Should Mom go back to school? Post-natal educational attainment
and parenting practices
Thurston Domina
a,⇑, Josipa Roksa
b aDepartment of Education, University of California, Irvine, 2062 Education, Irvine, CA 92697-5500, United States
b
Department of Sociology, University of Virginia, 2015 Ivy Road, 3rd Floor, P.O. Box 400766, Charlottesville, VA 22904, United States
a r t i c l e
i n f o
Article history:
Received 3 June 2011 Revised 26 October 2011 Accepted 15 December 2011 Available online 24 December 2011
Keywords: Educational attainment Parenting Cultural mobility Cultural reproduction
a b s t r a c t
Although the relationship between educational attainment and parenting practices is well documented, it is typically examined at only one point in time. What happens if mothers acquire more education after the birth of their children: do they alter their parenting prac-tices? Panel data models based on longitudinal data from ECLS-K indicate that changes in mother’s educational attainment are positively associated with increases in parental school involvement, having books in the home, and participating in non-academic family activi-ties, but not with attitudes toward discipline. Although post-natal maternal education does not change all aspects of parenting, our findings are broadly consistent with the theory of cultural mobility and provide insights into the extent of socio-cultural mobility in contem-porary American society.
Ó2011 Elsevier Inc. All rights reserved.
1. Introduction
Social class is related to many aspects of individuals’ lives, including parenting practices. From the classic works byKohn (1963, 1977)andBernstein (1971)to more contemporary research byLareau (2002, 2003)and others, scholars have docu-mented differences in parenting practices across social classes (Farkas, 2003; Hoff et al., 2002). This research typically views social class and parenting as static: studies in this tradition often examine the relationship between social class and parent-ing practices at a specific point in time. The implicit assumption in these studies is that parents’ social class, and thus their approach to parenting, does not change substantially over the life course.
We re-evaluate that assumption. The modal sequence for young adults transitioning to adulthood is to postpone child-bearing until after the completion of formal education, but the proportion of young adults following that modal sequence has declined as life course trajectories have become increasingly variable (Shanahan, 2000). One particularly notable change has involved the timing of child-bearing relative to formal education. Today, more than a quarter of undergraduates at Amer-ican institutions of higher education have dependents (NCES, 2002) and the share of American undergraduates who are sin-gle parents has more than doubled over the past 20 years (Goldrick-Rab and Sorensen, 2010). Given the intertwined sequences of schooling and parenting, do changes in mothers’ educational attainment alter their parenting practices?
Previous research demonstrates that highly educated mothers and fathers parent differently from less-educated parents (c.f.Attewell and Lavin, 2007; Carneiro et al., 2007; Hill and Stafford, 1980; Oreopoulos and Salvanes, 2009; Sayer et al., 2004). These findings imply that exposure to education changes parenting behaviors and attitudes. If parenting behaviors and attitudes are a manifestation of cultural capital, these findings suggest that educational access leads to socio-cultural mobility over the life course. However, since the vast majority of previous research is based on cross-sectional data, it is dif-ficult to separate the sources of educational attainment from its consequences.
0049-089X/$ - see front matterÓ2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ssresearch.2011.12.002
⇑Corresponding author.
E-mail address:[email protected](T. Domina).
Contents lists available atSciVerse ScienceDirect
Social Science Research
In this paper, we address this difficulty by tracing changes in mothers’ educational attainment and parenting behaviors during their children’s early elementary school years using data from the Early Childhood Longitudinal Study’s Kindergarten cohort. Since a substantial proportion of mothers in the sample continue to pursue education after their children are born, these data provide a unique opportunity to observe the extent to which parenting practices respond to educational attain-ment. After presenting cross-sectional models similar to those in the previous literature, we construct panel data models of the effects of changes in maternal education on change in parenting behavior. Unlike traditional cross-sectional models, our longitudinal models control for all characteristics of mothers and children that do not change over time. As a result, they generate relatively unbiased estimates of the effect of education on the parenting behavior of mothers who pursue post-natal education.
Our analyses suggest that post-natal maternal education influences several parenting practices which previous research has suggested are associated with children’s academic success. In particular, we find strong evidence to suggest that post-natal maternal educational attainment increases mothers’ involvement in their children’s elementary schools, number of books in the home, and the frequency with which mothers and children engage in non-academic family activities. However, changes in mother’s educational attainment are not associated with changes in attitudes toward discipline. These results provide new insights into the role of education in facilitating socio-cultural mobility.
1.1. Social class, education, and parenting
A long line of sociological research demonstrates a relationship between social class and parenting. While research in this tradition spans more than a century, employs diverse methods, and focuses on several different aspects of parent/child rela-tionships, it has consistently reported large and robust differences between upper-, middle-, and working-class parenting practices (c.f.Lynd and Lynd, 1956; Gans, 1962; Bronfenbrenner, 1958; Kohn, 1963; Lareau, 2000, 2003; Hart and Risely, 1995; Bradley et al., 2003; Bodovski and Farkas, 2008). Lareau’s ethnographic study of parenting practices of middle class and working class/poor families, in particular, has garnered much attention. Instead of examining specific parenting prac-tices individually,Lareau (2002, 2003)describes class-based differences in parenting styles. She proposes that middle-class parents engage in a ‘‘concerted cultivation’’ style of parenting, which is exemplified by a deliberate cultivation of children’s skills and talents, while working class parents engage in an ‘‘accomplishment of natural growth’’ style of parenting, which allows children to grow up in a more spontaneous manner. These differences in parenting styles help to reproduce class inequalities across generations. Lareau argues that concerted cultivation fosters a set of attitudes and behaviors among mid-dle class children that are rewarded by institutions such as schools and thus facilitate their educational success.
Although previous studies rely on varying definitions of social class, a subset of this research tradition focuses specifically on the association between parental education and parenting practices, which is particularly relevant for our study. Time-use data from the United States and elsewhere indicate that highly educated mothers and fathers spend more time on aver-age with their children than do less educated parents (Hill and Stafford, 1980; Sayer et al., 2004). Furthermore, there are important qualitative differences between the parenting practices of highly educated and less highly educated parents. Highly educated parents spend more time reading to their children than do less educated parents (Hill and Stafford, 1980; Huston and Aronson, 2005), and the children of highly educated parents spend more time reading to themselves and studying than do children of less educated parents (Bianchi and Robinson, 1997). Similarly, survey data suggest that highly educated parents have different approaches to parenting than do less highly educated parents.Oreopoulos and Salv-anes (2009), for example, demonstrate that college-educated adults are less likely to favor spanking and other forms of cor-poral discipline than less highly educated adults.
Taken together, these findings indicate that parental education is closely related to parenting behavior. However, the nat-ure of that relationship is less clear since parental education is associated with a host of factors that are likely also associated with parenting (including the practices and attitudes that parents learned from their own parents, family income and other economic and cultural resources, parental and child intelligence, motivation, and expectations.) Does formal education change the way individuals parent? Or is the relationship spurious, such that the same personal characteristics that lead individuals to pursue education also influence parenting behaviors?
Previous research on cultural stratification provides two different frameworks for considering these questions. Theories in the cultural reproduction tradition hold that schools primarily reproduce social inequalities, rather than create opportunities for social mobility (c.f.Bowles and Gintis, 1976). Bourdieu’s theory of cultural capital is perhaps the most fully articulated of these theories. Bourdieu argues that differences in family life and values lead children to develop ‘‘linguistic and cultural competencies’’ and ‘‘familiarity with culture’’ that are closely associated with their family’s class background (Bourdieu, 1973, p. 494; see alsoBourdieu and Passeron, 1977). Since educators and other socio-cultural gatekeepers typically come from relatively advantaged backgrounds, schools tend to recognize and reward the cultural inheritances that upper-class children share, rather than those shared by lower-class students (Lamont and Lareau, 1988; Lareau and Weininger, 2003). Upper-class cultural capital, therefore, facilitates school success for children from upper class families. In the process, it helps students from advantaged backgrounds acquire high-status careers and eventually pass their advantages onto their own children.
The cultural mobility model, on the other hand, emphasizes the possibility for individuals to acquire cultural capital out-side of their families and use that newly acquired cultural capital as a vehicle for upward mobility. Drawing onWeber’s (1968)work on status cultures,DiMaggio (1982)argues that an individual can acquire familiarity with the dominant culture
in settings other than the family and be rewarded for this familiarity. While this view acknowledges that family upbringing is one avenue for acquiring cultural capital, it holds that some individuals may change their social status by acquiring cultural capital during the life course (see alsoAschaffenburg and Maas, 1997), particularly in the course of formal schooling (Reay, 2004). Research in this tradition also suggests that individuals from less advantaged backgrounds may receive greater re-wards for the possession of cultural capital (De Graaf et al., 2000; DiMaggio, 1982; Dumais, 2006).
Following each of these theoretical frameworks, we can conceptualize parenting practices as a manifestation of parents’ cultural capital and a source of children’s cultural capital.1The two frameworks, however, differ in the way they understand the relationship between cultural capital and schooling. The cultural reproduction argument places an emphasis on family as a site of cultural capital production. Although in some of the later writings, Bourdieu acknowledged the possibility of individuals acquiring cultural capital through schooling, that possibility is limited because less advantaged children lack an appropriate habitus and thus are not likely to learn as quickly or acquire the ‘natural familiarly’ with schooling practices (Bourdieu, 1973). Education in this framework largely reproduces existing class differences that originate in the family. Acquiring more education would thus not necessarily lead to more cultural capital, or in the specific context we are examining, more schooling would not necessarily alter individual’s parenting practices. The cultural mobility argument is much more open to education acting as a mechanism of mobility, i.e., as providing less advantaged children with the necessary skills and attitudes as well rewarding them for acquisition of cultural capital. Cultural mobility scholars would thus be more inclined to expect that acquir-ing more education would play a role in alteracquir-ing individual cultural capital, and by extension shape their parentacquir-ing practices.
1.2. Effects of education on parenting
In this paper, therefore, we examine the extent to which formal education alters the way individuals parent. Our analyses pay close attention to the ways parenting practices change for mothers who acquire new levels of educational attainment while their children are in elementary school in order to estimate the effect of educational attainment on parenting. This approach has two distinct advantages: First, by comparing mothers who complete further education with themselves (rather than comparing highly educated mothers with less highly educated mothers), our analyses produce estimates of the effects of education on parenting that are not biased by the relationship between time-invariable maternal and child characteristics and maternal educational attainment.2Second, rather than estimating the average effects of maternal education across the
population, our analyses provide particular insight on mothers who are most likely to pursue formal education after childbirth. Several studies have considered the causal effects of educational attainment on a host of market outcomes (seeWolfe and Haveman (2003)for a recent review) and a growing literature investigates the effects of educational attainment on non-mar-ket outcomes such as civic engagement (c.f.Dee, 2004). However, relatively few existing studies explicitly model the causal effect of education on parenting, with some notable exceptions, such asAttewell and Lavin (2007). Using a propensity score matching approach, Attewell and Lavin demonstrate that maternal educational attainment has a positive effect on the cog-nitive stimulation and emotional support she provides for her children, the extent to which she is involved in her children’s school, and her educational attainment. Attewell and Lavin report, for example, that mothers who earn bachelor’s degrees score approximately one-fourth of a standard deviation higher on the HOME cognitive stimulation scale than they would have had they attained no further education after high school.Carneiro et al. (2007)reach similar conclusions using variation in the cost of education as an instrumental variable to estimate the effect of maternal educational attainment on parenting. Attewell and Lavin point to two primary mechanisms that may explain the relationship between maternal educational attainment and parenting practices. First, they hypothesize that educational attainment may influence parenting by increas-ing mothers’ knowledge. If formal education increases maternal appreciation of history, music, art, or other sorts of high cul-ture, educational attainment may encourage mothers to share these interests with their children, boosting their cultural capital. More broadly, formal schooling may also influence parenting behaviors by exposing mothers to new information about child development and education. Second, Attewell and Lavin hypothesize that education may influence parenting behaviors by exposing mothers to the behaviors and attitudes of higher status peers. Ethnographic studies of working class students in higher education clearly indicate that many students view formal education as a site for social mobility (c.f. Leh-mann, 2009; Reay et al., 2009). As the following quotation from one working class student enrolled in a research university suggests (Lehmann, 2009, p. 642), students may study their peers for clues about how to raise middle class children: ‘‘I’m trying to move up in society, tobecomethese people that have just lived with privilege all their lives and it’s normal for them. For me, it’s not normal, but it will become normal for my children.’’3
1Although early educational research on cultural capital tended to focus on high-brow cultural practices, recent research has broadened the definition of cultural capital to include a range of parenting practices, particularly those that may be beneficial in interaction with dominant institutions, such as schools (for a recent review, seeLareau and Weininger, 2003).
2Attewell and Lavin’s propensity score matched estimates of the effects of maternal education on parenting are unbiased if their propensity score matching model controls for each of the covariates that jointly predict maternal education and parenting. Similarly, Carneiro et al.’s estimates of the effect of maternal education on parenting assume that their instrumental variable (local education costs) is only associated with parenting practices via education. While both of these assumptions are plausible; neither is empirically verifiable. Our longitudinal estimates of the effect of maternal education on parenting only assume that the factors that jointly predict maternal education and parenting are time-invarying.
3
In addition, Attewell and Lavin hypothesize that education may influence parenting via marriage. They demonstrate that highly educated women are more likely to be in stable marriages than less highly educated women. Furthermore, they find that the spouses of highly educated women tend be better educated and tend to earn more than the spouses of less highly educated women.
Following these insights, we investigate the effects of post-natal educational attainment on a range of parenting practices, including involvement in children’s elementary schools, the number of children’s books in the home, attitudes toward dis-cipline, and parent–child interaction. Two of these parenting practices – parental school involvement and books in the home – may most directly influence children’s interaction with school as an institution, and are thus particularly relevant to the process of intergenerational inequality that Bourdieu describes. Moreover, all four parenting practices have commonly been considered in the research on the transmission of class advantage across generations. FollowingLareau (2003), a number of recent studies have aimed to capture different dimensions of the concerned cultivation style of parenting, including paren-tal-child interaction, parental involvement in schools, and child’s participation in extracurricular activities. Some studies have combined indicators of each of these aspects of parenting into composite measures (e.g.,Bodovski and Farkas, 2008; Cheadle, 2008; Roksa and Potter, 2011) while others have focused on a specific dimension, such as parental involvement in schools (e.g.,Parcel and Dufur, 2001; Crosnoe, 2004; Kelly, 2004; Domina, 2005). Similarly, the number of books in the home is often used as an indicator of parental efforts to enrich children’s educational experiences and produce a cognitively stimulating home environment (e.g.,Bodovski and Farkas, 2008; Cheadle, 2008, see alsoTeachman, 1987). Discipline – although less commonly considered in recent sociological research – constitutes an important aspect of parenting discussed byLareau (2003)and highlighted in classic studies of class inequality (e.g.,Kohn, 1977). Moreover, social psychologists, who have been more inclined to examine disciplinary practices, demonstrate the relevance of this aspect of parenting for chil-dren’s outcomes (e.g.,Dornbusch et al., 1987; Landry et al., 2000; Steinberg et al., 1989). While none of the quantitative mea-sures can reflect the totality of parenting practices, and while none of the datasets include all of the relevant aspects of parenting, previous studies have aimed to capture different dimensions of family contexts to the extent possible. We follow this practice and explore a range of different practices available in the dataset and shown by the previous research to be dif-ferentially distributed by social class and related to children’s educational outcomes.
If Attewell and Lavin’s hypothesis that education influences parenting via both knowledge acquisition and imitation is correct, we would expect educational exposure to lead mothers to be more involved in their children’s schools, to have more children’s books in home, to have a relatively communicative disciplinary style, and to interact more frequently with their children. However, the magnitude of this effect is far from clear. Women who pursue further education after childbirth tend to come from less advantaged backgrounds than traditional students. Disproportionately black, Hispanic, and Native Amer-ican, these women are more likely to mix schooling with work and are considerably less likely to graduate on time ( Goldrick-Rab and Sorensen, 2010). These demographic characteristics could either dilute or magnify the effects of education on par-enting. Furthermore,Elman and O’Rand (2004)suggest that delayed educational attainment conveys smaller positive labor market effects than on-time educational attainment. It is possible that the effects of education on parenting could be sim-ilarly contingent on the timing of education. If mothers of school-aged children already have firmly established parenting attitudes and routines, acquiring education after children enter schooling could be only weakly related to parenting prac-tices. On the other hand, these mothers may be particularly inclined to note the lessons implicit in their schooling for their own children and change their parenting behaviors accordingly.
2. Data and methods
We estimate the effects of changes in maternal educational attainment on changes in parenting attitudes and practices using longitudinal data from the Early Childhood Longitudinal Study – 1998 Kindergarten Cohort (ECLS-K). Sponsored by the US Department of Education’s National Center for Education Statistics and conducted by Westat, the ECLS-K follows a nation-ally representative sample of children from kindergarten through elementary school. The study drew a nationnation-ally-represen- nationally-represen-tative stratified sample of 21,260 kindergarteners from 1277 schools in 100 US counties in the fall of 1998, following up with them in spring 1999, spring 2000 (when most respondents were in the first grade), spring 2002 (when most were in the third grade), and spring 2004 (when most were in the fifth grade). The survey also collected data in the spring 2007 (when most were in eighth grade), but since the parenting measures reported in this last round do not match the earlier measures, we do not utilize this wave of data.
Our analyses rely on data from ECLS-K respondents who participated in the baseline student survey and assessments as well as parent surveys. Since mothers are typically more involved in child care activities than fathers (Monna and Gauthier, 2008), our analyses focus exclusively on maternal education and parenting. More than 90% of the children in the survey lived with their biological mother throughout the study. We transformed the ECLS-K data (which is organized around the child as the unit of analysis) to construct a sample of ECLS-K mothers, drawing data exclusively from students who lived with their biological mother in each of the survey rounds.4While our analyses include mothers who attrited from the sample or are
otherwise missing one or more survey rounds, it excludes mothers who did not provide educational attainment information in any round in which they participated in the survey. In addition, our sample excludes approximately 500 mothers who re-ported lower levels of educational attainment in later survey rounds than in the baseline survey.5The resulting analytic sample
4
In addition to dropping ECLS-K respondents who did not live with their biological mother in at least one survey round, our analysis drops 219 respondents who came from the same household (most of whom were siblings).
5
Some educational downward mobility is possible. For example, one could earn a GED after enrolling in a community college or vocational credential program. However, we suspect that measurement error is responsible for most of the discrepancies. Respondents could have either misreported their baseline attainment or misreported their attainment in a later round.
includes approximately 14,600 mothers. While the unit of analysis is the mother, rather than the child, our analytic sample is very similar demographically to the original ECLS-K sample.
2.1. Models
We begin by presenting preliminary analyses examining the relationship between maternal education and parenting practices at the beginning of the ECLS-K study period, when the children of the sampled mothers were beginning kindergar-ten. These preliminary analyses are designed to replicate the cross-sectional multivariate estimates of the association be-tween maternal education and parenting available inSayer et al. (2004)andOreopoulos and Salvanes (2009). To do so, we estimate cross-sectional OLS models with the following basic form:
Yik¼b1ikþb2ikðMother’s EdÞ þb3ikðcontrolsÞ þ
e
The dependent variable in these models,Yik, is one of four measures of parenting practices and attitudes, and the key
inde-pendent variables are a matrix of dummy variables indicating the highest level of education mothers attained by the time their child reached Kindergarten (Mother’s Ed). The next section of the paper provides a more detailed description of the dependent variables, maternal education variables, and all of the controls.
In addition, we take advantage of the ECLS-K’s longitudinal design to estimate repeated-observation fixed effects models of the effects of maternal education on parenting behaviors. In order to construct these models, we have reshaped the ECLS-K data into a long, or panel format, in which the unit of analysis is the respondent/observation. In this reshaped panel file, each ECLS-K mother has an observation from the study’s Kindergarten, First Grade, Third Grade, and Fifth Grad survey rounds. The fixed-effects panel models make it possible to isolate the causal effects of maternal education on parenting by focusing attention on mothers whose educational attainment changed between observations. In essence, we are estimating the
waychangesin maternal education change mothers’ parenting practices, net of the time-invariant maternal and child
char-acteristics and time-varying controls. These fixed-effects panel models take the following form: Yit¼b1itþb2itðMother’sEdÞ þb3itðtime-varying controlsÞ þhiþ
e
itwhereYitis a time-varying measure of maternal parenting practices measured at Kindergarten, First Grade, Third Grade, and
Fifth Grade;b2itis a set of time-varying dummy variables representing mother’s education;b3itis a vector of time-varying
controls, including dummy variables for survey wave; andhiis a respondent fixed-effect term. This respondent fixed effect is
equivalent to a dummy variable for each of the respondents in the analysis; it controls for all observable and unobservable time-invariant characteristics of mothers and their children (Schneider et al., 2007). While the cross-sectional models are built around the comparison of parenting practices between mothers of differing levels of educational attainment, the inclu-sion of the fixed effects term reorients the analyses around changes within individuals’ parenting practices.
We also estimate lagged random effects models of the relationship between maternal education and parenting practices. In the place of the maternal fixed effects term, these models include the same time-invarying controls utilized in the cross-sectional OLS model, as well as a lagged version of the dependent variable. More precisely, these models take the following form:
Yit¼b1itþb2itðMother’s EdÞ þb3itðtime-varying controlsÞ þb4itðtime-invarying controlsÞ þb5itðYit1Þ þ
e
itThese random effects models are more precise than the fixed effects models. However, if the time-invarying controls fail to capture all of the relevant characteristics of mothers and their children, these models may be biased. Traditionally, analysts use the Hausman test to choose between fixed effects and lagged random effects models. Since our models fail the asymp-totic assumptions associated with the Hausman test, we calculate more robust Sargan–Hansen statistics using Schaffer and Stillman’s program ‘‘xtoverid’’ Stata program (Schaffer and Stillman, 2010). This program tests the assumption that random effects estimators are uncorrelated with unmeasured maternal characteristics by re-estimating the random-effects model with additional variables derived from the original regressors (Wooldridge, 2002). The Sargan–Hansen test rejects the ran-dom effects model, which is thus not reported. However, we note that the ranran-dom effects models return results that are somewhat more conservative than the fixed effects results that we report. These models are available upon request.
AsTable 1reveals, a substantial number of mothers at all educational levels earned new degrees between each of the study’s survey rounds. These women are the focus of our longitudinal fixed-effects models. Between any given survey round, at least 5% of mothers who lacked high school diplomas earned a GED or equivalent. Similarly, approximately 5% of mothers with high school diplomas pursued some form of higher education between each of the survey rounds. While fewer mothers at higher levels of educational attainment pursued further schooling during the ECLS-K survey period, at least 1.6% of some-college mothers earned a BA between any given survey round and approximately 1% of BA mothers earned a Master’s degree or higher.Table 1indicates that the rate of new maternal educational attainment is particularly pronounced between the 1st and 3rd grade of the ECLS-K survey waves.
2.2. Variables
The central variable in our analyses is maternal education. At each survey round, the ECLS-K parental interview collects educational attainment data for each of the adults living in a child’s household. We recode the categorical maternal
educational attainment variables into a series of dummy variables for mothers who have not completed a high school diplo-ma; mothers whose highest degree is a high school diploma or equivalent; mothers who have an AA degree or other college short of a BA; mothers who have earned a BA; and mothers who have earned an MA, PhD or other graduate-level professional degree. Cross-sectional OLS models include mother’s educational attainment as of the focal child’s Kindergarten year, while the fixed-effects and random effects models include mother’s educational attainment at each of the survey rounds. Educa-tionalattainmentis an imperfect proxy for educationalexposure, but the only one available in the dataset. Since the ECLS-K data do not report maternal years of education or educational enrollment, our operationalization of maternal education does not allow us to measure the ways parenting changes while mothers are enrolled in school, nor does it allow us to account for mothers who attend postsecondary education but do not earn a degree during the study period.
We consider the relationship between maternal education and four measures of parenting practices and attitudes. As dis-cussed in the literature review, these measures have been considered in previous research and regarded as relevant in the transmission of class advantage across generations. By including a range of different measures, we are following recent re-search in aiming to capture different dimensions of family contexts. Each of these measures is derived from the ECLS-K parental interview which is typically administered to the child’s mother,6and each is standardized to have a mean of zero
and standard deviation of one:
– School involvement– Generated from a series of ECLS-K parent interview questions administered in the Kindergarten,
First Grade, Third Grade, and Fifth Grade waves,7this variable is a count of the number of different school involvement
activities that the mother or another adult in the household participated in the school year preceding the interview. The activities include: attending an open house or back-to-school night; attending a PTA meeting; participating in a parent/ teacher conference; attending a school event such as a play, sporting event, or science fair; and participating in a school fundraiser.
– # children’s books– Based on parental survey data from the Kindergarten, First Grade, Third Grade, and Fifth Grade
waves, this variable is a count of the number of children’s books in the child’s house, including library books. Since this variable’s distribution is skewed heavily to the right, we analyze the natural log of this count.
– Discipline– In the Kindergarten, First, Third, and Fifth Grade waves, the ECLS-K asked parents a series of hypothetical questions about how they would respond if their child hit them. Parents were allowed to pick as many of the following responses as applied: Spank, hit back, discuss, ignore, make fun, require the child to perform a chore, require an apol-ogy, take away a privilege, issue a warning, yell, or take other disciplinary action. Based on an exploratory factor anal-Table 1
Time-varying profile of educational attainment for ECLS-K mothers (unweightedNs).
Mother’s education at Kindergarten
<HS HS Some/AA BA MA+
Mother’s education at 1st grade <HS 1711 0 0 0 0
HS 59 4009 0 0 0
Some/AA 37 185 4417 0 0
BA 0 0 47 2340 0
MA+ 1 8 25 25 1156
Mother’s education at 1st grade
Mother’s education at 3rd grade <HS 1112 0 0 0 0
HS 146 2682 0 0 0
Some/AA 45 511 3504 0 0
BA 0 0 222 1937 0
MA+ 4 12 40 142 1065
Mother’s education at 3rd grade
Mother’s education at 5th grade <HS 819 0 0 0 0
HS 56 2129 0 0 0
Some/AA 23 91 3085 0 0
BA 0 0 49 1678 0
MA+ 0 5 19 43 1003
Mother’s education at 5th grade
Mother’s education at 8th grade <HS 632 0 0 0 0
HS 18 1688 0 0 0
Some/AA 10 83 2593 0 0
BA 0 4 57 1459 0
MA+ 0 5 18 50 958
6
The ECLS-K protocol requires the respondent for the parent interview to be: an adult (18 years of age or older), knowledgeable about the child’s care and education, and living in the same household with the child. More than 90% of the parent surveys we utilize were completed by mothers.
7
The ECLS-K also collects school involvement data in the 8th grade survey round, but since the survey item for this wave differs from earlier waves, we exclude it from analyses.
ysis, we constructed a scale measuring the odds that parents would discuss, require chores or an apology, take away a privilege, or issue a warning (alpha = .67). We consider this scale as a measure of the extent to which parents engage in authoritative parenting. This parenting style, which is often contrasted to ‘‘authoritarian’’ parenting, occurs when par-ents are highly responsive to their children as well as highly demanding (Steinberg et al., 1989).
– Family activities– This variable is a scale measuring how often parents tell stories, sing songs, do art projects, play games, do chores, explore nature, work on building projects, and play sports with their child (alpha = .70). The items composing this scale come from the Kindergarten, First Grade, and Third Grade parent interviews.
All models include time-varying controls for fathers or other adult males in the household, including a dummy variable for no father present and a series of dummy variables characterizing the educational attainment of resident father figures. These controls are important in light of the literature connecting family instability to parenting practices (c.f.Furstenberg and Cherlin, 1991; Berger et al., 2008). Moreover, all models control for maternal employment status at the time of the inter-view with a series of dummy variables distinguishing between mothers who work full-time, mothers who work part-time, mothers who are looking for work; mothers who do not work are the reference category. Finally, all models also include time-varying measures of household income (standardized); the number of times the family has moved; the number of sib-lings in the household, and whether or not the household receives AFDC or food stamps.
Cross-sectional Kindergarten OLS and longitudinal random effects models also include a series of time-invariant controls: the focal child’s race and gender; the focal child’s score on the math, reading, and general knowledge tests administered to ECLS-K children in the fall of their Kindergarten year; and mother’s age at the child’s birth. These time-invariant controls are not included in the fixed-effects models since those models control for all time-invariant characteristics of mothers and chil-dren by design. However, the fixed effects models do include a series of grade dummy variables to insure that our findings do not confound cohort trends in parenting practices with the effects of maternal educational attainment.
All models are estimated in Stata. The cross-sectional OLS models Kindergarten models utilize Stata’s ‘‘regress’’ command; the fixed-effects models use Stata’s ‘‘xtreg,fe’’ command; and the random-effects models use Stata’s ‘‘xtreg, re’’ command. To correct for oversampling and non-response in the initial survey, all models are weighted using the ECLS-K’s baseline child/ parent survey weight (bypw0). In addition, we use the Stata ‘‘cluster’’ subcommand on all of our multivariate analyses to correct statistical estimates account for the ECLS-K’s stratified design.8Item non-response rates in the ECLS-K are low, and approximately 95% of respondents with non-missing values on the parenting outcomes also have non-missing values on all con-trol variables. In instances of item non-response, the analyses use listwise deletion.
3. Results
Table 2provides a descriptive look at the relationship between maternal educational attainment and the parenting prac-tices of mothers of Kindergarteners. To ease interpretation, the parenting measures reported in this table are standardized to have a mean of zero and standard deviation of one. There is clear evidence to suggest that the parenting styles of highly edu-cated mothers differ from the parenting styles of less highly eduedu-cated mothers. Mothers with BA degrees, for example, score approximately two-thirds of a standard deviation higher on the school involvement measure than mothers who only com-pleted high school. In non-standardized terms, the mean number of school involvement activities for BA mothers is more than four, compared to approximately three for mothers with high school diplomas. The data reveal similar gaps in the num-ber of children’s books in the homes. The relationship between educational attainment and attitudes toward discipline and frequency of non-academic family activities also vary between more and less educated mothers, although these gaps are less pronounced. The differences in parenting that this table describes are similar in magnitude and direction to the descriptive differences between highly educated and less highly educated parents reported elsewhere (c.f.Attewell and Lavin, 2007; Oreopoulos and Salvanes, 2009).
The left-hand panel inTable 3reveals that parenting is not the only important way in which highly educated mothers differ from less educated mothers. Whites and Asians are over-represented among BA and MA mothers, while blacks and Table 2
Descriptive statistics for parenting practices in Kindergarten (standardized and weighted), by mother’s educational attainment.
<HS HS Some/AA BA MA+ School involvement 0.68 0.20 0.13 0.46 0.46 # Children’s books (ln) 0.99 0.14 0.18 0.48 0.64 Discipline 0.14 0.04 0.00 0.07 0.08 Family activities 0.28 0.05 0.08 0.12 0.21 N= 2095 N= 4729 N= 5036 N= 2528 N= 1220 8
All analyses cluster on the c245pstr variable. ECLS-K documentation suggests incorporating the strata and PSU via Taylor estimation or using bootstrapped replicate weights but Stata’s xtreg, fe fixed effects command does not allow for these design effects corrections. Versions of our OLS models run using Taylor estimation return nearly identical standard errors to those reported here.
Hispanics are over-represented among high school graduates and high school dropouts. College educated mothers are older than less-highly educated mothers. College-educated mothers are also considerably more likely to live with their child’s father or another man, and these male partners are much more likely to be highly educated themselves. This pattern, com-bined with the finding that college educated mothers are more likely to be employed than less educated mothers, helps to explain why the average household income for BA mothers is more than twice the average household income for mothers with high school diplomas. Nearly a third of high school dropout mothers collected AFDC support when their children were in Kindergarten and nearly half collected food stamps. By contrast, just 1% of BA mothers collected AFDC and 2% collected Food Stamps. Finally, consistent with previous literature, the descriptive results reveal a substantial Kindergarten test score gap between children of college educated mothers and less highly educated mothers. On average the children of BA mothers scored nearly three-fourths of a standard deviation higher than the children of mothers who only completed high school.
The right-hand panel ofTable 3focuses on the demographic characteristics of mothers who earned new degrees after their children enrolled in Kindergarten. One way to read this table is to compare the demographic background characteristics for mothers who earned new degrees with those of women whose educational attainment remained unchanged at one level lower in the educational attainment hierarchy. This comparison reveals that mothers whose educational attainment changed during the study period are advantaged relative to mothers who were similarly educated when their children entered kin-dergarten and pursued no further education during the study period. For example, while one-fourth of mothers who lack a high school diploma are white, nearly 40% of mothers who earn new high school diplomas during the ECLS-K study period are white. Likewise, the mean household income for mothers who earn new high school diplomas is $4000 higher at the Kin-dergarten survey round than the mean household income for mothers who remain high school dropouts throughout the study. Similar gaps exist elsewhere in the educational distribution (i.e. between mothers who earn AAs after their child en-ters Kindergarten and to mothers whose educational attainment remains stable at high school diplomas; between mothers who earn BAs after their child enters Kindergarten and mothers whose attainment remains stable at the some college/AA level; and between mothers who earn graduate degrees after their child enters Kindergarten and mothers whose attainment remains stable at the BA level), suggesting that ECLS-K mothers who earn degrees during the study period enjoy socio-eco-nomic advantages even before they earn their new degrees.
At the same time, mothers who earned new degrees during the study period were disadvantaged relative to mothers who had attained the same level of education before their children reached school age. For example, the mean household income level for new BA mothers is more than $25,000 a year lower than the mean household income level for BA mothers whose educational attainment remained stable during the study period (i.e., who have attained that level of education before their children entered Kindergarten). The same pattern holds for other characteristics, highlighting the disadvantages associated with delayed educational attainment. In summary, therefore,Table 3indicates the social status of mothers who earn new degrees while their children are in elementary school lies between that of mothers who start with similar educational
Table 3
Descriptive baseline statistics (weighted) by educational attainment for ECLS-K mothers whose attainment is constant over the study period and ECLS-K mothers who earned new degrees in the study period.
Mothers with same attainment, K-8 Mothers with new degrees, K-8
<HS HS Some/AA BA MA+ HS AA BA MA Time-invariant Female (child) 0.49 0.50 0.49 0.48 0.47 0.49 0.47 0.52 0.48 White 0.26 0.56 0.61 0.77 0.79 0.37 0.47 0.59 0.71 Black 0.19 0.18 0.16 0.07 0.05 0.20 0.20 0.18 0.08 Hispanic 0.49 0.18 0.16 0.07 0.06 0.34 0.25 0.14 0.15 Asian 0.03 0.02 0.03 0.06 0.06 0.06 0.03 0.04 0.03 Other 0.04 0.04 0.05 0.02 0.03 0.04 0.05 0.06 0.03
Mother’s age at birth 24.94 25.61 26.94 30.02 32.05 24.82 25.64 27.45 28.54
Time-varying No dad in HH 0.35 0.29 0.24 0.09 0.08 0.37 0.31 0.20 0.15 Dad < HS 0.21 0.38 0.24 0.09 0.08 0.24 0.32 0.22 0.18 Dad HS 0.39 0.11 0.05 0.01 0.01 0.27 0.12 0.05 0.03 Dad Some/AA 0.04 0.17 0.31 0.20 0.12 0.11 0.17 0.29 0.24 Dad BA+ 0.01 0.05 0.16 0.61 0.71 0.01 0.08 0.24 0.4 Mom works FT 0.31 0.48 0.50 0.42 0.51 0.40 0.50 0.44 0.50 Mom works PT 0.14 0.19 0.23 0.28 0.26 0.18 0.19 0.27 0.22
Mom looking for work 0.09 0.05 0.03 0.01 0.02 0.05 0.04 0.02 0.01
Mom not working 0.46 0.28 0.24 0.29 0.21 0.37 0.27 0.27 0.27
HH income $20,518 $37,373 $50,204 $85,299 $102,514 $24,744 $38,095 $58,922 $66,630 # Moves (birth-K) 2.39 2.29 2.24 1.90 1.85 2.14 2.23 1.99 1.96 Number siblings 1.89 1.43 1.36 1.35 1.33 1.69 1.55 1.41 1.33 AFDC 0.29 0.14 0.09 0.01 0.01 0.22 0.17 0.07 0.05 Food stamps 0.48 0.24 0.15 0.02 0.01 0.36 0.28 0.12 0.09 Test scores (std) 0.71 0.25 0.01 0.48 0.75 0.55 0.25 0.14 0.30 N= 1807 N= 4005 N= 4653 N= 2327 N= 1220 N= 184 N= 778 N= 303 N= 331
attainment levels and do not pursue further education and that of mothers who had reached similar levels of educational attainment before their children entered Kindergarten.
3.1. Relationship between education and parenting
Given the extent of the differences between highly educated and less highly educated mothers, it is reasonable to ask whether maternal education is truly the cause of differences in parenting styles, or whether these differences are due to other confounding factors. The analyses reported inTable 4provide an initial answer to this question, considering the rela-tionship between maternal education and parenting practices in the Kindergarten round of the ECLS-K. These models include controls for race, gender, paternal characteristics, maternal labor force status, number of siblings, household income and public assistance receipt, and child’s Kindergarten test scores. Three of the four OLS regression analyses reported inTable 4 suggest that the relationship between maternal education and parenting is robust even after taking this extensive list of con-trols into account. College-educated mothers are significantly more involved in their Kindergartener’s schools than are otherwise similar mothers who only finished high school. College educated mothers also have more children’s books in the house and participate in more non-academic family activities than mothers who only finished high school. While the magnitude of the positive relationship between maternal education and parenting varies somewhat from measure to mea-sure, it is substantial in each of these cases, with the gap between BA mothers and high school graduate mothers ranging from .13 standard deviations in the family activities measure to .23 standard deviations on the logged children’s books mea-sure. As would be anticipated from descriptive results, the OLS regression model of the relationship between maternal edu-cation and attitudes towards discipline returns less robust results. Mothers who only finished high school do not differ on the authoritative parenting scale from more highly educated mothers, net of controls.
Although the models inTable 4control for an extensive list of maternal and child characteristics, these models likely overstate the link between maternal education and parenting. We suspect that maternal education positively correlates with unobserved predictors of parenting behavior, such as intelligence, motivation, and approach to schooling. Just as these omit-ted variables introduce an upward bias to estimates of the effect education on earnings, we suspect they also introduce an Table 4
OLS regression coefficients; predictors of Kindergarten parenting practices for ECLS-K mothers.
School involvement # Children’s books (ln) Discipline Family activities
Mother’s education Mom < HS 0.25*** 0.46*** 0.06** 0.16*** Mom HS – – – – Mom some/AA 0.16*** 0.16***. 0.01 0.11*** Mom BA 0.23*** 0.18*** 0.04 0.13*** Mom MA+ 0.17*** 0.24*** 0.02 0.11*** Controls No dad in HH 0.13*** 0.01 0.05* 0.01 Dad < HS 0.14*** 0.27*** 0.08* 0.13*** (Dad HS) – – – – Dad Some/AA 0.11*** 0.09*** -0.05*** 0.08*** Dad BA 0.15*** 0.09*** 0.02 0.04 Dad MA+ 0.11*** 0.10*** 0.00 0.11 Mom works FT 0.14*** 0.02 0.01 0.08 Mom works PT 0.01 0.09** 0.04** 0.05
Mom looking for work 0.09***
0.11 0.05 0.01
(Mom not working) – – – –
HH income 0.06*** 0.05*** 0.01 0.00 # Moves (birth-K) 0.04*** 0.02* 0.00 0.01 Number siblings 0.01 0.01 0.01 0.04 AFDC 0.06* 0.04 0.02 0.10 Food stamps 0.21*** 0.12*** 0.00 0.01 Test scores (std) 0.13*** 0.18*** 0.03** 0.05 Female (child) 0.02* 0.07*** 0.07*** 0.05 (White) – – – – Black 0.16*** 0.68*** 0.02 0.03 Hispanic 0.09* 0.65*** 0.03 0.20 Asian 0.40*** 0.70*** 0.00 0.16 Other 0.09** 0.27** 0.05 0.03
Mother’s age at birth 0.02** 0.01 0.01 0.08
Constant 0.06 0.22*** 0.01 0.14 N 14,670 14,542 14,524 14,666 R-square 0.24 0.41 0.01 0.04 *p< 0.05. **p< 0.01. *** p< 0.001.
upward bias to our estimate of the effects of education on parenting. The models presented inTable 5address this bias by estimating the effects ofchangesin maternal education on change in parenting practices and behaviors. In addition to the time-varying controls for paternal characteristics, maternal work status, income, and family size, the fixed-effects panel models reported inTable 5control for all time-invariant characteristics of mothers and children, whether observable or unobservable. To simplify the presentation,Table 5reports only the estimates effects of maternal education on parenting; the full fixed-effects models are available in AppendixTable A1.
Overall, these models suggest that post-natal education attainment has a statistically significant and fairly large effect on maternal parenting. The fixed effects coefficient for earning a high school diploma on school involvement is positive but not statistically significant. However, the fixed effects models strongly suggest that exposure to higher education increases school involvement. Mothers who attain some college or AA degrees while their children are in elementary school increase their school involvement scores by a little more than one-tenth of a standard deviation; mothers who earn a BA increase school involvement scores by 0.17 standard deviations; and mothers who earn an MA or other graduate degree increase school involvement scores by one-fifth of a standard deviation.
The estimated effects of maternal education on in-home parenting practices are of even larger magnitude. Each increase in maternal education, from high school diplomas to graduate degrees, increases the logged number of children’s books in the home by approximately one-fourth of a standard deviation. Similarly, maternal educational attainment boosts the fre-quency of non-academic family activities such as singing and doing art projects. Earning a high school diploma boosts values on the family activities scale by one-fifth of a standard deviation; achieving some college, AA degrees, and BA degrees in-creases the family activity score by two-fifths of a standard deviation.
The highly statistically significant coefficients associated with maternal educational attainment in these models suggest that exposure to education leads women to adopt the types of parenting practices that are commonly associated with rel-atively affluent households in the American context. However, these models also point to an important limitation in the rela-tionship between education and parenting. There is no evidence to suggest that exposure to education changes mothers’ attitudes toward discipline. We find no effect of increasing education on mothers’ approach to discipline, as measured by the ‘‘authoritative parenting’’ scale.
InTable 6, we conduct two additional analyses to explore the timing of the effects of maternal education on parenting behavior. In the first model, we use fixed effects models to estimate the relationship between the time-varying maternal education variable and lagged parenting outcomes. In effect, this analysis asks whether changes in parenting practices occur
beforemothers complete their schooling and increase their educational attainment. While it is not plausible that maternal
educational attainment could have a retrospective effect on maternal parenting practices, it does seem likely that enrollment in formal education could influence parenting practices. Assuming that mothers are enrolled in school in the observation before they record increases in educational attainment, the first set of models inTable 6tests the extent to which educa-tional enrollment – instead of or in addition to degree attainment – influences parenting practices. In the second model, we estimate the effect of maternal educational attainment on parenting behaviors in the observation periodafterthe edu-cational attainment change occurs. This analysis is designed to provide insights into the extent to which mothers’ parenting practices continue to change in the years following their degree attainment. These lag-coded and lead-coded analyses are necessarily somewhat exploratory. Since executing these models eliminates one round of observations, these reduced panel data models have less statistical power and are likely more volatile than the models reported inTable 5.
The significant positive relationship between maternal education change and parenting behaviors in prior waves suggests that some of the increase in the number of children’s books in the home occursbeforemothers earn new degrees. While not statistically significant, the coefficients for some college and MA attainment in the lagged school involvement are also nearly as large as the estimates of the effects of maternal education on school involvement revealed inTable 5, although the coef-ficient for BA attainment is close to zero. One way to interpret these results is as an indication that exposure to education, not
Table 5
Fixed-effects coefficients; effects of education change on parenting practices for ECLS-K mothers (observations in K, 1st, 3rd, and 5th grades), selected results.
School involvement # Children’s books (ln) Discipline Family activities
Mother’s education (Mom < HS) – – – – Mom HS 0.08 0.26*** 0.03 0.20* Mom some/AA 0.13** 0.28*** 0.04 0.37*** Mom BA 0.17** 0.23* 0.17 0.42** Mom MA+ 0.20*** 0.27** 0.17 0.28 Rho 0.66 0.76 0.45 0.70 N(obs) 36,338 36,167 24,059 30,397 N(groups) 10,944 10,936 10,730 10,892 R-square (overall) 0.15 0.15 0.09 0.02
All models include a range of time-varying controls. For complete models, see Appendix A.
*p< 0.05. **
p< 0.01.
just degree attainment, changes maternal parenting behavior. However, since the ECLS-K does not provide reliable data on maternal school enrollment patterns, we are not able to assess this possibility more directly. Nonetheless, we note that the effects of educational attainment on parenting behaviors reported inTable 5are consistently larger than the effects of mater-nal educatiomater-nal attainment on lagged parenting reported inTable 6, suggesting that degree attainment itself exerts an inde-pendent effect on maternal parenting practices.
The second set of models reported inTable 6investigates the extent to which the effect of maternal education persists as mothers spend more time away from school. The parenting behavior variables in these models are lead-coded to represent parenting practices in the observationaftermaternal education is observed. Significant negative coefficients in these models would be an indication that the effects of maternal education on parenting that we reported inTable 5are transitory. How-ever, the findings reported inTable 6’s lead-coded models are largely statistically insignificant, and thus provide no evidence Table 6
Fixed-effects coefficients; predictors of parenting practices for ECLS-K mothers (observations in K, 1st, 3rd, and 5th grades).
School involvement # Children’s books (ln) Discipline Family activities
Lagged parenting practices
(Mom < HS) – – – – Mom HS 0.07 0.23 0.02 0.12 Mom some/AA 0.10 0.25 0.00 0.12 Mom BA 0.01 0.26 0.03 0.19 Mom MA+ 0.15 0.19 0.00 0.06 N(obs) 15,607 14,708 15,441 11,833
Forward-coded parenting practices
(Mom < HS) – – – – Mom HS 0.08 0.17 0.11 0.14 Mom some/AA 0.05 0.26 0.06 0.35 Mom BA 0.12 0.40⁄ 0.09 0.05 Mom MA+ 0.20 0.32 0.27 0.01 N(obs) 17,485 14,835 7802 11,878 Table A1
Fixed-effects coefficients; predictors of parenting practices for ECLS-K mothers (observations in K, 1st, 3rd, and 5th grades).
School involvement # Children’s books (ln) Discipline Family activities
Mother’s education (Mom < HS) – – – – Mom HS 0.08 0.26*** 0.03 0.20* Mom some/AA 0.13** 0.28*** 0.04 0.37*** Mom BA 0.17** 0.23* 0.17 0.42** Mom MA+ 0.20*** 0.27** 0.17 0.28 Controls No dad in HH 0.05** 0.06* 0.03 0.06 Dad < HS 0.00 0.00 0.03 0.03 (Dad HS) – – – – Dad Some/AA 0.02 0.02 0.07 0.00 Dad BA 0.01 0.04 0.03 0.01 Dad MA+ 0.04 0.12* 0.06 0.01 Mom works FT 0.04*** 0.01 0.01 0.04 Mom works PT 0.01 0.01 0.01 0.04
Mom looking for work 0.01 0.04 0.05 0.02
(Mom not working) – – – –
HH income 0.01 0.01 0.02 0.02
# Moves (since last interview) 0.00 0.02** 0.00 0.00
Number siblings 0.01 0.03* 0.04* 0.03 AFDC 0.00 0.02 0.09 0.00 Food stamps 0.04* 0.07** 0.03 0.01 1st Grade 0.10*** 0.17*** – 0.29*** 3rd Grade 0.15*** 0.31*** 0.69*** 0.39*** 5th Grade 0.11*** 0.15*** 0.76*** – Constant 0.29*** 0.28** 0.47*** 0.07 N(obs) 36,338 36,167 24,059 30,397 N(groups) 10,944 10,936 10,730 10,892 Rho 0.663 0.755 0.449 0.695 R-square (overall) 0.15 0.15 0.09 0.02 *p< 0.05. **p< 0.01. *** p< 0.001.
to suggest that the changes in maternal parenting fade away over time. Rather than fading way, the effects of maternal edu-cation on parenting seem to persist over time. In fact, there is some evidence to suggest that maternal eduedu-cational attain-ment starts a process of change in maternal behavior that continues in the years following degree attainattain-ment. Nearly all of the maternal education coefficients in these models are positive. The only significant maternal education coefficient in these lead-coded models is the positive coefficient associated with maternal BA attainment on the number of children’s books in the home. This coefficient indicates that BA mothers continue to acquire children’s books even after they finish higher education. Although not statistically significant, the lead-coded effects of maternal education on school involvement are nearly as large as the contemporaneous effect sizes reported inTable 5. Similarly, the lead-coded effects of maternal edu-cation on the lagged number of children’s books in the home are roughly comparable to the contemporaneous effects. These findings focusing on magnitude of the coefficients, as opposed to strictly statistical significance, provide a preliminary indi-cation that maternal parenting practices continue to evolve in the years after mothers reach new levels of eduindi-cational attainment.
4. Conclusion
Sociologists have dedicated much attention to understanding the extent to which education perpetuates social inequality or facilitates upward mobility. The cultural reproduction tradition has tended to emphasize the persistence of social inequal-ity across generations (Bourdieu, 1973; Bowles and Gintis, 1976), while the status attainment scholars have highlighted the possibilities of upward mobility (c.f.Blau and Duncan, 1967; Sewell et al., 1969). More recently, the cultural mobility tradi-tion has suggested that culture may not only serve to reproduce class structures but could also provide at least some avenues for mobility, particularly through formal schooling (DiMaggio, 1982).
We contribute to this long-standing debate by considering whether increases to mother’s educational attainment can al-ter parenting practices. Recent research has often conceptualized various dimensions of parenting practices as a form of cul-tural capital (for a recent review, seeLareau and Weininger, 2003), providing an adequate research base as well as rendering parenting a relevant setting for considering predictions from cultural reproduction and cultural mobility traditions. While the cultural reproduction tradition would expect limited or no effect of education on parenting practices, cultural mobility research implies that increasing schooling could alter parenting practices. Since specific parenting practices have been linked to children’s educational success (Bodovski and Farkas, 2008; Cheadle, 2008; Dornbusch et al., 1987; Landry et al., 2000; Lar-eau, 2003; Steinberg et al., 1989), the question of whether additional exposure to education can facilitate adoption of specific parenting practices is not only relevant for adjudicating between cultural reproduction and cultural mobility theories but could also have notable policy implications for the possibility of education to reduce social inequality.
While ample literature has noted that more educated parents engage in different parenting practices than less educated parents (c.f.Attewell and Lavin, 2007; Carneiro et al., 2007; Hill and Stafford, 1980; Oreopoulos and Salvanes, 2009; Sayer et al., 2004), that association could be spurious. More educated parents vary from less educated ones on many dimensions; therefore, variation in parenting practices may or may not reflect their differential exposure to schooling. Indeed, our descriptive results show that mothers with different levels of education vary on many individual and family characteristics. Previous research has ordinarily addressed this concern by controlling for a range of attributes available in the data. We ini-tially employ this strategy and find that a substantial portion of the gap in parenting practices between more and less edu-cated mothers is indeed a reflection of those observed differences. A challenge not addressed by including statistical controls in the model, however, is that mothers with more or less education also likely vary on a range of unobservable character-istics. One strategy for addressing this challenge is to examine whetherchangesin maternal education predictchangesin par-enting practices and attitudes. Using recent data from ECLS-K, we can identify mothers who acquire more education after their children enter elementary school and thus present a more robust assessment of the relationship between education and parenting practices.
Our results indicate thatchangesin maternal education indeed predictchangesin parenting practices. Earning a high school diploma increases the number of children’s books in the home and the frequency with which mothers participate in non-academic family activities with their children. In addition to positively influencing these parenting behaviors, expo-sure to higher education (whether or not it leads to a bachelor’s degree) also increases the extent to which mothers are in-volved in their children’s schools. These findings suggest that rather than simply reproducing and legitimating existing cultural capital inequalities, formal education can create avenues for cultural mobility for women who continue to pursue education after their children enter elementary school. More broadly, since fixed effects models effectively control for all time-invariant maternal and child characteristics, we view these findings as indication that education does indeed influence parenting, at least for women who pursue schooling while their children are in elementary school. These findings are broadly consistent with the predictions from the cultural mobility tradition.
Although the overall trend in the results points to changes in parenting practices as a result of schooling, our findings also point to important limitations to the potential effects of education on parenting behavior. We find very little evidence to sug-gest that education influences mothers’ approaches to child discipline. Mothers who lack a high school diploma score slightly lower than high school graduate mothers on the authoritative parenting scale in the cross-sectional Kindergarten analyses. However, we find no other education-based gaps in disciplinary attitude and none of our panel models indicate that earning new levels of educational attainment influences mothers’ disciplinary attitudes.
Considered jointly, these findings suggest that some parenting behaviors are more susceptible to educational influences than others. Maternal education has a particularly pronounced effect on parenting activities that are directly focused on chil-dren’s schooling, such as school involvement and books in the home. The link between maternal education and parenting activities that are less directly related to schooling is less pronounced. In particular, maternal attitudes toward disciplinary appear to be less malleable over the life course and less susceptible to educational influence than are other aspects of par-enting. Previous studies have implied that adult attitudes toward discipline are uniquely shaped by the disciplinary practices they were exposed to as children (c.f.Bower-Russa et al., 2001), which could help to explain the reported patterns. The per-sistence of maternal attitudes toward discipline points to the lasting character of cultural capital. While mothers use their exposure to education to develop new strategies for maximizing their children’s chances educational success, our findings suggest that these mothers tend to continue to draw upon their own experiences in parenting situations that are less directly related to their children’s schooling.
While we are able to examine the relationship between education and parenting, we are not able to explore potential mechanisms explaining that relationship. One hypothesis, emerging from the work ofAttewell and Lavin (2007)is that edu-cation influences women’s parenting practices both by giving them access to new information relevant to parenting and by exposing them to a wider range of peers. Future research is needed to more explicitly test these mechanisms. For example, in order to evaluate the role that access to information relevant to parenting plays in mediating the effect of education on par-enting, one might investigate the effects of education related to child development and education on parpar-enting, relative to the effects more traditional liberal arts of technical education. Similarly, in order to estimate the extent to which peers medi-ate the effect of education on parenting, one might compare the effect of elite or selective education on parenting compared to the effects of education at less selective institutions. Our data do not allow us to directly test the mechanisms through which maternal education influences parenting. However, several of our analyses indicate that the parenting effects of post-secondary educational attainment are larger than the effects of high school completion on parenting. Since students enrolled in secondary education – and particularly second-chance degree programs for high school dropouts – tend to have peers who are less socioeconomically advantaged than students at post-secondary institutions, this finding may provide pro-visional support for the notion that education influences parenting via peer exposure.
Although additional research is needed to explore the specific mechanisms involved, our findings suggest that education can serve as a vehicle of socio-cultural mobility. While these findings are consistent with the large and robust literature dem-onstrating that educational investments can generate economic mobility by boosting individual earnings and career trajec-tories, they provide new insights into the relationship between class and cultural capital. While the cultural reproduction tradition has tended to emphasize that schooling plays a key role in reproduction of social inequality across generations, since ‘‘the educational system demands of everyone alike that they have what it does not give (Bourdieu, 1973, p. 494),’’ our results imply that education could also serve as a source of cultural mobility. This finding has important implications for social policy and practice, suggesting that interventions that extend educational opportunities to mothers may have con-sequences that spill over across the generations. If maternal schooling changes parenting behaviors, it could change educa-tional prospects for their children, reducing social inequalities across generations.
Appendix A SeeTable A1. References
Aschaffenburg, K., Maas, I., 1997. Cultural and educational careers: The dynamics of social reproduction. American Sociological Review 62 (4), 573–587. Attewell, P., Lavin, David, 2007. Passing the Torch: Does Higher Education for the Disadvantaged Pay Off Across the Generations? Russell Sage Foundation,
New York.
Bernstein, Basil, 1971. Social class, language, and socialization. In: Class, Codes and Control. Routledge, London, pp. 193–205.
Berger, L.M., Carlson, M.J., Bzostek, S.H., Osborne, C., 2008. Parenting practices of resident fathers: The role of marital and biological ties. Journal of Marriage and Family 70 (3), 625–639.
Bianchi, Suzanne M., Robinson, John, 1997. What did you do today? Children’s use of time, family composition, and the acquisition of social capital. Journal of Marriage and the Family 59 (2), 332–344.
Blau, Peter M., Duncan, Ottis D., 1967. The American Occupational Structure. Free Press, New York.
Bodovski, Katerina, Farkas, George, 2008. ‘‘Concerted cultivation’’ and unequal achievement in elementary school. Social Science Research 37, 903–919. Bourdieu, Pierre, 1973. Cultural reproduction and social reproduction. In: Karabel, J., Halsey, A.H. (Eds.), Power and Ideology in Education. Oxford University
Press, New York, pp. 487–511.
Bourdieu, Pierre, Passeron, Jean-Claude, 1977. Reproduction in Education, Society and Culture. Sage Publications, Beverly Hills.
Bower-Russa, Mary E., Knutson, John F., Winebarger, Allen, 2001. Disciplinary history, disciplinary attitudes, and risk for abusive parenting. Journal of Community Psychology 29, 219–240.
Bowles, Samuel, Gintis, Herbert, 1976. Schooling in Capitalist America: Educational Reform and the Contradictions of Economic Life. Basic Books, New York. Bradley, Robert H., Corwyn, Robert F., McAdoo, Harriette Pipes, Coll, Cynthia Garcia, 2003. The home environments of children in the United States, part I:
variations by age, ethnicity, and poverty status. Child Development 72 (6), 1844–1867.
Bronfenbrenner, Urie, 1958. Socialization and social class through time and space. In: Maccoby, Eleanor (Ed.), Readings in Social Psychology. Holt, Rinehart & Winston, New York, pp. 400–425.
Carneiro, Pedro, Meghir, Costa, Parey, Mattias, 2007. Maternal Education, Home Environments, and the Development of Children and Adolescents. Discussion Paper 3072. Institute for the Study of Labor, Bonn.
Cheadle, Jacob E., 2008. Educational investment, family context, and children’s math and reading growth from kindergarten through the third grade. Sociology of Education 81 (1), 1–31.
