Validity of Big Five Personality Judgments in Childhood: A 9 Year Longitudinal Study

In a 9 year longitudinal study over childhood, the Big Five personality traits were assessed at ages 4–6 by teacher Q-sorts, at age 10 by parental Q-sorts, and at age 12 by parental and friend ratings on bipolar adjective scales. The Big Five Q-sort indices were based on definitions proposed by John, Caspi, Robins, Moffitt, and Stouthamer-Loeber (1994) for adolescent boys. They were related to judgments and behavioural observations of inhibition and aggressiveness, and to antecedents and consequences of school achievement such as IQ and cognitive self-esteem. Neuroticism and low extraversion correlated with social inhibition, low agreeableness and low conscientiousness with aggressiveness, and conscientiousness and/or culture/intellect/openness with antecedents and outcomes of school achievement. These correlations were consistently found throughout childhood. Copyright # 2002 John Wiley & Sons, Ltd. The Five-Factor Model (FFM) of personality description has attracted much interest over the past two decades. According to this model, five relatively independent, extremely broad dimensions (‘Big Five’) explain a major portion of judged interindividual differences in personality. These dimensions are extraversion, neuroticism (or the reversely coded dimension emotional stability), agreeableness, conscientiousness, and culture, intellect, or openness to new experience, which is the least agreed-upon factor. This model has been empirically supported by research mainly on North-American and Northern European adults’ selfand other-descriptions of personality (John & Srivastava, 1999). The Big Five resulted as the last step of a multi-step lexical approach procedure that started with the full lexicon of personality-descriptive terms, and reduced that information through item elimination and factor analysis (Goldberg, 1990; John, Angleitner, & Ostendorf, 1988). In addition, most of the variance captured by traditional personality questionnaires can be accounted for by the Big Five (McCrae & Costa, 1990). Received 22 November 2001 Copyright # 2002 John Wiley & Sons, Ltd. Accepted 28 May 2002 *Correspondence to: Jens B. Asendorpf, Institut für Psychologie, Humboldt-Universität zu Berlin, Oranienburger Strasse 18, 10178 Berlin, Germany. E-mail: [email protected] Building on this model of personality differences in adulthood and on early replications of the FFM through teacher ratings of children (Digman, 1989), researchers have begun to study the emergence of the Big Five structure in parental, teacher, and peer descriptions of children’s and adolescents’ personality (John, Caspi, Robins, Moffitt, & StouthamerLoeber, 1994; Kohnstamm, Halverson, Mervielde, & Havill, 1998; Mervielde, Buyst, & De Fruyt, 1995; Mervielde & De Fruyt, 2000; Scholte, van Aken, & Van Lieshout, 1997; Van Lieshout & Haselager, 1994). According to a recent review by Mervielde and Asendorpf (2000), the FFM has been supported best by parental and teacher ratings of personality. Studying the factor structure of the California Child Q-Set (Block & Block, 1980) instead of the factor structure of ratings, both Van Lieshout and Haselager (1994) and John et al. (1994), in addition to the original five dimensions, found evidence for two other dimensions in Dutch children and adolescents and ethnically diverse adolescent US boys, but these two dimensions were inconsistent across the two studies. The aim of the present study was to address two validity questions that are, in our view, at least as important. If the Big Five factors are operationalized by reasonable parental, teacher, and peer judgment measures, (i) how valid are such Big Five measures in terms of concurrently assessed important external variables and (ii) how consistent are these correlates of the Big Five across age? Concerning question (i), studies of the Big Five in childhood have focused on three main domains of personality and achievement: externalizing tendencies and internalizing tendencies, IQ and school performance, and domain-specific self-esteem. According to John et al. (1994), externalizing problems such as aggression, stealing and lying, inattention, impulsivity, and hyperactivity (Achenbach, Howell, Quay, & Conners, 1991) imply a pattern of low agreeableness and low conscientiousness. In contrast, internalizing problems such as anxiety, somatic complaints, social inhibition, and social withdrawal (Achenbach et al., 1991) imply a pattern of low extraversion and high neuroticism. Studying extreme groups of boys with externalizing or internalizing problems, John et al. (1994) confirmed the low agreeableness and low conscientiousness of externalizing boys, along with increased scores in extraversion, and also the high neuroticism in boys with internalizing problems, but not the expected low extraversion in these boys. This finding may be specific to the sample because other studies have consistently found an association between both high neuroticism and low extraversion with internalizing difficulties in childhood (Asendorpf, Borkenau, Ostendorf, & van Aken, 2001; Rubin & Asendorpf, 1993). Several other studies have reported associations between the Big Five and measures of externalizing and internalizing problems, confirming most of the relations between extraversion, agreeableness, and conscientiousness and externalizing problems, and between neuroticism and internalizing problems, but also presenting some incidental findings. Ehrler, Evans, and McGhee (1999) found that children with low scores on agreeableness and conscientiousness exhibited social problems, conduct problems, attention deficits, and hyperactivity. Children with low scores on openness exhibited problems in social behavior, conduct, and attention, whereas neuroticism was associated with anxiety and depression. Victor (1994) reported various associations between the Big Five and several measures of problem behavior. Conduct disorder was related to extraversion, and low agreeableness and conscientiousness; socialized aggression to extraversion, and low agreeableness and openness; anxiety-withdrawal to low extraversion, neuroticism and low openness; attention problems to low conscientiousness and openness; and motor excess to low agreeableness, to neuroticism and to low openness. 2 J. B. Asendorpf and M. A. G. van Aken Copyright # 2002 John Wiley & Sons, Ltd. Eur. J. Pers. 17: 1–17 (2003) In a study on the development of behavioural and emotional problems in clinic-referred children, Huey and Weisz (1997) found that extraversion and low agreeableness were independent predictors of externalizing problems, whereas only neuroticism predicted internalizing problems. In a similar study, van Aken and Heutinck (1998) found that teacher ratings of low agreeableness at ages 7, 10, and 12 predicted antisocial behaviour at age 20. In several other studies on agreeableness, Graziano and colleagues (Graziano & Ward, 1992; Jensen-Campbell & Graziano, 2001) found that this factor was related to processes and outcomes during interpersonal conflicts, and also to teacher-rated school adjustment. Shiner (2000) found that in 8–12 year olds a higher order factor agreeableness (a combined parent and child description) was related to concurrent rule-abiding conduct (combined peer and parent judgments) and peer nominations on social competence. A second higher order factor, surgent engagement (including extraversion, expressiveness, and attention), was also related to social competence. Finally, in a sample of 4–12 year olds, Goedhart, Treffers, and Kohnstamm (1994) found that agreeableness, conscientiousness, extraversion, and openness differentiated between clinically depressive and control children. John et al. (1994) also studied the relation between their Q-sort based Big Five indices and IQ and school performance. As they expected on the basis of similar findings by Digman (1989), both IQ and school grades were consistently related to conscientiousness and openness. Conscientiousness and openness, or intellect, were also reported in several other studies to have relations with measures of IQ and/or school achievement. Mervielde et al. (1995) found that conscientiousness and openness, but also extraversion, significantly predicted average grades across three primary school age levels. Judge, Higgins, Thoresen, and Barrick (1999) found relations between general mental ability (measured with the Stanford–Binet intelligence test) and openness and conscientiousness, but also neuroticism. Shiner (2000) reported that the higher order factor surgent engagement (including extraversion) was related to concurrent academic functioning and IQ-scores. Regarding motivational school factors, Lay, Kovacs, and Danto (1998) found associations between conscientiousness and teacher ratings of procrastination in a sample of 7–11 year olds. Concerning domain-specific self-esteem, Graziano and Ward (1992), using the Harter scales, reported that all Big Five factors were clearly related to domains of self-esteem. Agreeableness was significantly related to self-esteem in the domains of academics, appearance, and conduct but not social acceptance or athletics. Extraversion was related to self-esteem regarding academics, appearance, athletics, and appearance but not conduct. Conscientiousness was related to self-esteem regarding academics and ahtletics. Neuroticism was related to self-esteem in the domains of academics, social acceptance, and athletics, but not appearance or conduct, and openness was related to self-esteem concerning academics, social acceptance, athletics, and appearance, but not conduct. In this highly diverse literature, no age shifts in the correlates of the Big Five are easily detectable over childhood. Therefore, we assumed with regard to question (ii) that external correlates of the Big Five are constant over childhood. Together, this literature review suggested the following hypothesis: neuroticism and low extraversion correlate with social inhibition, low agreeableness and low conscientiousness with aggressiveness, and conscientiousness and culture/intellect/openness with antecedents and outcomes of school achievement. These correlations are consistently found all throughout childhood. We tested this hypothesis with available data from the Munich Longitudinal Study on the Genesis of Individual Competencies (LOGIC; Weinert & Schneider, 1999). We Big five in childhood 3 Copyright # 2002 John Wiley & Sons, Ltd. Eur. J. Pers. 17: 1–17 (2003) operationalized the Big Five in early to middle childhood through teacher and parental Q-sort indices, following the approach of John et al. (1994), and in middle to late childhood through parental and friend ratings on Big Five scales that were derived from a large pool of German personality-descriptive adjectives (Ostendorf, 1990). Although this design is not optimal for the analysis of the stability of interindividual differences because both the instruments and the judges changed over age, the use of multiple measures and judges and the longitudinal sample provide a particularly strong test for the continuity of external correlates of the personality judgments. Concerning these correlates, we studied social inhibition and aggressiveness through parental judgments and, if available, also behavioural observations in the preschool and kindergarten peer group. For achievement, we studied IQ, cognitive self-esteem, and deviation from expected grade. In Germany, deviation from expected school grade can be due to late schooling, repetition of a class (a regular intervention of the school in case of poor grades), or advancing to a higher grade (in the case of gifted children). Negative deviations indicate problems with schooling and can be considered a major real-life developmental outcome of German children’s cognitive and social–emotional competence.