A questionnaire for measuring the Big Five in late childhood

Abstract

A study is presented which aims at measuring the Big Five factors in late childhood through self-report as well as parent and teacher ratings. First, several factor analyses examined self-report and teacher and parent ratings on a 65-item questionnaire developed for assessing the Big Five. Five clear factors emerged from these analyses conducted on self-report and other ratings of elementary and junior high school children. Factors showed a high degree of congruence. Self-reports, parent and teacher ratings resulted moderately although significantly convergent. Second, as a validation step, the Big Five factors were used as concurrent predictors of academic achievement and of Externalizing and Internalizing problematic behavior syndromes. Intellect/Openness and Conscientiousness resulted as important predictors of Academic Achievement. Externalizing problems were associated to low Conscientiousness and low Emotional Stability, Internalizing problems to low Emotional Stability. Finally, also the correlations of the Big Five factors with the dimensions of Sybil Eysenck's Junior Personality Questionnaire further corroborated the construct validity of the questionnaire.

Introduction

In recent years, an impressive body of research has accumulated supporting a five-factor structure to describe personality (the so-called “Big Five”), confirming the early structure proposed by Fiske (1949), Tupes and Christal (1961), and Norman (1963). These five robust factors emerged irrespective of factor analytical techniques (Goldberg, 1990), rating procedures (Botwin and Buss, 1989, Digman and Takemoto-Chock, 1981, McCrae and Costa, 1987), language (DeRaad, Perugini, Hrebickova, & Szarota, 1998). Although there are some divergences among various authors regarding the interpretation of each factor (see Block, 1995, Digman, 1990, Goldberg, 1990, John, 1990, John et al., 1988, McCrae, 1990), there is substantial agreement on using the following labels for these five factors: I. Extraversion, II. Agreeableness, III. Conscientiousness, IV. Neuroticism, and V. Intellect (or Openness to Experience).

The Big Five are the meeting point of two traditions of research: the lexicographic and the factorial tradition. While lexical studies, moving from the “sedimentation hypothesis” (e.g. Cattell, 1943), examined the emergence of the five factors through trait terms (adjective, nouns, verbs) extracted from the vocabulary, factorial studies examined the emergence of the same factors through the analysis of the descriptive phrases contained in personality questionnaires. A number of studies suggested the plausibility of reducing to the five factors the dimensions measured by several questionnaires such as Cattell's 16PF, Guilford's GZTS, Eysenck's EPQ, and Comrey's CPS (Krug and Johns, 1986, McCrae, 1989, McCrae and Costa, 1985, Ostendorf and Angleitner, 1992).

Notwithstanding the impressive number of research studies that confirmed its validity and replicability, the Five Factor Model (FFM) has recently come under severe criticism (Block, 1995, Church and Burke, 1994, McAdams, 1994, Parker et al., 1993, Pervin, 1994). For some of the critics, the answers are clearly dependent on further research. With regard to other issues, some counterarguments have been raised by FFM supporters (e.g. McCrae & Costa, 1999). We believe the Five Factors represent useful constructs especially for their practical utility; they provide a common language for self report and other ratings (see McCrae & Costa, 1987), and can help in reducing the distance among different informants and in enhancing interrater convergency.

As recently noted by Shiner (1998), there is a substantial increase of studies exploring personality structure in middle childhood and early adolescence. A main problem of these studies, however, is in the absence of a common framework for interpreting the results. There is no agreement on the nature and the number of dimensions needed to describe personality. While the model developed by Cattell comprises 16 factors (see Coan & Cattell, 1966), the model developed by Sybil Eysenck comprises only three dimensions (Eysenck, 1975). Concluding her review, Shiner (1998) proposed a theoretical taxonomy for the classification of personality dimensions in middle childhood. This taxonomy comprised four general dimensions which can be traced back to four of the Big Five: (1) Positive Emotionality (corresponding to Extraversion); (2) Negative Emotionality (corresponding to Neuroticism); (3) Aggressiveness versus Prosocial Tendencies (corresponding to Agreeableness); and (4) Constraint (corresponding to Conscientiousness). In light of these considerations, we may wonder if the Big Five Model can be extended from adult personality to children's personality, serving thus as a reference structure for study comparison and results generalizability (see in this regard Mervielde and De Fruyt, 1999, Mervielde and De Fruyt, 2000).

Although many studies have investigated the Big Five in adulthood, researchers only recently began to study the Big Five in late childhood. Digman and Inouye (1986) found five factors very similar to the “adult” Big Five in factor analyses of teacher ratings of about 500 junior high school children, using 43 adjectives scales. Mervielde (Mervielde, 1994, Mervield et al., 1995) analyzed teacher ratings on four different age groups (from 4 to 12 years old), identifying most of the time a five factor solution consistent with the Big Five. In this study, moreover, Conscientiousness and Intellect/Openness showed high correlations with Academic Achievement.

Extensive work has been conducted by several researchers within the framework of an international project aimed at investigating parental free descriptions of children's personality (Kohnstamm et al., 1998, Kohnstamm et al., 1995). Results of this project evidenced that 76–85% of parental descriptions of children aged 3 to 12, across countries, could be assigned to the Big Five. These free descriptions provided the basis for the development of the Hierarchical Personality Inventory for Children (HiPIC; Mervielde & De Fruyt, 1999). Mervielde and De Fruyt (1999) showed that children aged 5–13 years can be described by their teachers and their parents by means of the HiPIC, whose psychometric properties turned out to be stable across informants and age levels. The HiPIC was also valid when used as a self-report instrument on young adolescents (from 12 to 17 years), evidencing a high degree of convergent and discriminant validity with the NEO-PI-R (De Fruyt, Mervielde, Hoekstra, & Rolland, 2000). This last study demonstrated also that an “adult” measure of the Big Five such as the NEO-PI-R (Costa & McCrae, 1992) can be applied to assess personality traits in adolescence.

Similar results were found by John, Caspi, Robins, Moffitt, and Stouthamer-Loeber (1994), who developed scales for the Big Five from Q-sorts of 12–13 year old boys rated by their mothers. In particular, teacher reports of school performance correlated with Conscientiousness and with Openness. Verbal, performance and full scale IQ of WISC-R correlated with Openness. High levels of delinquency were associated with low levels of Agreeableness and of Conscientiousness. Externalized and Internalized disorders measured by the Achenbach Child Behavior Checklist (Achenbach & Edelbrock, 1983) were associated respectively with low Agreeableness, low Conscientiousness and high Extraversion, and with high Neuroticism and low Conscientiousness. More recently, Graziano, Jensen-Campbell, and Finch (1997) assessed the Big Five from self-report of fifth to eighth graders using Goldberg's (1992) bipolar self-rating markers. The resulting Big Five correlated with several measures of adjustment. In particular, Intellect was associated with both self-report and teacher ratings of Academic Adjustment, Agreeableness was associated to both self-report and teacher ratings of classroom behavior and teacher and peer relations.

The studies examined in the previous section clearly evidenced the possibility of using the Big Five to obtain personality ratings of children, especially from adults (parents and teachers). We may question, at this point, if the same five big factors can be replicated reliably also in the case of children's self-report. The study of Graziano and colleagues (1997) seems to represent a positive answer to this issue. The question is particularly crucial since children's self-reports have been neglected not only within the Big Five tradition, but in general in the field of child personality assessment, where only teacher and parent ratings have so far been used. The questionnaires developed by Sybil Eysenck (Junior Personality Questionnaire, JPQ, 1965) and by Coan and Cattell (1966) are outstanding exceptions of this reductive approach to children's personality assessment. The subject who gives a description of her/his own personality represents an accurate source of much information that can be considered as less accessible for observers (consider, for example, ratings of depression), or that can be “distorted” in other ratings through response biases such as halo, severity, central tendency and the like.

As a matter of fact, research on sociometric assessment by means of peer nomination show clear evidence of children's competence in using trait terms for describing their peers (see Coie et al., 1982, Newcomb et al., 1993). Peer nomination based measures of sociometric status as well as of prosocial behavior, aggression and impulsivity showed great predictive validity in studies aimed at individuating children at risk for maladaptive behavior (see Caprara et al., 2000, Dodge et al., 1990). In a recent study Mervielde and De Fruyt (2000) showed that peer nominations from children from 8 to 12 years on 25 bipolar traits can be interpreted using the Big Five as a framework. The “space” derived from peer nominations, however, is reduced to only three dimensions: In fact, factor analyses of the 25 scales showed a three-factor solution with Conscientiousness and Intellect “merging” in a single factor, as well as Extraversion and Emotional Stability, and Agreeableness emerging as a separate smaller third factor.

All these results clearly demonstrate that children have the capacities for giving reliable and valid descriptions of their peers personalities: as a matter of fact, it is reasonable to think that they have also the competence for describing their own personality. As recommended by Shiner (1998), the more the instrument that is used to assess personality is consistent with children's cognitive capacities and cultural characteristics, the more the information gathered will be accurate and reliable. Following these recommendations, and capitalizing on long experience in the construction of assessment instruments for elementary and junior high school children (e.g. see Caprara & Pastorelli, 1993) the aim of this research was to develop and validate an instrument useful for measuring the Big Five through self-report in children from 8 to 9 years old. In particular, we wanted to examine the emergence of the Big Five factors in self-reports of 9–13 year old children, to compare this structure with the structure emerging from teacher and parent ratings on the same personality descriptors, and to validate personality descriptions against different criteria.