مقایسه سه عامل بزرگ و پنج عامل بزرگ در کودکان و نوجوانان

This study examines the robustness of the PEN model and the Big Five model in self-reported personality data of 419 early adolescents, aged 12–14 years. Adolescents filled out the Eysenck Personality Questionnaire for Youth-Revised (JEPQ-R) and a questionnaire designed to assess the Big Five factors. Factor analysis of the combined item set revealed that three factors best represented the data. Direct comparison of both models by exploratory analyses at the scale level seemed to be more in favour of the PEN model, and lent support for the assumption that Agreeableness and Conscientiousness are aspects of Psychoticism. Confirmatory factor analyses, however, did not favour one model over the other. The position of Openness/Intellect is discussed and suggestions for future personality research are described.

In the past decades two models have dominated in the area of personality research. The first model is Eysenck's hierarchical three factor model, recently described as the Giant Three model (Eysenck, 1994). The second model assumes that personality is best described in terms of five factors, or the Big Five, and is usually referred to as the Big Five model (John, 1990). Initially, Eysenck distinguished two basic dimensions of personality, which he labelled Extraversion (E), reflecting extraverted tendencies such as sociability, activity, liveliness, and sensation seeking, and Neuroticism (N), reflecting neurotic tendencies such as anxiousness, tenseness, guilt feelings, and depression. Subsequently, a third dimension, labelled Psychoticism (P), was added (Eysenck & Eysenck, 1975). This factor, which was also referred to as ‘tough-mindedness', included traits like aggression, egocentrism and impulsiveness. The original P-scale in the Eysenck Personality Questionnaire (Eysenck & Eysenck, 1975) showed major psychometric weaknesses, most important a low reliability (Block, 1977), and was revised in the EPQ-R for adults (Eysenck & Eysenck, 1991), and the Y- EPQ-R for youth (Cloninger et al., 1993, De Bruyn et al., 1995 and Scholte & De Bruyn, 2001). The EPQ(-R) is now one of the most widely used instruments for measuring the three PEN factors. These factors have been replicated in a great number of studies employing the EPQ(-R) in a wide variety of settings and countries including Finland (Eysenck & Haapasalo, 1989), Japan (Hosokawa & Ohyama, 1993), Italy (Saggino, 2000), and The Netherlands (Sanderman, Eysenck, & Arrindell, 1991). In the Big Five model of personality, five factors, or the Big Five, are regarded as the basic dimensions of personality organisation. Factor analysis of personality descriptive adjectives (Goldberg, 1981 and Goldberg, 1992) as well as personality questionnaire items (Costa & McCrae, 1995 and McCrae & Costa, 1985) have revealed more or less the same five factors. Although slightly different terms are sometimes used for these five factors, there is general agreement regarding the first four: I: Extraversion (E), II: Agreeableness (A), III: Conscientiousness (C), and IV: Emotional Stability or Neuroticism (ES or N). There has been some debate about the fifth factor, which has been labelled “Intellect” by Goldberg (1990) and “Openness” (O) by McCrae and Costa (1985). This debate, however, did not question the appropriateness of the Big Five model as a whole. The Big Five factors have been shown to be robust across methods of factor extraction and rotation. For example, Goldberg (1990) found virtually identical Big Five representations in factor solutions based on five variants of orthogonal and five variants of oblique rotations. Support for the model was also found in a multitrait- multimethod study by Barbaranelli and Caprara (2000). Cross-cultural evidence for the existence of the Big Five factors has come from studies within a number of languages, including Western languages such as German and Dutch (Hofstee, Kiers, De Raad, Goldberg, & Ostendorf, 1997), Spanish (Benet-Martı́nez & John, 1998), and Italian (Caprara, Barbaranelli, Borgogni, & Perugini, 1993). The Big Five factors have also been found in Asian languages like Chinese (Trull & Geary, 1997). Even though the Big Five model is probably predominant in personality assessment research to date, it is recognised that the model has limitations (see Block, 1995, John & Robbins, 1994 and McCrae & John, 1992) and that the “Big Five model is not fully developed” (Goldberg & Saucier, 1995, p. 221). Indeed, some studies have reported on the existence of seven factors or the Big Seven (Almagor, Tellegen, & Waller, 1995). Two of these factors were evaluative or valence factors but the other five factors corresponded to the Big Five. Other studies question the cross-cultural generalizability of the Big Five factors. In a number of Italian studies on the Big Five factors, some factors with slightly different meanings tended to emerge, even if the same samples and variable selection procedures were used as in other studies. For example, in studies by Caprara and Perugini, (1994) and Di Blas and Forzi (1998) a Big Five factor structure emerged in which Emotional Stability was only marginally represented. Perugini and Ercolani (1998) reported that the fifth factor in their study was better characterised as Autonomy than as Openness. 1.1. Comparing the Giant Three and the Big Five models A number of studies have compared the Giant Three and Big Five models. In these studies, the attention has mostly been focused on the level of the broad personality dimensions or scale levels. Many of these studies report clear correspondences between Eysenck's E and the Big Five E, and between Eysenck's N and the Big Five N (or the opposite of Emotional Stability) (e.g., Goldberg & Rosolack, 1994, McCrae & Costa, 1985 and Saggino, 2000). Theorists from both models now seem to agree that these two factors are basic dimensions of personality. On the other hand, disagreement remains to exist over the number and meaning of the other basic dimensions of personality. Most of the debate has focused on the nature of the Psychoticism factor and the extent to which it overlaps with A and C from the Big Five. In addition, there is some debate concentrating on Openness or Intellect. Openess/Intellect is the fifth factor in the Big Five model, but is not viewed as a separate personality dimension in the PEN model, in which it is considered a component of Psychoticism (Eysenck, 1991, Eysenck, 1992 and McKenzie, 1988). Several methods have been applied to examine whether three or five factors best represent the basic dimensions of personality and to what extent P should be regarded as a combination of A, C, and O, as Eysenck (1991) claims. First, some studies used multiple questionnaires, including ones that were not designed to measure only one of the two models, and by means of joint exploratory factor analysis examined which factors were present. For example, Noller, Law, and Comrey (1987) tested the presence of the Big Five factors using the Cattell 16 Personality Factor Questionnaire, the Comrey Personality Scales, and the Eysenck Personality Inventory. The authors found evidence for the first four of the Big Five factors, while O did not emerge. In a reanalysis of the data set and using a different analytical approach, Boyle (1989) confirmed the results, although some of the factors differed slightly across the studies. Noller et al. (1987) suggested that, given the differences between the three instruments they used, similar results were likely to have occurred if combinations of other personality instruments would have been used. Indeed, using self-report and parent-report versions of the NEO-FFI (Costa & McCrae, 1992), Adjective Checklist (Gough & Heilbrun, 1983), Myers-Briggs Type Indicator (Myers & McCaulley, 1985) and the California Child Q-set (Block & Block, 1980) in a sample of academically talented youth, Parker and Stumpf (1998) found evidence that largely supported the Big Five factor model. Other studies used instruments that were more explicitly designed to assess the PEN and the Big Five factors, and simultaneously analysed the three PEN and five Big Five scales. This method seems more able to directly compare the two models and to answer the question whether P is a combination of A, C, and O. For example, Draycott and Kline (1995) jointly factor analysed the scales of the EPQ-R (Eysenck & Eysenck, 1991) and NEO-PI (Costa & McCrae, 1995) and found that P, A, and C loaded on one factor. In a replication study, Saggino (2000) analyzed the scales of the Italian edition of the EPQR and the Big Five Questionnaire (Caprara et al., 1993). In the factor analysis of these scales, EPQR-P turned out to form one factor with BF-Friendliness— which corresponded with Agreeableness— but not with BF-C. In a subsequent canonical correlation analysis, BF-C tended to form a canonical variate with BF-O, BF-Friendliness and EPQR-P. Other studies showed that P subsumed A and C (McKenzie, Tindell, & French, 1997). In contrast, however, some investigations seemed to provide evidence for the view that A and C are basic, higher-order factors of personality as the Big Five model claims and suggest that five instead of three factors constitute the basic dimensions of personality. By correlating the PEN and Big Five scales derived from the Eysenck Personality Inventory and the Extended NEO Inventory, respectively, McCrae and Costa (1985) found P to be negatively related to A (−0.41) and C (−0.31) and argued that these correlations indicated that the P scale was a blend or a relatively arbitrary conflation of the independent dimensions A and C. In addition, A and C were found to have different patterns of correlations with external criteria, suggesting that A and C should be regarded as independent dimensions of personality instead of lower-order factors of P. A decade later, Costa and McCrae (1995) examined the scales of the Eysenck Personality Profiler (EPP) and the NEO-PI-R, and concluded that the EPP scales represented five rather than three scales. Using confirmatory factor analysis on the EPP scales, Jackson, Furnham, Forde, and Cotter (2000) did not confirm Costa and McCrae's findings, but instead showed that the three factor solution was at least as psychometrically valid as the Big Five factor solution. They concluded that the Eysenckian tradition had still much to offer in the discussion on the fundamental structure of personality. 1.2. Features of the present study The present study has a number of specific features that may contribute to broadening our knowledge about the basic dimensions of personality. First, it will focus on the personality structure in a normal sample of early adolescents and compare the PEN model and the Big Five model in adolescents' personality data. Although both the PEN model (Cloninger et al., 1993, De Bruyn et al., 1995, Francis, 1996 and Scholte & De Bruyn, 2001) and the Big Five model (Graziano & Ward, 1992 and Scholte et al., 1997) have been examined separately in early adolescents, almost all of the studies in which both models have been compared were based on adults. One of the exceptions seems to be a study by Ng, Cooper, and Chandler (1998), who argued that P was not a superfactor consisting of A and C. Barrett (1999) pointed out, however, that because of the study's insufficient sample size and the lack of sufficient data for forming and testing the hypotheses used, Ng et al's conclusions should be treated with caution, and more research was warranted. Second, although in The Netherlands studies on the Big Five (e.g., Scholte et al., 1997) and the PEN factors (e.g., Scholte & De Bruyn, 2001) in adolescents self-descriptions have been performed, direct comparisons of both models in one study are currently lacking. Research has shown that differences in personality and personality perceptions may exist between cultures, even between countries in Europe. For example, in The Netherlands, Scholte and De Bruyn (2001) found that Dutch boys and girls scored markedly lower on Neuroticism assessed with the JEPQ-R than did British adolescents (Corulla, 1990). In Italian studies (Di Blas & Forzi, 1999), Emotional Stability did not emerge as an independent factor, which may have been due to cultural differences in personality perception (Caprara & Perugini, 1994, Di Blas & Forzi, 1998 and Di Blas & Forzi, 1999). Other studies (e.g., De Raad et al., 1997 and Yang & Bond, 1990) also suggest cultural differences in the perception of personality traits, and findings from studies performed in one country can not readily be generalised to studies performed in other countries. Third, almost all studies that compare both models have performed analyses at the scale level. The scales refer to the higher level of the personality structure (ie., dimensions), whereas items refer to the lower levels (ie., traits or habits, cf. Eysenck, 1947) and are thus also meaningful to examine. The present study will therefore focus on the item level as well as on the scale level. At the item level, factor analysis will be carried out on the separate as well as on the combined item sets of the JEPQ-R and Big Five instruments. At the scale level, exploratory and confirmatory factor analysis of the scales of both instruments will be performed.