Cross-cultural assessment of childhood temperament - Springer Link

European Child & Adolescent Psychiatry 11:101–107 (2002) DOI 10.1007/s00787-002-0248-4

I. Gasman D. Purper-Ouakil G. Michel M.-C. Mouren-Siméoni M. Bouvard F. Perez-Diaz R. Jouvent

Accepted: 21 August 2001

I. Gasman · D. Purper-Ouakil () · G. Michel · M.-C. Mouren-Siméoni Service de Psychopathologie de l’Enfant et de l’Adolescent Hôpital Robert Debré 48 bd Sérurier 75019 Paris, France E-Mail: [email protected] M. Bouvard Hôpital Charles Perrens 121 rue de la Bêchade 33000 Bordeaux F. Perez-Diaz · R. Jouvent CNRS UMR 7593 Personnalité et conduites adaptatives Pavillon Clérambault, Hôpital de la Salpêtrière 47 bd de l’hôpital 75013 Paris, France

ORIGINAL CONTRIBUTION

Cross-cultural assessment of childhood temperament A confirmatory factor analysis of the French Emotionality Activity and Sociability (EAS) Questionnaire

■ Abstract The Emotionality Activity Sociability (EAS) questionnaire focuses on heritable individual differences in reactivity and behavior which are often referred to in developmental temperament research. Psychometric properties of the French version of EAS were examined in a sample of 197 school-children aged six to 12 years. Parents, teachers and children aged nine years and more completed parallel forms of the EAS questionnaire. Confirmatory factor analysis was used to examine the fit between the original factors and the data. Internal consistency of each subscale, inter-rater and external validity were also examined. Children-rated EAS showed the best indices of fit between the four hypothesized factors and the data,

Introduction

■ Key words temperament – child – confirmatory analysis – questionnaire

measured and the existence of published reliability and validity data. Emotionality, activity level, and sociability appear early in development and are among the most stable and heritable temperament traits (7, 8). Unlike most temperament questionnaires, the items of the EAS are not restricted to a specific age range and even if the available information concerns mostly young children (6, 22, 24), the behaviors and reactions measured by this instrument can also be studied in adolescents and young adults. Emotionality refers to negative quality of emotional style and to a great intensity of emotional reactions. It is close to Eysenck’s dimension of neuroticism (13) and to Cloninger’s dimension of harm avoidance (11) and has been identified as a risk factor for the onset of major depression in adolescents (17) and for an in-

ECAP 248

Individual differences in reactivity and behavior have often been associated with developmental outcomes such as social adjustment, mental health and resilience to stress. Relationships between temperament and psychopathology, especially the predictive value of temperament attributes for the vulnerability and onset of mental disorders have become an active area of study within child psychology and psychiatry (17, 21, 25, 29). Among the wide range of available temperament questionnaires, the EAS survey (7, 8) is one of the most interesting for cross-cultural adaptation because of its extensive use, the type of temperament dimensions

but internal consistency was generally lower than in adult-rated questionnaires. Shyness and sociability showed significant overlap in both parent and teacher-rated EAS. The low concordance between childand adult-ratings indicates that temperament evaluation and interpretation of items may be influenced by subjective and/or developmental factors. Results are discussed in the perspective of validity versus cross-cultural comparability of temperament measurement. The theoretical four-factor structure was not completely replicable in our sample.

102

European Child & Adolescent Psychiatry, Vol. 11, No. 3 (2002) © Steinkopff Verlag 2002

creased risk of comorbidity (20). Activity level is a characteristic which refers to frequency and intensity of motor responses. Sociability is another higher-order factor in several approaches to personality; it can be defined as a preference for being with others rather than alone. It consists of seeking and being gratified by social rewards and is associated with positive emotionality, approach and adaptability. Recent evolution in temperament concepts suggest that sociability should be distinguished from shyness which refers to inhibited or tense behavior with strangers and a tendency to escape from social interaction (22). Shyness has been associated with fearfulness (10) and with vulnerability to anxiety disorders (18, 19). In their initial theory, Buss and Plomin proposed four temperaments: emotionality, activity, sociability and impulsivity (7). The EASI was the first instrument designed for these specific dimensions, but the impulsivity factor did not show sufficient stability and was therefore surrendered as a temperament (8). Rove and Plomin (26) used the EASI items in a joint factor analysis with 54 items derived from the New York Longitudinal Study (NYLS). For each of the factors extracted, the five highest-loading items coming from both the EASI and the NYLS were included in the Colorado Childhood Temperament Inventory (CCTI). The Emotionality, Activity and Sociability scales of the CCTI became part of the EAS. The Sociability scale was later redefined as a measure of shyness and an experimental Sociability scale was added to the instrument (8). Several psychometric studies were conducted with early versions of the EAS (15, 16) but the lacking cross-validation of the current EAS (parental ratings) was carried out by Boer and Westenberg (6) who also examined the psychometric properties of the experimental Sociability scale in a sample of Dutch children between four and 13 years of age. The average of inter-correlations of the Emotionality, Activity and Shyness scales was 0.16, proving the independence of the scales whereas the Sociability scale was significantly related to Shyness and Activity. An age trend was also detected toward a closer association between the Sociability and the Activity scales for the oldest age cohort (mean age = 10.5 years). Even though the factorial position of Sociability was more ambiguous, the authors concluded that Sociability should not be equated with Shyness. They also raised the possibility that Sociability and Shyness may be undistinguishable in a younger age group. Reliability in terms of inter-rater agreement (between the parents) and internal consistency was satisfactory (average Cronbach alpha for Emotionality, Activity, Shyness was 0.78). Recently, Mathiesen and Tambs (22) examined the psychometric properties of EAS in a sample of Norwegian children aged 18, 30 and 50 months. The study confirmed the separate usability of the Sociability scale and stability of factor structure over three years. Average al-

pha coefficients increased from 0.62 at 18 months to 0.70 at 50 months indicating that children may become more predictable in their behavior and reactions with older age. Gender differences were moderate and increased with age, boys having higher scores for Activity and girls for Shyness in the older age cohorts. Significant age trends were found for higher scores of Emotionality and Shyness and lower scores of Activity and Sociability with increasing age. These findings emphasize the issue of the cross-cultural comparability of temperament measurement and the need for a precise analysis of the factorial and correlational concordance between different language versions of the same instrument. In previous studies, interrater agreement was studied in terms of agreement between father and mother ratings (6), but the convergence of perceptions between parents and other raters (children, teachers) has not been explored. The aims of this study were • To compare the psychometric properties of the French version of EAS to those described in anterior validation studies by examining the fit between the hypothesized factor structure and the data • To study inter-rater agreement between parents, children and teachers • To examine correlations between temperament and related behavioral variables

Methods ■ Sample Subjects, their parents and teachers were recruited in a French elementary school attended by children from middle-class families. One hundred and ninety-seven children (112 boys, 85 girls) aged six to 12 (mean age 9.2) participated in the study. Parents received a letter explaining the goals of the study and gave written informed consent. Parents and teachers were asked to independently complete the EAS questionnaire (parental and teacher rating). Parents also received the CBCL (Children Behaviour Check-List) (1, French version 14). A pilot study was undertaken with twelve children selected into three age groups (6–9; 9–10; 10–12) and differing with regard to academic performance in order to test their understanding of the EAS questionnaire (children rating). The comprehension of items was correct for children aged nine and above; children-rated questionnaires were thus given only in the older age cohort (119 EAS child-rated questionnaires were analyzed).

I. Gasman et al. Cross-cultural assessment of childhood temperament

■ Instruments The EAS temperament survey Temperament was assessed by means of three different versions of the EAS Temperament Survey: parent-, teacher- and child-rated questionnaires (8). Each EAS questionnaire has 20 items,five items for each of the four scales. The items are rated on a five-point Likert scale ranging from 1 (not typical) to 5 (very typical) with six items negatively keyed. Scores are obtained by summing the items of each scale and dividing the result by five. The questionnaire was translated from English to French by an English-speaking psychologist and then back translated into English. If necessary, the French formulation of items was discussed by translators and authors until a consensus was achieved.

The Children Behavior Checklist (CBCL) (1) (French version, 14) The CBCL is a 113-item parent-report questionnaire on general behavioral and affective characteristics of children between four and 11 years. The CBCL has two subscales: internalized disorders with a predominance of anxious or depressive symptoms; externalized disorders combining motor instability, impulsivity and oppositional manifestations. A total score of 41 and more is an indicator of psychopathology. French adaptation of the CBCL has been carried out by Fombonne et al. (14).

■ Statistical analysis As the EAS questionnaire was constructed with regard to specific theoretical concerns, we first carried out a confirmatory factor analysis on the hypothesized factor structure (6, 22, 24). In this kind of analysis, a factor structure is explicitly hypothesized and is tested for its fit with the observed covariance structure of the measured variables. Indicators of fit include a χ2 test to examine the hypothesis that the model is consistent with the pattern of covariation among the observed variables. A significant χ2 implies that a significant amount of the observed covariance between measures is not explained by the model. The ratio between the model χ2 and the degrees of freedom (df) has been used as an indicator of fit with a threshold value of 2 (9). Other criteria are the goodness of fit index (GFI) which estimates the extent to which the sample covariances are reproduced by the hypothesized model, and the GFI adjusted for degrees of freedom (AGFI). A value of GFI exceeding 0.90 and a AGFI higher than 0.85 indicate a good fit (28). The root mean square error of approximation (RMSEA) (27), which has to be 0.05 or less, is also used to measure the discrepancy per degree of freedom. The comparative

103

fit index (CFI) (5) permits a comparison between the hypothesized model and the null model; a value of CFI greater than 0.90 indicates a satisfactory fit. Independence between the four scales of EAS and inter-item correlations were examined. Finally, inter-rater agreement and links with CBCL scores were studied by means of correlation analyses.

Results ■ Confirmatory factor analyses The fit indices of the confirmatory analyses carried out with both three and four factor solutions of the EAS Temperament Survey are presented in Table 1. Only the χ2/dl ratio reached significance for the EAS children-rated questionnaire. Other indices of fit were just below standard criteria. Generally, EAS parent- and teacher-rated versions yielded lower indices of fit than the children-rated questionnaire.

EAS children-rated questionnaire Three items in the French version of the EAS (children rating) did not load on the hypothesized factor when a threshold of 0.30 was considered statistically significant. Item 19 “Reacts intensely when upset” did not load on the a priori factor “Emotionality”, and two items did not load on the hypothesized factor “Shyness” (item 8 “Makes friends easily” and item 12 “Is very sociable”). Internal consistency estimated with Cronbach’s alphas ranged from 0.47 (Shyness) to 0.62 (Sociability) (see Table 3). Inter-factor correlations as presented in Table 4 were moderate.

EAS parent-rated questionnaire Three items did not load on the expected factors: item 6 “Tends to be somewhat emotional” loaded beyond the threshold value on the factor “Emotionality”, item 18 “When alone, child feels isolated” did not load significantly on the factor “Sociability” and item 1 “Tends to be Table 1 Summary of fit indices

χ2/dl GFI AGFI CFI

EAS children

EAS parents

EAS teachers

4-Factor 3-Factor

4-Factor 3-Factor

4-Factor 3-Factor

1.31 0.85 0.81 0.81

2.81 0.76 0.70 0.73

4.89 0.64 0.54 0.73

1.41 0.83 0.79 0.75

GFI Goodness of fit index AGFI GFI adjusted for degrees of freedom CFI Comparative fit index

2.82 0.76 0.69 0.72

4.88 0.63 0.64 0.72

104


Table 2 Confirmatory factor loadings for the EAS children, parent and teacher versions A priori factors and items

EAS children

Shyness Item 1: Tends to be shy Item 8: Makes friends easily* Item 12: Is very sociable* Item 14: Takes a long time to warm up to strangers Item 20: Is very friendly with strangers*

EAS teachers

0.50 0.10 1 0.65 0.44

0.57 0.28 0.81 0.87 0.70

0.52

0.61

0.48

–0.36

–0.33

–0.54

0.53 0.31

0.61 0.76

0.85 0.96

–0.65

–0.45

–0.60

■ Inter-item correlations

0.46 –0.11 –0.20

0.22 –0.72 –0.83

0.42 –0.63 –0.89

0.61 –0.53

0.42 –0.46

0.48 –0.46

0.77

0.83

0.75

0.88

0.75 –0.59 0.26

0.68 –0.54 0.65

In both parent and teacher versions of the EAS, the items belonging to the hypothesized factor structure showed stronger positive correlations compared to childrenrated questionnaires (see Table 4). Several items show low correlations with all other items: item 6 “Tends to be somewhat emotional” in both parent and children-rated EAS, and items 8 “Prefers playing with others rather than alone” and 19 “Reacts intensely when upset”. Interitem correlations also confirm strong overlap between sociability and shyness in the adult-rated questionnaires.

Sociability Item 3: Likes to be with people 0.79 Item 5: Prefers playing with others rather than alone 0.54 Item 10: Finds people more stimulating than anything else 0.47 Item 16: Is something of a loner* –0.29 Item 18: When alone, child feels isolated 0.34 * scores reversed

■ Inter-rater and inter-scale correlations

Table 3 Reliability for the four Temperament scales of EAS children-, parent- and teacher-rated versions (Cronbach’s alpha) French sample

Emotionality Activity Shyness Sociability

Only one item (“Tends to be somewhat emotional”) of the EAS teacher-rated questionnaire did not load on the expected factor “Emotionality”. Internal consistency coefficients were highest for Sociability (0.85) and Activity (0.82) but also satisfactory for Emotionality (0.79) and Shyness (0.72). Inter-factor correlations showed similar trends as observed with the parent-rated version: no correlation of the Emotionality factor, a significant positive correlation between Activity and Sociability (r = 0.70, p < 0.01), a negative correlation (r = –0.79, p < 0.01) between Activity and Shyness and an overlap between low Sociability and high Shyness (r = –0.98, p < 0.01).

EAS parents

Emotionality Item 2: Cries easily 0.76 Item 6: Tends to be somewhat emotional 0.57 Item 11: Often fusses and cries 0.65 Item 16: Gets upset easily 0.31 Item 19: Reacts intensely when upset –0.02 Activity Item 4: Is always on the go Item 7: When child moves about, he/she moves slowly* Item 9: Is off and running as soon as he/she wakes up Item 13: Is very energetic Item 17: Prefers quiet, inactive games to more active ones*

EAS teacher-rated questionnaire

Mathiesen and Tambs, 1999

EAS Child

EAS Parent

EAS Teacher

EAS (mothers ratings)

0.60 0.60 0.47 0.62

0.70 0.70 0.69 0.77

0.79 0.82 0.72 0.84

0.67 0.75 0.79 0.60

shy” did not load significantly on the expected factor “Shyness”. Cronbach’s alphas for the parent-rated EAS ranged from 0.69 to 0.77 (see Table 3). Inter-factor correlations indicated independence of “Emotionality”, whereas “Activity” showed marked correlations with “Sociability” (r = 0.67, p < 0.01) and “Shyness” (–0.62, p < 0.01). Between “Sociability” and “Shyness” there was a strong negative correlation (r = –0.91, p < 0.01) suggesting an overlap between low sociability and shyness in the parent-rated EAS.

Pearson’s correlation coefficients for inter-rater correlations between children and parents were 0.30 for emotionality, 0.11 for activity, 0.15 for shyness and 0.09 for sociability. They ranged from 0.24 (activity) to 0.002 (sociability) between children and teachers. Generally, concordance was low between children- and adult-rated versions and modest between parent- and teacher-rated questionnaires (0.17 for emotionality, 0.35 for activity, 0.24 for shyness, 0.29 for sociability.), the highest value of the Pearson correlation coefficient being observed between parents and teachers for activity. Correlations between EAS and CBCL were significant between internalized symptoms and teacher-rated shyness (r = 0.21, p < 0.01) and between externalized symptoms and parent-rated activity (r = 0.25, p < 0.01). No significant correlations were observed between childrated EAS and CBCL scores (Table 5).


105

Table 4 Inter-item correlation matrix for children and parent EAS EAS Parents items 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

1

2

3

4

1 –0.07 –0.01 0.10 0.32 0.26 –0.16 –0.09 –0.05 0.56 –0.03 –0.10 0.19 0.29 0.04 0 0.01 0.16 –0.09

1 0.39 0.67 0.21 –0.10 0.70 0.45 0.63 –0.12 0.75 0.43 –0.31 –0.10 –0.55 –0.10 0.15 –0.03 0.38

1 0.46 –0.03 –0.03 0.24 0.41 0.50 0.10 0.37 0.52 –0.23 0.23 –0.36 –0.37 0.23 0.24 0.21

1

2

3

4

5

6

0.06 0.17 –0.22 0.15 0.23 0.39 –0.26 –0.30 –0.03 0.14 0.15 –0.06 0.28 0.23 0.08 0.02 0.09 –0.11 –0.23

1 0.07 –0.27 0.18 0.53 0.08 0.05 –0.25 0.03 0.06 0 –0.24 0.10 0.25 0.05 0.24 0.22 –0.04 –0.25

1 0.23 0.57 0.30 0.04 0.23 –0.03 0.56 0.16 0.49 0.32 0.23 –0.02 –0.38 –0.29 0.32 –0.15 –0.26

1 0.21 –0.16 –0.27 –0.10 0.28 0.10 –0.10 0.18 0.28 –0.07 –0.05 –0.23 –0.35 –0.02 0.05 0.16

1 0 0.07 0.06 –0.03 0.45 0.36 0.27 0.16 0.22 0.11 –0.21 –0.08 0.36 –0.03 –0.33

1 0.09 –0.03 –0.24 0.15 0.39 0.15 –0.02 0.06 0.11 0.14 0.17 0.22 0 –0.18

1 0.28 –0.10 –0.16 0.08 0.19 0.22 –0.29 –0.13 0.01 0.15 –0.17 –0.21 0.44 0.16 0.10 0.26 0.02 0.06. –0.52

5

6

7

1 0.20 1 –0.08 0.11 1 0.48 0.14 –0.08 0.34 0.22 –0.32 0.72 0 –0.03 0.14 0.13 0.08 0.70 0.21 –0.07 0.45 0.14 –0.33 –0.10 0 0.25 0.15 –0.08 0.20 –0.55 0 0.36 –0.32 0.07 0.30 0.36 0.06 0.10 0.26 0.07 –0.08 0.14 0.02 –0.06

8

9

10

1 0.34 0.53 –0.11 0.65 0.39 –0.48 –0.16 –0.51 –0.12 0.09 0.08 0.51

1 0.36 –0.03 0.32 0.59 –0.39 –0.04 –0.24 –0.17 0.04 –0.08 0.32

8

9

11

12

13

14

15

16

17

1 –0.01 1 0.69 –0.08 1 0.33 –0.09 0.51 1 –0.23 0.17 –0.29 –0.37 1 0.06 0.73 –0.15 –0.11 0.20 1 –0.57 0.17 –0.57 –0.36 0.37 0.18 1 –0.30 0.10 –0.27 –0.48 0.37 0.05 0.51 1 0.37 0.19 0.24 0.11 0.05 0.22 0.01 –0.09 0.12 0.53 0.09 0.17 0 0.45 0.03 –0.07 0.34 –0.18 0.35 0.27 –0.65 –0.19 –0.18 –0.05

18

19

1 0.29 0

20

1 0.02

1

EAS-Childen items 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

7

10

11

12

13

14

15

16

17

18

19

20

1 –0.09 1 –0.27 0.13 1 00 0.18 –0.04 1 –0.04 0.03 –0.12 0.19 1 –0.05 0.04 –0.10 0.32 0.01 1 –0.19 0.26 0.15 0.06 –0.03 0.21 1 0.33 –0.10 –0.17 0.04 0.23 0.06 –0.10 1 –0.10 0.03 –0.12 0.02 0.38 0.11 –0.06 0.14 1 0.07 –0.22 –0.26 –0.14 0.19 –0.11 0.04 0.01 0.16 1 0.33 –0.10 –0.47 0 0.15 –0.08 –0.32 0.26 –0.06 0.30 1 –0.07 0.03 –0.04 0.25 0.31 0.06 –0.10 0.25 0.14 –0.05 –0.07 1 –0.07 0.12 0.22 –0.19 0.06 0 0.10 –0.08 0.14 –0.12 –0.03 0.26 1 –0.10 0.08 0.05 0.02 –0.14 –0.16 –0.12 –0.44 –0.17 –0.12 –0.06 –0.10 –0.04

Table 5 Means and standard deviations for EAS temperament dimensions EAS Child EAS Parent EAS Teacher * p < 0.05; ** p < 0.01

Girls n=50 Boys n=67 Girls n=73 Boys n=89 Girls n=84 Boys n=107

Emotionality Mean (SD)

Activity Mean (SD)

Sociability Mean (SD)

Shyness Mean (SD)

2.99 (0.87)* 2.64 (0.91) 2.98 (0.83) 2.91 (0.83) 2.86 (0.46) 2.94 (0.47)

3.38 (0.80) 3.77 (0.78)** 3.70 (0.77) 3.96 (0.72)* 3.06 (0.51) 3.28 (0.63)**

3.87 (0.80) 3.84 (0.78) 3.71 (0.79) 3.69 (0.80) 2.68 (0.47) 2.65 (0.47)

2.77 (0.65) 2.63 (0.70) 2.38 (0.89) 2.21 (0.74) 2.73 (0.79) 2.59 (0.85)

1

106


■ Age and gender differences In our sample, there were no significant differences in temperament scores between age groups. Higher activity scores in boys were reported with parent-, teacherand child-rated EAS. Girls obtained significantly higher scores for Emotionality in the child-rated EAS, whereas no differences were observed in the adult-rated versions.

Discussion The confirmatory factor analysis of the French version of EAS revealed that most of the indices of fit were just below the standard criteria. Compared to the adult-rated questionnaires, the child-rated EAS had better indices of fit, but only one reaching standard criteria. Moreover, several items did not load significantly on the a priori factors.For child- and parent-rated EAS,this was mostly observed with items initially belonging to the Sociability or to the Shyness scale. Our findings indicate that the theoretical four-factor structure including Sociability was not completely replicable in our sample. Inter-scale correlations show a marked overlap between high Shyness and low Sociability scores especially for parent- and teacherrated questionnaires. However, a confirmatory factor analysis carried out with a three-factor solution did not improve the indices of fit. Boer and Westenberg (6) who examined the psychometric properties of EAS in a Dutch sample aged four to 13 years, found that the Sociability items were significantly related to both Shyness and Activity. They observed an age trend, from an association between high Sociability and low Shyness in the youngest age cohort (mean age = 6.4 years) toward a closer relation between Sociability and Activity in the oldest age cohort (mean age = 10.5 years). They hypothesized a developmental interpretation of these findings, suggesting that in a very young population of toddlers and infants, Sociability and Shyness may be undistinguishable. Mathiesen and Tambs (22) tested this hypothesis in a Norwegian sample of infants aged 18, 30 and 50 months. They found a four-factor structure including the Sociability scale to be the best fitting model across all age groups. A four-factor structure has also been hypothesized by Rowe and Plomin (26) but the Sociability scale was not examined in their sample of American children. The divergent results concern the position of the experimental Sociability scale and the hypothesis of a clearer differentiation, with increasing age between Sociability and Shyness. Even though the mean age of our sample was greater (mean age 9.3 years) than the mean age of the Norwegian and the Dutch children, our results show an overlap between Shyness and Sociability. Both scales were moderately correlated to Activity and only Emotionality was an independent factor.

Teacher- and parent-rated EAS questionnaires yielded good internal consistency with average alpha coefficients of respectively 0.79 and 0.71. For the children-rated questionnaire, the average alpha coefficient was 0.57 indicating poor reliability of the EAS temperament questionnaire in this age group. The weak interitem correlations of item 8 “Prefers playing with others rather than alone” and item 19 “Reacts intensely when upset”, could be due to lack of understanding and random responses, in particular because both items contain two clauses. In all rater groups, the Shyness scale had the lowest reliability estimates. Compared to previous findings, internal consistency, except for the children-rated questionnaires, is similar to the values observed in the Dutch sample (average value of 0.78) (6) and higher than the alpha coefficients obtained with the Norwegian children (0.62 at 18 month, 0.66 at 30 month and 0.70 at 50 month) (22). Considering the low internal consistency of the child-rated EAS, its psychometric properties should be further examined in older children and adolescents, less likely to be hampered by lacking comprehension of the EAS items. For the adult-rated questionnaire, good reliability estimates and positive correlations with related behavioral variables support the validity of the French version of EAS. However, results of the confirmatory factor analysis indicate that cross-cultural comparability is limited. Cultural differences in interpretation of children’s temperamental characteristics by adults or developmental variations in the expression of temperament traits may account for the lacking fit between the data and either three- or four-factor models of EAS. These factors may account for the overlap between sociability and shyness in both parents’ and teachers’ questionnaires. On the other hand, the week inter-item correlations of item 6 “Tends to be somewhat emotional” in children, parent and teacher questionnaires may be better explained by translation bias. The interpretation of the weak inter-rater correlations should take into account the psychometric limitations of the child-rated EAS. However, the results are consistent with findings in clinical samples showing that emotional symptom scores are greater in children- compared to parent-ratings of children’s psychopathology (3, 4, 12) while the contrary has been observed for activity-level scores (2, 3). Similar results have been obtained in longitudinal temperament studies by Maziade et al. (23, 24): emotional symptoms were identified in self-reports but not in diagnostic interviews in subjects with difficult temperament in childhood. Our findings indicate that subjective, cultural and developmental factors might play an important part in the interpretation of temperament dimensions. These results emphasize the need for multiple evaluations of temperament characteristics based on ratings of different observers and careful examination of cross-cultural comparability.


Studies with larger and older samples are needed to further verify the measurement models.

107

■ Acknowledgements The authors wish to thank M. Brodin, B. Falissard for their contributions to the research, S. Cook for comments on the manuscript and P. Cador for technical assistance.

References 1. Achenbach TM, Edelbrock C (1983) Manual for the Child Behavior Checklist and Revised Child Behavior Profile, University of Vermont, Department of Psychiatry, Burlington 2. Andrews VC, Garrison CZ, Kirby LJ, Addy CL, McKeon RE (1993) Motheradolescent agreement of the symptoms and diagnosis of adolescent depression and conduct disorders. Journal of the American Academy of Child and Adolescent Psychiatry 32:731–737 3. Angold A, Weissman MM, John K, Merikangas KR, Prusoff BA,Wickramaratne P, Gammon GD, Warner V (1987) Parent and child reports of depressive symptoms in children at low and high risk of depression. Journal of Child Psychology and Psychiatry 28: 901–915 4. Barrett ML, Berney TP, Bhate S, Famuyiwa OO, Fundudis T, Kolvin I, Tyrer S (1991) Diagnosing childhood depression. Who should be interviewed – Parent or child? The Newcastle child depression project. British Journal of Psychiatry. 159 (suppl 0.11):22–27 5. Bentler PM (1980) Multivariate analysis with latent variables: causal modeling. Annual Review of Psychology 31: 419–456 6. Boer F, Westenberg PM (1994) The factor structure of the Buss and Plomin EAS Temperament Survey (parental ratings) in a Dutch sample of elementary school children. Journal of Personality Assessment 62, 537–551 7. Buss AH, Plomin R (1975) A temperament theory of personality development. Wiley, New York 8. Buss AH, Plomin R (1984) Temperament: early developing temperament traits. Lawrence Erlbaum Associates, Hillsdale, NJ 9. Carmines EG, McIver SP (1981) Analyzing models with unobserved variables: analysis of covariance structures. In: Bohrnstedt GW, Borgatta EF (eds), Social measurement: Current issues, Sage, Beverly Hills, CA, pp 65–115

10. Cheek JM, Buss AH (1981) Shyness and sociability. Journal of Personality and Social Psychology 41:330-339 11. Cloninger CR (1987) A systematic method of clinical description and classification of personality variants: a proposal. Archives of General Psychiatry 44:573–588 12. Edelbrock C, Costello AJ, Dulcan MK, Conover NC, Kalas R (1986) Parentchild agreement on child psychiatric symptoms assessed via structured interview. Journal of Child Psychology and Psychiatry 27:181–190 13. Eysenck HJ (1984) Models of personality. Clinical Neuropharmacology 7 (suppl 1):744–745 14. Fombonne E, Chehdan F, Carradec AM, Achard S, Navarro N, Reiss S (1988) Le Child Behavior Checklist: un instrument pour la recherche en psychiatrie de l’enfant. Psychiatrie et Psychobiologie 3:409–418 15. Gibbs MV, Reeves D, Cunningham CC (1987) The application of temperament questionnaires to a British sample: issues of reliability and validity. Journal of Child Psychology and Psychiatry 28:61–77 16. Goldsmith HH, Rieser-Danner LA, Briggs S (1991) Evaluating convergent and discriminant validity of temperament questionnaires for preschoolers, toddlers and infants. Developmental Psychology 27:566–579 17. Goodyer IM, Ashby L, Altham PL, Vize C, Cooper PJ (1993) Temperament and major depression in 11 to 16 years olds. Journal of Child Psychology and Psychiatry 34:1409–1423 18. Kagan J, Reznick JS, Snidman N (1987) The physiology and psychology of behavioral inhibition in children. Child Development 58:1459–1473 19. Kagan J, Snidman N, Arcus D (1995) The role of temperament in social development. Annals of the New York Academy of Science 771:485–490

20. Kelvin RG, Goodyer IM, Altham PME (1996) Temperament and psychopathology amongst siblings of probands with depressive and anxiety disorders. Journal of Child Psychology and Psychiatry 37:543-550 21. Masse LC, Tremblay RE (1997) Behavior of boys in kindergarten and the onset of substance use during adolescence. Archives of General Psychiatry 54:62-68 22. Mathiesen KS, Tambs K (1999) The EAS temperament questionnaire – factor structure, age trends, reliability and stability in a Norwegian sample. Journal of Child Psychology and Psychiatry 40:431-439 23. Maziade M, Caron C, Côté R, Mérette C, Bernier H, Laplante B, Boutin P, Thivierge J (1990) Psychiatric status of adolescents who had extreme temperaments at age 7. American Journal of Psychiatry 147:1531–1536 24. Maziade M, Caron C, Cote R (1990) Extreme temperaments and diagnosis. Archives of General Psychiatry 47:477–484 25. Michel G, Mouren-Siméoni MC, PerezDiaz F, Falissard B, Carton S, Jouvent R (1999) Validation and construction of a sensation-seeking scale for adolescents. Personality and Individual Differences 26:159–154 26. Rove DC, Plomin R (1977) Temperament in early childhood. Journal of Personality Assessment 41:150-156 27. Steiger JH (1990) Structural model evaluation and modification: an interval estimation approach. Multivariate Behavioral Research 25:173–180 28. Stevens J (1996) Exploratory and confirmatory factor analysis. In: Stevens J (ed) Applied multivariate statistics for the social sciences. 3rd edn. Lawrence Erlbaum Associates, Mawah, NJ, pp 362–427 29. Wills TA, Vaccaro D, Mc Namara G (1994) Novelty seeking, risk taking and related constructs as predictors of adolescent substance use: an application of Cloninger’s theory. Journal of Substance Abuse 6:1–20