b (SE) .57** (.13) 10.98 .22 .22

P
.10

b
(SE) .28t (.15) 13.38 .00 .00

P
.04

b (SE) .10 (.13) 11.83 .21 .21

P
.02

Note: For women, n = 1,694. For men, n = 1,829. In Equation1 original,two-parentfamily statusis enteredalone. Equation 2 includes original, two-parentfamily status and other family backgroundcharacteristics,including logged numberof siblings, Catholic,logged parentaleducation,White. tp < .05. *p < .025. **p < .005. One-tailedtest.

TABLE OCCUPATIONAL REGRESSEDFAMILY 3. ON STATUS STRUCTURE BACKGROUND 1), (E (EQUATION FAMILY


AND EDUCATION (EQUATION 3) FOR WOMEN AND MEN

Women Equation1 Variables Independent Original,two-parentfamily Numberof siblings Catholic Parents'education White Educationa 0-11 years 12 years (omittedcategory) 13-15 years 16 years 17+ years Constant R2 AdjustedR2 38.56 .01 .01 24.83 .14 .14 b (SE) 4.31** (1.15) P .09 Equation2 b (SE) Pb (SE) .71 .01 (.92) -.30* -.04 (.14) 3.14** .07 (.84) .82** .07 (.27) 3.50** .07 (1.08) -10.48** (1.23) 6.20** (.95) 17.44** (1.25) 26.76** (1.23) 30.17 .40 .39 -.18 .14 .30 .47 40.89 .00 -.00 Equation3 Equation1 b (SE) .94 (1.28) 3 p .02

3.39** .07 (1.08) -.70** -.10 (.16) 2.05* .05 (1.00) 3.68** .30 (.29) 3.43** .07 (1.28)

Note: For women, n = 1,694. For men, n = 1,829. Educationdummiesare statisticallysignificantas a group.Women, F = 174.83, significantF .0000. Men, F = 279.23, significantF .0000 *p < .025. **p < .005. One-tailedtest.

Family Structure, Gender, and Earnings Attainment

427

ON TABLE LOGGED 4. EARNINGS REGRESSED ORIGINAL, STATUS TWO-PARENT FAMILY BACKGROUND 1), (EQUATION FAMILY STATUS (EQUATION EDUCATION 2), (EQUATION MARITAL 3), 4), (EQUATION ANDWORK(EQUATION FORWOMEN 5)

Equation1 Independent Variables Original,two-parent family Logged numberof siblings Catholic Logged parents' education White Educationb 0-11 years 12 years (omitted) category) 13-15 years 16 years 17+ years Married Logged occupational status Logged years worked since 18 Logged hours worked Constant R2 AdjustedR2 9.34 .00 .00 b (SE) .11' (.06) t ratioa 1.84

Equation2 b (SE) t ratio

Equation3 b (SE) .03 (.06) -.01 (.02) .12* (.05) .04 (.06) .01 (.07) t ratio .45 -.85 2.31 .68 .08

Equation4 b (SE) .03 (.06) -.01 (.02) .12* (.05) .02 (.06) .10 (.07) t ratio .55 -.73 2.36 .42 1.52

Equation5 b (SE) t ratio

1.77 .1ll (.06) -.04* -2.19 (.02) 1.66 .09t (.06) .29** 5.19 (.06) .03 .41 (.07)

.03 .89 (.04) -.02t -1.92 (.01) .08* 2.21 (.04) .04 1.06 (.04) .07 1.51 (.05) -.15** -2.74 (.05) .07t 1.71 (.04) .27** 4.99 (.05) .36** 6.43 (.06) -.11** -3.10 (.04) .37** 9.68 (.04) .27** 8.39 (.03) 1.00 fixed (.00) parameter -.15 .61 .60

-.49** -6.29 (.08) .12* 1.96 (.06) .35** 4.14 (.08) .60** 7.71 (.08)

-.52** -6.74 (.08) .12* 1.97 (.06) .33** 4.19 (.08) .59** 7.73 (.08) -.33** -6.37 (.05)

8.97 .03 .02

9.25 .09 .09

9.43 .12 .11

Note: n = 1,694 women. at ratios are shown instead of standardized ps. dummies are statisticallysignificantas a group, F = 31.63, significantF .0000. bEducation tp .05. *p <.025. **p< .005. One-tailedtest.

Regression Analysis

Table 2 shows that original, two-parentfamily statusduringchildhoodhas positiveeffects on education for girls that do not diminish with the introductionof controls for other family background variables. Women from original, twofamiliesachieveover a half yearmoreeduparent cation than women from other family structures when otherfamily background characteristics are held constant.In contrast,coming from an original, two-parent family is not a significantpredictor of educational attainment men, once relefor vantbackground arecontrolled. predictors Table 3 suggeststhat for women, growingup with both naturalparentshas positive effects on occupationalstatus that operatethrougheduca-

tional attainment. significanteffects of origiThe nal, two-parentfamily statusremainwhen other family backgroundvariables are controlled, alof thoughthe magnitude the effect dropsand becomes insignificantwhen educationis held constant. Completionof college and graduatework are the strongestpredictorsof high occupational status for women. The education dummy variableshave the expectedeffect. The category0-11 has yearsof education a negativeimpacton occupationalstatus,andthereareincreasingly positive effects associatedwith moreeducation. An original,two-parent is family background never a positive, significantpredictorof occupational status for men. Notice, however, that although original,two-parent family status has no statusin Equasignificanteffect on occupational

428

Journal of Marriage and the Family

TABLE5. LOGGED EARNINGS REGRESSED ORIGINAL, ON FAMILY TWO-PARENT STATUS(EQUATION FAMILY BACKGROUND 1), STATUS (EQUATION EDUCATION 2), (EQUATION MARITAL 3), (EQUATION ANDWORK(EQUATION FORMEN 4), 5)

Equation1 Independent Variables Original,two-parent family Logged numberof siblings Catholic Logged parents' education White Educationb 0-11 years 12 years (omitted category) 13-15 years 16 years 17+ years Married Logged occupational status Logged years worked since 18 Logged hours worked Constant
R2

Equation2 b (SE) .02 (.05) -.05** (.01) .32** (.04) .31** (.04) .24** (.06) t ratio .44 -3.90 7.65 7.18 3.87

Equation3 b (SE) -.00 (.05) -.03* (.01) .30** (.04) .09* (.05) .22** (.06) t ratio -.01 -2.14 7.52 2.08 3.59

Equation4 b (SE) -.00 (.04) -.02* (.01) .29** (.04) .11* (.04) .17** (.06) t ratio -.06 -2.00 7.36 2.54 2.82

Equation5 b (SE) t ratio

b (SE) .08t (.05)

t ratioa 1.76

-.03 -.69 (.04) -.02* -1.97 (.01) .23** 6.43 (.04) 1.78 .07t (.04) 1.70 .09t (.05) -.20** -3.72 (.05) .04 1.02 (.04) .20** 3.87 (.05) .18** 3.12 (.06) .11* 2.51 (.04) .33** 8.60 (.04) .20** 4.46 (.05) 1.00 fixed (.00) parameter .38 .33 .32

-.23** -3.79 (.06) .16** 3.45 (.05) .43** 8.39 (.05) .53** 9.55 (.06)

-.22** -3.64 (.06) .17** (.05) .43** (.05) .54** (.05) .29** (.05) 3.69 8.44 9.95 6.06

AdjustedR2

10.12 .00 .00

9.53 .08 .08

9.66 .16 .16

9.43 .18 .17

Note: n = 1,829 men. at ratios are shown instead of standardized3s. dummiesare statisticallysignificant as a group,F = 41.51, significant F .0000. bEducation tp <05. *p <.025. **p< .005. One-tailedtest.

tion 1, it has significant negative effects on occupational status in Equation 3. We discuss possible explanations for this finding later. As with women, the strongest predictors of high occupational status are college graduation and graduate work. Not surprisingly, failure to complete high school has negative implications for occupational status. The effect of growing up in an original, twoparent family on adult earnings differs for men and women. Women who grow up with both parents in the family earn more than those who do not. This difference holds when other family background variables are controlled. As Table 4 indicates, about two thirds of this effect is accounted for by education, rendering the twoparent family coefficient not significant. We re-

call from Table 2, however, that growing up in a disrupted family has a negative impact on the education of women, which is later reflected in earnings. The education variables operate as expected. The lowest category of education has significantly negative effects, and the categories of 13-15 years of education and more have positive effects. College completion signified a dramatic gain over 13-15 years of education, and returns to post-baccalaureate work are quite strong. Table 5 suggests chat men's adult earnings, in contrast to women's, are not significantly affected by family structure. The positive effect of growing up in an original, two-parent family on earnings disappears when other family background variables are considered. The education dummies show the same effects as those for women, but the

Family Structure, Gender, and Earnings Attainment effects of graduate education are minimal for men. Men receive relatively more benefits from a 4-year college degree than for higher levels of education. Women receive relatively more returns from graduate work. Being married works to men's advantage as far as earnings are concerned. So do higher occupational status and years of work experience. For theoretical reasons mentioned earlier, we fixed the coefficient for hours worked to 1 for the earnings analyses. Although the coefficients for effects more clearly predict wage attainment in our model than in an unrestricted model, significance of predictors is virtually identical in both models. Additional Analyses Interactions. Growing up in a home without both naturalparents appears to disrupt women's educational attainment but not men's. In order to determine if there is a significant interaction between gender and two-parent family status for educational attainment, we combined men and women in the sample and created a multiplicative gender x two-parent family interaction variable. When added to the equation, the interaction is significant at p < .05, two-tailed test. Clearly, education is more disrupted for women who did not live with both parents during childhood than for men. There is no significant interaction of two-parent family status and gender on occupational status or earnings. In addition, we created a multiplicative term composed of race and two-parent family sta-

429 tus. This term was not significant for any of the dependent variables. The finding of no effects of growing up in an original, two-parent family on educational achievement, occupational status, or earnings attainment for men is provocative because both theory and prior empirical work indicate that living in two-parent families during childhood should have positive implications for adult outcomes. In response to these findings, we divided our sample of men into two groups: those born prior to the baby boom (before 1946) and who entered a relatively friendly labor market, and baby boomers, born from 1946 through 1959, who entered a tighter labor market. This analysis suggests that two-parent family status had negative effects on earnings for men in the older group and positive effects for the younger men (see Table 6). It is possible that in the whole group of men, these effects cancel each other out, resulting in the finding that two-parent family status has no significant effect on outcomes for men. Although there are opposite effects on earnings for the two groups of men, for both groups the effect is explained when occupational status, years worked, and hours worked are added to the model. Effects do not occur through education, as is the case for women. These results suggest that the finding of no significant effect of two-parent family status during childhood for men as a whole should be reexamined. A similar analysis by cohort for the women (not shown) indicated the effect of family

TABLE 6. COMPARISON OF YOUNGER MEN AND OLDER MEN, LOGGED EARNINGS REGRESSED ON ORIGINAL, TWO-PARENT FAMILY STATUS (ABBREVIATED EQUATIONS)

Equation1 Original,TwoParentFamily b (SE) Youngermen (n = 1,114) Oldermen (n = 715) All men (reference) .24** (.06) -.16* (.08) .08t (.05)

Equation2

Equation3

Equation4 Equation3 and MaritalStatus b (SE)


.09t

Equation5 Equation4 and Work Variables b (SE) .09 (.05) -.08 (.06) -.03 (.04)

Equation1 and Family Background Equation2 and Characteristics Education b (SE) .12* (.06) -.13t (.08) .02 (.05) b (SE) .10t (.06) -.16* (.07) -.00 (.04)

(.06) -.16* (.07) -.00 (.04)

Note: Equation 1: Original,two-parentfamily statusis enteredalone. Equation2: Original,two-parentfamily status and other family backgroundcharacteristics,including logged numberof siblings, Catholic, logged parents' education, White. Equation3: Original,two-parentfamily, other family backgroundcharacteristics,and education.Equation4: Original,twoparent family, other family backgroundcharacteristics,education, and married.Equation 5: Original, two-parent family, other family background characteristics, education, married, and work variables, including logged occupational status, logged years worked,logged hours worked (set to 1). tp .05. *p <.025. **p <.005. One-tailedtest.

430 structure during childhood on attainment outcomes was similar for both groups. Earners versus nonearners. One might argue that a sample including only those with earnings who declare occupations would fail to reveal the entire effect of original, two-parent family status during childhood on adult outcomes. Suppose, for example, that children raised in single-parent families are less likely to become earners. To determine whether the effects of growing up in a two-parent family differ for those who are excluded from the sample because of their status as nonearners, we conduct a hazard analysis. We construct a variable to represent the hazard of being excluded from the analysis because of nonearner status, given that respondents who are included are earners. The hazard of nonemployment measures a latent trait that varies among those who are employed. Leaving it out of the model could bias some of the observed effects. In this case, the hazard of nonemployment is constructed from a variable indicating the respondent's assessment of personal health because there is a positive correlation between health and earnings. We believe that the effect occurs simply because poor health affects the chance of being excluded from the sample. We first add all nonearning heads and wives back into the sample (n = 4,576) and perform a probit analysis that predicts the probability of being in or out of the sample because of earner status. This equation includes key independent variables as well as the indicator for health. These results are used to compute the hazard of nonemployment. We then run the regressions for earners, including the hazard variable. This analysis is performed with Limdep 6.0, which does not permit the restriction of the coefficient of hours worked to 1, so we have estimated another model for comparability (details available from the first author). Our results indicate that the hazard of being excluded differs for women and men. The hazard is insignificant for women, indicating that limiting our sample to women who are earners does not bias our estimates. For men, the hazard of being a nonearner is significant, but the significance levels of most of the independent variables remain stable. The equation controlling for sample selection bias reveals that the positive effects of parents' education, being married, and being White in the original analysis are due to sample selection. In addition, once we control for the hazard of nonemployment, low levels of educa-

Journal of Marriage and the Family tion no longer have a statistically negative effect on earnings for men. In sum, respondents from higher socioeconomic backgrounds, those who are White or married are more likely to be earners, hardly a surprising finding. Because some of the differences in family structure act through education, we included all nonearners with earners in a regression on education for men and women. The results were essentially the same as in the main analyses. For example, in the main analysis, women from original, two-parent families had significantly higher levels of education than those from other family forms, even when other family background characteristics were controlled. That was not the case for men. For men, the effect of two-parent family status on education was explained by other family background variables. When nonearners were included in the analysis, the same pattern appeared. DISCUSSION Our interest was to explore the effects of growing up in an original, two-parent family on the earnings attainment process and to explore specifically how this process differs by gender. We predicted that men and women from two-parent families would fare better than those from other family structures. This is supported. Further, we found that there were gender differences in the effects of family structure during childhood on adult outcomes. When all earners between the ages of 30 and 59 in 1989 are considered, growing up in a two-parent family is more important for women than for men. Women's education suffers when they are raised in disrupted families, and those disadvantages carry through to occupational and earnings outcomes. The same effect is not present for men. The positive association between twoparent family status and education for men is explained by other family background variables. In addition, when the men are divided into two cohort groups, growing up in a two-parent family has positive implications for the earnings of the men born during the baby boom, and it has negative effects on that outcome for men born prior to the baby boom. Although our data set does not allow the investigation of the cause of disruption, we know that most family disruption for the older group was caused by death rather than divorce. We speculate that because the pre-baby boom and baby boom cohorts entered the labor market under dramatically different economic conditions, the interplay of family structure and economic

Family Structure, Gender, and Earnings Attainment opportunity when the labor market was entered may have had subsequent effects on their earnings attainment. In conclusion, these findings support status attainment and human capital theories for the women in the study. Two-parent family status can be viewed as a valuable family background characteristic that enhances educational opportunities for daughters and has subsequent implications for occupational status and earnings. The value of two-parent family status as a resource for men is much less certain. Overall, for the men in the study, being raised in a two-parent family offers no apparent benefit for earnings attainment outcomes. However, the results of the cohort study suggest that the state of the labor market may explain differences in the effect of family structure on the results for men. For the younger men, the cohort analysis supports earlier findings of Biblarz and Raftery (1993) that men whose families are disrupted by divorce are disadvantaged in their occupational endeavors. What is clear from these data, however, is that men's education is not compromised by disrupted family status during childhood the way women's education is. The cohort analysis also suggests that the positive and negative effects of various family structurescould be historically specific. Furtherresearch is needed on social supports available to disrupted families of earlier times. It is useful to consider what implications these findings might have for successive cohorts of youths who experience divorce during childhood. The divorce rate is likely to remain quite high in the near future (Martin & Bumpass, 1989), and great numbers of children will spend part of their childhood in single-parent families (Bumpass & Rindfuss, 1979; Furstenberg & Nord, 1985). If the processes we have modeled here hold for younger cohorts, women from disrupted families may be more at risk for attaining fewer years of schooling than are men from disrupted families. The specific mechanisms at work could include women not receiving the financial support that men do to attain additional schooling. In addition, women may not receive the parental encouragement and social support needed to aspire to higher levels of education or to complete degrees. Given the strong impact of schooling on earnings, lower levels of educational attainment have clear implications for financial well-being. Although girls not raised in original, two-parent families lose, on average, only slightly more than a half year of education, as Krein (1986) points out, which half a

431 year is lost is important. If that time makes the difference between completing and not completing high school, for example, it could make a big difference in later socioeconomic outcomes. For these reasons, parents and policymakers need to consider the possibility that girls from disrupted families may be especially vulnerable to compromising educational aspirations and attainment. This possibility suggests the need for parents to explicitly consider how marital dissolution might affect the educational attainment of their children, particularly their daughters. These youths especially may need encouragement from parents and schools to pursue higher levels of education. Many of today's children will be raised in single-parent and blended families, and it is vital that we have as clear an understanding of the short- and long-range implications of these changes as possible. This study alerts us to the long-range implications of family structure and reminds us that the mechanisms and the impact can differ for women and men. We also must remain cognizant of the difficulties families and children face when there is conflict or lack of resources that may bring disadvantages to those children concerned. NOTE An earlierversion of this article was presentedat the 1995 annualmeetingof the AmericanSociologicalAssociationin Washington, D.C. We thankRobertKaufJohnMirowsky,Catherman,ElizabethG. Menaghan, ine Ross, and two anonymous reviewersfor comments andadviceon earlierdrafts. REFERENCES Acock,A. C., & Kiecolt,J. (1989). Is it familystructure or socioeconomic status? Family structureduring adolescenceand adultadjustment. Social Forces, 68, 553-571. divorceand Amato,P. R., & Keith,B. (1991). Parental adult well-being:A meta-analysis. Journalof Marriageand theFamily,53, 43-58. Ambert,A. (1983). Separatedwomen and remarriage behavior: comparison financiallysecurewomen A of and financiallyinsecurewomen.Journalof Divorce, 6, 43-54. Astone, N. M., & McLanahan,S. S. (1991). Family and structure, practices, high school compleparental tion.American SociologicalReview,56, 309-320. and Becker,G. S. (1975). Humancapital:A theoretical to empirical analysiswithspecialreference education Press. (2nded.). New York:Columbia University Becker,G. S., & Tomes,N. (1986). Humancapitaland the rise and fall of families. Journalof LaborEconomics,4, S1-S39.

432 to Berk,R. A. (1983). An introduction sampleselection


bias in sociological data. American Sociological Review, 48, 386-398. and economic hardship: The short-run picture for

Journal of Marriage and the Family Holden, K. C., & Smock, P. J. (1991). The economic costs of maritaldissolution:Why do women bear a 17, 51-78. for Consortium Politicaland Social ReInter-university 1994. Ann Arbor,MI: Inter-university Consortium forPoliticalandSocialResearch. Kobrin,F. E., & Waite,L. J. (1984). Effects of childon hood family structure the transitionto marriage. famiKrein,S. F. (1986) Growingup in a single-parent ly: The effect on educationand earningsof young attainKrein,S. F., & Beller,A. H. (1988). Educational ment of childrenfrom single-parent families:Differences by exposure, 25, genderandrace.Demography, 221-234. Martin,T. C., & Bumpass,L. L. (1989). Recenttrends in marital 26, disruption. Demography, 37-51. S. and McLanahan, (1985). Familystructure the repro90, 873-901. famiMcLanahan, & Booth,K. (1991). Mother-only S., lies: Problems,prospects,and politics. In A. Booth MN: National ing back (pp. 405-428). Minneapolis, Councilon FamilyRelations. McLanahan,S., & Bumpass, L. (1988). Intergenerational consequencesof family disruption. American McLanahan,S., & Sandefur,G. (1994). Growing up Press. bridge,MA:Harvard University G. Morgan,S. P., Lye, D. N., & Condran, A. (1988). Sons, daughters,and the risk of maritaldisruption.
American Journal of Sociology, 94, 110-129. Mott, F. (1993). Absent fathers and child development: Emotional and cognitive effects at ages 5-9. Unpubwith a single parent: What hurts, what helps. CamJournal of Sociology, 94, 130-152. (Ed.), Contemporaryfamilies: Lookingforward, lookduction of poverty. American Journal of Sociology, men. Family Relations, 35, 161-168. Journal of Marriage and the Family, 46, 807-816. search (1993). Guide to resources and services, 1993disproportionate cost? Annual Review of Sociology,

E. Bianchi,S., & McArthur, (1991). Familydisruption children. (CurrentPopulationReports, Household EconomicStudies,SeriesP-70, No. 23). Washington, DC:U.S. Department Commerce. of Biblarz,T. J., & Raftery,A. E. (1993). The effects of on American Sociofamilydisruption socialmobility.
logical Review, 58, 97-109. Blake, J. (1989). Family size and achievement. Berke-

of Press. ley: University California L. and Bumpass, L. (1984). Children marital disruption: a replication update. and 21, Demography, 71-82. Bumpass,L. L., & Rindfuss,R. G. (1979). Children's of American Journalof experience marital disruption. Coleman,J. S. (1988). Social capitalin the creationof S95-S120. Corcoran, M., Gordon, R., Laren, D., & Solon, G. betweenmen's economicsta(1991).The association tus and theirfamily and community origins.Journal relations:Assessing Demo, D. H. (1992). Parent-child 54, 104-117. Demo, D. H., & Acock, A. C. (1991). The impactof divorce on children.In A. Booth (Ed.), Contemporary MN: NationalCouncilon Family 191). Minneapolis, Relations. FamiDowney,D. B. (1995). Whenbiggeris not better: and resources, children'seducational ly size, parental 761. Duncan,B., & Duncan,O. D. (1969). Family stability 285. Duncan,G. J., & Hoffman,S. D. (1985). A reconsiderationof the economicconsequences marital of dissoluF. Furstenberg, F., & Nord, C. W. (1985). Parenting Patterns childrearing marital of after apart: disruption. D., Greenberg, & Wolf, D. (1982).Theeconomicconsemaritaldisruptions. quencesof experiencing parental
Children and YouthServices Review, 4, 141-162. Journal of Marriage and the Family, 47, 893-904. tion. Demography, 22, 485-497. and occupational success. Social Problems, 16, 273performance. American Sociological Review, 60, 746families: Looking forward, looking back (pp. 162recent changes. Journal of Marriage and the Family, of Human Resources, 27, 575-601. human capital. American Journal of Sociology, 94, Sociology, 85, 49-65.

lished monograph for prepared the NationalInstitute of ChildHealthand HumanDevelopment, Bethesda, MD. Mueller,C. W., & Pope, H. (1977). Maritalinstability: A study of its transmission between generations. Parcel,T. L. (1979). Race, regionallabormarkets,and 279. and Parcel,T. L., & Mueller,C. W. (1983). Ascription New York:AcademicPress.
labor markets: Race and sex differences in earnings. Peterson, R. R. (1989). Women, work, and divorce. Alearnings. American Sociological Review, 44, 262Journal of Marriage and the Family, 39, 83-93.

on in Hauser,R. L. (1980). Comments "Stratification a 702-712. J. Heckman, (1980). Sampleselectionbias as a specificationerror.In J. Smith(Ed.),Female labor supply:
Hill, M. S. (1992). The Panel Study of Income Dynamics: A user's guide. Newbury Park, CA: Sage. dual economy." American Sociological Review, 45,

Princeton Press. University

Theory and estimation (pp. 206-248). Princeton, NJ:

of bany:StateUniversity New YorkPress. R. of Peterson, R. (1996). A re-evaluation the economic


consequences of divorce. American Sociological Review, 61, 528-536.

Hill, M. S., Augustyniak, & Ponza, M. (1987). EfS., lished manuscript, SurveyResearchCenter,Institute for SocialResearch, AnnArbor, MI. Hoffman,S. D., & Duncan,G. J. (1988). Whatare the economicconsequences divorce? of 25, Demography, 641-645.
fects of parental divorce on children's attainments: An empirical comparison of five hypotheses. Unpub-

T. Peterson, (1989). The earningsfunctionin sociological studies of earningsinequality:Functionalform andhoursworked.In A. L. Kalleberg (Ed.),Research CT: Greenwich, JAIPress. of Seltzer,J. A. (1994). Consequences maritaldissolution for children. Annual Review of Sociology, 20, in social stratification and mobility, 8 (pp. 221-250).

235-266.

Family Structure, Gender, and Earnings Attainment


Shaw, L. B. (1982). High school completion for young women: Effects of low income and living with a sinSteelman, L. C., & Powell, B. (1991). Sponsoring the 1505-1529. Stevens, G., & Featherman, D. L. (1981). A revised socioeconomic index of occupational status. Social Sci-

433
Weiss, R. S. (1984). The impact of marital dissolution on income and consumption in single-parent house127.

Journalof FamilyIssues,3, 147-163. gle parent.

holds.Journalof Marriageand the Family,46, 115parental structure during childhood and adolescence.

next generation: Parental willingnessto pay for higher education.American Journal of Sociology, 96,

R. Wojtkiewicz, A. (1992). Diversityin experiencesof


Wojtkiewicz, R. A. (1993). Simplicity and complexity in the effects of parental structure on high school graduation.Demography, 30, 701-717.

29, Demography, 59-68.

enceResearch,10, 364-395.

:k

\c.9

ll;iijC 4

fromthe 1996 Purchase videoandaudiotapesof majorsessions tools! Greatsupplemental Conference.

Family Policy Minority/International Families and in #96V1.Families Politics theNewSouth #96V5. FromPlessy vs. Ferguson,Brownvs. and N. Bd. Africa, Tutu Topeka of Education BackAgain: A and to Implications African mericans Interest #96V9.Reaction Special for of Groups M. the1996Election, Caccamo, Ness,B. Their J. N. Families, Burgess,K. Hardy Franklin Violence Violence: Gender Issues/Family Roles #96V3. Intimateand Interpersonal and Practice, R. Gelles vs. #96V2.TheWomen's Politics, Policy, Agenda theFamily or J. Agenda: Conflict Concordance? Welfare Reform N. Stacey, Glenn Health
Health Carefor #96V7. Collaborative

#96V 12. Welfare Reform,Poverty,Children,A.

Collins

Families: We It Together? Can Do J. reactor MD.; Rolland, panel Parenting #96V6. Effective,EnjoyableParenting,M. Rossmann
National Council on Family

Work and Family


Parentsin #96V4. TheMissingMiddle: Working

T. and U.S.Democracy SocialPolicy, Skocpol


#96V10. TimeAllocationto Family,Home, Work, A. and Community, Rossi

Relations, 3989 Central Ave. NE,


#550, Minneapolis, MN 55421 Phone: 612-781-9331 Toll Free: 888-781-9331 FAX: 612-781-9348

E-mail: ncfr3989@ncfr.com

Video Tapes: NCFR members: $39.95 each!Nonmembers: $49.95. Audio Tapes: $8.50each.Pricesincludeshippingand to payable NCFR. U.S. handling.Checks fundsdrawnon U.S. banksonly. MN orders residents 6.5%salestax. Canadian add add7%GST(123-830-465). JMF597

You might also like