Journal of Marriage and Family 64 (August 2002): 703–716 703 FRIEDAFOWLER* The University of Nebraska—Lincoln
l
Parenting Practices, Child Adjustment, and Family Diversity
The authors used data from Waves 1 and 2 of the National Survey of Families and Households (NSFH) to test the generality of the links between parenting practices and child outcomes for chil- dren in two age groups: 5–11 and 12–18. Par- ents’ reports of support, monitoring, and harsh punishment were associated in the expected di- rection with parents’ reports of children’s adjust- ment, school grades, and behavior problems in Wave 1 and with children’s reports of self-esteem, grades, and deviance in Wave 2. With a few ex- ceptions, parenting practices did not interact with parents’ race, ethnicity, family structure, educa- tion, income, or gender in predicting child out- comes. A core of common parenting practices ap- pears to be linked with positive outcomes for children across diverse family contexts.
Support, monitoring, and discipline are central di- mensions of parental behavior that are linked with children’s adjustment, development, and well-be- ing. Parental support is reflected in behaviors such as helping with everyday problems, praising chil- dren’s accomplishments, and showing affection.
Department of Sociology, The Pennsylvania State Univer- sity, University Park, PA 16802 ([email protected]).
*Department of Sociology, University of Nebraska—Lin- coln, Lincoln, NE 68588-0324.
Key Words: diversity, family structure, gender, parenting, race, social class.
Monitoring involves supervising children’s activ- ities, keeping track of children’s school work and peer relationships, and requiring conformity to family and community norms. When children misbehave, some parents turn to coercive forms of discipline (such as spanking), whereas other parents rely on noncoercive methods (such as dis- cussing the consequences of misbehavior). A large body of research indicates that the optimal com- bination of parental behavior involves a high level of support, a high level of monitoring, and the avoidance of harsh punishment (Baumrind, 1968, 1978; Darling & Steinberg, 1993; Maccoby &
Martin, 1983; Rollins & Thomas, 1979).
A focus on family diversity shifts our attention to the fact that most studies in this literature have been based on samples of White, two-parent, mid- dle-class families. For this reason, it is not clear whether the dimensions of effective parenting identified in previous research are linked with pos- itive child outcomes among African Americans or Mexican Americans, single parents, parents with low levels of education, and poor parents. Fur- thermore, because most studies have focused on the parenting practices of mothers, it is not clear whether effective parenting takes the same form among fathers as it does among mothers. Similar- ly, few studies consider whether the same parent- ing practices are equally beneficial for sons and daughters. The purpose of our study is to deter- mine whether the dimensions of effective parent- ing identified in prior research apply mainly to
children in relatively advantaged families (that is, White, two-parent, middle-class families) or whether these benefits extend to children in di- verse families as defined by race and ethnicity, family structure, and socioeconomic status. To ac- complish this goal, we rely on data from Waves 1 and 2 of the National Survey of Families and Households (NSFH).
PARENTINGPRACTICES ANDFAMILYDIVERSITY
Research consistently shows that parental support, monitoring, and (avoidance of) harsh punishment are associated with positive outcomes among chil- dren, including higher school grades, fewer be- havior problems, less substance use, better mental health, greater social competence, and more pos- itive self-concepts. Furthermore, these benefits ap- pear to extend to children of all ages, including children in preschool (Crockenberg & Litman, 1990), children in primary school (Jackson, Hen- riksen, & Foshee, 1998; Pratt, Green, MacVicar,
& Bountrogianni, 1992), teenagers (Gunnoe, Hetherington, & Reiss, 1999; Jackson et al., 1998), and college-age offspring (Strage &
Brandt, 1999). These studies have relied on two measurement strategies. Some studies have ex- amined associations between child outcomes and continuous measures of parental behavior, such as support, involvement, warmth, approval, control, monitoring, and harsh punishment (e.g., Amato, 1989; Gray & Steinberg, 1999; Kurdek & Fine, 1994). Other studies have adopted a categorical approach, with the most common scheme involv- ing four parenting styles: authoritative, authoritar- ian, indulgent, and neglecting (e.g., Lamborn, Mounts, Steinberg, & Dornbusch, 1991; Radzisz- ewska, Richardson, Dent, & Flay, 1996). Regard- less of whether researchers have used a dimen- sional or a categorical strategy, however, the results typically lead to similar conclusions. That is, children appear to do best when parents are warm and supportive, spend generous amounts of time with children, monitor children’s behavior, expect children to follow rules, encourage open communication, and react to misbehavior with discussion rather than harsh punishment.
This body of evidence does not imply a simple unidirectional model in which parents’ behavior causes children’s behavior. Indeed, most family scholars assume that children and parents influ- ence one another in a reciprocal fashion (e.g., Am- bert, 1992; Belsky, 1990; Maccoby, 2001). For example, aggressive behavior among children
may elicit harsh behavior from parents, and harsh parenting, in turn, may provoke further misbehav- ior among children (Patterson, Bank, & Stool- miller, 1990). We assume that parenting practices and positive (or negative) child behavior are best viewed as two components of a dynamic system.
The assumption of reciprocal influence, however, does not undermine the conclusion that a combi- nation of strong support, a high level of monitor- ing, and the avoidance of harsh punishment rep- resents an optimal form of parenting.
As noted earlier, most studies in this literature have relied on samples of White, two-parent, mid- dle-class families. This lack of attention to family diversity may be a serious limitation. According to Bronfenbrenner’s (1979) ecological theory, family processes (such as parental behavior) and contextual factors (such as parents’ social class or race) often interact in affecting children’s devel- opment. Similarly, sociological perspectives sug- gest that optimal socialization practices depend on a family’s location within the social structure (e.g., Kohn, 1977). More generally, attention to how parenting practices and child outcomes are linked in different types of families is consistent with the call for greater attention to diversity in family re- search (Demo, Allen, & Fine, 2000). How might parenting practices interact with family context in shaping offspring outcomes? We consider two po- sitions. The first position assumes that optimal parenting practices vary across family contexts, whereas the second position assumes that optimal parenting practices apply to most children, irre- spective of family context.
Optimal Parenting Varies Across Family Contexts
Some researchers have argued that disadvantaged children may benefit less from authoritative par- enting than from a more restrictive, tough style of parenting. For example, Baldwin, Baldwin, and Cole (1990) argued that poor parents living in dangerous neighborhoods need to exercise a high level of control over their children—a level of control that parents living in affluent neighbor- hoods would find neither necessary nor desirable.
Deater-Deckard and Dodge (1997) claimed that a moderate level of physical punishment is not problematic for African American children be- cause this disciplinary style is normative in Black communities. Steinberg, Dornbusch, and Brown (1992) claimed that factors in the social environ- ments of some minority children (such as peer
groups that devalue academic attainment) can at- tenuate the otherwise positive effects of authori- tative parenting on children’s school success. Lar- eau (1989) and McNeal (1999) argued that because middle and upper-class parents possess more cultural capital than working-class or poor parents, parental involvement is especially bene- ficial to children in middle- and upper-class fam- ilies, at least in terms of children’s educational success and occupational attainment.
Consistent with these views, some studies sug- gest that the consequences of different parenting styles vary across family contexts. In one large- scale study, Dornbusch, Ritter, Leiderman, Rob- erts, and Fraleigh (1987) and Steinberg, Mounts, Lamborn, and Dornbusch (1991) found that au- thoritative parenting was positively associated with school grades among White and Latino ad- olescents but not among Asian or African Amer- ican adolescents. In another report from the same study, Lamborn et al. (1991) concluded that au- thoritative parenting was associated with fewer somatic complaints among adolescents but only if their parents had college degrees. With respect to family structure, Lamborn and colleagues reported that authoritative parenting was associated with less internal distress among adolescents with con- tinuously married parents but not among adoles- cents with single or remarried mothers. In a 1-year follow up, Steinberg, Lamborn, Dornbusch, and Darling (1992) found that the generally positive effects of parental involvement and encourage- ment did not apply to school performance among African Americans.
Other studies based on different data sets yield comparable results. McNeal (1999) reported that parent-child discussions about school were posi- tively related to science achievement among White and African American adolescents but not among Latino or Asian adolescents. In addition, parental monitoring was related to less truancy and dropping out of school among most groups but not among Asian Americans. Similarly, Dea- ter-Deckard, Dodge, Bates, and Pettit (1996) found that physical discipline was correlated with externalizing behavior among White primary school children but not among their Black coun- terparts. Baldwin et al. (1990) found that among poor minority students living in single-parent fam- ilies, restrictive and autocratic parenting was as- sociated with the highest level of school achieve- ment. Taken together, these studies suggest that the pattern of optimal parenting described earlier (high support, monitoring, and avoidance of harsh
punishment) is more applicable to White, middle- class children living with two parents than to chil- dren in other circumstances.
Optimal Parenting Does Not Vary Across Family Contexts
An alternative position—and the most parsimo- nious one theoretically—is that a common core of parental behaviors is associated with positive child outcomes, irrespective of race, ethnicity, family structure, or class. Consistent with this position, some studies uncover no evidence to suggest that parenting practices and child outcomes are linked differently across diverse types of families. For example, Rowe, Vazsonyi, and Flannery (1994) found that covariance matrices reflecting parent- ing practices and child outcomes were nearly identical across samples of Black, White, Hispan- ic, and Asian families. Pilgrim, Luo, Urberg, and Fang (1999) reported that parents’ authoritative behavior predicted low rates of drug use among European American, African American, and Chi- nese (Beijing) adolescents. Radziszewska, Rich- ardson, Dent, and Flay (1996) found that author- itative parenting was associated with fewer depressive symptoms, less smoking, and higher academic achievement among adolescents irre- spective of race or ethnicity. Bradley and Corwyn (2000) reported that parental responsiveness was positively associated with self-esteem among Eu- ropean American as well as Chinese American ad- olescents. McLoyd and Smith (2002) found that mothers’ use of spanking, especially in the context of low emotional support, was associated with similar increases in behavior problems among Af- rican American, European American, and Hispan- ic children. Finally, Kim and Ge (2000), in a study of parenting practices and children’s depressive symptoms, found few differences between Chi- nese American and European American families.
In summary the evidence is mixed, with some studies suggesting that optimal parenting practices vary across different types of families, and other studies suggesting that the fundamental dimen- sions of effective parenting are applicable to fam- ilies across a wide range of social contexts. This mixed set of findings illustrates the complexity and nuances of recent family research (Demo &
Cox, 2000; McLoyd, Cauce, Takeuchi, & Wilson, 2000.) Nevertheless, because the number of stud- ies that have explored parenting in diverse family contexts is relatively small, many authors have called for additional work to resolve this issue
(e.g., Darling & Steinberg, 1993; Deater-Deckard et al., 1996; Demo & Cox, 2000; Jackson et al., 1998; McNeal, 1999; Pilgrim et al., 1999).
THEPRESENTSTUDY
Our goal was to determine whether the associa- tions between parenting practices and forms of child adjustment vary across a range of social contexts. This study advances previous research in several ways. First, the majority of prior studies have relied on samples drawn from a single lo- cation, such as a city or school. The study by Steinberg and his associates, perhaps the largest and best known in this genre, was based on ado- lescents from two states. In contrast, we used a large nationally representative data set: the NSFH.
The NSFH is well-suited to address issues of fam- ily diversity because it contains oversamples of African Americans, Latinos, and single parents.
Second, the majority of studies in this literature have been cross-sectional and have relied on a sin- gle source for data (usually parents or children).
In contrast, we employ two waves of data to see whether parents’ reports of parenting practices at Time 1 predict children’s reports of adjustment at Time 2. The use of different sources of informa- tion for independent and dependent variables avoids the problem of common-method variance, which typically leads researchers to overestimate the strength of association between variables. Fi- nally, although most studies have focused on chil- dren in a single age group, the focal children in the NSFH1 ranged in age from 5 to 18. Because of developmental differences between children within this wide age range, we conducted separate analyses for children between the ages of 5 and 11 and for children between the ages of 12 and 18.
Our research involved four steps. First, we se- lected items from the NSFH1 parent interview that appeared to measure parental support, moni- toring, and harsh punishment, and we subjected these items to a confirmatory factor analysis. Sec- ond, we examined the associations between these parenting practices and parents’ reports of three child outcomes: adjustment, grades, and behavior problems. This step was necessary to determine whether we could replicate the general pattern of findings from prior research with the measures de- rived from the NSFH. Third, to see if parenting dimensions were related to child outcomes simi- larly (or differently) across different types of fam- ilies, we examined interactions between parenting
practices and parents’ race and ethnicity, family structure, education, poverty status, and gender. In choosing these variables, we were guided by Demo, Allen, and Fine’s (2000, p. 3) view that the best way to characterize family diversity involves the three major dimensions of social stratification (race, socioeconomic status, and gender) plus family structure. For exploratory purposes, we also included children’s gender. We accomplished this step by comparing structural equation models in which the paths between parenting practices and measures of children’s adjustment were (a) allowed to vary across groups and (b) constrained to be identical across groups. Finally, we turned to the Wave 2 data and used parents’ reports of parenting behavior at Time 1 (T1) to predict chil- dren‘s reports of outcomes (grades, self-esteem, and deviance) at Time 2 (T2). We also tested to see whether parenting practices at T1 interacted with dimensions of family diversity in predicting children’s adjustment at T2. Because children’s age, gender, and number of siblings could affect parents’ behavior as well as children’s adjustment, we controlled for these variables in all analyses.
METHOD Sample
The first wave of the NSFH was conducted in 1987–88 and involved a multistage probability sample of 13,017 adult respondents (Sweet, Bum- pass & Call, 1988). In the main interview, parents of children aged 5 to 18 answered a series of ques- tions about a randomly-chosen focal child (N 5 3,808). Cases missing data on parenting or key demographic characteristics were dropped from the analysis. This procedure resulted in the omis- sion of 408 cases (11%) and an effective sample of 3,400 families.
Interviewers contacted respondents again in 1992–94. In addition to parents, children between the ages of 10 and 17 were interviewed in that year. After deleting cases that were missing data on key variables (n585), our analysis of Wave 2 data was based on a total of 1,331 children (see Sweet and Bumpass, 1996, for details on Wave 2). Because some families dropped out of the analysis between waves (about 22%), we relied on Heckman’s (1979) widely used method to as- sess the existence of attrition bias. We used a probit analysis to predict whether the focal child provided an interview in 1992–94, based on a va- riety of 1987–88 demographic characteristics. We
TABLE1. CONFIRMATORYFACTORANALYSISBASED ONTHREELATENTVARIABLESREPRESENTINGPARENTALSUPPORT, MONITORING,ANDUSE OFHARSHPUNISHMENT
Item
Children age 5–11
Support Monitoring
Harsh Punishment
Children Age 12–18
Support Monitoring
Harsh Punishment 1. Spend time in leisure activities
2. Work on projects or playing 3. Have private talks
4. Help with reading or homework 5. How often praise child 6. How often hug child
.56 .78 .70 .66 .45 .43
—
—
—
—
—
—
—
—
—
—
—
—
.62 .79 .67 .69 .44 .54
—
—
—
—
—
—
—
—
—
—
—
— 7. Child not allowed home alone
8. Know child’s whereabouts 9. Rules about amount of television 10. Rules about types of television pro-
grams
11. How often slap or spank child 12. How often yell at child
—
—
—
—
—
—
.30 .25 .47 .44
—
—
—
—
—
— .40 .97
—
—
—
—
—
—
.67 .36 .48 .54
—
—
—
—
—
— .76 .51 Note: Modelx25614.18; df5100; GFI5.95; CFI5.93; RMSEA5.04. N51,693 children between the ages of 5 and 11 and 1,707 children between the ages of 12 and 18. All paths between latent variables and observed indicators are significant at p,.001.
then used the resulting equation to calculate lamb- da, or the probability of not being interviewed, for all cases. Lambda was not associated significantly with the independent (parenting) variables or the dependent (child outcome) variables and using it as a control variable did not result in substantively important changes in the conclusions. We found little evidence, therefore, that attrition biased the results of our longitudinal analyses.
Variables
Parenting practices. We used items from the Wave 1 interviews that appeared to measure pa- rental support, monitoring, and harsh punishment.
Six items served as indicators of support. The first four questions asked parents how often they spend time with their child in leisure activities away from home (picnics, movies, sports, etc.), at home working on a project or playing together, having private talks, and helping with reading or home- work (15never or rarely, 65almost every day).
The fifth and sixth questions asked parents how often they praised or hugged their children (1 5 never, 4 5very often). Four items served as in- dicators of parental monitoring. First, parents re- ported on times when they allowed the child to be at home alone (before school, in the afternoon af- ter school, all day when there is no school, at night, and overnight). Responses to this question ranged from 05child allowed to be alone all of these times to 55child not allowed to be alone during any of these times. The second item dealt
with how often children were expected to tell par- ents where they are when away from home (15 hardly ever, 4 5 all the time). Finally, parents reported on whether they restricted the amount of television (1 5yes, 05no) or the types of tele- vision programs the child watches (15yes, 05 no). To measure harsh discipline, we askded par- ents how often they (a) yell at their children and (b) spank or slap their children (1 5never, 4 5 very often).
To determine the factorial validity of the 12 parenting items, we conducted a confirmatory fac- tor analysis using Analysis of Moment Structures (AMOS; Arbuckle, 1997) software. Item loadings appear in Table 1. The three latent variables ap- peared to represent parental support, monitoring, and harsh punishment, respectively. The fit of the model was acceptable, as reflected in values of .97 for the Goodness-of-Fit Index (GFI), .93 for the Comparative Fit Index (CFI), and .039 for the Root Mean Square Error of Approximation (RMSEA). Furthermore, the item loadings were generally similar for children between the ages of 5 and 11 and for adolescents between the ages of 12 and 18. The one exception involved the load- ings of the monitoring items, which were lower for younger children than for adolescents. This difference may reflect the fact that parents report- ed greater variability in their monitoring of ado- lescents than of younger children.
Child outcomes. As part of the NSFH1 interview, parents rated the focal child on 10 items drawn
TABLE2. MANDSDFORCHILDADJUSTMENTVARIABLES, FAMILYDIVERSITYVARIABLES,ANDCONTROLVARIABLES
Child Age 5–11
M SD
Child Age 12–18
M SD
Child well-being Adjustment (87–88) Grades (87–88)
Behavior problems (87–88) Self-esteem (92–94)
1.45 4.01 .34 3.25
.27 .93 .69 .38
2.56 6.82 .75
.29 1.66 1.15 Grades (92–94)
Deviance (92–94) Diversity variables
Parent race–ethnicity African American
5.98 .66
.21
1.60 1.16
.19 Mexican American
White Other Family structure
Married biological parents
.08 .65 .06 .59
.05 .72 .04 .56 Single biological parent
Other Parent education Income (log base 10) Respondent is father
.28 .13 12.99 4.46 .34
2.78 .44
.29 .15 12.93 4.53 .32
2.98 .43 Control variables
Child is son Child age Number of siblings
.49 7.83 1.21
1.97 1.04
.52 15.14 .95
1.99 1.11 Note: Standard deviations are not included for dichotomous variables. All variables are measured in 1987–1988, except for child self-esteem, grades, and deviance. For variables measured in 1987–1988, the sample size is 1,693 children between the ages of 5 and 11 and 1,707 adolescents between the ages of 12 and 18. For child self-esteem, grades, and deviance (measured in 1992–1994) the sample size is 1,331.
from a larger inventory of child behavior (Ach- enbach and Edelbrock, 1981). Sample items in- cluded ‘‘loses temper easily,’’ ‘‘bullies or is cruel or mean to others,’’ ‘‘does what you ask,’’ and
‘‘gets along well with other kids.’’ Response op- tions were 15not true, 25sometimes true, and 3 5 often true. Items were scored so that high scores indicated negative behavior, and the mean of the 10 items served as the measure of children’s adjustment (a 5.68). Table 2 provides descriptive statistics for this variable and for other variables in the analysis.
Parents also provided information on children’s school success. For children age 5–11, parents rat- ed how well the child was doing in school (15 near the bottom, 5 5 one of the best students).
For children age 12–18, parents reported on typ- ical school grades (15 mostly F’s, 9 5 mostly A’s). Finally, parents indicated whether the focal child had ever experienced one of the following behavior problems: repeated a grade, met with a teacher or principal because of behavior problems at school, was suspended or expelled from school, ran away from home, was in trouble with the po- lice, saw a doctor for an emotional or behavioral
problem, and was especially difficult to raise. The sum of the number of problems reported by par- ents served as the measure of behavior problems.
To assess outcomes from the child’s perspec- tive, we relied on Wave 2 interviews with focal children. The child’s self-esteem was based on four items, such as ‘‘I am a person of worth,’’ and
‘‘I am satisfied with myself.’’ Response options ranged from 15strongly disagree to 45strong- ly agree, and the mean response across the four items served as the measure of self-esteem (a 5 .65). Children also reported on their school grades, with responses ranging from 15 mostly F’s to 9 5 mostly A’s. Although it is probable that some children misrepresent their grades, Dornbusch et al. (1987) found a correlation of .76 between high-school students’ self-reported grades and official grade point averages. Finally, children reported on five deviant activities: being truant from school, having sexual intercourse, smoking cigarettes in the last month, drinking al- cohol during the last month, and using marijuana.
The five items correlated positively, indicating that children who had engaged in one activity were more likely to have engaged in the other
FIGURE1. MODELRELATINGPARENTALSUPPORT, MONITORING,ANDHARSHPUNISHMENT TOCHILDREN’SWELL-BEING
activities. The sum of the activities (which could range from 0 to 5) served as the measure of de- viance. (Because the distributions for parent-re- ported behavior problems and child-reported de- viance were positively skewed, we replicated all analyses with log transformations to normalize the distributions and the results were identical to those reported later.)
Family diversity variables. Data on parents’ race and ethnicity, family structure, education, income, and gender were obtained from the Wave 1 inter- view with parents. Race was coded into four cat- egories representing Whites (n52,329), Blacks (n 5 680), Mexican Americans (n 5 221), and other (n5170). We decided not to combine Mex- ican Americans (the largest Latino group) with other Latinos, such as Cubans and Puerto Ricans, because of cultural differences between these groups. Family structure was coded into three cat- egories: households in which the child lived with two married, biological (or adoptive) parents (n5 1,996); households in which the child lived with a single parent (n5969); and other family struc- tures, such as stepfamilies and unmarried cohab- iting couples with children (n5 435). Although the single-parent category excluded single parents cohabiting with a partner, it included a variety of other family forms, such as single-mother house- holds, single-father households, children living with one never-married parent, children living with a divorced parent, children living with a wid- owed parent, and multigenerational households.
The average parent had about 13 years of educa- tion. Sixty percent of parents had levels of edu- cation corresponding to a high-school diploma or less (n52,056) and the remaining 40% had some education beyond high school (n 5 1,344). The average household income was about $30,000 per year in 1987–88. Twenty-three percent of respon- dents had household incomes below 150% of the poverty line (n 5774), and the other 77 percent of respondents had household incomes above this threshold (n52,626). About one third of respon- dents were fathers (n 51,122) and the rest were mothers (n 52,278).
Control variables. Because they could be associ- ated with parents’ behavior as well as children’s behavior, we controlled for the focal child’s gen- der (15son, 05daughter), the focal child’s age, and the number of siblings living in the house- hold. The typical household had about one child in addition to the focal child, and the gender of the focal child was almost evenly split between boys and girls.
RESULTS
Parenting Practices and Parents’ Reports of Child Well-Being in Wave 1
We relied on AMOS software with maximum likelihood estimation for all analyses (Arbuckle, 1997). Figure 1 shows the analytic model for the analyses based on data from Wave 1. The three
TABLE3. ASSOCIATIONSBETWEENLATENTPARENTINGVARIABLES ANDCHILDADJUSTMENTRATINGS FORCHILDREN
BETWEEN THEAGES OF5AND11AND FORADOLESCENTSBETWEEN THEAGES OF12AND18 (STANDARDIZEDCOEFFICIENTS)
Independent Variables
Children 5–11
Adjustment Grades
Behavior Problems
Adolescents 12–18
Adjustment Grades
Behavior Problems Parenting dimensions
Support Monitoring Harsh punishment
2.13***
2.01 .39***
.12***
2.03 2.13***
.02 2.05
.20***
2.28***
2.05 .37***
.18***
2.01 2.28***
2.14***
2.01 .22***
Son Child age Siblings R2
2.01 2.14***
2.02 .18***
2.15***
2.10**
2.02 .07***
.13***
.18***
2.07*
.10***
.00 .02 2.04
.16***
2.18***
2.11*
.06*
.12***
.18***
.04 2.11**
.09***
Note:x251,277.42; df5206, GFI5.96; CFI5.91; RMSEA5.04. N51,693 children between the ages of 5 and 11 and 1,707 adolescents between the ages of 12 and 18.
*p,.05. **p,.01. ***p,.001.
parenting variables (support, monitoring, and harsh punishment) appear on the left side of the figure as circles (latent variables), and the control variables (grouped together) appear in the lower left side of the figure. The three child outcomes (adjustment, grades, and behavior problems) ap- pear on the right side of the figure in rectangles (observed variables). Paths between the parenting variables (along with the control variables) and the three child variables from 1987–88 were es- timated simultaneously. The e terms in the figure represent the errors (or residuals) for each of the three child measures. The model also incorporated correlations between the three child outcomes, as represented by curved arrows between the error terms, as well as correlations between the latent variables and the control variables.
Table 3 shows the standardized coefficients for children in both age groups. Column 1 reveals that, among children between the ages of 5 and 11, child adjustment was negatively associated with support and positively associated with harsh punishment. Similarly, Column 2 shows that chil- dren’s school performance was positively associ- ated with support and negatively associated with harsh punishment. Finally, Column 3 reveals that children’s behavior problems were positively as- sociated with harsh punishment. Overall, these re- sults are consistent with prior literature in showing that children appeared to be doing best when their parents were supportive and avoided harsh pun- ishment. Contrary to prior studies, however, we found no evidence that parental monitoring was related to children’s functioning.
The results in Table 3 were similar for children between the ages of 12 and 18. Parental support
was negatively associated with adjustment, posi- tively associated with grades, and negatively as- sociated with behavior problems. Parents’ use of harsh punishment was positively associated with adjustment, negatively associated with grades, and positively associated with behavior problems.
Once again, parental monitoring was not related to any child measure.
The results presented thus far indicate that the parenting scales based on NSFH items (with the exception of monitoring) were linked with child outcomes in ways consistent with prior research.
Our next step was to see if the associations be- tween dimensions of parenting and children’s well-being varied with parents’ race and ethnicity, family structure, education, poverty status, and gender. We used multigroup models for this pur- pose. To illustrate the procedure, consider parents’
race. We split the sample into Black parents and White parents and estimated all nine parameters linking parenting (support, control, and harsh pun- ishment) and child well-being (adjustment, grades, and behavior problems) simultaneously for both groups. In one model, we allowed these nine pa- rameters to differ for Blacks and Whites. In a sec- ond model, we constrained these nine parameters to be identical for Blacks and Whites. We then compared the chi-square values of the two mod- els. If the difference in chi-square values were sig- nificant, then we would conclude that the uncon- strained model fits the data significantly better than the constrained model. (As a rule, a con- strained model fits the data less well than an un- constrained model.) This result would indicate that the overall pattern of associations between parenting practices and children’s well-being dif-
TABLE4. DIFFERENCE INCHI-SQUAREVALUESBETWEENMODELS INWHICHPATHSBETWEENPARENTINGPRACTICES ANDCHILDADJUSTMENTVARIABLESAREFREE TOVARYACROSSGROUPS(UNCONSTRAINED)AND
CONSTRAINED TO BEEQUALACROSSGROUPS
Comparison Groups
Parents’ Reports of Children in
1987–1988
Parents’ Reports of Adolescents
in 1987–1988
Children’s Self-Reports in 1992–1994 Blacks vs. Whites
Mexicans vs. Whites One- vs. two-parent Low vs. high education Poor vs. nonpoor Fathers vs. mothers Sons vs. daughters
1.44 10.28 11.06 13.37 11.18 24.15**
6.52
15.48 16.02 21.48*
4.93 14.27 7.32 70.64***
1.92 16.48 16.66 11.39 9.77 8.46 7.69
Note: N51,693 children in 1987–1988; 1,707 adolescents in 1987–1988; and 1,331 children in 1992–1994. For the significance of the difference in chi-square values, df59.
*p,.05. **p,.01. ***p#.005.
fers for Blacks and Whites. If the difference inx2 values were not significant, however, then we would conclude that the constrained model fits the data about as well as the unconstrained model.
This result would suggest that the overall pattern of associations between parenting practices and children’s well-being does not differ between Blacks and Whites. This procedure is conceptually equivalent to an omnibus test for interactions be- tween parenting practices and race (or any other diversity variable) in analysis of variance or re- gression analysis (Arbuckle, 1997). (Note that in all of these analyses, we constrained the paths be- tween the latent parenting variables and the ob- served indicators of these variables to be the same, so that the same latent variable was being tested across groups.)
Table 4 summarizes the results of these mul- tigroup analyses by showing the differences inx2 values between the unconstrained models and the constrained models. We tested thesex2 statistics for significance with degrees of freedom equal to nine, which reflects the number of parameters constrained to be equal. When Black parents with young children and White parents with young children were compared (Row 1), the difference inx2values between the unconstrained model and the constrained model was 1.44 (Column 1), which was not significant. A comparable result was obtained for adolescents (Column 2). When Mexican American parents and White parents were compared (Row 2), the difference inx2val- ues was not significant for young children (Col- umn 1) or adolescents (Column 2). These results provide no support for the notion that the associ- ation between parenting practices and children’s functioning varies with race or ethnicity.
With respect to family structure, thex2statistic indicated a significant overall difference between single-parent families with adolescent children and two-parent families with adolescent children.
To determine the specific parameters that differed between the two groups, we examined additional multigroup models in which each of the nine pa- rameters was allowed to vary or was constrained to be equal. These analyses revealed two signifi- cant differences in parameters (both p , .05).
Among single parents, parental monitoring was positively associated with grades (b 5 .14, p , .10) and negatively associated with behavior prob- lems (b 5 2.14, p,.10). Among married par- ents, in contrast, the coefficients for parental mon- itoring were low and nonsignificant. These difference may reflect the fact that although all adolescents benefit from monitoring, consistent monitoring tends to be difficult for single mothers to provide on their own.
With respect to education, the sample was split into parents with a high-school degree or less ver- sus parents with at least some college. No signif- icant interactions involving education were appar- ent. With respect to income, the sample was split into low-income parents (those with household in- comes at 150% of the poverty line or less) and high-income parents (those with household in- comes greater than 150% of the poverty line). No significant interactions involving income were ap- parent. Taken together, these results suggest that the basic pattern of relations between parenting practices and children’s adjustment did not vary with socioeconomic status.
With respect to parents’ gender, a significant overall difference emerged among parents with young children. Additional multigroup analyses
revealed that the association between harsh pun- ishment and children’s behavior problems was stronger among mothers (b 5 .23, p,.01) than among fathers (b 5.02, ns). Finally, with respect to children’s gender, a significant overall differ- ence emerged for young children. Additional mul- tigroup analyses revealed that the association be- tween harsh punishment and behavior problems was stronger among sons (b 5.49, p,.001) than among daughters (b 5.28, p ,.001). No other gender differences emerged from the analysis.
In general, the results in Table 4 (Columns 1 and 2) provide relatively little support for the ex- istence of group differences in the associations be- tween parenting practices and children’s adjust- ment. Of the 14 comparisons, only 3 were significant. Two of these differences involved gen- der and these differences did not suggest a con- sistent pattern.
Parenting Practices and Adolescents’
Self-Reports of Adjustment in Wave 2 The next stage of our analysis turned to adoles- cents’ self-reports of adjustment in 1992–1994. In general, the associations between parents’ reports of support, monitoring, and harsh punishment in 1987–1988 and adolescents’ self-reports of out- comes in 1992–1994 were comparable to those shown Table 3. Parental support predicted chil- dren’s self-esteem (b 5.06, p,.10) and grades (b 5.07, p,.05). Parental monitoring predicted lower levels of children’s deviance (b 5 2.17, p ,.01). And parents’ use of harsh punishment pre- dicted lower self-esteem (b 5 2.05, p,.10) and grades (b 5 2.08, p, .05). These coefficients were weaker than those reported in Table 3, pre- sumably for two reasons: (a) different sources of data were used for independent (parenting) vari- ables and dependent (child adjustment) variables, and hence, common method variance could not inflate the associations, and (b) children’s behav- ior at T2 was measured 5 years after parents’ be- havior at T1.In a supplementary analysis, we add- ed parents’ ratings of children’s well-being at T1
as additional predictors in the model but respon- dents’ ratings of their parenting behavior contin- ued to have significant associations with children’s well-being at T2 comparable to those just noted.
Overall, these associations—although modest—
are consistent with the general model of parenting practices described earlier.
The third column of Table 4 shows the results of analyses based on multigroup models for the
1992–1994 data. In no comparison were models with constrained parameters significantly different from models with unconstrained parameters.
These results provide no evidence that the longi- tudinal associations between parenting practices and children’s outcomes depended on parents’
race, family structure, education, income, or gen- der.
Supplementary Analysis
One might argue that more is meant by social con- text than simply being Black rather than White, or being a married parent rather than a single parent.
Defining an ecological niche in these broad strokes almost certainly underestimates the degree of variability within general social categories.
Consequently, it may be necessary to consider the intersection of multiple dimensions of diversity (gender, race, family structure, and class) to delin- eate families that occupy distinctly different po- sitions in the social structure. A difficulty of this approach, of course, is that as the number of di- mensions used to specify groups increases, the number of cases in one’s sample decreases cor- respondingly.
Nevertheless, in an attempt to capture the no- tion of context more accurately, we constructed two specific groups from the 1987–1988 data for a more rigorous comparison: White married moth- ers and Black single mothers. The sample sizes for the two groups were 509 and 145, respective- ly, among young children, and 517 and 150, re- spectively, among adolescents. Although we did not rely on income to form these groups, the mean income in the former group was about $37,000 compared with about $9,000 in the latter group.
In spite of the obvious differences between these two groups, a multigroup analysis (not shown) re- vealed no significant difference in the overall pat- tern of associations between parenting variables and child outcomes among young children (x2dif- ference59.2, df59, ns) or adolescents (x2dif- ference 515.48, df59, ns). These results indi- cate that even within these two ostensibly different groups, the general pattern of linkages between parenting practices and child outcomes was comparable.
DISCUSSION
The goal of this study was to see if the associa- tions between parenting practices and aspects of child functioning vary across basic dimensions of
family diversity. We considered two perspectives.
The first states that the widely accepted model of effective parenting (high support, high monitor- ing, and avoidance of harsh punishment) is appli- cable in some, but not in all, family contexts. A stronger form of this first perspective holds that the model of effective parenting identified in prior research is relevant mainly to advantaged children (that is, to White, middle-class children living with both biological parents). The second per- spective, in contrast, holds that the benefits of ef- fective parenting are shared widely by children, irrespective of family context.
We began by constructing measures of three dimensions of parenting based on parents’ an- swers to questions on the NSFH1 interview. Con- sistent with expectations, these items formed three factors that appeared to correspond to parental support, monitoring, and harsh punishment. Also consistent with prior research, scores on these scales (with the exception of monitoring) were as- sociated in the expected direction with parents’
reports of children’s adjustment, school grades, and behavior problems. We also examined asso- ciations between parents’ reports of parental be- havior in 1987–1988 and children’s reports of out- comes in 1992–1994 using the NSFH2. Because common method variance could not inflate these correlations, this set of analyses provided a more stringent test of the hypothesis that parenting prac- tices are linked with child outcomes. Although the latter analysis yielded associations that were mod- est in magnitude, the general pattern of results was consistent with the parenting literature: Children were doing best when parents exhibited a high level of support, monitored their children’s behav- ior, and avoided harsh punishment.
Although the analysis restricted to the 1987–
88 data did not reveal evidence that parental mon- itoring was related to children’s adjustment (Table 3), parental monitoring emerged as a significant longitudinal predictor of children’s deviance in 1992–1994. That is, when parents engaged in high levels of monitoring when children were between the ages of 5 and 11, children reported less de- viance (being truant from school, having sexual intercourse, smoking cigarettes, drinking alcohol, and using marijuana) approximately 5 years later.
These results suggest that early parental monitor- ing may help children to refrain from troubling and potentially dangerous behaviors during their adolescent years, perhaps because firm parental control facilitates the internalization of parental values and the capacity to engage in self-control.
Having shown that the three parenting vari- ables performed in ways congruent with prior re- search, our next goal was to examine interactions between the parenting dimensions and parents’
race, ethnicity, family structure, socioeconomic status (education and income), and gender. Per- haps the most interesting finding to emerge from this analysis was a trend for parental monitoring to be related to positive adolescent functioning more strongly among single parents than married parents. The importance of monitoring among sin- gle parents presumably reflects the fact that only one parent, rather than two, is available in the household. Hence, single parents may need to be especially vigilant to protect their children from engaging in deviant behavior.
Overall, however, our analysis revealed little evidence that the associations between parenting practices and child outcomes differed across groups. Moreover, we found little evidence of contextual effects even when groups as different as (a) White, married, non-poor mothers and (b) Black, single, poor mothers were compared. Our data, therefore, provide the strongest support for the second perspective described earlier, that is, that the dimensions of effective parenting can be generalized across a range of social contexts. This conclusion does not mean that optimal parental behavior is identical for every child. It is likely, for example, that parents living in dangerous neighborhoods need to exercise more caution with their children than do parents living in safe neigh- borhoods. Also, it seems plausible that well-edu- cated parents have to expend less effort to facili- tate their children’s academic success than do poorly educated parents. Nevertheless, our anal- ysis provides little support for the notion that a model of effective parenting based on support, monitoring, and the avoidance of harsh punish- ment is appropriate only for White, married par- ents in middle-class households. The search for group differences in the effects of parenting (or any other family process) may require studies that target more narrowly defined groups occupying more specialized niches in the social order than we were able to accomplish in the project de- scribed in this article.
We believe that this conclusion is generally consistent with prior literature taken as a whole.
Although some previous studies (contrary to the present study) report variations in the links be- tween parenting practices and child outcomes, these variations appear only for some groups, some parenting practices, and some outcomes,
with relatively little consistency across or even within studies. For example, Deater-Deckard et al.
(1996) found a significant positive association be- tween parents’ use of harsh punishment and chil- dren’s externalizing behavior for European Amer- ican children but not for African American children, at least when children’s behavior was re- ported by teachers or peers. But the association was significant and positive for European Amer- ican and African American children when chil- dren’s behavior was reported by mothers. Fur- thermore, a racial difference was apparent when the investigators focused on parents’ use of mod- erately harsh discipline but not when the investi- gators focused on parents’ use of abusive disci- pline, which was linked to elevated levels of externalizing behavior in both groups of children (Deater-Deckard & Dodge 1997). Similarly, Stein- berg et al. (1991) found that authoritative parent- ing was positively associated with school grades among White and Latino adolescents, but not among Asians or African Americans. Other anal- yses based on their data, however, revealed that the links between parenting practices and chil- dren’s externalizing and internalizing behavior did not vary with racial, social class, or family struc- ture (Steinberg, Lamborn, Darling, Mounts, &
Dornbusch, 1994). Support for the hypothesis of contextual effects requires the demonstration of consistent, replicable trends across groups. Al- though a meta-analysis of this literature may clar- ify the current state of knowledge, available evi- dence suggests relatively few consistent, replicable contextual effects. In the absence of strong evidence otherwise, researchers should ac- cept the most parsimonious model consistent with the data.
The study reported in this article was not with- out limitations. First, the NSFH interview was not designed to measure parenting in its full complex- ity. For this reason, we were forced to rely on a less-than-ideal pool of items to construct our mea- sures. Although a confirmatory factor analysis in- dicated that the NSFH items clustered into com- ponents that were consistent with prior literature, measurement would have been stronger with a more comprehensive battery of items. For the younger children in our study, the item loadings for monitoring were modest, which may account for the small number of significant findings for this factor. Furthermore, we were forced to omit some dimensions of parenting from consideration, such as induction (providing explanations to chil- dren), support for autonomy, and democratic con-
trol (Barber, Olson, & Shagle, 1994; Furstenberg, Cook, Eccles, Edler, & Sameroff, 1999; Gray &
Steinberg, 1999). Dimensions of parenting un- measured in the present study may yield stronger evidence of contextual effects than we were able to uncover.
Our study also was limited with respect to the range of families that could be incorporated into the design. Even with a large sample, the number of Mexican Americans was relatively low, result- ing in a decline in statistical power to detect group differences. Moreover, there were not enough cas- es of Cuban Americans, Puerto Ricans, or Asian American families for analysis. Group differences might have emerged if a larger range of racial and ethnic groups had been available. In addition, some of our categories, such as single-parent fam- ily, were defined broadly, and the consequences of some parental behaviors might vary with fac- tors such as the presence of extended kin in the household.
In summary, our results suggest that a core of parental practices benefits (or harms) children across a variety of contexts. The importance of specific parenting behaviors for child development does not appear to depend on whether parents are poorly educated or well-educated, Black or White, married or single, mothers or fathers. When par- ents spend time with children, help with home- work, talk about problems, provide encourage- ment, and show affection, children do well. When parents provide a high level of monitoring and expect their children to follow family rules, ado- lescents engage in less deviant behavior. Finally, when parents rely on hitting and yelling as fre- quent methods of responding to children’s mis- behavior, children’s well-being declines. As Demo, Allen, and Fine (2000, p. 2) pointed out, family research tends to emphasize differences in the experiences of various groups, but studying family diversity also requires that we attend to the commonalities shared by all types of families. The extent to which a common core of optimal par- enting practices appears across diverse social and cultural settings can be determined through addi- tional studies that explore large, national data sets, as well as studies that focus specifically on fam- ilies occupying particular ecological niches in the social structure.
NOTE
We thank Nan Crouter and David Johnson for helpful advice on an earlier draft of this article.
REFERENCES
Achenbach, T. M., & Edelbrock, C. S. (1981). Behavior problems and competencies reported by parents of normal and disturbed children age four through six- teen. Monograph of the Society for Research in Child Development, 46, 82.
Amato, P. R. (1989). Family processes and the compe- tence of adolescents and primary school children.
Journal of Youth and Adolescence, 18, 39–53.
Ambert, Anne-Marie. (1992). The effect of children on parents. New York: Haworth Press.
Arbuckle, J. L. (1997). Amos users’ guide, version 3.6.
Chicago: Smallwaters.
Baldwin, A., Baldwin, C., & Cole, R. E. (1990). Stress- resistant families and stress-resistant children. In J. J.
Rolf, A. S. Masten, D. Ciccetti, & S. Weintraub (Eds.), Risk and protective factors in the development of psychopathology (pp. 257–280). Cambridge, UK:
Cambridge University Press.
Barber, B. K., Olson, J. E., & Shagle, S. C. (1994).
Associations between parental psychological and be- havioral control and youth internalized and external- ized behavior. Child Development, 65, 1120–1136.
Baumrind, D. (1968). Authoritarian versus authoritative parental control. Adolescence, 3, 255–272.
Baumrind, D. (1978). Parental disciplinary patterns and social competence in youth. Youth and Society, 9, 236–276.
Belsky, J. (1990). Parental and nonparental child care and children’s socioemotional development. In A.
Booth (Ed.), Contemporary families: Looking for- ward, looking back (pp. 122–140). Minneapolis, MN:
National Council on Family Relations.
Bradley, R. H., & Corwyn, R. F. (2000). Moderating effects of perceived amount of family conflict on the relation between home environmental processes and the well-being of adolescents. Journal of Family Psy- chology, 14, 349–364.
Bronfenbrenner, U. (1979). The ecology of human de- velopment. Cambridge, MA: Harvard University Press.
Crockenberg, S., & Litman, C. (1990). Autonomy as competence in 2-year-olds: Maternal correlates of child defiance, compliance, and self-assertion. Devel- opmental Psychology, 26, 961–971.
Darling, N., & Steinberg, L. (1993). Parenting style as context: An integrative model. Psychological Bulle- tin, 113, 487–496.
Deater-Deckard, K., Dodge, K. A., Bates, J. E., & Pettit, G. S. (1996). Physical discipline among African American and European American children: Links to children’s externalizing behaviors. Developmental Psychology, 32, 1065–1072.
Deater-Deckard, K., & Dodge, K. A. (1997). External- izing behavior problems and discipline revisited:
Nonlinear effects and variations by culture, context, and gender. Psychological Inquiry, 8, 161–175.
Demo, D. H., & Cox, M. J. (2000). Families with young children: A review of research in the 1990s. Journal of Marriage and the Family, 62, 876–895.
Demo, D. H., Allen, K. & Fine, M., (1999). Handbook of family diversity. New York: Oxford University Press.
Dornbusch, S. M., Ritter, P. L., Liederman, P. H., Rob- erts, D. F., & Fraleigh, M. J. (1987). The relation of
parenting style to adolescent school performance.
Child Development, 58, 1244–1257.
Furstenberg, F. F., Jr., Cook, T. D., Eccles, J., Elder, G.
H., & Sameroff, A. (1999). Managing to make it:
Urban families and adolescent success. Chicago:
University of Chicago Press.
Gray, M. R., & Steinberg, L. (1999). Unpacking au- thoritative parenting: Reassessing a multidimensional construct. Journal of Marriage and the Family, 61, 574–587.
Gunnoe, L. M., Hetherington, E. M., & Reiss, D.
(1999). Parental religiosity, parenting style, and ad- olescent social responsibility. Journal of Early Ado- lescence, 19, 199–225.
Heckman, J. J. (1979). Sample selection bias as a spec- ification error. Econometrica, 47, 153–161.
Jackson, C., Henriksen, L., & Foshee, V. A. (1998). The authoritative parenting index: Predicting health risk behaviors among children and adolescents. Health Education and Behavior, 25, 319–337.
Kim, S. Y., & Ge, X. (2000). Parenting practices and adolescent depressive symptoms in Chinese Ameri- can families. Journal of Family Psychology, 14, 420–
435.
Kohn, M. L. (1977). Class and conformity: A study of values (2nd ed.). Chicago: University of Chicago Press.
Kurdek, L. A., & Fine, M. A. (1994). Family acceptance and family control as predictors of adjustment in young adolescents: Linear, curvilinear, or interactive effects? Child Development, 65, 1137–1146.
Lamborn, S. D., Mounts, N. S., Steinberg, L., & Dorn- busch, S. M. (1991). Patterns of competence and ad- justment among adolescents from authoritative, au- thoritarian, indulgent, and neglectful families. Child Development, 62, 1049–1065.
Lareau, A. (1987). Social class differences in family- school relationships: The importance of cultural cap- ital. Sociology of Education, 60, 73–85.
Maccoby, E. E. (2001, December). The gender of child and parent as a factor in family dynamics. Paper pre- sented at the Annual Family Issues Symposium, Pennsylvania State University, University Park, PA.
Maccoby, E. E., & Martin, J. A. (1983). Socialization in the context of the family: Parent-child interaction.
In P. H. Mussen & E. M. Hetherington (Eds.), Hand- book of child psychology. Volume 4: Socialization, personality, and social development (pp. 11–101).
New York: Wiley.
McLoyd, V. C., Cause, A. M., Takeuchi, D., & Wilson, L. (2000). Marital processes and parental socializa- tion in families of color: A decade review of research.
Journal of Marriage and the Family, 62, 1070–1093.
McLoyd, V. C., & Smith, J. (2002). Physical discipline and behavior problems in African American, Euro- pean American, and Hispanic Children: Emotional support as a moderator. Journal of Marriage and Family, 64, 40–53.
McNeal, R. B., Jr. (1999). Parental involvement as so- cial capital: Differential effectiveness on science achievement, truancy, and dropping out. Social Forc- es, 78, 117–144.
Patterson, G. R., Bank, L., & Stoolmiller, M. (1990).
The preadolescent’s contributions to disrupted family process. In R. Montemayor & G. R. Adams (Eds.), From childhood to adolescence: A transitional peri-
od? Advances in adolescent development (Vol. 2, pp.
107–133). Thousand Oaks, CA: Sage.
Pilgrim, C., Luo, Q., Urberg, K. A., & Fang, X. (1999).
Influence of peers, parents, and individual character- istics on adolescent drug use in two cultures. Merrill- Palmer Quarterly, 45, 85–107.
Pratt, M. W., Green, D., MacVicar, J., & Bountrogianni, M. (1992). The mathematical parent: Parental scaf- folding, parent style, and learning outcomes in long- division mathematics homework. Journal of Applied Developmental Psychology, 13, 17–34.
Radziszewska, B., Richardson, J. L., Dent, C. W., &
Flay, B. R. (1996). Parenting style and adolescent de- pressive symptoms, smoking, and academic achieve- ment: Ethnic, gender, and SES differences. Journal of Behavioral Medicine, 18, 289–305.
Rollins, B. C., & Thomas, D. L. (1979). Parental sup- port, power, and control techniques in the socializa- tion of children. In W. R. Burr, R. Hill, F. I. Nye, &
I. L. Reiss (Eds.), Contemporary theories about the family. Volume 1: Research-based theories (pp. 317–
364). Glencoe, NJ: The Free Press.
Rowe, D. C., Vazsonyi, A. T., & Flannery, D. J. (1994).
No more than skin deep: Ethnic and racial similarity in developmental process. Psychological Review, 101, 396–413.
Steinberg, L. Dornbusch, S., & Brown, B. (1992). Eth- nic differences in adolescent achievement: An eco-
logical perspective. American Psychologist, 47, 723–
729.
Steinberg, L., Lamborn, S. D., Darling, N., Mounts, N.
S., & Dornbusch, S. M. (1994). Over-time changes in adjustment and competence among adolescents from authoritative, authoritarian, indulgent, and ne- glectful families. Child Development, 65, 754–770.
Steinberg, L., Lamborn, S. D., Dornbusch, S. M., Dar- ling, N. (1992). Impact of parenting practices on ad- olescent achievement: Authoritative parenting, school involvement, and encouragement to succeed. Child Development, 63, 1266–1281.
Steinberg, L., Mounts, N. S., Lamborn, S. D., & Dorn- busch, S. M. (1991). Authoritative parenting and ad- olescent adjustment across varied ecological niches.
Journal of Research on Adolescence, 1, 19–36.
Strage, A., & Brandt, T. S. (1999). Authoritative par- enting and college students’ academic adjustment and success. Journal of Educational Psychology, 91, 146–
156.
Sweet, J. A., & Bumpass, L. L. (1996). The National Survey of Families and Households—Waves 1 and 2:
Data description and documentation. Madision, WI:
Center for Demography and Ecology, University of Wisconsin—Madison.
Sweet, J. A., Bumpass, L. L., & Call, V. (1988). The design and content of the National Survey of Families and Households. Madison, WI: Center for Demog- raphy and Ecology, University of Wisconsin–Madi- son.