Predicting early academic achievement: The role of higher- versus lower-order personality traits

The study explored the role of children’s (N = 193) individual differences and parental characteristics at the beginning of the first year of schooling in predicting students’ attainment of academic standards at the end of the year. Special attention was paid to children’s personality as perceived by the teachers’ assistants. Along with parents’ education, parenting practices and first-graders’ cognitive ability, the incremental predictive power of children’s higher-order (robust) personality traits was compared to the contribution of lower-order (specific) traits in explaining academic achievement. The specific traits provided a somewhat more accurate prediction than the robust traits. Unique contributions of maternal authoritative parenting, children’s cognitive ability, and personality to academic achievement were established. The ratings of first-graders’ conscientiousness (a higher-order trait) improved the prediction of academic achievement based on parenting and cognitive ability by 12%, whereas assistant teacher’s perceived children’s intelligence and low antagonism (lower-order traits) improved the prediction by 17%.

Identification and understanding of the role of students' individual differences in their academic achievement is an important concern of educationalists because knowledge of the relevant factors influencing academic performance has notable implications for learning and education (e.g., Chamorro-Premuzic & Furnham, 2003;O'Connor & Paunonen, 2007).Personality traits, enduring tendencies to feel, think, and act in a relatively consistent way over time and across situations (Funder, 2001) have been suggested to play a significant role in the development of knowledge for several reasons.They may direct individuals' choices and level of persistence to engage in intellectually stimulating activities and academic settings (Ackerman, 1996), affect certain Corresponding author: maja.zupancic@ff.uni-lj.sihabits associated with academic success (Rothstein, Paunonen, Rush, & King, 1994), and predict what the students will do rather than what they can do as reflected in ability test scores (Furnham & Chamorro-Premuzic, 2004).
Recent studies of the links between personality traits and scholastic achievement, which have been mainly investigated in adolescent and emerging adult students, generally operate under the framework of the Five Factor Model (FFM) of personality structure (O'Connor & Paunonen, 2007).The five personality factors (higher-order traits) of extraversion, agreeableness, conscientiousness, neuroticism, and openness reside at the highest level of personality trait hierarchy (McCrae & Costa, 1997).Although compelling evidence has been provided for a similar higher-order personality trait structure in children, the precise number of factors in child personality organization, the grouping of questionnaire items and patterns of factor loadings appear dependent on children's age, informants (parent-, teacher report), and assessment tools employed (e.g., Digman, 1990;Halverson et al., 2003;Knyazev, Zupančič, & Slobodskaya, 2008;Mervielde, DeFruyt, & Buyst, 1995;Mervielde & DeFruyt, 2002;Zupančič, Sočan, & Kavčič, 2009).Quite often, child personality structure appears less differentiated than adolescent/adult structure with conscientiousness and openness combined into a common factor (Mervielde et al., 1995;Zupančič & Kavčič, 2009).In general, extraversion in children includes lower-order traits such as activity, outgoingness, sociability, and positive emotionality.Agreeableness is characterized by kindness, cooperation, amiability, and pro-sociality.Conscientiousness is represented by self-discipline, organized behaviour, persistence, diligence, sustained attention, and achievement orientation.Neuroticism encompasses anxiety, fears, irritability, moodiness, shyness, and frequent experiences of insecurity.Openness refers to curiosity, tendency to explore new things, imagination, and informant perceived child intelligence.
Some of the FFM traits have been found better predictors of children's academic achievements than others.Conscientiousness and openness (or the combined factor) have been most consistently demonstrated as significant contributors (e.g., Barbaranelli, Caprara, Rabasca, & Pastorelli, 2003;Laidra, Pullmann, & Allik, 2007;Mervielde et al., 1995;Zupančič & Kavčič, 2007).Extraversion, agreeableness and low neuroticism also seem to predict higher grades in children according to a few studies (for a review see Laidra et al., 2007).However, only a few studies have examined the relationship between academic achievement and personality at the specific (lower-order) trait level and all of them have been performed with adolescents or emerging adults (Chamorro-Premuzic & Furnham, 2003;De Fruyt & Mervielde, 1996;Gray & Watson, 2002;Smrtnik Vitulić & Zupančič, 2010).Correlations of several traits with academic achievement were obtained, particularly of dutifulness, achievement orientation, and subjectively perceived intelligence.Also, anxiety, impulsiveness, sociability, activity level, and antagonism have been reported to be negatively associated with academic performance but the findings are not strongly consistent across these studies.An investigation of the predictive value of lower-order traits is important because it provides information about specific traits and processes that may explain the associations of higherorder traits with academic success.Considering lower-order traits may also improve the prediction of academic performance relative to the prediction based on higher-order traits as students with similar robust trait scores may have different specific trait scores.An examination of lower-order traits would thus reduce speculative interpretations about the predictive nature of higherorder traits, indicating which aspects of the robust traits are actually related to academic achievement (e.g., Chamorro-Premuzic & Furnham, 2003;O'Connor & Paunonen, 2007).
In the present study we focused on predictability of first-graders' teacher assessed academic achievement by both the higher-order personality traits and their constituent lower-order traits as perceived by the children's assistant teachers (teachers with a college degree in early childhood education, assisting school teachers in the first grade of Slovene obligatory schools).We also accounted for several demographic characteristics (child gender, attendance of pre-school, parental education), parenting practices, and children's cognitive ability in our predictive analyses.Along with previously documented links of these variables with academic achievement, the most recent studies show that children's cognitive ability (Grimm et al., 2010;Laidra et al., 2007), parenting practices (Bodovski & Youn, 2010;Junttila, Vauras, & Laakkonen, 2007), attendance of a quality pre-school prior to school entry (Vandell et al., 2010), parental education (Sektnan et al., 2010) and child gender (De Fruyt et al., 2008) are significantly associated with academic success.The academic success in our inquiry was defined in terms of students' attainment of performance standards (APS) at the end of the first grade because descriptive academic assessment is used over the first three grades of Slovene obligatory schooling (teachers' qualitative evaluations of students' progress).

PROBLEM AND NOVELTY OF THE RESEARCH
Our study extends previous research at least in three ways: (a) it examines predictive relations of relevant parental characteristics, children's cognitive ability and their personality with academic attainment (APS scores) jointly, (b) investigates the extent to which the children's personality trait ratings at the beginning of obligatory schooling explain unique variation in their APS at the end of the school year, over and beyond the prediction based on demographics, parenting and cognitive ability, and (c) explores the relation between teacher assessments of the first-graders' APS and assistant teacher ratings of their lowerorder personality traits along with the association between the APS and higherorder traits; this allows us to find out which of the specific trait constituents actually contribute to the students' APS, and whether the prediction of the criteria by personality is more accurate at the higher-or lower-order trait level.

Method
Participants.The participants were 193 first-graders (46% boys) attending one of 49 elementary schools in different regions of Slovenia.The children's age at their entry into school ranged from 5 years and 7 months to 6 years and 8 months (M = 72.2;SD = 2.7 months; data missing for 4 children); 77% of them attended government funded non-obligatory pre-schools prior to school entry.Data about the target children was also provided by their mothers and fathers, 70 school teachers and assistant teachers.Parental education ranged from completed elementary school (8 years) to masters degree with mean maternal and paternal education of 12.9 years (SD = 2.5) and 12.4 years (SD = 2.3), respectively.
Measures.An adult report form of the Inventory of Child Individual Differences (ICID; Halverson et al., 2003;Slovene version by Zupančič & Kavčič, 2009) was used to measure child personality as perceived by their assistant teachers.Written instructions indicated that the assistant teachers were to rate the extent to which each of 108 personality characteristics is expressed in the target child in comparison to the average same-age child.The informants were asked to respond to all of the items using a 7-point rating scale (1 = the characteristic is present in the child much less than in the average child or not at all; 7 = ...much more than in the average child).The items combine into 15 internally consistant mid-level scales (all αs over .62),representing lower-order (specific) personality traits.With Slovene children rated by teachers, the mid-level scales form four moderately intercorrelated robust dimensions, representing higher-order personality traits: extraversion (including Activity Level, Considerate, Positive Emotions, Open to Experience, and Sociable scales), conscientiousness (Achievement Orientation, Compliant, (un)Distractible, Intelligent, and Organized), disagreeableness (Antagonistic, Negative Affect, and Strong Willed), and neuroticism (Fearful/Insecure, and Shy).The dimensional scores show high internal consistencies (all αs over .83), a moderate to high stability over early childhood, and a high interrater agreement between teachers and assistant teachers (Zupančič & Kavčič, 2009).The ICID scores also show meaningful and differential relations to child temperament (Halverson et al., 2003) and social behaviour (Slobodskaya & Akhmetova, 2010;Zupančič & Kavčič, 2006).
Children's general cognitive ability was assessed by the Coloured Progressive Matrices (CPM; Raven, Raven, & Court, 1999).The test is standardized in a representative sample of Slovene children.It includes three sets of 12 incomplete patterns increasing in difficulty.The child's task is to identify the missing item that completes a pattern.Splithalf coefficients of reliability of CPM scores in childhood range from .65 and .90;retest reliability coefficients from .81 to .95,.68 to .92 and .71 to .95 over 10 days, one month and one year, respectively.
The Family Environment Questionnaire (FEQ-II, Zupančič & Kavčič, 2007) was used as a self-report measure of maternal and paternal parenting practices.It includes 52 items tapping parental behaviour towards their offspring in middle childhood.The items are rated along a 6-point scale (1 = almost never, 6 = almost always) and they are combined into four dimensions: authoritative parenting (setting limits, responsiveness, sensitivity, warmth, induction), ineffective control (inconsistency in demands and rule setting, giving in to a child's unreasonable requests), power assertion (punishment, withholding of privileges, requiring immediate compliance), and stimulation of child's cognitive development (enhancing fine motor skills, introducing mathematical concepts, explaining causes and consequences of natural phenomena and interpersonal relationships).With Slovene 6-year-olds internal consistencies of the dimensions range from 0.72 to 0.91 and from 0.67 to 0.92 for mothers and fathers, respectively.Maternal and paternal ratings are moderately inter-correlated and moderately stable over a three year time interval.
The teachers assessed the children's academic achievement in Slovene and mathematics using the Templates for evaluation of attained performance standards in the first grade of elementary school (Zupančič, 2006).The templates for each of the two subjects are based on the curriculum (grade 1) and outlines for evaluation books with explanations (The National Education Institute of the Republic of Slovenia 2001Slovenia , 2002)).The instrument was designed previously in collaboration with teachers.It captures 124 and 50 items in Slovene (e.g.Writes simple words by himself/herself.)and math (e.g.Understands mathematical symbols + andand uses them with examples.),respectively.The items are rated according to a 6-point rating scale (1 = the child does not attain the standard, 3 = attains the standard, 6 = attains much more than required by the standard).The average score for the two school subjects was used in the present study.
Procedure.Parents, teachers and assistant teachers of first-graders across the Slovene obligatory schools were asked for informed consent to participate in the study at the beginning of a school year.Approximately two months later, questionnaires were distributed to assistant teachers and parents who agreed to take part.The parents provided demographic data and reports on their parenting practices, and the assistant teachers reported on children's personality.Written instructions on how to fill-in the questionnaires were provided in the introductory section of each instrument used.The filled-in questionnaires were returned to the research team in sealed envelopes.The children's cognitive ability was assessed individually in a quiet room of the child's school by previously trained research assistants.Finally, the teachers evaluated the children's attainment of performance standards (APS) in Slovene and mathematics at the end of the school year.The teachers' evaluations were also returned to the researchers in sealed envelopes.

Results
First, we explored if first-grader's APS scores differed with respect to their gender and attendance of pre-school prior to school entry.Neither of these two variables had a significant effect on the first-graders' APS (F 1,193 = 1.84 and 1.90 for gender and pre-school attendance, respectively, both ps > .05).
Next, we examined associations of family and child characteristics with the APS scores.The correlations (Pearson's rs) and descriptive statistics are presented in Table 1.The APS was moderately related with children's extraversion, conscientiousness, and low neuroticism as perceived by assistant teachers.The remaining correlations were low and suggested that the firstgraders' APS is associated positively with their cognitive ability, maternal authoritative parenting and her cognitive stimulation but shows negative links with paternal power assertion and children's disagreeableness.Two sets of three-step hierarchical linear regressions were conducted in order to estimate a relative contribution of children's higher-order personality traits vs. lower-order traits to their APS at the end of the school year.In the first set of the analysis, three blocks of variables were entered into regressions: (i) family characteristics, i.e. maternal and paternal education and their selfreported parenting practices were entered in the first step; (ii) child cognitive ability was considered in the second step; and (iii) assistant teachers' reports on child personality (four higher-order traits) were added in the third step.In the second set of regressions, blocks (i) and (ii) remained the same, but instead of children's higher-order personality traits, 15 lower-order personality traits were entered as block (iii).
Table 2 shows that family characteristics explained 10% of the variance in the first-graders' APS with maternal authoritative parenting emerging as a single statistically significant predictor.Children's cognitive ability contributed significantly to their APS over and above the family variables (6% of unique variance explained).Finally, assistant teachers' ratings of the children's higherorder personality traits improved the prediction by additional 12%, with child conscientiousness appearing to be a significant predictor.We re-run the analysis with a reduced set of predictors, taking into account only the statistically significant ones in order to explore their individual, specific contributions.Maternal authoritative parenting was entered in the first step (ΔR 2 = .06,ΔF 1, 191 = 12.68, p < .001),children's cognitive ability in the second one (ΔR 2 = .05,ΔF 1, 190 = 11.09,p < .01),and assistant teacher-rated child conscientiousness in the third step (ΔR 2 = .12,Δ F 1, 189 = 30.20,p < .001).The Adj. R 2 showed that the three variables together explained 22% of criteria variance (F 3, 189 = 19.46,p < .001)with βs .19(p < .01),.16(p < .05)and .36(p < .001)for authoritative parenting, cognitive ability and conscientiousness, respectively.
In the second set of the regression analysis, 15 lower-order child personality traits were entered into regression instead of the four higher-order traits.The model explained 28% of variance (F 26, 166 = 3.86, p <.001) in the first-graders' APS.The results of the first two steps of the analysis remained the same as those shown in Table 2 (family variables explained 10% of the criteria variance and children's cognitive ability accounted for 6% of unique variance in the APS beyond family variables).Nevertheless, children's lower-order personality traits as perceived by assistant teachers improved the prediction based on family characteristics and children's ability by 22% (ΔF 15, 166 = 3.81, p < .001).With respect to children's personality, antagonism (a constituent trait of disagreeableness; β = -.29,p < .05)and intelligence (a constituent trait of conscientiousness; β = .56,p < .01)were signifi cant single contributors to their APS.Again, the analysis was re-run with a reduced set of predictors, i.e. the statistically signifi cant ones.Maternal authoritative parenting was entered in the fi rst step (ΔR 2 = .06,ΔF .1,191 = 12.68, p < .001),cognitive ability in the second one (ΔR 2 = .05,ΔF 1, 190 = 11.09,p < .01),and assistant teacher-rated child antagonism and intelligence were added in the fi nal step (ΔR 2 = .17,ΔF 2, 188 = 22.75, p < .001).The following βs were obtained: .17(p < .05),.12(.10 > p > .05),-.11 (.10 > p > .05),and .39(p < .001)for authoritative parenting, cognitive ability, antagonism, and intelligence, respectively.The four variables together explained 27% of the variance (F 4, 188 = 18.88, p < .001) in children's APS in Slovene and mathematics at the end of the academic year.

DISCUSSION AND CONCLUSIONS
The present study examined predictive relations of several family variables, children's cognitive ability and their personality with attainment of performance standards (APS) in the end of the first school year.Children's personality traits appeared to be the strongest predictors of their APS scores.Among other variables accounted for, children's cognitive ability and maternal authoritative parenting also contributed significantly to the APS, a finding consistent with prior studies.The joint effect of all of the variables considered was substantial (explaining 28% of variance in children's APS).However, the information on maternal authoritative parenting, children's non-verbal cognitive ability and conscientiousness (or specifically, assistant teachers' perceived child intelligence and low antagonism) provides the prediction of almost the same accuracy than the prediction based on all the variables.Our results add to mounting evidence about a beneficial role of authoritative parenting in students' academic outcomes (e.g., Jeynes, 2005;Simpkins et al., 2006).Intelligence has also been previously well documented to play an important (causal) role in students' academic achievement (e.g., Deary, Strand, Smith, & Fernandes, 2007;Rinderman & Neubauer, 2004;Watkins, Lei, & Canivez, 2007).
Our study also suggests that children's personality trait ratings have a unique value in predicting academic attainment over and beyond demographic and parenting variables, and cognitive ability.The results support previous findings (e.g., Laidra et al., 2007;Mervielde et al., 1995) on the predictive power of conscientiousness in explaining academic achievement in children.In extension to recent research, we provided information on lower-order traits which explain the links of the higher-order personality traits with school success.
The relation between the higher-order trait of conscientiousness and academic achievement has often been interpreted in terms of motivation, classroom behaviour, and study habits (Chamorro-Premuzic & Furnham, 2003).Conscientious students may be more motivated to perform well academically than their less conscientious counterparts.Also, teachers often consider students' academic effort and cooperation in the classroom when evaluating academic performance (Peček, Valenčič Zuljan, Čuk, & Lesar, 2008).An examination of lower-order trait ratings in our study reduced speculative interpretations about the predictive nature of conscientiousness and offered an insight into a more specific aspect of the higher-order trait which is actually related to first-graders' APS.Namely, assistant teachers' perceived child intelligence substantially contributed to the prediction of APS.First-graders who are perceived as learning oriented, eager to learn, and quick to understand may tend to acquire academic content more effectively and therefore perform well in school.In addition, children who are rated high in intelligence by the assistant teachers are also likely to be perceived as more intelligent by their teachers (note a high inter-rater agreement of personality ratings between teachers and assistant teachers reported in the Method section).Thus, teachers' may generalize their perceptions of children's intelligence to the evaluations of APS.This supposition is also consistent with our finding that subjectively perceived intelligence in the classroom has a stronger predictive power for the first-graders' APS than their objectively measured cognitive ability.Relatively low predictive power of children's cognitive ability for their APS may partly be due to the fact that a non-verbal cognitive ability measure was used in our study.
Although educational implications of agreeableness are not clear and our prediction based on higher-order traits suggested the respective trait as non-significant, ratings of children's low antagonism (a constituent of low disagreeableness) were marginally predictive of the first-graders' APS.Antagonistic children, prone to confrontational behaviour indicated by being uncooperative, discourteous, rude and aggressive may exhibit behavioural problems in the classroom (Asendorpf & Van Aken, 2003;Zupančič & Kavčič, 2007), making learning more difficult for them.They are also less pleasurable to work with in the classroom, are more disliked by peers and teachers (Rubin, Bukowski, & Parker, 1998), and may have difficulties due to opposing teachers in performing academic tasks.All these may contribute to somewhat poorer APS in antagonistic children as compared to their less negativistic and more cooperative peers.
In accordance with findings of Chamorro-Premuzic and Furnham (2003) and the meta-analysis of O' Connor and Paunonen (2007), our results indicate that the accuracy of the prediction of the first-graders' APS based on personality ratings could be increased by considering lower-order rather than higherorder traits.This may occur because children with identical higher-order trait scores may have different lower-order trait scores but only some of the latter encapsulate "pure" personality variance shared with academic achievement.Some limitations of our study should be mentioned.The data set relied on a single overall indicator of children's academic achievement which reflects teachers' evaluation of students' APS.In future, objective tests of knowledge should also be included, along with children's verbal cognitive ability and other non-cognitive variables (e.g., motivation, interests and study habits).We would also like to point out a possible overestimation of some socially relevant traits because the assistant teachers who assessed child personality worked continuously with the children for about two months only.As personality refers to relatively stable traits expressed over situations, more precise reports may be expected from informants who daily interact with the targets for a longer period of time.However, the ICID standardization study (Zupančič & Kavčič, 2009) offered ample evidence that (assistant) teachers provide accurate, reliable and valid ratings of children's personality after approximately two months into the school year.
We believe that increasing knowledge of the factors associated with early academic achievement has important implications for education: it allows educators to predict who will perform better or worse in school and enables teachers to adjust their classroom work to individual characteristics of children in order to optimize their academic performance, i.e. to compensate for identified children's weaknesses related to academic achievement and to encourage their strengths.