at SciVerse ScienceDirect

Learning and Instruction 23 (2013) 43e51

Contents lists available

Learning and Instruction

journal homepage: www.elsevier .com/locate/ learninstruc

Parental involvement and general cognitive ability as predictorsof domain-specific academic achievement in early adolescence

Julia Karbach*, Juliana Gottschling, Marion Spengler, Katrin Hegewald, Frank M. SpinathDepartment of Psychology, Saarland University, Germany

a r t i c l e i n f o

Article history:Received 14 May 2012Received in revised form13 September 2012Accepted 24 September 2012

Keywords:Parental involvementGeneral cognitive abilityAcademic achievementTwin study CoSMoS

* Corresponding author. Department of EducationalCampus A4 2, D-66123 Saarbruecken, Germany. Tel.:681 302 58341.

E-mail address: j.karbach@mx.uni-saarland.de (J. K

0959-4752/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.learninstruc.2012.09.004

a b s t r a c t

Numerous studies showed that general cognitive ability (GCA) is a reliable predictor of academicachievement. In addition, parental involvement in their children’s academic development is ofmajor importance in early adolescence. This study investigated the incremental validity of parentalinvolvement over GCA in the prediction of academic performance within the domains of math andlanguage. We examined four dimensions of perceived parental involvement: autonomy supportingbehavior, emotional responsivity, structure, and achievement-oriented control. Results from a sample of334 adolescents (mean age ¼ 12.4, SD ¼ .9, range ¼ 10e14 years) showed that GCA was the strongestpredictor of achievement in both domains. While autonomy support and emotional responsivity had nopredictive value over GCA, high levels of achievement-oriented control and structure were detrimental toacademic success. These findings provide new evidence for the significance of parental involvement intheir children’s achievement in school even after the most powerful predictor of academic success hasbeen accounted for.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Achievement in school is considered as a critical prerequisitefor subsequent academic and vocational success (e.g., Jimerson,Egeland, & Teo, 1999; Schmidt & Hunter, 1998; Williamson,Appelbaum, & Epanchin, 1991). As a consequence, a lot of researchhas been dedicated to the identification of factors that contribute toschool achievement. This research indicates that academic perfor-mance in middle childhood and adolescence is determined bya complex interplay of numerous variables. Most authors distin-guish between cognitive predictors, such as general cognitive abilityor working memory (e.g., Gathercole, Pickering, Knight, &Stegmann, 2004; Lu, Weber, Spinath, & Shi, 2011) and non-cognitive predictors, such as motivation (e.g., Gottfried, 1985;Gottschling, Spengler, Spinath, & Spinath, 2012; Greven, Harlaar,Kovas, Chamorro-Premuzic, & Plomin, 2009; Schicke & Fagan,1994; Spinath, Spinath, Harlaar, & Plomin, 2006) or characteristicsof the family and the school environment (e.g., Hill & Tyson, 2009;Son & Morrison, 2010). The importance of cognitive variables,

Science, Saarland University,þ49 681 302 3875; fax: þ49

arbach).

All rights reserved.

particularly the role of general cognitive ability in academic success,is well documented in the literature on individual differences (e.g.,Gottfredson, 2002; cf. Gustafsson & Undheim, 1996). Beyond that,the field of educational psychology has increasingly acknowledgedthe impact of the home environment on students’ learning anddevelopmental processes (e.g., Seginer, 2006; Son & Morrison,2010). One particular aspect that has received increasing attentionover the last years is the degree of parental involvement in theirchildren’s education (formeta-analyses see Fan & Chen, 2001; Hill &Tyson, 2009). Although this research identified parental involve-ment as a robust predictor of academic success, it is still an openquestion whether it can explain additional variance after the mostpowerful predictor of school achievement has been accounted for.This issue is particularly important considering that parentalinvolvement is an influence on children’s academic developmentthat can be considered modifiable, for instance by means of coun-seling or intervention. Thus, identifying which types of parentalinvolvement are particularly beneficial (or detrimental) to chil-dren’s academic success is of major importance in the educationalcontext. Therefore, the aim of our study was to integrate previousfindings from research on individual differences and educationalscience and to investigate the incremental validity of parentalinvolvement over general cognitive ability in the prediction ofacademic achievement in early adolescence.

J. Karbach et al. / Learning and Instruction 23 (2013) 43e5144

1.1. General cognitive ability and academic achievement

Despite the wide range of definitions of intelligence, mostresearchers agree that general cognitive ability iswell representedbyageneral factorof intelligence, theg-factor (e.g., Jensen,1993;Neisseret al., 1996). When it comes to academic success, there is no doubtthat g is the single most important predictor (e.g., Deary, Strand,Smith, & Fernandez, 2007; Fraser, Walberg, Welch, & Hattie, 1987;Gottfredson, 2002; Humphreys & Stark, 2002; Kuncel, Hezlett, &Ones, 2004; Laidra, Pullmann, & Allik, 2007; Rhode & Thompson,2007).

Regardless of the specific academic context, correlations around.50 are typically reported between measures of educationalachievement and general cognitive ability (cf. Gustafsson &Undheim, 1996; Neisser et al., 1996), explaining 25% of the totalvariance in academic performance. The importance of generalcognitive ability is also supported by the range of differentachievement criteria that can be predicted by one general measureof g, from school achievement to academic performance inuniversity students and vocational success in later life. Furthermore,general cognitive ability is related to a number of variables that areassociated with academic success, such as socioeconomic status(SES), level of education, and income (e.g., Neisser et al., 1996).

Nevertheless, even though general cognitive ability is the mostpowerful individual predictor of achievement, it leaves muchvariance unaccounted for. Given that numerous studies pointed tothe importance of parenting practices for achievement in school(for reviews see Desforges & Abouchaar, 2003; Spera, 2005), thepresent study focuses on one specific aspect of parenting, namelythe role of parental involvement in their adolescent children’sacademic success.

1.2. Parental involvement and academic achievement

Parental involvement (PI) generally refers to parents’ behaviorat home and in school settings meant to support their children’seducational progress (e.g., El Nokali, Bachman, & Votruba-Drzal,2010; Fan & Chen, 2001; Gonzales-DeHass, Willems, & Holbein,2005). Consistent with this rather broad definition, the term hasbeen loosely applied to a variety of activities and the parentalbehavior subsumed under the construct of PI has been veryheterogeneous. It ranges from parents’ attendance of school activ-ities to homework assistance and parenting styles (Gonzales-DeHass et al., 2005; Maegi, Lerkkanen, Poikkeus, Rasku-Puttonen,& Nurmi, 2011; Shumov & Miller, 2001), rendering it difficult tocompare empirical findings across studies.

In an attempt to provide a classification of involvement strate-gies, Epstein (1987) and Comer (1995) differentiated between totwo basic types of PI: School-based strategies, such as the commu-nication betweenparents and teachers or their attendance of schoolevents, and home-based strategies, such as educational activitiesand parental supervision, support, and reinforcement of learning athome. In their multidimensional concept Grolnick and Slowiaczek(1994) introduced three types of involvement: Behavioral involve-ment refers to both home-based and school-based involvementstrategies, for instance active communication between home andschool, volunteering at school, and assisting with homework.Cognitive-intellectual involvement reflects that parents expose theirchildren to educationally stimulating activities and experiences,while personal involvement describes parental attitudes andexpectations about the value and the utility of school and education.This definition of PI as a multidimensional concept not only allowsthe simultaneous assessment of different aspects of parentalbehavior, but it also facilitates the analysis of differential associa-tions between specific components of involvement and important

aspects of academic development, such as motivation and schoolperformance (Fan & Chen, 2001; Gonzales-DeHass et al., 2005).

Over the last decade, there has been increasing interest in the roleof PI for academic success in adolescence. Hill and Tyson (2009)studied the effects of different types of involvement on achieve-ment in a meta-analytic approach that differentiated betweenschool-based involvement, home-based involvement, and academicsocialization. The latter can be mapped onto Grolnick andSlowiaczek’s (1994) concept of personal involvement: It refers toparentechild communication creating an understanding for thegoals and purposes of academic performance, discussing learningstrategies, or communicating parental expectations for educationand achievement. Results of the meta-analysis showed an overallpositive relationship between PI and academic skills (cf. Jeynes,2007; for similar results in elementary school see Fan & Chen,2001). Interestingly, the type of involvement parents engaged inmodulated this association:Whereas different types of school-basedinvolvement were moderately related to achievement, home-basedinvolvement was not consistently associated with achievementwhen it pertained to homework assistance. Other types of home-based involvement, however, were positively related to schoolsuccess. Finally, academic socialization proved to be the bestpredictor for academic achievement. It subsumes parental behaviorthat supports the student’s autonomy and independence, buildsupon the development of internalized motivation for achievement,and provides a link between school work and future goals.

Yet, it should be noted that not all types of PI seem to fosterschool success. A number of studies have shown that parentalcontrol and achievement-related pressure can have detrimentaleffects on academic performance (e.g., Levpuscek & Zupancic, 2009;Rogers, Theule, Ryan, Adams, & Keating, 2009). In these studies, theuse of commands, coercive interactions, criticism or punishmentwas associated with lower academic performance (e.g., Niggli,Trautwein, Schnyder, Ludtke, & Neumann, 2007; Pomerantz &Eaton, 2001), possibly because this type of behavior is perceivedas over-controlling and thereby undermines the students’ sense ofcompetence and autonomy. This effect appears to be particularlystrong in adolescence (Gonzales-DeHass et al., 2005).

Age-related changes in the relation between PI and academicachievement are most prominent between elementary andsecondary school (Hill & Tyson, 2009; Stevenson & Baker, 1987).School-based involvement is of particular importance in theelementary school context, because parental visits to the classroomand interactions with children’s teachers increase parents’ knowl-edge about the curriculum and support the effectiveness ofinvolvement at home (Comer, 1995; Hill & Taylor, 2004). Insecondary school, however, home-based involvement plays anincreasingly important role, providing assistance with homework,enhancing motivation, and structuring free time and homeworktime (Cooper, 1989; Fan & Chen, 2001) while promoting indepen-dence and autonomy (Desforges & Abouchaar, 2003; Hill & Tyson,2009). This shift away from direct school-based involvement isrelated to changes in the children’s school environment, whichbecomesmore complex and therefore challenges the parents’ abilityto stayactively involved in their children’s schoolwork (e.g., Sanders& Epstein, 2000). In addition, early adolescence is associated withmajor developments in terms of cognition and self-concept (Adams& Berzonsky, 2003; Lerner & Steinberg, 2004). That is, adolescentsare increasingly better able at integrating knowledge derived fromown previous successes and failures and at coordinating the waytheypursuemultiple educational and personal goals (Byrnes,Miller,& Reynolds, 1999). Therefore, they are more and more able tounderstand how present school achievement is related to futureacademic success and tomake decisions regarding their educationalprocess (e.g., course selections). As the student’s sense of autonomy

J. Karbach et al. / Learning and Instruction 23 (2013) 43e51 45

increases, direct school-based PI is needed less and becomes lesseffective (Seginer, 2006).

1.3. The present study

There is growing evidence indicating that multiple aspects of PIcan be considered relevant for students’ academic success. Giventhat PI is neither a unitary nor a consensus concept, it is particularlyimportant to identify the extent to which specific dimensions of PIare most effective in promoting academic achievement in earlyadolescence. In this study, we relied on a multidimensionalframework suggested by Lorenz and Wild (2007) in order to assessthe benefits and risks associated with different types and levels ofinvolvement. This framework is based on three assumptions: First,the definition of PI is restricted to home-based learning involvingdirect parentechild interactions which seem to be more relevant inadolescence than parenteteacher communication (cf. Hill & Tyson,2009). Second, PI not only incorporates practices that are explicitlysupposed to have positive effects on learning and achievement butincludes a wide range of activities that can be considered relevantparental behavior. Third, the authors assume that PI is not restrictedto direct instructional interactions. Instead, it can provide oppor-tunities for the development of effective study habits and positiveattitudes toward learning and education (i.e., to foster self-regulated learning, cf. Deci & Ryan, 1985; Ryan & Deci, 2000).Based on these assumptions, Lorenz andWild (2007) proposed fourdimensions of PI: (1) Autonomy supportive practices refer to theencouragement of self-initiated learning activities, for instance bymeans of scaffolding or contingent shift rule (Pratt, Green, MacViar,& Bountrogianni, 1992). (2) Achievement-oriented control and pres-sure are manifested as controlling teaching strategies, the exertionof excessive pressure on children to complete assignments,extrinsic performance-contingent rewards, or direct instructionsthat undermine intrinsic motivation and autonomous behavior. It isassumed that achievement-oriented pressure and control is onlysuited to a limited extent and is otherwise detrimental to learningand achievement (cf. Levpuscek & Zupancic, 2009; Niggli et al.,2007; Pomerantz & Eaton, 2001; Rogers et al., 2009). (3) Structureis related to the parents’ organization of the environment in orderto provide clear and consistent guidelines, rules, and expectationsregarding learning and school work. According to Lorenz and Wild(2007), a high degree of structure can support achievement ifchildren are able to differentiate between parental structure andcontrol; otherwise, it can also undermine autonomy and therebyhave negative effects on achievement. Finally, (4) emotionalresponsivity describes the parents’ readiness to acknowledge thechild’s feelings associated with learning but also to spend conso-lation and to encourage the student in failure situations. Ina longitudinal study, Lorenz and Wild (2007) showed that thesefour components represented distinct but intercorrelated dimen-sions of PI that were related to student motivation in the domain ofMath (see also Knollmann & Wild, 2007).

In summary, general cognitive ability is the strongest predictorfor academic achievement (e.g., Deary et al., 2007; Gottfredson,2002; Gustafsson & Undheim, 1996; Humphreys & Stark, 2002;Neisser et al., 1996) but multiple dimensions of PI may as wellcontribute to academic success (Fan & Chen, 2001; Hill & Tyson,2009). Surprisingly, to our knowledge no empirical study hasdirectly assessed their relative predictive value. Although there isno evidence for a close association between both variables, incre-mental validity of PI over cognitive abilities would strengthen theimportance of parental behavior for their children’s achievement inschool. Therefore, our aimwas to investigate the predictive value ofPI for academic achievement in early adolescence after generalcognitive ability had been accounted for.

Based on previous findings, we expected cognitive ability to bethe strongest predictor for achievement (cf. Dearyet al., 2007; Fraseret al., 1987; Gottfredson, 2002; Humphreys & Stark, 2002; Kuncelet al., 2004; Laidra et al., 2007; Rhode & Thompson, 2007)(Hypothesis 1). In terms of PI, we predicted that autonomy sup-porting practices (Hypothesis 2) and emotional responsivity(Hypothesis 3) should have positive effects on achievement (cf. Hill& Tyson, 2009), while achievement-oriented control should havedetrimental effects on academic performance (cf. Levpuscek &Zupancic, 2009; Niggli et al., 2007; Pomerantz & Eaton, 2001;Rogers et al., 2009) (Hypothesis 4). Finally, parental behaviorproviding structure e if not perceived as controlling behavior e

should be positively related to achievement (cf. Lorenz & Wild,2007). Nonetheless, considering that our focus was on adoles-cents, it seemed likely that students would regard rules andguidelines as restrictions of their autonomy and self-determination,in which case parental structure may also be associated with lowerachievement scores (Hypothesis 5).

2. Method

2.1. Participants

The sample investigated in this study was part of the twinstudy on Cognitive Ability, Self-reported Motivation, and SchoolAchievement (CoSMoS, blinded reference) that focused on theprediction of academic performance by means of individualand environmental variables in a genetically sensitive design. Thepresent investigation is based on data from 334 twins that partic-ipated in the CoSMoS study in 2009 (mean age ¼ 12.4, SD ¼ .9;age range ¼ 10e14 years; 50.6% female). Most of them (64%)attended university-track secondary school (levels I þ II; German‘Gymnasium’), 24% were in secondary school (level I; German‘Realschule’), 3% in vocational-track secondary general school (levelI; German ‘Hauptschule’), and 9% were enrolled in comprehensiveschool. The sample predominantly consisted of middle-classfamilies (in 58% of the families at least one parent held a generalqualification for university entrance and in 34% of the families atleast one parent held a university degree). Most childrenwere fromintact two-parent families (85% of the parents were married orshared a residence).

2.2. Procedure

Participants performed the tests at their homes. Parentsprovided written informed consent, demographic information, andschool grades. The children completed a questionnaire assessingchild-perceived PI. Prior to testing, the families were given theopportunity to choose whether they preferred to complete thequestionnaire in a paper-and-pencil version (in which case thematerial was mailed to their home address; 14.35% of the partici-pants) or whether they preferred a computerized online version ofthe questionnaires (which they completed using their homecomputer; 85.65% of the participants). Regarding the assessment ofcognitive abilities, the procedure was different: Given that thesetests are time sensitive and because they include specific start andstop criteria, trained experimenters administered them over thephone (cf. Gottschling et al., 2012). This procedure that has previ-ously been shown to yield adequate internal consistencies and topredict general cognitive ability measured in face-to-face testingsituations in elementary school children (Petrill, Rempell, Dale,Oliver, & Plomin, 2002), adolescents (Kent & Plomin, 1987), andadults (Legree, Fischl, Gade, & Wilson, 1998). Prior to the assess-ments, the test sheets were mailed to the families in sealed enve-lopes but were not to be opened before the experimenter called.

J. Karbach et al. / Learning and Instruction 23 (2013) 43e5146

2.3. Measures

2.3.1. Perceived parental involvementChild-perceived PI was assessed with a 21-item questionnaire

modified from Wild and colleagues (e.g., Lorenz & Wild, 2007) andbased on the Children’s Perceptions of Parents Scale (Grolnik, Ryan,& Deci, 1991). Given that Lorenz and Wild (2007) had investigatedinvolvement only in the domain of Math and because the presentstudy also included German as a domain of school achievement,several items specifically referring to “grades/performance inMath” were rephrased so that they referred to “grades/perfor-mance in school”. The 21 items served as proximal indicators for PI,that is, they referred to the child’s learning context at home. Chil-dren were to respond to each one of them on a four-point Likert-type response scale ranging from (1) I don’t agree at all to (4) Icompletely agree. The following four dimensions of PI wereassessed: (1) Autonomy supportive practices (six items, e.g., “Whenmy parents help me with my school work, they always encourageme to find the correct answer bymyself.”), (2) achievement-orientedcontrol (three items, e.g., “When I get bad grades, my parents scoldme and tell me to study more.”), (3) structure (six items, e.g., “Iknow exactly how good my parents expect me to do in school.”),and (4) emotional responsivity (six items, e.g., “My parentsencourageme and help me if I have problems at school.”). Thus, thePI questionnaire applied in this study included a total of 21 items(in contrast to the one applied by Lorenz and Wild (2007), whichincluded five items for autonomy supportive practices, six items forachievement oriented pressure, five items for structure and threeitems for emotional responsivity, yielding a total of 19 items).

2.3.2. General cognitive abilityWe assessed general cognitive ability (g) by means of two

adapted verbal subtests (vocabulary, 25 items, and general knowl-edge, 18 items) and two adapted nonverbal subtests (figural classifi-cation and figural reasoning, 25 items, respectively) from theGermanCognitive Ability Test (KFT 4-12þR; Heller & Perleth, 2000) and theWechsler Intelligence Scale for Children (WISC-III; Wechsler, 1991).

2.3.3. Academic achievementSchool grades served as the indicator for academic achievement.

Parents provided mid-term and full-term grades for Math andGerman from 2008 and the mid-term grades from 2009. All gradeswere initially coded according to the German grade system, rangingfrom (1) very good to (6) insufficient. In order to facilitate theinterpretation of the data analyzed in this study, we reversed thecoding of the grades (i.e., higher values represent better perfor-mance). Correlations between the grades at the three test occasionswere high (German: r ¼ .59e.71; Math: r ¼ .63e.73).

2.3.4. Control variableWe included the level of parental education (i.e., the highest

educational degree held both by the mother and the father;1 ¼ none, 2 ¼ vocational-track secondary school, 3 ¼ secondaryschool attendance without degree, 4 ¼ secondary school,5 ¼ university-track secondary school attendance without degree,6 ¼ university-track secondary school, 7 ¼ university attendancewithout degree, 8 ¼ university degree) into the model to accountfor differences in children’s academic achievement that may be dueto parental education rather than PI (cf. Jimerson et al., 1999;Neisser et al., 1996; Stevenson & Baker, 1987).

2.4. Data analysis

In order to investigate the predictive value of perceived PI foracademic success in the domains of German and Math, we chose

a latent variable approach using structural equationmodeling. Priorto model fitting, we carefully analyzed the missing data patterns.Little’s MCAR test (missing completely at random; Little & Rubin,2002) indicated that the missing data in this study occurredcompletely at random (all p’s > .05). We therefore imputed missingdata using the expectation-maximization (EM) algorithm in SPSS18. Imputations were restricted to the four dimensions of parentalinvolvement. In addition, raw scores were corrected for age and sexeffects by means of a regression procedure.

Structural equation modeling was performed with the softwarepackage MPlus (Muthén & Muthén, 2004). Given that the power ofthe chi-square test severely depends on the sample size and thesize of the correlations (Byrne, 2001; Kline, 2010), it has beensuggested to rely on the chi-square/degrees of freedom ratioinstead of the chi-square p-value. This index should be smaller thantwo (Schermelleh-Engel, Mossbrugger, &Müller, 2003). In addition,we report the Comparative Fit Index (CFI) and the Root MeanSquare Error of Approximation (RMSEA) as measures of incre-mental fit. For the CFI, values larger than .90 are desirable (Bentler,1989) and the RSMEA should be smaller than .06 (Hu & Bentler,1999).

Since our data consisted of twin pairs, it may be argued thatstandard errors were underestimated in our sample (McGue,Wette,& Rao, 1984). In order to adjust for this potential bias, we appliedaggregated analysis under complex sampling according to Muthénand Satorra (1995). This approach assumes that the twin pairs areindependent and relies on themeans of the values for the membersof each pair. Thus, the variances (and standard errors) of theparameter estimates can be calculated by using stratified randomsampling formulae (cf. Johnson, Hicks, McGue, & Iacono, 2007).

3. Results

Means, standard deviations, and reliability as well as intercor-relations of PI, general cognitive ability, level of parental education,and academic achievement in the domains of German andMath areshown in Table 1. In order to test the factorial validity of the PImodel (cf. Lorenz & Wild, 2007), we first computed a confirmatoryfactor analysis (CFA). Second, we performed structural equationmodeling in order to predict academic achievement by PI andgeneral cognitive ability. In order to control for effects of parentaleducation, we also considered the highest degree held by bothfathers and mothers.

3.1. Confirmatory factor analysis for parental involvement

The CFA was based on the 21 items serving as indicators for thefour dimensions of PI (autonomy support, emotional responsivity,achievement-oriented control, and structure). Because of thesubstantial correlations between autonomy support and emotionalresponsivity (r ¼ .66, p < .001) on the one hand and achievement-oriented control and structure (r ¼ .50, p < .001) on the other hand,we specified a model with four first-order factors (autonomysupport, emotional responsivity, structure, and achievement-oriented control) and two second-order factors (support &responsivity and control & structure). The CFA yielded an accept-able model fit (c2 ¼ 186.94, df ¼ 53, p < .01, c2/df ¼ 3.53; CFI ¼ .91;RMSEA¼ .09, CI¼ [.07; .10]). All factor loadings were significant (allp’s < .01) and we found a negative correlation between the twosecond-order factors (r ¼ �.64, p < .001).

3.2. Prediction of academic achievement

For each one of the achievement domains (Math and German),we specified a model including the four first-order factors of PI

Table 1Means (M), standard deviations (SD), reliability (Cronbach’s a), and correlations between cognitive abilities, the level of parental education, parental involvement (autonomysupport, achievement-oriented control, structure, emotional responsivity), and academic achievement (German, Math).

Descriptives Correlations

PE Parental involvement Achievement

M SD a AUT RES STR CON GER MAT

Cognitive abilities .00 .72 e .29** �.02 .02 �.02 �.05 .38** .38**Parental education (PE) 5.04 1.85 e e .00 .02 .01 .04 .17** .21**Autonomy support (AUT) 3.41 .46 .80 e .66** �.06 �.40** .15** .07Emotional responsivity (RES) 3.37 .45 .83 e �.01 �.37** .06 .02Structure (STR) 2.59 .51 .74 e .50** �.10 .00Achievement-oriented control (CON) 1.89 .78 .79 e �.25** �.20**German achievement (GER) 4.29 .72 .85 e .53**Math achievement (MAT) 4.24 .85 .86 e

Note. N ¼ 334. The scales for parental involvement (autonomy support, achievement related control, structure, emotional responsivity) ranged from 1 to 4 with 1 indicatinghigher perceived involvement. The scale for the grades ranged from 1 (insufficient) to 6 (very good). The scale for parental education referred to the highest educational degreeheld by the mother and the father and ranged from 1 (no degree) to 8 (university degree). Correlations refer to bivariate correlations on the manifest level.**p < .01 (two-tailed).

J. Karbach et al. / Learning and Instruction 23 (2013) 43e51 47

(autonomy support, emotional responsivity, achievement-orientedcontrol, and structure), the two second-order factors of PI (support& responsivity and control & structure), general cognitive ability,parental education, and the dependent variable (achievement) (seeFigs. 1 and 2). For this more complex model, we parceled the PIitems according to the item-to-construct balance technique (Little,Cunningham, Shahar, & Widaman, 2002), resulting in three two-item parcels for autonomy support, emotional responsivity, andstructure. The standardized scores in the two verbal and the twononverbal tests served as indicators for general cognitive ability,and the highest educational degree of both mother and father asindicators for the level of parental education. In the full model, norestrictions were applied. We compared results of the full model tonested models by successively fixing non-significant paths to zero.The comparisonwas based on the criterion of parsimony, fit indices,and c2 difference tests.

Fig. 1. Latent variable model for the prediction of German academic achievement. Cogniticlassification, nonverb2 ¼ figural reasoning. Academic achievement: Grade1 ¼ domaingrade3 ¼ domain-specific mid-term grade 2009. e1ee26: Error terms. Dashed lines indicat

3.2.1. German achievementThe full model provided a good fit to the data (c2 ¼ 324.46,

df ¼ 182, p < .01, c2/df ¼ 1.78, CFI ¼ .94; RMSEA ¼ .05, CI ¼ [.04;.06]), but the paths for support & responsivity and for the level ofparental education failed to reach significance in the full model(p ¼ .54 and p¼ .94, respectively). The most parsimonious model isshown in Fig. 1. This model did not fit significantly worse than thefull model (c2diff ¼ .11, dfdiff ¼ 2, p¼ .85). In sum, the latent variablesexplained 34% of the variance in German achievement. Whilecognitive ability had the strongest predictive value (b ¼ .52),control & structure (b ¼ �.26) also had a significant influence onGerman achievement.

3.2.2. Math achievementFor Math achievement, the full model also yielded a good fit to

the data (c2 ¼ 336.33, df ¼ 182, p < .01, c2/df ¼ 1.84, CFI ¼ .93,

ve ability tests: Verb1 ¼ vocabulary, verb2 ¼ general knowledge, nonverb1 ¼ figural-specific mid-term grade 2008, grade2 ¼ domain-specific end-term grade 2008,e paths with insignificant coefficients.

Fig. 2. Latent variable model for the prediction of Math academic achievement. Cognitive ability tests: Verb1 ¼ vocabulary, verb2 ¼ general knowledge, nonverb1 ¼ figuralclassification, nonverb2 ¼ figural reasoning. Academic achievement: Grade1 ¼ domain-specific mid-term grade 2008, grade2 ¼ domain-specific end-term grade 2008,grade3 ¼ domain-specific mid-term grade 2009. e1ee26: Error terms. Dashed lines indicate paths with insignificant coefficients.

J. Karbach et al. / Learning and Instruction 23 (2013) 43e5148

RMSEA ¼ .05, CI ¼ [.04; .06]). Consistent with German achieve-ment, the paths for support & responsivity and for the level ofparental education were not significant (p ¼ .66 and p ¼ .53,respectively). The reduced model, which did not fit significantlyworse than the full model (c2diff ¼ .11, dfdiff ¼ 2, p ¼ .76), is shown inFig. 2. It explained 33% of the total variance in Math achievement.Again, cognitive ability was the strongest predictor for academicachievement (b ¼ .54) and we also found a significant effect forcontrol and structure (b ¼ �.20).

4. Discussion

The purpose of this study was to examine the incremental val-idity of perceived PI in the prediction of academic achievementover general cognitive ability. We investigated school achievementin the domains of German and Math in a sample of 334 adolescentsbetween the ages of 10 and 14. The assessment of PI included thedimensions autonomy supporting practices and emotionalresponsivity as well as achievement-oriented control and structure.Our findings replicated previous research on the relation betweengeneral cognitive ability and achievement, and more importantly,extended prior knowledge regarding the role of PI for academicsuccess after cognitive abilities had been controlled for.

4.1. General cognitive ability and academic achievement

Consistent with a variety of previous findings, general cognitiveability was the strongest predictor for both German and Mathachievement (Hypothesis 1). It explained 27% of the total variance inGerman achievement and 29% of the total variance in Mathachievement, which is completely consistent with results previ-ously reported in the literature (e.g., Deary et al., 2007; Gottfredson,2002; Gustafsson & Undheim, 1966; Humphreys & Stark, 2002;Neisser et al., 1996; Rhode & Thompson, 2007). Also in line withprior evidence (e.g., Neisser et al., 1996; Stevenson & Baker, 1987),

wefoundanassociationbetween the level of parental education andtheir children’s academic performance (see Table 1). It is widelyacknowledged that children of highly educated parents strive tomeet their parents’ expectations regarding their success in schooland to reach the same educational level held by their parents. Inaddition, these parents may have the intellectual and financialresources to provide stimulating and supporting learning environ-ments for their children (Bronstein & Bradley, 2003). Interestingly,thedirect associationbetweenparental education and school gradesdisappeared when we predicted achievement in a more complexmodel allowing correlations between cognitive ability and the levelof parental education. Instead, we found an indirect effect ofparental education via general cognitive ability (see Figs. 1 and 2).

However, the main focus of this study was to investigate theinfluence of PI on academic success. Since we know that achieve-ment in school is of major importance for adolescents’ furthereducation and career opportunities (e.g., Jimerson et al., 1999;Levpuscek & Zupancic, 2009; Schmidt & Hunter, 1998; Williamsonet al., 1991), it seems particularly important to identify variablesthat have the potential to increase academic success and to explainvariance in achievement unaccounted for by intelligence.

4.2. Parental involvement and academic achievement

The present study was the first one to analyze the influence ofperceived PI on academic success in early adolescence whilecontrolling for general cognitive ability. Although a number ofprevious studies have shown that high levels of PI can havesubstantial effects on school achievement (Fan & Chen, 2001; Hill &Tyson, 2009), they also indicated that it is not only of high relevancehow much parents are involved, but also what kind of involvementthey engage in. The present study focused on two intercorrelateddomainsof PI, namelyautonomysupport andemotional responsivityon the one hand and achievement-oriented control and structure onthe other hand. According to Lorenz and Wild (2007), autonomy

J. Karbach et al. / Learning and Instruction 23 (2013) 43e51 49

support and emotional responsivity are types of PI that generallyrefer to parents’ willingness and ability to take their children’sperspective and respond to their needs. In contrast, control andstructure refer to excessive control and pressure on children tocomplete assignments as well as to clear and consistent guidelines,rules, and expectations regarding learning and school work. Therelation between achievement-oriented pressure and structureappears to depend on the way children perceive parental rules andguidelines (Lorenz &Wild, 2007). In the present study, childrenwereapparently not able to differentiate between structure and control(r ¼ .50). Although this effect may be specific for the particular agegroup investigated in this study (in the sense that adolescents arelikely to regard rules and guidelines as restrictions of theirautonomy), it also suggests that a refinement of the items applied inthis study may be needed in order to clearly set the concepts ofperceived controlling behavior and supporting structure apart.Nevertheless, as itwas, the dimensions assessed in the present studyreflected two theoretically opposite dimensions of PI (i.e., support &responsivity vs. control & structure) (r ¼ �.64). Importantly, thisconceptualization of PI allowed the simultaneous assessment ofqualitatively different types of PI, their reciprocal relations, and theirrelative contribution to domain-specific achievement (cf. Hill &Taylor, 2004).

Our data revealed that for both Math and German achievement,control and structure were the most powerful dimensions of PIexplaining variance over general cognitive ability (7% for Germanand 4% for Math achievement) (Hypotheses 4 and 5). That is, higherlevels of control and structure were associated with pooreracademic achievement, which is consistent with a number ofprevious findings (e.g., Koutsoulis & Campbell, 2001; Levpuscek &Zupancic, 2009; Niggli et al., 2007; Pomerantz & Eaton, 2001;Rogers et al., 2009). The fact that certain types of PI e even if theyare well meant e may be detrimental to academic success againillustrates the importance of understanding the differentialconsequences of qualitatively different types of involvement.

In the literature, at least two possibleways to explain this findingare discussed. On the one hand, one may expect an indirect influ-ence of achievement-oriented control and pressure on academicsuccess that is mediated by other variables, such as self-efficacy ormotivation (e.g., Levpuscek & Zupancic, 2009; Rogers et al., 2009).That is, perceived parental distrust, criticism, and punishment maybe detrimental to the child’s beliefs about his or her own ability tolearn or to perform certain tasks. As a consequence, negative self-efficacy affects academic performance. On the other hand, parentsmay bemore likely to assert control and provide structure regardingtheir children’s academic development if the children have troublelearning and performing in school. Thus, the increased parentalinvolvement may rather be a response to than a predictor of theirchild’s academic failure (cf. Campbell &Mandel, 1990; Grolnik et al.,1991; Levpuscek & Zupancic, 2009; see also Silinskas, Leppaenen,Aunola, Parilla, & Nurmi, 2010). While the cross-sectional findingsof the present study are generally consistent with both accounts,they certainly call for future longitudinal research in order to clarifythe nature of the relation between achievement-oriented control/structure and academic success.

In contrast to our expectations, we found no influence ofautonomy and responsivity on academic success after cognitiveability had been accounted for (see Figs. 1 and 2) (Hypotheses 2 and3). Furthermore, there were only very small or insignificant mani-fest correlations between autonomy support or emotional respon-sivity and either one of the achievement variables (see Table 1).These findings are in contrast to a number of previous findingsindicating that parental academic support and responsivity hadbeneficial effects on school achievement (e.g., Hill & Tyson, 2009;Koutsoulis & Campbell, 2001; Rogers et al., 2009). In these studies, it

was expected that supporting and encouraging styles of involve-ment provide the students with a sense of initiative and confidenceregarding their academic development. However, the presentresults suggest that particularly in adolescence, highly supportiveparents may have been perceived as overprotective and interferingwith the student’s need for autonomy (cf. Levpuscek & Zupancic,2009). As a result, the student’s academic initiative and persis-tence may have decreased (Ginsburg & Bronstein, 1993) and thepositive effect of parental support dissolved. This interpretation issupported by the fact that the children in our study perceived theirparents as highly supportive (M ¼ 3.41 on a scale from 1 to 4) andemotionally responsive (M ¼ 3.37 on a scale from 1 to 4). Finally, itshould be noted that the majority of previous studies pointing topositive effects was not based on German samples, suggesting thatdifferences with respect to the educational system or culture-specific differences in parental attitudes toward their children’sacademic development may also have contributed to differences inempirical findings. In addition, previous studies have applieda number of differentmeasures for parental involvement, a fact thatprobably also hampers their comparability to the results of thepresent investigation.

Thus, achievement-oriented PI appears to be a type of parentalbehavior that can be directly related to children’s performance inschool, indicating that involvement continues to make a differencein secondary school despite the popular stereotype that adoles-cents might be impervious to the influence of their parents(Steinberg, Lamborn, Dornbusch, & Darling, 1992). It is alsoconsistent with the assumption that in contrast to elementaryschool, home-based involvement may be more relevant thanschool-based involvement at this age (cf. Hill & Tyson, 2009;Seginer, 2006). Importantly, the influence of PI on achievementpertained to a negative association between control/structure andacademic success (Hypothesis 4 and 5). This pattern, which wasconsistent across the domains of German and Math, suggests thatparents may not be aware of the disadvantageous effects of theircontrolling, punishing, and pressuring behavior. It therefore seemsessential to provide appropriate information for parents addressingthe effects of their parenting practices or to implement traininginterventions that help analyzing and improving parentechildinteractions in the context of academic work at home.

4.3. Limitations and outlook

Although the present study yielded important new findingshighlighting the importance of PI for school achievement in earlyadolescence, several limitations may be addressed in future studies.In contrast to the majority of previous studies on PI, we did notanalyze parents’ self-reported behavior. Instead, we focused onchild-perceived PI, reflecting the parental behavior toward theacademicdevelopment fromthe child’s perspective.However, itmayalso beworth consideringother sources of information (e.g., parent’sself-reports, teacher ratings) and to address other relevant factors inorder to provide amore comprehensive assessment of the children’sindividual characteristics and family background. This aspect seemsparticularly important in the light of recent findings indicating thatthe influence of PI may also be moderated by the student’s self-efficacy (Levpuscek & Zupancic, 2009), their self-concept (Rogerset al., 2009), and the parenting style (Steinberg et al., 1992).

In addition, future studies may want to investigate whether thepresent findings are valid across different age groups and insamples that are more diverse in terms of SES and cultural back-ground. In terms of cognitive abilities, it may also be worthincluding measures of basic cognitive processing, such as workingmemory capacity, executive control, and attention that may be lessstrongly related to SES than measures of general intelligence.

J. Karbach et al. / Learning and Instruction 23 (2013) 43e5150

Finally, longitudinal data are needed in order to determine thecausal direction of the relationship between PI and achievementand to test for reciprocal effects. Analyzing data of at least two testoccasions would allow addressing the question, whether, forinstance, a high level of parental control and pressure to do well inschool is a reaction to the child’s academic failure or whether itactually triggers poor academic performance. Although Steinberget al. (1992) reported data indicating that academic performancefollows from (and does not precede) PI, it will be a key issue forfuture studies to shed more light on this issue.

4.4. Conclusion

In sum, the results of the present study added to previousevidence indicating that general cognitive ability is a very reliablepredictor for academic performance in early adolescence. However,our data also shows that the specific dimensions of home-basedparental behavior with respect to their children’s academic devel-opment are a major influence on achievement both in the domainsof Math and German. Specifically, we found that children performworse in school if their parents engage in high levels of controllingbehavior and if they tend to set strict rules, guidelines and expec-tations. These findings provide essential insights into the differ-ential effects of parental behavior on their children’s academicdevelopment, which is particularly important in order to provideappropriate counseling and to design effective interventions (cf.Rogers et al., 2009).

Acknowledgment

The authors received no financial support for the research,authorship, and/or publication of this article.

References

Adams, G. R., & Berzonsky, M. D. (2003). Blackwell handbook of adolescence. Malden,MA: Blackwell.

Bentler, P. M. (1989). Fit indexes, lagrange multipliers, constraint changes,and incomplete data in structural models. Multivariate Behavioral Research, 25,163e172. http://dx.doi.org/10.1207/s15327906mbr2502_3.

Bronstein, M. H., & Bradley, R. H. (2003). Socioeconomic status, parenting, and childdevelopment. Mahwah: Lawrence Erlbaum Associates Publishers.

Byrne, B. M. (2001). Structural equation modeling with AMOS, basic concepts,applications, and programming. Hillsdale, New Jersey: Lawrence ErlbaumAssociates.

Byrnes, J. P., Miller, D. C., & Reynolds, M. (1999). Learning to make good decisions:a self-regulation perspective. Child Development, 70, 1121e1140. http://dx.doi.org/10.1111/1467-8624.00082.

Campbell, J. R., & Mandel, F. (1990). Connecting math achievement to parentalinfluences. Contemporary Educational Psychology, 15, 64e74. http://dx.doi.org/10.1016/0361-476X(90)90006-M.

Comer, J. P. (1995). School power: Implications of an intervention project. New York:Free Press.

Cooper, H. (1989). Homework. New York: Longman.Deary, I. J., Strand, S., Smith, P., & Fernandez, C. (2007). Intelligence and educational

achievement. Intelligence, 35, 13e21. http://dx.doi.org/10.1016/j.intell.2006.02.001.

Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in humanbehavior. New York: Plenum Press.

Desforges, C., & Abouchaar, A. (2003). The impact of parental involvement, parentalsupport, and family education pupil achievements and adjustment: A literaturereview, (Research Report 443). Nottingham: DfES Publications.

El Nokali, N. E., Bachman, H. J., & Votruba-Drzal, E. (2010). Parent involvement andchildren’s academic and social development in elementary school. ChildDevelopment, 81, 988e1005. http://dx.doi.org/10.1111/j.1467-8624.2010.01447.x.

Epstein, J. L. (1987). Toward a theory of familyeschool connections: teacher prac-tices and parent involvement. In K. Hurrelman, F. X. Kaufman, & F. Losel (Eds.),Social intervention: Potential and constraints (pp. 121e136). Berlin: de Gruyer.

Fan, X., & Chen, M. (2001). Parental involvement and students’ academic achieve-ment: a meta-analysis. Educational Psychology Review, 13, 1e22. http://dx.doi.org/10.1023/A:1009048817385.

Fraser, B. J., Walberg, H. J., Welch, W. W., & Hattie, J. A. (1987). Syntheses ofeducational productivity research. International Journal of Educational Research,11, 145e252. http://dx.doi.org/10.1016/0883-0355(87)90035-8.

Gathercole, S. E., Pickering, S. J., Knight, C., & Stegmann, Z. (2004). Working memoryskills and educational attainment: evidence from national curriculum assess-ments at 7 and 14 years of age. Applied Cognitive Psychology, 18, 1e16. http://dx.doi.org/10.1002/acp.934.

Ginsburg, G. S., & Bronstein, P. (1993). Family factors related to children’s intrinsic/extrinsic motivational orientation and academic performance. Child Develop-ment, 64, 1461e1474. http://dx.doi.org/10.2307/1131546.

Gonzales-DeHass, A. R., Willems, P. P., & Holbein, M. F. (2005). Examining the rela-tionship between parental involvement and student motivation. EducationalPsychology Review, 17, 99e123. http://dx.doi.org/10.1007/s10648-005-3949-7.

Gottfredson, L. S. (2002). Where and why g matters: not a mystery. HumanPerformance, 15, 25e46. http://dx.doi.org/10.1207/S15327043HUP1501&02_03.

Gottfried, A. E. (1985). Academic intrinsic motivation in young elementary schoolchildren and junior high school students. Journal of Educational Psychology, 77,631e645. http://dx.doi.org/10.1037/0022-0663.77.6.631.

Gottschling, J., Spengler, M., Spinath, B., & Spinath, F. M. (2012). The prediction ofschool achievement from a behavior genetic perspective: results from theGerman twin study on Cognitive Ability, Self-Reported Motivation, and SchoolAchievement (CoSMoS). Personality and Individual Differences, 53, 381e386.http://dx.doi.org/10.1016/j.paid.2012.01.020.

Greven, C. U., Harlaar, N., Kovas, Y., Chamorro-Premuzic, T., & Plomin, R. (2009).More than just IQ. School achievement is predicted by self-perceived abilities ebut for genetic rather than environmental reasons. Psychological Science, 20,753e762. http://dx.doi.org/10.1111/j.1467-9280.2009.02366.x.

Grolnick, W. S., & Slowiaczek, M. L. (1994). Parents’ involvement in children’sschooling: a multidimensional conceptualization and motivational model. ChildDevelopment, 65, 237e252. http://dx.doi.org/10.2307/1131378.

Grolnik, W. S., Ryan, R. M., & Deci, E. L. (1991). Inner resources for schoolachievement: motivational mediators of children’s perceptions of their parents.Journal of Educational Psychology, 83, 508e517. http://dx.doi.org/10.1037/0022-0663.83.4.508.

Gustafsson, J.-E., & Undheim, J. O. (1996). Individual differences in cognitive func-tions. In D. C. Berliner, & R. C. Calfee (Eds.), Handbook of educational psychology(pp. 186e242). New York: Prentice Hall International.

Heller, K. A., & Perleth, C. (2000). Kognitiver Fähigkeitstest für 4. bis 12. Klassen,Revision (KFT4-12þR). Weinheim: Beltz.

Hill, N. E., & Taylor, L. C. (2004). Parental school involvement and children’sacademic achievement: pragmatics and issues. Current Directions in Psycho-logical Science, 13, 161e164. http://dx.doi.org/10.1111/j.0963-7214.2004.00298.x.

Hill, N. E., & Tyson, D. F. (2009). Parental involvement in middle school: a meta-analytic assessment of the strategies that promote achievement. Develop-mental Psychology, 45, 740e763. http://dx.doi.org/10.1037/a0015362.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structureanalysis: conventional criteria versus new alternatives. Structural EquationModeling, 6, 1e55. http://dx.doi.org/10.1080/10705519909540118.

Humphreys, L. G., & Stark, S. (2002). General intelligence: measurement, correlates,and interpretations of the culture-genetic construct. In R. J. Sternberg, &E. Grigorenko (Eds.), The general factor of human intelligence. How general is it?(pp. 87e115). Mahwah: Erlbaum.

Jensen, A. R. (1993). Psychometric g and achievement. In R. Gifford (Ed.), Policyperspectives on educational testing (pp. 117e227). Boston: Kluwer Academic.

Jeynes, W. H. (2007). The relationship between parental involvement and urbansecondary school student academic achievement. A meta-analysis. UrbanEducation, 42, 82e110. http://dx.doi.org/10.1177/0042085906293818.

Jimerson, S., Egeland, B., & Teo, A. (1999). A longitudinal study of achievementtrajectories: factors associated with change. Journal of Educational Psychology,91, 116e126. http://dx.doi.org/10.1037/0022-0663.91.1.116.

Johnson, W., Hicks, B. M., McGue, M., & Iacono, W. G. (2007). Most girls are alright,but some aren’t: personality trajectory groups from ages 14 to 24 and someassociations with outcomes. Journal of Personality and Social Psychology, 93,266e284. http://dx.doi.org/10.1037/0022-3514.93.2.266.

Kent, J., & Plomin, R. (1987). Testing specific cognitive abilities by telephone andmail.Intelligence, 11, 391e400. http://dx.doi.org/10.1016/0160-2896(87)90019-5.

Kline, R. B. (2010). Principles and practice of structural equation modeling. New York:Guilford Press.

Knollmann, M., & Wild, E. (2007). Quality of parental support and student’semotions during homework: moderating effects of students’ motivationalorientations. European Journal of Psychology of Education, 22, 63e76.

Koutsoulis, M. K., & Campbell, J. R. (2001). Family processes affect students’ moti-vation, and science and math achievement in Cypriot high schools. StructuralEquationModeling,1, 108e127. http://dx.doi.org/10.1207/S15328007SEM0801_6.

Kuncel, N. R., Hezlett, S. A., & Ones, D. S. (2004). Academic performance, careerpotential, creativity, and job performance: can one construct predict them all?Journal of Personality and Social Psychology, 86, 148e161. http://dx.doi.org/10.1037/0022-3514.86.1.148, Special Section, Cognitive Abilities: 100 Years afterSpearman (1904).

Laidra, K., Pullmann, H., & Allik, J. (2007). Personality and intelligence as predictorsof academic achievement: a cross-sectional study from elementary tosecondary school. Personality and Individual Differences, 42, 441e451. http://dx.doi.org/10.1016/j.paid.2006.08.001.

Legree, P. J., Fischl, M. A., Gade, P. A., & Wilson, M. (1998). Testing word knowledgeby telephone to estimate general cognitive aptitude using an adaptive test.Intelligence, 26, 91e98. http://dx.doi.org/10.1016/S0160-2896(99)80056-7.

Lerner, R. M., & Steinberg, L. (2004). Handbook of adolescent psychology. Hoboken:John Wiley & Sons.

J. Karbach et al. / Learning and Instruction 23 (2013) 43e51 51

Levpuscek, M. P., & Zupancic, L. (2009). Math achievement in early adolescence: therole of parental involvement, teacher’s behavior and student’s motivationalbeliefs about math. Journal of Early Adolescence, 29, 541e570. http://dx.doi.org/10.1177/0272431608324189.

Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002). To parcel or notto parcel: exploring the question, weighing the merits. Structural EquationModeling, 9, 151e173, doi:10.1.1.118.6289.

Little, R. J. A., & Rubin, D. B. (2002). Statistical analysis with missing data. Hoboken:John Wiley & Sons.

Lorenz, F., & Wild, E. (2007). Parental involvement in schooling e results concerningits structure and impact on students’ motivation. In M. Prenzel, & L. Allolio-Näcke (Eds.), Studies on the educational quality of schools. The final report on theDFG Priority Program (pp. 299e316). Münster: Waxmann.

Lu, L., Weber, H. S., Spinath, F. M., & Shi, J. (2011). Predicting school achievementfrom cognitive and non-cognitive variables in a Chinese sample of elementaryschool children. Intelligence, 39, 130e140. http://dx.doi.org/10.1016/j.intell.2011.02.002.

Maegi, K., Lerkkanen, M.-K., Poikkeus, A.-M., Rasku-Puttonen, H., & Nurmi, J.-E.(2011). The cross-lagged relations between children’s academic skill develop-ment, task-avoidance, and parental beliefs about success. Learning & Instruction,21, 664e675.

McGue, M., Wette, R., & Rao, D. C. (1984). Evaluation of path analysis throughcomputer simulation: effect of incorrectly assuming independent distributionof familial correlations. Genetic Epidemiology, 1, 255e269. http://dx.doi.org/10.1002/gepi.1370010305.

Muthén, L. K., & Muthén, B. O. (2004). Mplus: Statistical analysis with latent variables(version 3) [Computer software]. Los Angeles: Authors.

Muthén, B. O., & Satorra, A. (1995). Complex sample data in structural equationmodeling. Sociological Methodology, 25, 267e316.

Neisser, U., Boodoo, G., Bouchard, T. J., Boykin, A. W., Brody, N., Ceci, S. J., et al.(1996). Intelligence: knowns and unknowns. American Psychologist, 51, 77e101.http://dx.doi.org/10.1037/0003-066X.51.2.77.

Niggli, A., Trautwein, U., Schnyder, I., Ludtke, O., & Neumann, M. (2007). Parentalhomework support can be beneficial, but parental intrusion is detrimental:family background, parental homework supervision, and performance gains.Psychologie in Erziehung und Unterricht, 54, 1e14.

Petrill, S. A., Rempell, J., Dale, P. S., Oliver, B., & Plomin, R. (2002). Testing cognitiveabilities by telephone in a sample of 6-to 8-year olds. Intelligence, 30, 353e360.http://dx.doi.org/10.1016/S0160-2896(02)00087-9.

Pomerantz, E. M., & Eaton, M. M. (2001). Maternal intrusive support in the academiccontext: transactional socialization processes. Developmental Psychology, 37,174e186. http://dx.doi.org/10.1037/0012-1649.37.2.174.

Pratt, M. W., Green, D., MacVicar, J., & Bountrogianni, M. (1992). The mathematicalparent: parental scaffolding, parenting style, and learning outcomes in long-division mathematics homework. Journal of Applied Developmental Psychology,13, 17e34. http://dx.doi.org/10.1016/0193-3973(92)90003-Z.

Rhode, T. E., & Thompson, L. A. (2007). Predicting academic achievement with cogni-tive ability. Intelligence, 35, 83e92. http://dx.doi.org/10.1016/j.intell.2006.05.004.

Rogers, M. A., Theule, J., Ryan, B. A., Adams, G. R., & Keating. (2009). Parentalinvolvement and children’s school achievement: evidence for mediating

processes. Canadian Journal of School Psychology, 24, 34e57. http://dx.doi.org/10.1177/0829573508328445.

Ryan, R. M., & Deci, E. L. (2000). Self-determination theory and the facilitation ofintrinsic motivation, social development, and well-being. American Psychologist,55, 68e78. http://dx.doi.org/10.1037/0003-066X.55.1.68.

Sanders, M. G., & Epstein, J. L. (2000). Building schoolefamilyecommunity part-nerships in middle and high schools. In M. G. Sanders (Ed.), Schooling studentsplaced at risk: Research, policy, and practice in the education of poor and minorityadolescents (pp. 339e361). Mahwah, NJ: Erlbaum.

Schermelleh-Engel, K., Mossbrugger, H., & Müller, H. (2003). Evaluating the fit ofstructural equation models: tests of significance and descriptive goodness-of-fitmeasures. Methods of Psychological Research Online, 8, 23e74.

Schicke, M., & Fagan, T. K. (1994). Contributions of self-concept and intelligence tothe prediction of academic achievement among grade 4, 6, and 8 students.Canadian Journal of School Psychology, 10, 62e69.

Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods inpersonnel psychology: practical and theoretical implications of 85 years ofresearch findings. Psychological Bulletin, 124, 262e274. http://dx.doi.org/10.1037/0033-2909.124.2.262.

Seginer, R. (2006). Parents’ educational involvement: a developmental ecologicalperspective. Parenting: Science and Practice, 6, 1e48. http://dx.doi.org/10.1207/s15327922par0601_1.

Shumov, L., & Miller, J. D. (2001). Parents’ at-home and at-school academicinvolvement with young adolescents. Journal of Early Adolescence, 21, 68e91.http://dx.doi.org/10.1177/0272431601021001004.

Silinskas, G., Leppaenen, U., Aunola, K., Parilla, R., & Nurmi, J.-E. (2010). Predictors ofmothers’ and fathers’ teaching of reading and mathematics during kindergartenand grade 1. Learning & Instruction, 20, 61e71.

Son, S.-H., & Morrison, F. J. (2010). The nature and impact of changes in homelearning environment on development of language and academic skills inpreschool children. Developmental Psychology, 46, 1103e1118. http://dx.doi.org/10.1037/a0020065.

Spera, C. (2005). A review of the relationship among parenting practices, parentingstyles, and adolescent school achievement. Educational Psychology Review, 17,125e146. http://dx.doi.org/10.1007/s10648-005-3950-1.

Spinath, B., Spinath, F. M., Harlaar, N., & Plomin, R. (2006). Predicting schoolachievement from general cognitive ability, self-perceived ability, and intrinsicvalue. Intelligence, 34, 363e374. http://dx.doi.org/10.1016/j.intell.2005.11.004.

Steinberg, L., Lamborn, S. D., Dornbusch, S. M., & Darling, N. (1992). Impact ofparenting practices on adolescent achievement: authoritative parenting,school involvement, and encouragement to succeed. Child Development, 63,1266e1281. http://dx.doi.org/10.1111/1467-8624.ep9301210142.

Stevenson, D., & Baker, D. (1987). The family-school relation and the child’s schoolperformance. Child Development, 58, 1348e1357. http://dx.doi.org/10.1111/1467-8624.ep8591320.

Wechsler, D. (1991). Manual for the Wechsler intelligence scale for children (3rd ed.).San Antonio: The Psychological Corporation.

Williamson, G., Appelbaum, M., & Epanchin, A. (1991). Longitudinal analyses ofacademic achievement. Journal of Educational Measurement, 28, 61e76. http://dx.doi.org/10.1111/j.1745-3984.1991.tb00344.x.