Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Jan 5.
Published in final edited form as: Soc Sci Q. 2013 Dec;94(4):933–955. doi: 10.1111/ssqu.12019

Self-Esteem and the Reproduction of Social Class*

Spencer L James 1, Paul R Amato 2
PMCID: PMC4283772  NIHMSID: NIHMS498110  PMID: 25568500

Abstract

Objective

Although prior research has demonstrated the multiple pathways through which socioeconomic attainment occurs, one unexplored avenue regards the role of psychological mechanisms such as self-esteem in this process.

Method

Using three waves of data from the National Survey of Families and Households (N = 1,952), we employed structural equation models to examine the relationship between parenting practices and attitudes, socioeconomic status, offspring's self-esteem, and the likelihood of offspring college attendance.

Results

Self-esteem was positively related to the likelihood of offspring's college attendance. Additionally, self-esteem was found to be a modest mediator of the relationship between parental educational expectations and parental income, respectively, and the likelihood of offspring completing or being currently enrolled in college.

Conclusion

Self-esteem may constitute one previously unconsidered mechanism for reproducing the class structure in the United States.

Prior research has demonstrated the multiple pathways through which social class reproduction occurs, with children1 tending to attain social statuses similar to those of their parents—a trend attributed to parents’ socioeconomic resources and, their parenting practices and attitudes, as well as social, human, and cultural capital (Breen, 2005; Devine, 2004; Erikson and Goldthorpe, 1992; Lareau, 2003). However, sociological considerations of social class reproduction have largely ignored the role of psychological mechanisms in this process. Only a handful of studies have examined this possibility (Bynum and Durm, 1996; Demo, Small, and Savin-Williams, 1995; Gecas, 2003; Rosenberg et al., 1995), and such studies have not been able to employ large, national data sets. Thus, whether and how psychological mechanisms are related to the reproduction of the contemporary class structure in the United States remains an unanswered question. This article aims to begin to fill this gap in the literature by examining the role one important psychological mechanism—self-esteem—plays in social class outcomes.

To address the role that self-esteem plays in social class attainment, we used data from three waves of the National Survey of Families and Households (NSFH) to evaluate a model that focused first on the influence of parenting practices and attitudes on offspring's self-esteem. We then estimated how these variables affected college attendance. Because offspring were aged 18 to 35 at the final wave, we employed college attendance (completion or current enrollment) as an indicator of social class attainment, due to the prominent role of tertiary education in future occupational and financial prospects (Blau and Duncan, 1967; Fischer and Hout, 2006).

Figure 1 displays the theoretical model that guides this article. Note that this figure can be regarded as a visual representation of our expectations for the statistical model. The model mirrors the traditional socioeconomic attainment model, with its focus on indicators of parents’ socioeconomic attainment, such as education and income (Blau and Duncan, 1967), but builds on it in two important ways. First, it hypothesizes that offspring's self-esteem plays a role in the reproduction of class advantage, both through a direct effect on college completion or current enrollment and by potentially mediating the influence of both social class and parenting practices and attitudes. Second, it explicitly describes how family processes influence children's social class attainment. Thus, although social class and childrearing practices and parents’ attitudes are assumed to have a direct influence on educational outcomes (Lareau, 2000), they also may have indirect effects through children's self-esteem.

FIGURE 1.

FIGURE 1

Open in a new tab

Hypothesized Model of Social Class Reproduction via Self-Esteem

In what follows, we touch briefly on social reproduction theory, and then focus on the delineated paths of Figure 1. One section examines how parental characteristics influence children's self-esteem, followed by another section that explores the role self-esteem may play in social class outcomes, particularly the attainment of a college education.

Social reproduction theory seeks to explain the intergenerational persistence of social class positions by focusing on how families equip children with the skills that, when coupled with individual effort, make socioeconomic attainment possible (Althusser, 2006; Bourdieu and Passeron, 1990; Erikson and Goldthorpe, 1992; Willis, 1981). Parents’ educational attainment, class location, and parenting practices and attitudes play a critical role in education and eventual social class outcomes. Scholars have identified a host of factors associated with the likelihood of social class reproduction, and particularly of obtaining a college degree (Erikson and Goldthorpe, 1992; Fischer and Hout, 2006; Lareau, 2003). Much research on the intergenerational reproduction of social class has focused on how children of parents with higher levels of education, income, and occupational status tend to have better socioeconomic outcomes than children whose parents enjoy fewer economic and social advantages (Blau and Duncan, 1967; Devine, 1997, 2004; Goldthorpe, 1996), particularly because parents with more socioeconomic resources are better able to support and finance their children's educations. Prior research has also identified other characteristics that impinge on social class outcomes, including school-based activities (Choi et al., 2008; Dumais, 2008; Southgate and Roscigno, 2009), peer networks (Riegle-Crumb and Callahan, 2009), and neighborhood effects (Sharkey, 2008), although these are beyond the scope of the current study.

Thus, research has demonstrated that parental social class substantially influences patterns of educational attainment and social class outcomes—this claim appears uncontroversial (Nakhaie, 2000); it is rendered further salient in light of the growing income disparity between those with a high school and college education (Krueger, 2006).

Parenting and Children's Self-Esteem

In light of such pathways, our focus is on how parental attitudes and practices influence children's self-esteem, which then influences the likelihood of completing or enrolling in college (we describe the distinction between completion and enrollment later). Thus, our focus simultaneously emphasizes the important role of parental education and income in the intergenerational transfer of social class and examines how parental practices and self-esteem influence subsequent outcomes.

One influential context for the development of self-esteem, often conceptualized as how favorably people think of themselves, is the family (Amato and Ochiltree, 1986; Demo, Small, and Savin-Williams, 1987; Felson and Zielinski, 1989; Hughes and Demo, 1989). Prior literature has established the importance of parental social class for subsequent educational attainment and social class outcomes. One aspect of this process that has received considerably less attention is how self-esteem may influence these outcomes, as well as whether self-esteem mediates the relationship between parents’ characteristics, behaviors, and attitudes and the social class outcomes of their children.

There is suggestive evidence that parental social class, as measured by parental income and education, is related to children's self-esteem. A recent meta-analysis, for example, found that individuals with greater levels of socioeconomic status tended to report higher levels of self-esteem, particularly during young adulthood (Twenge and Campbell, 2002), when individuals make choices about college attendance. Similarly, circumstances where parental support is lacking due to financial uncertainty and instability may also lead to low levels of self-esteem among adolescents (Mayhew and Lempers, 1998). The stress and uncertainty often tied to low socioeconomic standing can also result in low self-esteem (Bradley and Corwyn, 2002). Thus, prior research leads us to expect a positive relationship between parental income and education levels and children's self-esteem.

Beyond parental social class, however, social psychologists have identified two crucial mechanisms in the development of self-esteem: reflected appraisals and the facilitation of an engaging learning environment.2 Although each proposed mechanism is based on distinct theoretical traditions, a significant degree of conceptual overlap exists between these two constructs. First, reflected appraisals (children's perceptions of significant others’ reactions and evaluations) anchor their self-concept (Burke, 2004; Cooley, 1902; Goffman, 1959). These perceptions lead children to construct positive or negative self-conceptions. When the reflected appraisals of significant others (such as parents) are positive, children are likely to internalize positive self-appraisals; when the reflected appraisals of significant others are negative, children are likely to internalize negative self-appraisals (Bouchey and Harter, 2005). Second, parents help children to develop self-esteem by facilitating an engaging and responsive learning environment through encouraging problem solving, exploration, and experimentation. Such an environment, when combined with inductive control (relying on reasoning more than coercion), high expectations, and parental warmth, is likely to foster high levels of self-esteem (Sillick and Schutte, 2006).

One way to think about how reflected appraisals and the facilitation of an engaging learning environment can be conceptualized is via specific parenting practices. Particularly, this article examines four parenting practices and attitudes, including educational expectations, harsh discipline, involvement, and investment. Available evidence suggests that these parenting practices and behaviors may help children develop self-esteem.

Parental educational expectations can influence actions that in turn influence children's self-esteem (Darolia andWydick, 2011). Parental expectations regarding how far their children will go in school can have long-term influences on both self-esteem and other subsequent outcomes (Kohn and Schooler, 1983; Lareau, 2003), and research suggests that children whose parents hold high expectations for their educational future tend to have higher levels of self-esteem (Darolia and Wydick, 2011). We therefore expect parental expectations to be linked to higher levels of self-esteem.

In terms of parental disciplinary practices, prior research suggests that both paternal and maternal authoritarian parenting practices, often characterized by harsh discipline, may be linked to lower levels of self-esteem in offspring (Furnham and Cheng, 2000). In contrast, Amato and Fowler (2002) found that the relationship between parents’ use of harsh punishment (spanking, yelling, etc.) and children's self-esteem failed to reach traditional significance levels.

Prior research has also examined the link between parental involvement and adolescent self-esteem (Baumrind, 1971). For instance, parental neglect and rejection have been shown repeatedly to be associated with low levels of self-esteem in children, whereas parents of children with high levels of self-esteem tended to be more warmly accepting of their children (Haque, 1988; Rice, 1990). Similarly, both parental involvement and parental psychological autonomy were significantly related to self-esteem in a sample of Mexican-American adolescents (Koydemir-Özden and Demir, 2009).

Similarly, greater parental investment, often defined by the amount of time parents spend with their children, along with higher levels of parental involvement, has been tied to children's high self-esteem, parental involvement having the greater influence (Ho, 2003). In contrast, Gecas and Schwalbe (1986) found little evidence that the amount of support parents reported offering their children, such as helping them with tasks, showing affection, and expressing approval for their actions, predicted adolescent self-esteem. Further, they found little evidence that either the amount of time parents spent or sharing activities with their children were related to self-esteem after accounting for parental control. Similarly, Amato and Fowler (2002) found that parental involvement (they termed it parental support) was not significantly related to children's self-esteem.

What is plain, then, from past literature on harsh parental discipline, parental involvement, and parental investment is that there is conflicting evidence regarding the role these parenting practices play in the development of children's self-esteem, making it difficult to formulate specific, directional hypotheses. Therefore, we aim to shed additional light on how parenting practices and attitudes influence offspring's self-esteem by examining the relationship between parenting practices and attitudes and offspring's self-esteem and eventual college completion or current enrollment in a model based on national longitudinal data.

Although we have focused primarily on the ways in which parents aid in developing and facilitating children's self-esteem, other sources also may be influential, such as peers and school groups (Gecas, 2003). Despite this fact, the family-centered focus of this analysis is appropriate for three reasons. First, studies of social class attainment demonstrate that demographic characteristics of the family of origin are associated with offspring's educational and labor market outcomes (Breen and Jonsson, 2005). Second, the parenting styles of middle-class parents (vs. working-class or poor parents) have been linked to children's school success (Lareau, 2003). And third, parents affect children's choices by helping them weigh the expected objective and subjective benefits of higher education relative to the likely costs (Becker, 2003).

Self-Esteem in Social Class Attainment

Self-esteem can shape, in turn, the course of children's lives through its influence on the type of activities and social contexts that children engage in and occupy. These activities and contexts may reinforce and promote children's proficiencies, ideals, and interests (Bandura, 2001). Children with positive views of themselves may emphasize behaviors and activities that foster academic attainment and improved social class outcomes. Conversely, those with lower self-esteem may pursue activities such as watching television or playing video games, which are associated with lower mathematics achievement test scores and grade point averages, and research has demonstrated social class differences in the type of activities pursued (Dumais, 2008; Southgate and Roscigno, 2009).

Two mechanisms are instrumental in this process. Self-affirmation (action undertaken to demonstrate the accuracy of one's self-concept) and self-regulation (efforts aimed at minimizing inconsistencies between one's actions and self-concept) enable individuals to maintain positive views about themselves and their abilities despite perceived inadequacies (Bandura, 2001). Children with high levels of self-esteem, for example, may view academic achievement as a form of personal validation (Gecas, 2003; Gecas and Schwalbe, 1983, 1986). Conversely, children with unfavorable views of themselves may engage in problematic behaviors, such as procrastination, truancy, early sexual activity, and nonmarital childbearing that reduce the likelihood of academic success and socioeconomic attainment (Fomby and Cherlin, 2007; Parker and Benson, 2004). Children with high levels of self-esteem may adopt an active approach to academics, resulting in high levels of achievement (Valentine, DuBois, and Cooper, 2004). Indeed, research has found that highly motivated children find ways to reach desired outcomes, even in the face of significant structural constraints (Chin and Phillips, 2004).

The development of self-esteem may also lead to what has been called mastery motivation (Smith, 2006) or planful competence (Clausen, 1993; Elder, 1974). Clausen (1993) and Elder (1974) demonstrated how self-esteem, developed in childhood and adolescence, had long-term implications for young adults. The ability to contribute to family well-being during times of economic hardship during the transition to adulthood had positive effects on young men's educational pursuits in later years—self-esteem was crucial in these young men's social class outcomes. We therefore expect self-esteem and college completion or current enrollment to be positively related, with individuals with higher levels of self-esteem more likely to have favorable college attendance and/or completion.

Method

Sample

In the present study, data are collected over a 14-year period, drawn from the NSFH. The first wave of the NSFH, conducted in 1987–1988, involved a probability sample of 13,007 adults, with follow-ups in 1992–1994, and 2001–2002 (Sweet, Bumpass, and Call, 1988) and was representative of the noninstitutionalized adult population living in the United States in 1987–1988. Respondents answered a series of questions about a focal child, randomly selected from household children, and these same focal children responded to separate questionnaires in Waves 2 and 3. Of the 3,915 focal children eligible at Wave 2 (when data on self-esteem were collected), 1,952 were interviewed in Wave 3, the point at which college attendance or current enrollment was measured, resulting in approximately 50 percent attrition. Prior work on sample attrition in panel studies of similar size and scope to the NSFH suggests that 50 percent attrition over the course of a study has only a minimal effect on the quality of the estimates (Fitzgerald, Gottschalk, and Moffitt, 1998), and we reached similar conclusions in our own analyses (available upon request) using Heckman's well-known method.3 This study advances previous research in several ways. Because we use data drawn from a longitudinal, national sample, we can draw stronger inferences about temporal processes than previous studies that employed cross-sectional data (Amato and Ochiltree, 1986; Demo, Small, and Savin-Williams, 1987; Hughes and Demo, 1989). We can also draw stronger inferences because we use both parent and child reports for our analyses. This makes our estimates less prone to common-method variance, which results when researchers obtain information for both the independent and dependent variables from the same source (e.g., the parents), which can lead to overestimation of the relationship between variables.

Variables

College Completion or Current Enrollment

Completion of at least a four-year degree or current enrollment in college (1 = yes, 0 = no) served as the dependent variable. Among offspring old enough to have finished a college degree (i.e., 26 or older), only those who had completed a degree were coded as 1. Some focal children, however, were not yet old enough to have finished a degree by Wave 3 (i.e., less than 26) but were enrolled in college. Offspring 25 or younger were coded as 1 if they had either completed a college degree or were currently enrolled in a degree-seeking program. Thus, focal children (the offspring) were coded as 1 if they (a) had completed a college degree or (b) were enrolled in a degree-seeking college program and were less than 25 years old. (The phrase “college completion or current enrollment” refers to this coding.) After assessing alternative ways of measuring the dependent variable and following prior research (Colclough and Beck, 1986), we used college completion or current college enrollment as the measure of future class attainment.4 Operationalizing social class reproduction in this way allowed us to retain those who were not yet old enough to have completed their degree in our sample. Note that those who enrolled, then subsequently dropped out, were coded as 0, since college attendance itself is unlikely to be as influential in determining social class outcomes as obtaining a college degree. Although it would have been ideal to have measures of offspring income and occupational status, the comparatively young age of the sample (18–35) means that indicators of income and occupational status are not likely to be indicative of future socioeconomic status. This is because social class outcomes for young adults, who characterize the focal children in the NSFH sample, are less reliably measured, since many of them choose to pursue higher education, join themilitary, or begin their careers at “entry-level” positions. None of these options reliably indicates future social class status by themselves, since all of these are only initial steps and are often ephemeral and intermediary statuses.

Parenting Practices and Attitudes

Parenting practices and attitudes included four variables: parents’ educational expectations, harsh discipline practices, parental investment, and parental involvement. Previous research has shown that these dimensions of parenting are associated with children's self-esteem, as noted earlier (Sillick and Schutte, 2006). Parental educational expectations were measured by a single item asking parents how much education they believed their child would obtain (1 = not finish high school to 7 = complete a master's or doctorate degree). Harsh discipline was measured by a single item asking how often a parent spanked the child (1 = never or seldom, 2 = sometimes, 3 = very often). Parental investment, a latent construct, was measured by computing the number of hours the more involved parent (i.e., whichever parent had spent the greatest number of hours) spent in an average week in youth sporting, academic (via the local parent-teacher association), and community groups as a participant, advisor, coach, or leader. Responses ranged from 0 to 20 hours (responses greater than 20 were given a value of 20 to deal with skewness). Parental involvement with the child was a latent construct consisting of five variables. The first question asked about how often the adult respondent hugged the child (coded 1 = never to 4 = very often). The next four asked how often the respondent spent time having private talks, playing, engaged in leisure activities, and working on homework together (all range from 1 = never or rarely to 6 = almost every day). All variables were coded in the direction of positive parental behavior and standardized prior to the estimation of the measurement and structural portions of the model.

Parents’ Social Class

Social class is generally thought of as a combination of multiple variables, such as income, education, and occupation. For our purposes, we operationalized social class using two commonly used measures—income and education, both measured at the first wave. Household income was coded in $10,000 groups, with a cap at $110,000 (to deal with skewness). Parental education was measured in continuous years and ranged from 0 to 20.

Self-Esteem

Self-esteem is a composite measure of four constituent variables (Rosenberg, 1965): (a) I feel that I am a person of worth, at least on an equal basis with others, (b) I am able to do things as well as other people, (c) On the whole, I am satisfied with myself, and (d) I can do just about anything I put my mind to. The variables were coded in the direction of positive self-esteemand ranged from 1 strongly disagree to 5 strongly agree.

The factorial validity of parental involvement, parental investment, and self-esteem was determined via confirmatory factor analysis. Table 1 displays the results. The factor loadings were acceptable and all paths between the latent variables and their observed indicators were significant. The model thus supported our conceptualization by indicating that the variables represent parental involvement, parental investment, and self-esteem, respectively.

TABLE 1.

Unstandardized (Standard Errors) and Standardized Loadings for Confirmatory Factor Analysis Based on Three Latent Variables Representing Parent Investment and Involvement and Focal Children's Self-Esteem, NSFH Waves 1 and 2, N = 1,952

Parental Involvement Parental Investment Offspring Self-Esteem
b S.E. β b S.E. β b S.E. β
How often do you hug your child? 1.00 (–) 0.53
How often do you talk with your child? 2.81 (0.17) 0.65
How often do you play with your child? 3.35 (0.20) 0.76
Amount of time spent in leisure activities 2.40 (0.16) 0.58
Amount of time spent on homework help 3.67 (0.24) 0.70
# Hours in PTA 1.00 (–) 0.53
# Hours in community youth groups 0.81 (0.05) 0.59
# Hours in sports/athletic clubs 1.17 (0.05) 0.53
I am a person of worth 1.00 (–) 0.55
I am able to do things as well as others 1.13 (0.06) 0.63
I am satisfied with my life 1.15 (0.06) 0.61
I can do anything I set my mind to 1.05 (0.06) 0.57
Correlations Between Latent Variables
Parental Involvement Parental Investment Self-Esteem
Parental Involvement 1
Parental Investment 0.28 1
Self-esteem 0.05 0.09 1
Open in a new tab

Note: χ2(36) = 104.03, CFI = 0.97, TLI = 0.97, RMSEA = 0.03. Robust standard errors in parentheses. All paths between latent variables and observed indicators are significant at p < 0.001.

Control Variables

Other variables in themodel included child's age (measured at Wave 3), parents’ marital status, parents’ racial-ethnic status (children's race-ethnicity was not available), urban residence, and child gender. Children were between the ages 5 and 18 at Wave 1, 11 and 25 at Wave 2, and 18 and 35 at Wave 3.

Missing Data

Because prior research has it to be the optimal mechanism for dealing with missing data in structural equation models (Enders and Bandalos, 2001), full-information maximum likelihood (FIML) techniques were used to deal with missing data. Although the amount missing was not greater than 10 percent for any variable, patterns of missing data can affect the estimation procedures used to obtain coefficients and standard errors (Acock, 2005). Under the assumption of missing at random (MAR) and to mitigate against possible bias induced by missingness, we used all valid cases on a given x variable in the equation to estimate the coefficients and standard errors representing the relationship between the independent or control (x) variables and the outcome variable (y). However, this implies that each coefficient in the model could potentially be based on a different sample size, meaning that each coefficient would have a different reliability estimate. To account for this, the standard errors were adjusted by the different reliabilities, thereby correcting for potential bias introduced when using FIML. See Myung (2003) for more information.

Analytical Approach

In the present study, data are drawn from a national sample of families and households, with data collected over a 14-year period. To ensure proper temporal ordering, parenting and social class measures (obtained from the parents) are taken from Wave 1 of the NSFH, self-esteem (obtained from the children) from Wave 2, and our measure of college completion or current enrollment (also obtained from the child) is from Wave 3. Although the nonexperimental nature of the data precludes causal inference, the design of the present study addresses several limitations of previous research, making it a more robust test of the relationship between parenting practices, social class, self-esteem, and educational attainment.

After selecting and validating our measures of parenting practices and attitudes and self-esteem (see Table 1), we created a dummy variable from NSFH3 that measured whether the focal child of the NSFH main respondent had completed a college degree if he or she was over age 25. If he or she was under the age of 26 and currently enrolled in college, he or she also was coded as 1, an indication of being “on-track” to complete a college degree. Finally, we employed structural equation modeling to assess the relationship between parenting practices and attitudes, social class, self-esteem, and college completion or current enrollment. We also controlled for children's age, gender, living in an urban area, marital status, and race/ethnicity. Note that we allowed variables tapping social class and parenting practices and attitudes to be correlated throughout the analysis, since social class and parenting practices and attitudes are interrelated (Lareau, 2003).

Findings

Table 2 presents the means, proportions, standard deviations, and ranges for all variables used in the analysis. We employed survey weights to estimate descriptive statistics. Weights were not used in the other models, for reasons described by Winship and Radbill 1994.5 Levels of self-esteem were high, with the average focal child reporting a 13 on the 16-point scale. The average child's parents had at least a high school education, made $43,000 per year (in 1988 dollars), were married, were relatively involved and invested in their child's life, and held high educational expectations for the child.

TABLE 2.

Descriptive Statistics of All Analyzed Variables, NSFH Waves 1-3 (N = 1,952)

M SD Range
College 0.48 0.50 0–1
Self-esteem 13.08 1.57 1–16
(a) I am a person of worth 3.25 0.55 1–4
(b) I am able to do things as well as others 3.24 0.53 1–4
(c) I am satisfied with myself 3.23 0.56 1–4
(d) I can do just about anything I put my mind to 3.36 0.55 1–4
Parental educational expectations 4.98 1.72 1–7
Parental discipline 1.83 0.78 1–4
Parental investment 3.05 5.76 0–60
(a) # Hours in community youth groups 0.54 2.03 0–20
(b) # Hours in sports/athletic clubs 1.36 3.26 0–20
(c) # Hours in PTA 1.23 2.80 0–20
Parental involvement 19.95 4.90 5–28
(a) How often hug child? 3.66 0.63 1–4
(b) How often talk with child? 4.23 1.42 1–6
(c) How often play with child? 4.28 1.45 1–6
(d) How often spend time in leisure activities? 3.60 1.36 1–6
(e) How often spend time helping with homework? 4.16 1.72 1–6
Parental income 3.90 2.48 0–11
Parental education 13.08 2.53 0–20
Age (Wave 3) 25.78 4.44 18–34
Parents' marital status
Married (%) 72 0–1
Not married (%) 22 0–1
Never married (%) 03 0–1
Cohabiting (%) 03 0–1
Race/ethnicity
White (%) 80 0–1
Black (%) 14 0–1
Hispanic (%) 05 0–1
Other race (%) 01 0–1
Urban residence (%) 69 0–1
Child is female (%) 54 0–1
Open in a new tab

Notes: All estimates are weighted. The table reports means for variables treated continuously and percentages for those treated as categorical (marital status, race/ethnicity, urban residence and gender). Standard deviations are not included for dichotomous variables.

Structural equation modeling (SEM) techniques were used to examine the relationship between parenting practices and attitudes, social class, self-esteem, and college completion or current enrollment. We used Mplus because it allows structural equation modeling with both continuous and categorical outcomes, accomplished via a weighted least squares estimation algorithm with mean and variance adjustment. Table 3 displays the results of the SEM based on Figure 1.

TABLE 3.

Regression (Self-Esteem) and Logistic Regression (College) Coefficients from a Structural Equation Model Estimating the Odds of College Completion or Current Enrollment, NSFH, N = 1,952

Child Outcome
Predicting Self-Esteem
 Predicting College Completion or Current Attendance
b Robust S.E. β b Robust S.E. OR
Self-esteem – – – 0.44*** 0.12 1.55
Parental practices and attitudes
Educational expectations 0.04*** 0.01 0.19 0.174*** 0.02 1.19
Harsh discipline –0.02 0.02 –0.05 –0.06 0.04 0.94
Investment 0.01 0.01 0.06 –0.03 0.03 0.97
Involvement 0.05 0.04 0.06 0.02 0.11 1.02
Social class
Parental income 0.013* 0.005 0.10 0.04** 0.01 1.04
Parental education –0.00 0.01 –0.00 0.08*** 0.01 1.08
Offspring is female 0.03 0.02 0.04 0.10± 0.05 1.11
Open in a new tab

Note: b = unstandardized coefficient, B = standardized coefficient, OR = odds ratio. Robust standard errors in parentheses. All models control for age, urban residence, parental marital status, and race/ethnicity. χ2(88) = 319.53, CFI = 0.94, TLI = 0.93, RMSEA = 0.04.

±

p < 0.10

*

p < 0.05

**

p < 0.01

***

p < 0.001.

We employed three commonly used measures provided by the Mplus program to assess model fit, the CFI (comparative fit index), the TLI (Tucker-Lewis index), and the RMSEA (root mean square error of approximation). Each varies between 0 and 1 and constitutes a different way of quantifying the proportion of improvement in fit from the null model to the theoretical or proposed one. Values above 0.9 for the CFI and TLI and below 0.05 are considered indicators of good model fit. The model presented in Table 3 fit the data well (CFI = 0.94, TLI = 0.93, RMSEA = 0.04).

The first equation presents the results obtained when predicting self-esteem with parenting and social class variables. The second equation includes all variables used in the previous model along with self-esteem to predict the likelihood of college completion or current attendance.

Equation (1) shows that only one of the parenting measures, parental educational expectations, was significantly and positively associated with children's self-esteem. Increases in the educational expectations parents held for their children in 1988 were associated with increases in the likelihood that off-spring would attend or have finished college nearly 15 years later. In fact, we found that an increase of one standard deviation in parental educational expectations was associated with one-fifth (β = 0.19) of a standard deviation increase in self-esteem. The remaining variables tapping parental practices and attitudes did not appear to be significantly associated with self-esteem. A similar pattern was found for the social class variables, where parental income was also positively related to self-esteem. The effect of a one standard deviation increase (about $25,000) in income was linked to an increase in self-esteem of one-tenth (β = 0.10) of a standard deviation. We found no evidence of a link between parental education and self-esteem. Thus, a key finding is that parental educational expectations are positively related to their children's self-esteem.

The second model added self-esteem to an equation predicting the odds of college completion or current enrollment. In line with our expectations, focal children with higher levels of self-esteem were especially likely to be currently attending or to have completed a college degree. Each one-unit increase in self-esteem was linked to a 55 percent (e0.44 = 1.55) increase in the odds of obtaining a college degree or of being currently enrolled. Thus, offspring with more self-esteem are expected to have a higher probability of attending college, after accounting for parenting practices and attitudes, social class, and the other variables in the model. Additionally, we again found evidence of the robust role parental educational expectations plays in college attainment. Each one unit increase in parental educational expectations for their children increases the odds that the focal child attends college by about 20 percent. Parents’ social class variables (income and education) also were associated with greater odds of college attendance. We also found suggestive gender differences, with females appearing more likely to attend college or be currently enrolled therein (p = 0.063). Thus, one of the key findings of this article is that self-esteem appears to exert a direct effect on the likelihood that offspring will complete or be currently enrolled in college. This effect persists even after allowing for the independent effect of parental practices and attitudes, social class, marital status, age, urban residence, and race-ethnicity.

The results displayed in Table 3, however, do not provide all relevant pieces of information because the table displays estimated direct effects only. One advantage of structural equation modeling is the ability to examine mediating effects via an analysis of the total, direct, and indirect effects of each variable on the outcome of college completion or current college enrollment. Table 4 displays these effects for each variable included in Table 3. The most notable finding from this table is the way self-esteem mediates the relationships between college completion or current enrollment and two independent variables, parental educational expectations and parental income. For educational expectations, the standardized total effect of 0.33 is composed of a direct effect of 0.30 and an indirect effect of 0.03, indicating that about 9 percent of the impact educational expectations have on college outcomes is mediated through the way parent's expectations for their children's educational futures is translated into higher self-esteem. Similarly, about 18 percent of the total effect of parental income is mediated through self-esteem. Thus, at least some part of the effects of previously established mechanisms for producing social class, such as educational expectations and parental income, is mediated through the conversion of parental expectations and income into higher levels of offspring self-esteem. However, emphasis must be placed on the fact that the mediating role of self-esteem is modest (9 percent and 18 percent, respectively, for educational expectations and income). Thus, another key finding of this article is that self-esteem plays a modest mediating role in the relationship between parental characteristics such as parenting practices and attitudes and socioeconomic status.

TABLE 4.

Standardized Direct, Indirect, and Total Effects of Variables Predicting College Completion or Current Enrollment, NSFH, N = 1,952

Direct Indirect Total
Parental practices and attitudes
Educational expectations 0.30*** 0.03 0.33
Harsh discipline –0.04 –0.01 –0.05
Investment –0.04 0.01 –0.03
Involvement 0.01 0.01 0.02
Social class
Parental income 0.09** 0.02 0.11**
Parental education 0.21*** –0.00 0.21***
Offspring is female 0.05± 0.01 0.06*
Open in a new tab
±

p < 0.10

*

p < 0.05

**

p < 0.01

***

p < 0.001.

Direct and indirect effects may not add up to total effect due to rounding.

Sensitivity Analyses

We conducted several additional analyses to test the sensitivity of the results to alternative model specifications and research decisions. First, because the effect of parenting practices and attitudes may not be uniform across different developmental stages (e.g., the teenage years vs. the early school years), we conducted two additional analyses. In the first, we estimated models separately for those between the ages 5 and 11 and those between ages 12 and 18 at the first wave. No meaningful differences were found. We then examined age interactions with all variables in the model, along with squared and cubic terms for age. These analyses yielded no significant results for self-esteem or college completion or current enrollment. These results supported the findings of a recent meta-analysis that children's age is not significantly associated with effect sizes between self-esteem and academic achievement (Valentine, DuBois, and Cooper, 2004).

Second, because we combined focal children who had a college degree with those still enrolled in college, we employed multinomial logistic regression and examined whether results differed between those still enrolled in college (but under age 26) and those who had already finished their degrees. (Youth without degrees and not currently enrolled served as the reference group.) The results were both substantively and statistically similar for the two college groups. In addition, we examined possible differences between those reporting having earned a bachelor's versus an associate's degree. Again, these analyses revealed no substantive deviations from the results presented above.

The third set of sensitivity analyses examined whether the effect of self-esteem on college completion or current enrollment differed by gender and race-ethnicity. To do so, we held the effect of self-esteem on the college outcome constant across gender and race-ethnic groups, respectively, and compared the model fit statistics to a model wherein these variables were allowed to vary. We found no evidence that the effect of self-esteem varied by gender or race-ethnicity.

The final set of sensitivity analyses pertained to our strategy for missing data. We reestimated all models using three other well-known ways of dealing with missing data: multiple imputation, list-wise deletion, and mean replacement (with an accompanying dichotomous indicator variable). The results were substantively similar to those reported here. In summary, all of the sensitivity analyses (available upon request) provided evidence that the results described here are robust to alternative specifications.

Discussion

The goal of this study was to examine the role of self-esteem in social class attainment (as reflected in college attendance). Importantly, we found evidence that self-esteem is indeed implicated in the likelihood of college completion or current enrollment, a critical indicator of social class reproduction. Children who had higher levels of self-esteem at Wave 2 of the NSFH had a higher likelihood of having either completed or being enrolled in college at Wave 3 a little less than a decade later. Thus, this study provides evidence of the role that one psychological mechanism, self-esteem, plays in social class outcomes.

Our expectations regarding the role of parental practices and attitudes on both self-esteem and college completion or current enrollment were only modestly supported. One of the four variables tapping parental practices and attitudes was related to college outcomes. Parents’ educational expectations for their children exerted a consistent and clear relationship to self-esteem. The more education a parent expects his or her offspring to obtain, the higher the offspring's self-esteem. Parental expectations also mattered for the likelihood that children obtained a college degree if old enough to have finished (over 25 years old) or be currently enrolled in college if not. However, we found no evidence that harsh discipline, parental investment, or parental involvement were related to either offsprings’ self-esteem or the likelihood of children having completed or being currently enrolled in college.

In contrast to parenting practices and attitudes, our expectations regarding the influence of parental social class were largely supported. The analysis provided further evidence of the enduring impact of social class. We found that parental income was associated with higher self-esteem among offspring, and that both parental education and income were linked to greater chances of offspring completing or attending college. Although this observation is far from striking, the importance of underlining the way that parental income and education pave the way for offsprings’ future socioeconomic success can hardly be understated, especially if current trends in income inequality continue, to say nothing of the increasing disparity between the returns to a high school and college education (Krueger, 2006). This study, therefore, supported previous studies demonstrating the importance of parental education and income in children's educational outcomes (Blau and Duncan, 1967; Bowles and Gintis, 1972; Kohn and Schooler, 1983).

The mediating role of self-esteem in the link between parental characteristics and children's college completion or current enrollment was not as strong as projected. On the one hand, self-esteem partially mediates the relationship between two prominent mechanisms for reproducing the American class structure—income and parent's educational expectations. Nearly one-fifth (about 18 percent) of the impact of parental income is mediated through self-esteem, where offspring with higher income parents appeared to benefit not only from increased access to resources but also from their parents’ ability to convert income and other resources into higher levels of self-esteem for their children and ultimately a college education. Because a college degree is a pathway to jobs providing a high income and occupational status, it represents a vital credential for obtaining socioeconomic success in the contemporary economic structure (Choi et al., 2008; Dumais, 2008; Jackson et al., 2007; Pascarella and Terenzini, 2005). In contrast, the mediating (or indirect) effect of parental educational expectations was smaller, with self-esteem mediating approximately 9 percent of the effect of parents’ educational expectations on college completion or current enrollment. Although this effect reached traditional levels of statistical significance, the substantive importance of this mediation may be modest. Yet, given the national, longitudinal panel data we employ and the wealth of factors that have been shown to predict college outcomes, the fact that self-esteem mediates these relationships is noteworthy. To summarize, our expectations regarding the relationship between parental practices and attitudes, parental social class, offspring self-esteem, and offspring's college completion or current enrollment received modest support. Of the four parenting measures examined, one, parental educational expectations, was associated with children's self-esteem and eventual college outcomes, and self-esteem's mediating role in the process was not large. It therefore appears that self-esteem and parental expectations are generally independent of one another with respect to their estimated influence on college completion or current enrollment.

It should be stressed, however, that this article suggests that psychological mechanisms such as self-esteem may play a role in social class reproduction. Children's self-esteem during adolescence was found to be associated with higher levels of college completion or current enrollment during the transition to adulthood. This effect was relatively large, with each increase in self-esteem associated with an increase of 55 percent in the odds of college completion or current enrollment. Further, nothing in our model negates the strong influence of established predictors of social class reproduction. Indeed, the effects of sociostructural forces such as household income and parental educational attainment are pervasive across historical time and individual cohort (Blanden et al., 2004; Erikson and Goldthorpe, 2010; Goldthorpe and Jackson, 2007). Interestingly, self-esteem also mediated a proportion of the influence of parental income on college outcomes.

This study benefited from several improvements over prior work. First, because we use data from both parents and offspring, our study is considerably less subject to artificially inflated correlations due to common-method variance, which can occur when data are obtained from the same source (e.g., parents). Furthermore, our data come from a national longitudinal study, enabling us to ensure proper temporal ordering of the variables, with parenting and social class variables taken from the first wave, self-esteem from the second wave, and our measure of college outcomes from the third. We also employed structural equation modeling techniques and confirmatory factor analysis to deal with measurement error on the latent variables, along with maximum likelihood techniques to deal with missing data.

Although this study gained considerably from the advantages outlined, the current study also has several limitations. First, we examined only college completion and current enrollment.How self-esteem affects subsequent occupational trajectories, earnings, family formation patterns, and other outcomes should be explored, particularly if the effect of the family of origin decreases with age, as Breen and Jonsson (2005) have suggested. One possibility is that the effects of self-esteem become amplified over the life course, as new accomplishments and positive self-esteem reciprocally reinforce one another. The second limitation deals with our measure of self-esteem. A developmental perspective points out that many factors can influence self-esteem during childhood and adolescence, including peers; physical, emotional, biological, and cognitive changes; and school environments (academic, social, and athletic). These factors are crucial and future models should incorporate them to gain a clearer picture of the pathways through which self-esteem influences educational and social class outcomes. They were, however, beyond the focus of the current study. Third, parental influence extends well beyond the four measures of parenting used here. Parental choices and actions, for example, often determine children's neighborhood and school contexts. Parents can also influence the types of peer and friendship networks children construct, along with the nature of extracurricular activities in which children engage. It is likely that we would have observed a greater number of significant associations involving parental behavior if we had been able to cast a broader net. Fourth, the reciprocal nature of the ties between parental educational expectations and children's self-esteem should be underlined. For example, parental expectations could be the consequence, not cause, of children's self-esteem. If parents recognize in the child the confidence and potential for high achievement, they may then express higher expectations for her. Unfortunately, data limitations preclude the inclusion of children's prior self-esteem.

This research also gives rise to new questions regarding self-esteem's role in social class reproduction. Mobility researchers and social class scholars, for example, may be interested in exploring self-esteem as a possible mechanism for explaining secondary effects. Secondary effects are observed when children with similar academic achievement levels but divergent backgrounds experience disparate future educational trajectories (Jackson et al., 2007). If children who hold more positive opinions of themselves are more likely to complete college or be currently enrolled therein, as this study suggests, the encouragement and facilitation of self-esteem may constitute one way to mitigate, though certainly not obviate, sociostructural constraints.

Social class reproduction theory, in seeking to explain the link between parents’ and offsprings’ social statuses, has primarily focused on parents’ financial, human, social, and cultural capital (Erikson and Goldthorpe, 1992). We move the literature forward by illustrating the importance of psychological mechanisms such as self-esteem in this process. Prior literature has hinted that children's beliefs about themselves may play a role in their socioeconomic outcomes (Chin and Phillips, 2004; Devine, 2004). The current study provides support for the notion that children with high self-esteem have more favorable outcomes than those who do not, even after accounting for measures of parents’ social class and parenting practices and attitudes. These findings suggest that attempts to foster high self-esteem may prove fruitful in facilitating children's educational attainment. The growing income disparity between those with high school and college educations highlights the importance of obtaining tertiary qualifications in the current economy (Krueger, 2006). Parents’ ability to utilize available resources, and children's capacity to make proper use of those same resources, is an important step toward endowing children with a sense of both purpose and possibility.

Footnotes

*

The first author will share all data and coding for replication purposes. We wish to thank Tim Heaton, Ralph Brown, Lance Erickson, and Claire Altman for helpful input.

1

We use offspring and children interchangeably throughout the article to denote a relationship to parents rather than a developmental stage.

2

Note that we do not claim that reflected appraisals and the facilitation of an engaging learning environment are the onlymechanisms for developing self-esteem. We do not, however, have sufficient space here to discuss other mechanisms for developing self-esteem. See Rosenberg et al. (1995) for a more in-depth treatment of the topic.

3

Heckman's test involves saving the predicted scores (called lambda) from an equation using attrition as the outcome (1 if the respondent attrited, 0 otherwise). When used as a control variable in an equation predicting the outcome of interest, a significant lambda is evidence of potential attrition bias in the model results

4

Thirty-seven percent of those coded as “1” were enrolled in college at Wave 3.

5

In instances, such as this one, where sampling weights are a function of the independent variables included in the model, unweighted estimates are preferable to weighted ones because they are less biased and more consistent and efficient than weighted ones. See Winship and Radbill (1994) for further information on the topic.

Contributor Information

Spencer L. James, Brigham Young University

Paul R. Amato, The Pennsylvania State University

REFERENCES

  1. Acock AC. Working with Missing Values. Journal of Marriage and Family. 2005;67(4):1012–28. [Google Scholar]
  2. Althusser L. Ideology and Ideological State Apparatuses (Notes Towards an Investigation). The Anthropology of the State: A Reader. 2006 [Google Scholar]
  3. Amato PR, Fowler F. Parenting Practices, Child Adjustment, and Family Diversity. Journal of Marriage and Family. 2002;64(3):703–16. [Google Scholar]
  4. Amato PR, Ochiltree G. Family Resources and the Development of Child Competence. Journal of Marriage and the Family. 1986;48(1):47–56. [Google Scholar]
  5. Bandura A. Social Cognitive Theory: An Agentic Perspective. Annual Review of Psychology. 2001;52(1):1–26. doi: 10.1146/annurev.psych.52.1.1. [DOI] [PubMed] [Google Scholar]
  6. Baumrind D. Current Patterns of Parental Authority. Developmental Psychology. 1971;4(1):1–103. [Google Scholar]
  7. Becker R. Educational Expansion and Persistent Inequalities of Education: Utilizing Subjective Expected Utility Theory to Explain Increasing Participation Rates in Upper Secondary School in the Federal Republic of Germany. European Sociological Review. 2003;19(1):1–24. [Google Scholar]
  8. Blanden J, Goodman A, Gregg P, Machin S. Changes in Intergenerational Mobility in Britain. In: Corak M, editor. Generational Income Mobility in North American and Europe. Cambridge University Press; Cambridge: 2004. pp. 122–46. [Google Scholar]
  9. Blau PM, Duncan OD. The American Occupational Structure. Wiley; New York: 1967. [Google Scholar]
  10. Bouchey HA, Harter S. Reflected Appraisals, Academic Self-Perceptions, and Math/Science Performance During Early Adolescence. Journal of Educational Psychology. 2005;97(4):673–86. [Google Scholar]
  11. Bourdieu P, Passeron JC. Sage; Thousand Oaks, CA: 1990. Reproduction in Education, Society, and Culture. [Google Scholar]
  12. Bowles S, Gintis H. IQ in the US Class Structure. Social Policy. 1972;3–5:65–96. [Google Scholar]
  13. Bradley RH, Corwyn RF. Socioeconomic Status and Child Development. Annual Review of Psychology. 2002;53(1):371–99. doi: 10.1146/annurev.psych.53.100901.135233. [DOI] [PubMed] [Google Scholar]
  14. Breen R, Jonsson JO. Inequality of Opportunity in Comparative Perspective: Recent Research on Educational Attainment and Social Mobility. Annual Review of Sociology. 2005;31:223–43. [Google Scholar]
  15. Burke PJ. Identities and Social Structure: The 2003 Cooley-Mead Award Address. Social Psychology Quarterly. 2004;67(1):5–15. [Google Scholar]
  16. Bynum MK, Durm MW. Children of Divorce and Its Effect on Their Self-Esteem. Psychological Reports. 1996;79(2):447–50. doi: 10.2466/pr0.1996.79.2.447. [DOI] [PubMed] [Google Scholar]
  17. Chin T, Phillips M. Social Reproduction and Child-Rearing Practices: Social Class, Children's Agency, and the Summer Activity Gap. Sociology of Education. 2004;77(3):185–210. [Google Scholar]
  18. Choi KH, Kelly Raley R, Muller C, Riegle-Crumb C. Class Composition: Socioeconomic Characteristics of Coursemates and College Enrollment. Social Science Quarterly. 2008;89(4):846–66. doi: 10.1111/j.1540-6237.2008.00587.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Clausen JA. American Lives: Looking Back at the Children of the Great Depression. Free Press; New York: 1993. [Google Scholar]
  20. Colclough G, Beck EM. The American Educational Structure and the Reproduction of Social Class. Sociological Inquiry. 1986;56(4):456–73. [Google Scholar]
  21. Cooley CH. Human Nature and the Social Order. Charles Scribner's Sons; New York: 1902. [Google Scholar]
  22. Darolia R, Wydick B. The Economics of Parenting, Self-Esteem and Academic Performance: Theory and a Test. Economica. 2011;78(310):215–39. [Google Scholar]
  23. Demo D, Small S, Savin-Williams R. Family Relations and the Self-Esteem of Adolescents and Their Parents. Journal of Marriage and the Family. 1987;49(4):705–15. [Google Scholar]
  24. Devine F. Class Counts: Comparative Studies in Class Analysis. British Journal of Sociology. 1997;48(4):713–14. [Google Scholar]
  25. Devine F. Class Practices: How Parents Help Their Children Get Good Jobs. Cambridge University Press; Cambridge, UK: 2004. [Google Scholar]
  26. Dumais SA. Adolescents’ Time Use and Academic Achievement: A Test of the Reproduction and Mobility Models. Social Science Quarterly. 2008;89(4):867–86. [Google Scholar]
  27. Elder GH. Children of the Great Depression. University of Chicago Press; Chicago: 1974. [Google Scholar]
  28. Enders CK, Bandalos DL. The Relative Performance of Full Information Maximum Likelihood Estimation forMissing Data in Structural EquationModels. Structural Equation Modeling: A Multidisciplinary Journal. 2001;8(3):430–57. [Google Scholar]
  29. Erikson R, Goldthorpe JH. The Constant Flux: A Study of Class Mobility in Industrial Societies. Clarendon Press; Oxford: 1992. [Google Scholar]
  30. Erikson R. Has SocialMobility in Britain Decreased? Reconciling Divergent Findings on Income and Class Mobility. British Journal of Sociology. 2010;61(2):211–30. doi: 10.1111/j.1468-4446.2010.01310.x. [DOI] [PubMed] [Google Scholar]
  31. Felson RB, Zielinski MA. Children's Self-Esteem and Parental Support. Journal of Marriage and the Family. 1989;51(3):727–35. [Google Scholar]
  32. Fischer CS, Hout M. Century of Difference: How America Changed in the Last One Hundred Years. Russell Sage Foundation Publications; New York: 2006. [Google Scholar]
  33. Fitzgerald J, Gottschalk P, Moffitt R. An Analysis of the Impact of Sample Attrition on the Second Generation of Respondents in the Michigan Panel Study of Income Dynamics. Journal of Human Resources. 1998;33(2):300–44. [Google Scholar]
  34. Fomby P, Cherlin A. Family Instability and Child Well-Being. American Sociological Review. 2007;72(2):181–204. doi: 10.1177/000312240707200203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Furnham A, Cheng H. Perceived Parental Behaviour, Self-Esteem and Happiness. Social Psychiatry and Psychiatric Epidemiology. 2000;35(10):463–70. doi: 10.1007/s001270050265. [DOI] [PubMed] [Google Scholar]
  36. Gecas V. In: Self-Agency and the Life Course. Handbook of the Life Course. Mortimer JT, Shanahan MJ, editors. Kluwer Academic/Plenum; New York: 2003. pp. 369–90. [Google Scholar]
  37. Gecas V, Schwalbe ML. Beyond the Looking-Glass Self: Social Structure and Efficacy-Based Self-Esteem. Social Psychology Quarterly. 1983;46(2):77–88. [PubMed] [Google Scholar]
  38. Gecas V, Schwalbe ML. Parental Behavior and Adolescent Self-Esteem. Journal of Marriage and the Family. 1986;48(1):37–46. [Google Scholar]
  39. Goffman E. The Presentation of Self in Everyday Life. Doubleday; New York: 1959. [Google Scholar]
  40. Goldthorpe JH. Class Analysis and the Reorientation of Class Theory: The Case of Persisting Differentials in Educational Attainment. British Journal of Sociology. 1996;47(3):481–505. doi: 10.1111/j.1468-4446.2009.01248.x. [DOI] [PubMed] [Google Scholar]
  41. Goldthorpe JH, Jackson M. Intergenerational ClassMobility in Contemporary Britain: Political Concerns and Empirical Findings. British Journal of Sociology. 2007;58(4):525–46. doi: 10.1111/j.1468-4446.2007.00165.x. [DOI] [PubMed] [Google Scholar]
  42. Haque A. Relationship Between Perceived Maternal Acceptance-Rejection and Self-Esteem Among Young Adults in Nigeria. Journal of African Psychology. 1988;28:787–93. [Google Scholar]
  43. Ho E. Students’ Self-Esteem in an Asian Educational System: Contribution of Parental Involvement and Parental Investment. School Community Journal. 2003;13(1):65–84. [Google Scholar]
  44. Hughes M, Demo D. Self-Perceptions of Black Americans: Self-Esteem and Personal Efficacy. American Journal of Sociology. 1989;95(1):132–59. [Google Scholar]
  45. Jackson M, Erikson R, Goldthorpe JH, Yaish M. Primary and Secondary Effects in Class Differentials in Educational Attainment. Acta Sociologica. 2007;50(3):211–29. [Google Scholar]
  46. Kohn ML, Schooler C. Work and Personality: An Inquiry into the Impact of Social Stratification. Ablex; Norwood, NJ: 1983. [Google Scholar]
  47. Koydemir-Özden S, Demir A. The Relationship Between Perceived Parental Attitudes and Shyness Among Turkish Youth: Fear of Negative Evaluation and Self-Esteem as Mediators. Current Psychology. 2009;28(3):169–80. [Google Scholar]
  48. Krueger AB. Inequality: Too Much of a Good Thing. In: Grusky DB, Szel'enyi S, editors. The Inequality Reader: Contemporary and Foundational Readings in Race, Class, and Gender. Westview; Boulder, CO: 2006. p. 23. [Google Scholar]
  49. Lareau A. Home Advantage: Social Class and Parental Intervention in Elementary Education. Rowman &amp; Littlefield; Lanham, MD: 2000. [Google Scholar]
  50. Lareau A. Unequal Childhoods: Class, Race, and Family Life. University of California Press; Berkeley, CA: 2003. [Google Scholar]
  51. Mayhew KP, Lempers JD. The Relation Among Financial Strain, Parenting, Parent Self-Esteem, and Adolescent Self-Esteem. Journal of Early Adolescence. 1998;18(2):145–72. [Google Scholar]
  52. Myung J. Tutorial on Maximum Likelihood Estimation. Journal of Mathematical Psychology. 2003;47:90–100. [Google Scholar]
  53. Nakhaie RM. Social Origins and Educational Attainment in Canada: 1985 and 1994. Review of Radical Political Economics. 2000;32:577–609. [Google Scholar]
  54. Parker JS, Benson J. Parent-Adolescent Relations and Adolescent Functioning: Self-Esteem, Substance Abuse, and Delinquency. Adolescence. 2004;39(155):519–30. [PubMed] [Google Scholar]
  55. Pascarella ET, Terenzini PT. How College Affects Students: A Third Decade of Research. Jossey-Bass; Indianapolis, IN: 2005. [Google Scholar]
  56. Rice KG. Attachment in Adolescence: A Narrative and Meta-Analytic Review. Journal of Youth and Adolescence. 1990;19(5):511–38. doi: 10.1007/BF01537478. [DOI] [PubMed] [Google Scholar]
  57. Riegle-Crumb C, Callahan RM. Exploring the Academic Benefits of Friendship Ties for Latino Boys and Girls. Social Science Quarterly. 2009;90(3):611–31. doi: 10.1111/j.1540-6237.2009.00634.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Rosenberg M. Society and the Adolescent Self-Image. Princeton University Press; Princeton, NJ: 1965. [Google Scholar]
  59. Rosenberg M, Schooler C, Schoenbach C, Rosenberg F. Global Self-Esteem and Specific Self-Esteem: Different Concepts, Different Outcomes. American Sociological Review. 1995;60(1):141–56. [Google Scholar]
  60. Sharkey P. The Intergenerational Transmission of Context. American Journal of Sociology. 2008;113(4):931–69. [Google Scholar]
  61. Sillick TJ, Schutte NS. Emotional Intelligence and Self-Esteem Mediate Between Perceived Early Parental Love and Adult Happiness. E-Journal of Applied Psychology. 2006;2(2) [Google Scholar]
  62. Smith MB. Social Psychology and Human Values. Aldine; Piscataway, NJ: 2006. [Google Scholar]
  63. Southgate DE, Roscigno VJ. The Impact of Music on Childhood and Adolescent Achievement. Social Science Quarterly. 2009;90(1):4–21. [Google Scholar]
  64. Sweet J, Bumpass L, Call V. The Design and Content of the National Survey of Families and Households. Center for Demography and Ecology, University of Wisconsin-Madison; 1988. [Google Scholar]
  65. Twenge JM, Campbell WK. Self-Esteem and Socioeconomic Status: A Meta-Analytic Review. Personality and Social Psychology Review. 2002;6(1):59–71. [Google Scholar]
  66. Valentine JC, DuBois DL, Cooper H. The Relation Between Self-Beliefs and Academic Achievement: A Meta-Analytic Review. Educational Psychologist. 2004;39(2):111–33. [Google Scholar]
  67. Willis PE. Learning to Labor: How Working Class Kids Get Working Class Jobs. Columbia University Press; New York: 1981. [Google Scholar]
  68. Winship C, Radbill L. Sampling Weights and Regression Analysis. Sociological Methods and Research. 1994;23(2):230–57. [Google Scholar]

RESOURCES