Categorii

Editii

NORMATIVE DATA IN THE ROMANIAN POPULATION FOR THE TEMPERAMENT AND CHARACTER INVENTORY (T.C.I.) PERSONALITY QUESTIONNAIRE - Mircea Al. BIRT, Aura VAIDA

Mircea Al. BIRT Aura VAIDA

INTRODUCTION The Temperament and Character Inventory personality questionnaire was conceived by Cloninger et al. (1994), who reached the conclusion that the personality model elaborated by Eysenck was inacceptable because it was too general and could be applied to both "normal" and "pathological" subjects. Starting from the theory of operative learning, Cloninger implemented a neurobiological model underlying the rational development of temperament descriptors (Cloninger, 1987, 1991).

The temperament and character inventory (TCI) represents a test battery intended to measure the differences between individuals according to the 7 dimensions of temperament and character. Cloninger attempted to describe personality by several fundamental variables based on temperament, supposed to be genetically transmitted and controlled by specific neurobiological systems (Bouvard, 1999). Temperament refers to automatic emotional responses that occur in various situations, which are to a certain extent inborn and stable in the course of life. In this sense, temperament was identified by three independent dimensions as early as 1987: novelty seeking (NS), harm avoidance (HA) and reward dependence (RD). In 1993, Cloninger et al. isolated a fourth dimension of temperament – persistence (P), starting from a less stable dimension, reward dependence. People with high scores for novelty seeking (NS) tend to be more curious, disorderly, they get bored rapidly, are more impulsive, prefer change and are nonconformist. Subjects with a high score for harm avoidance (HA) appear as shy, pessimistic, fearful, hesitating and fatigable. In contrast, those with high scores for reward dependence (RD) are sensitive, warm and fragile under conditions of social isolation. Subjects with a high score for P are ambitious, perseverent, determined and perfectionist. Novelty seeking (NS) would be under the control of the behavioral activation system sensitive to unknown and reward stimuli, and implicitly of dopaminergic activity.

This dimension corresponds to Eysenck's extrovesion and Zuckerman's sensation seeking. Harm avoidance (HA) is dependent on the system of behavioral inhibition to signals of punishment or frustration induced by serotoninergic activity, being similar to neuroticism described by Eysenck. Reward dependence (RD) is tributary to the system of behavioral maintenance in the absence of continuous reinforcement and would be conditioned by central noradrenergic activity. This dimension involves social attachment. At the same time, Cloninger et al. isolated 3 dimensions of character, as it is known that this represents the acquired component of personality. Character refers to personal beliefs and individual differences regarding the values and aims that influence voluntary choices (decisions), intentions and the significance of events experienced in the course of life. Differences in terms of character are moderately influenced by sociocultural aspects and the process of maturation during life. The 3 dimensions of character are: self-directedness (S), cooperativeness (C) and self-transcendence (ST). All these dimensions reflect the subject's maturity and adaptation acquired during the subject's development under the influence of environment and learning (Bouvard, 1999).

It is important that these 3 dimensions of character are susceptible to evolve in time depending on the subject's maturity and especially according to possible mental disorders. Self-directedness (S) is characterized by individual maturity (self-awareness); cooperativeness (C) is defined by social maturity (acquisition of social conscience), and self-transcendence (ST) refers to spiritual maturity (universal conscience, meditation or prayer). This questionnaire explores both the genetic component and the acquired component of personality by the identification of 4 temperament dimensions and 3 character dimensions. All these aspects of personality interact and contribute to the process of adaptation to the events experienced in the course of life, also influencing the vulnerability to emotional and behavioral disorders. The inventory was conceived so as to allow the evaluation of differences in both normal subjects and subjects with deviant behavioral patterns. The inventory has proved its efficacy in clinical practice, being useful for the establishment of clinical diagnosis, differential diagnosis, the therapeutic plan in different mental disorders, and also for the dynamic evaluation of patients attending psychotherapy sessions. The area of applicability is extremely wide: anxiety disorders, eating disorders, alcohol and other addictions, personality disorders. TCI is also useful in the evaluation of psychopathological comorbidity, as clinical differences between and within the different psychopathological syndromes (e.g. anxiety, affectivity, eating disorders, substance abuse) are expressed by differences existing in TCI profiles. The differences in the response to psychoactive substance treatment, including andidepressant and antianxiolytic agents, are tightly correlated with temperament variables of TCI.

OBJECTIVES The TCI personality questionnaire has been so far translated into several languages (French, Italian,Czech, Spanish, Japanese, etc.) and there are also validity studies, as well as standards specific for these populations. In the majority of the studies, factorial analysis has been confirmed, being identical to that obtained in the USA population (Kozeny et al., 1989; Svrakic et al., 1991; Manfredonia et al., 1991; Armani et al., 1994). In the study carried out in the Japanese population (Takeuchi et al., 1993), the original factorial structure based on 7 factors was not confirmed; in contrast, a 6 factor structure was obtained, while the original number of items was maintained. The validation study performed in the French population (N=602 subjects) only identified 3 factors for temperament and 3 factors for character (Pélissolo et al., 1998). The authors of these studies positively appreciate the efficacy and utility of this questionnaire for both the normal population and psychiatric population, being a useful instrument in the making of diagnosis, the elaboration of the therapeutic plan and the evaluation of the efficacy of drug therapies and psychotherapies.The utility of this questionnaire being demonstrated by various studies, this investigation attempted to establish normative data for the TCI questionnaire, as well as to carry out the confirming factorial analysis in a group of 350 subjects extracted from the general population. In order to achieve these aims, the first step consisted in the translation of the questionnaire and the performance of a preliminary study for the identification of ambiguous, unclear or insufficiently defined terms.

SUBJECTS Normative data were established for a group of 350 subjects extracted from the general population. Age ranged between 18 and 82 years, with a mean ageMa=12.75 (SD=11.61). Women represented 69.14% (N=242) and men 30.86% (N=108). The frequency of distribution according to education is shown in Figure 1

METHOD The first step consisted of the translation of the questionnaire by several persons (specialists and nonspecialists), after which the translations were compared and only the concordant items were retained. Then, the test in the Romanian version was translated back into English, and the resulting version was compared to the original variant.

The items that did not correspond to the original variant were reevaluated. The final variant of the test in Romanian was administered to a group of 30 subjects (pilot study) in order to establish whether the items were easy to understand, sufficiently clear and well defined. There were several items that needed to be redefined in order to improve comprehension and clarity, while maintaining the original meaning as accurate as possible. The final version of the questionnaire was subsequently administered to a group of 350 subjects, the results obtained representing the database necessary for the establishment of norms and the reperformance of factorial analysis. TCI is a self-evaluation questionnaire by which subjects are asked to answer by "true" or "false" to 240 statements depending on their own interests, emotional reactions, attitudes, aims and values.

The analysis of the questionnaire allows the calculation of the main scores corresponding to the 4 temperament dimensions and the 3 character dimensions, as well as of 25 subscores corresponding to the 3 to 5 subscales of each main dimension:

- novelty seeking (NS) contains 40 items
- harm avoidance (HA) = 35 items
- reward dependence (RD) = 24 items
- persistence (P) = 8 items
- self-directedness (S) = 44 items
- cooperativeness (C) = 42 items
- self-transcendence (ST) = 33 items.

A total of 226 items result that are subsumed under the 7 dimensions; the other items up to 240 are not evaluated. Results can be expressed in raw or standard scores (T scores).

RESULTS AND DISCUSSION The questionnaire was studied in a general population by Cloninger et al. (1993) and in a psychiatric population by Svrakic et al. (1993). The French study was performed in students, control subjects, depressive and anxious patients. There are quite few validity studies performed for the TCI questionnaire. In contrast, for temperament evaluation scales (questionnaire termed TPQ), there has been a greater number of such studies.

Regarding the verification of the factorial structure of the questionnaire, the high number of items should be taken into consideration, and the investigation of extremely large subject samples is required. The results were processed using the SPSS, Excel and NCSS programs, which allowed descriptive analysis (calculation of means, standard deviations, standard T scores, percentiles) for each dimension, as well as inferential analysis (differences between means according to sex, age, education). The reperformance of factorial analysis, the calculation of Cronbach alpha coefficients, of correlations between scales, were also initiated.

Frequency of distribution according to education
Table 1. Descriptive analysis (means, standard deviations) for the 7 TCI dimensions (350 subjects)
Table 2. Descriptive analysis (means, standard deviations) and inferential analysis (Student "t" test)

The confirming factorial analysis was also performed in order to see if the factors existing in the original test (amounting to 7) were also found in our sample. Normative data (means, standard deviations) are shown in Table 1, and "T" standard scores and percentiles are shown in Table 2. A comparison of the means obtained for the 7 dimensions of the questionnaire and the normative data established by Cloninger in the American population (300 subjects), shows differences in the mean values, with small variations depending on sex and age groups. For our sample, the means, standard deviations and differences between the means depending on sex, age and education are presented in Tables 2, 3, 4 and 5. Statistically significant differences are found between men and women for dimensions (see Table 2):
- NS (and for subscales NS2 and NS4) with higher values in men;
- HA (and for subscales HA1, HA2 and HA4) with higher values in women;
- RD (for all subscales: RD1, RD3, RD4) with higher values in women;
- C (with all subscales C1, C2, C3, C4, C5) with significantly higher values in women; - for subscales S5 and ST3 also with significantly higher values in the case of women.

Table 3. Descriptive analysis (means, standard deviations) and inferential analysis (Student "t" test) by age groups

Our results are in accordance with the results obtained by Cloninger, in which women seem to have higher scores for harm avoidance scales (HA), reward dependence (RD), cooperativeness (C) and self-transcendence (ST). Men have higher scores for novelty seeking (Cloninger, 1992, 1993). The French version finds high scores in women for the HA and RD scales, and in men for the self-directedness scale (S) (Pélissolo, 1998; Le Bon, 1998). The mean scores of the French version are significantly different compared to those of the American version (Pélissolo, 1998). As a conclusion, we may ascertain the presence of significant differences between sexes, women having a more marked tendency to avoid dangers, to be more anxious, more sensitive to social rewards and more sociable compared to men, who tend to adopt more frequently behaviors that involve certain risks, having an increased need for novelty seeking, stimulation and avoidance of monotony. The data of Table 3 show the presence of significant differences between the different age groups for some dimensions, these differences being more obvious between young subjects and middle aged or elderly subjects. Differences between the 3 age groups appear for dimensions NS (a characteristic of temperament) and C and ST (which represent character dimensions). These differences can be explained by the fact that young people search more for sensations and stimulations as compared to adults, and the conceptions, ideas and values that influence their priorities and guide them are much different from those of middle aged and elderly people. Table 5 shows intermean differences depending on education, with statistically significant differences for some temperament and character dimensions: NS, HA, S, C, ST. The higher the level of education of these persons, the more they will search for stimulation and the more they will be involved in various activities, trying to avoid monotony, tending to avoid dangers and not to act on the impulse of the moment, they will be more sociable and communicative.

Table 4. Descriptive analysis (means, standard deviations) according to education (training)
Table 5. Inferential analysis (Student “t” test) according to education

Table 6 presents raw scores, standard "T" scores and percentiles, these data being necessary for the interpretation of results and in order to establish if the data obtained by a subject are at average level, below or above the average. Cloninger proposes the following value groups:

- 0-16.7% extremely low scores
-17-33% low scores
- 34-66.7% medium scores
- 67-88.3% high scores
- 84-100% extremely high scores

Table 6. Standard scores and percentiles for the 7 TCI dimensions (350 subjects)-1
Table 6. Standard scores and percentiles for the 7 TCI dimensions (350 subjects)-2
Table 6. Standard scores and percentiles for the 7 TCI dimensions (350 subjects)-3

The reliability of the questionnaire was evaluated by the calculation of Cronbach alpha internal consistency coefficients. These were calculated for the whole sample of 350 subjects for each dimension, but also depending on sex. In the case of our sample, internal consistency coefficients are situated between 0.47 and 0.66, while in the case of the American sample, the internal consistency of the 7 main dimensions varies from 0.65 (Persistence) to 0.87 (Cloninger et al., 1993).

The internal consistency coefficients are presented in Table 7, for each dimension and for the whole questionnaire, both in the 350 subjects and depending on sex. Since the questionnaire is conceived so that the temperament and character evaluation scales can also be used independently, in what follows we attempt to perform factorial analysis both separately for each of these two scales and for the whole questionnaire.

STRUCTURE OF TEMPERAMENT EVALUATION SCALES

In what follows, we attempt to reperform factorial analysis for the scales that evaluate temperament dimensions and separately, for those that evaluate character. The first step in the performance of confirming factorial analysis is represented by the visual inspection of the correlation matrix, which is shown in Table 8. The absence of null correlations between scales is noted, so there is no identical matrix, which allows the performance of factorial analysis.

After the inspection of the correlation matrix, Bartlett's test of sphericity is calculated, similarly to the chi-square test. If the value of this test is sufficiently high, we can definitely reject the null hypothesis according to which the correlation matrix is an identical one. The value of Bartlett's test is 741.58, which warrants the performance of factorial analysis. For temperament evaluation scales, the main components were extracted, with a resulting number of 3 factors. The first factor is responsible for 46.57% of total variance, the second factor for 32.25% of variance, and the third one for 21.20% of total variance.

The cumulative percentage of the 3 factors represents 100% of total variance. After the extraction of the 3 factors, in order to identify the scale structure, but also for a more accurate interpretation of factors, we used the performance of rotation by the VARIMAX method. The results obtained are presented in Tables 9 and 10. An inspection of Tables 9 and 10 indicates that factorial analysis only evidences 3 factors compared to 4 in the original variant of the questionnaire. Also, the structure of factors is found to be slightly different, in the sense that for factor 1, the NS1 subscale is added to the four specific subscales (HA1, HA2, HA3, HA4), and the loading of the P scale, which in the original variant represented a separate factor, is high in our sample within factor 2, factor that also includes subscales NS2, NS3 and NS4. The presence of higher correlations is also found between P and the 3 subscales (NS2, NS3, NS4) (see Table 8). Some differences are noted in the structure of factors between men and women: NS2 is subsumed under factor 3 in the case of men, and factor 2 lacks P, which has a higher loading within factor 1, but not high enough to be subsumed under factor 2. Otherwise, the structure of factors differentiated by sexes follows the general tendency noted in the whole sample. The NS1 subscale evaluates exploratory excitability versus rigidity. In the case of our sample, it seems that subjects perceive this characteristic as being associated with a tendency to avoid dangers, to feel anxiety about changes, the mean scores for this characteristic being relatively high. This shows the presence of a certain rigidity, of a certain fear of changes, with a tendency to consider in particular the negative aspects of changes. By reperforming factorial analysis, Bagby et al. (1992) also found the presence of 3 factors for the temperament evaluation scale, but in their case, the P subscale is found with subscales RD1, RD3 AND RD4.

Table 7. (Cronbach alpha) internal consistency coefficients
Table 8. Correlation matrix (x100) for temperament evaluation scales (350 subjects)
Table 9. Saturation of factors for temperament evaluation scales after the
Table 10. Loading of factors for temperament evaluation scales after the performance of Varimax
Table 11. Correlation matrix (x100) for character evaluation scales (350 subjects)
Table 12. Saturation of factors for character evaluation scales after Varimax
Table 13. Loading factors for character evaluation scales after Varimex rotation, for the whole smaple
Table 14. Correlation between temperament and character scales (N = 350)
Table 15. Saturation of factors for the TCI questionnaire after Varimax rotation (N = 350)
Table 16. Loading of factors for the TCI questionnaire after Varimax rotation for the whole sample (N = 350) and

Pelissolo et al. (1998) also found 3 factors for temperament evaluation scales, without the P dimension being an independent factor.

STRUCTURE OF CHARACTER EVALUATION SCALES

The factorial analysis for character dimensions was performed following the same steps. The inspection of the correlation matrix and of Bartlett's test value (1072.99), which are shown in Table 11, indicates that there is no identical matrix, so we can proceed to the extraction of the main components. The factorial analysis for the character evaluation scale confirms the existence of the 3 factors, like in the original variant, but the loading of the S4 subscale is found to be insufficient for it to be subsumed under any factor. The comparison of the structure of factors between the two sexes shows differences in the structure of certain factors. Thus, subscales S1, S2 and S5 are part of factor 1 in the case of men, and in women, the subscale S4 along with the 5 subscales of the C dimension are subsumed under this factor, except for the C4 subscale in the case of men. In contrast, factor 2 maintains the same composition, without any sex differences. At the level of factor 3, there are again differences between the two sexes in terms of subscales that are subsumed under this factor.

FACTORIAL STRUCTURE FOR THE TCI QUESTIONNAIRE The factorial analysis performed separately for the temperament and character evaluation scales did not confirm the presence of the original factorial structure with 4 factors for temperament and 3 factors for character. For our sample, factorial analysis showed the presence of 3 factors for temperament evaluation scales and 3 factors for character evaluation scales. Next, we performed a factorial analysis for all the scales of the questionnaire in order to see if the presence of the 7 factors of the original variant was confirmed. The correlations between temperament and character dimensions (according to the original variant) are shown in Table 14. Values higher than 0.40 are in bold. Correlations higher than 0.40 were obtained between HA and S (-0.44), RD and C (0.47), and S and C (0.41). In the case of Cloninger's sample, high correlations were obtained between the same dimensions: r (HA, S) = -0.47; r (RD, C) = 0.54; r (S, C) = 0.57. The correlations between the 7 dimensions obtained in our sample are close to those found by Cloninger, unlike the French study, in which these correlations are poor. The extraction of the main components shows the existence of only 5 factors whose saturation is higher than 1. The results of the factorial analysis are presented in Tables 15 and 16. The reperformance of factorial analysis indicates the presence of only 5 factors, not 7, as in the original structure of the questionnaire. Differences in the structure of factors are found depending on sex, but also compared to the structure obtained for the whole sample.The high correlations obtained between dimensions HA and S, RD and C, and C and S explain why these subscales show a higher saturation within the same factors. For example, factor 4 includes both the subscales of the C dimension and RD dimension. This can be explained by the fact that the subjects of our sample present a poorer differentiation between the two dimensions: the socializing dimension and the reward dependence dimension.

CONCLUSIONS The TCI personality questionnaire appears to be an instrument with acceptable internal consistency and good construct validity, which is proved by factorial analysis. The reperformance of factorial analysis separately for the two subtests (temperament and character evaluation) does not confirm the same factorial structure as in the original version for temperament, but the number of factors is maintained in the case of character, with small differences explained in the section "Results and discussion". The fact that the factorial analysis performed on our sample did not confirm the presence of the 7 factors of the original version can also be explained by the existing cultural differences, by the interpretations and significances given by subjects to various situations, by the way in which certain behaviors considered "normal" or “pathological" are defined and evaluated.

The dimensional model of personality elaborated by Cloninger has not been completely confirmed, which requires further validation studies performed on different populations, both control subjects and subjects with mental disorders. Our investigation regarding the establishment of normative data for the Romanian population represents just the starting point for a much more complex study whose objectives are:

- the continuous updating of these standards and the extension of the subject group so as to include a significantly high number of subjects from different regions of Romania;
- the estimation of the stability of results in time by the calculation of test-retest reliability coefficients;
- the estimation of concurrent validity by the establishment of correlations with other tests that evaluate certain personality aspects;
- the estimation of the predictive validity of the questionnaire;
- the evaluation of the extent to which this questionnaire is useful for the making of differential diagnosis, as well as for the choice of the most efficient class of drug substances in the treatment of various mental disorders.

We wish to thank Prof. Dr. Cloninger (Center for Psychobiology of Personality, Washington University) for the consent obtained for the performance of this study (on 21.09.2002).

1. Bouvard M. (1999): "Questionnaires et l&srquo;echelles d&srquo;evaluation de la personnalite", Masson, Paris, p. 29-35.
2. Cloninger C.R. (1987): "A systematic method for clinical description and classification of personality variants", Arch. Gen.Psychiatry 44, p. 573-588.
3. Cloninger C.R., Przybeck T.R, Svrakic D.M si Wetzel R.D. (1994): "The Temperament and Character Inventory (TCI): A Guide to Its Development and Use", Center of Psychobiology of Personsality, Washington University, St. Louis, Missouri.
4. Cloninger C.R., Svrakic D.M. si Przybeck T.R. (1993): "A psychobiological model of temperament and character", Arch. Gen. Psychiatry 50, p. 975-990.
5. Le Bon O., Staner L., Tecco J., Pull Ch. si Pelc, I. (1998): "Questionnaire de personnalite tridimensionnel (TPQ): validation chez une population controle francophone", Encephale XXIV, p. 40-45.
6. Lepine J.P., Pelissolo A., Teodorescu R. si Teherani M. (1994): "Ëvaluation des proprietes psychometriques de la version française du questionnaire tridimensionnel de la personalite", Encephale XXI, p. 747-753. 7. Pelissolo A. si Lepine J.P.(1997): "Traduction française et premieres etudes de validation du questionnaire de personnalite TCI", Ann. Med. Psychol. 155, p. 497-508.
8. Pelissolo A., Guillem E., Oram S., Lepine J.P. (1998): "The Temperament and Character Inventory (TCI) in a French general population sample: comparative study with US normative data", Communication presentee a l&srquo;APA, Toronto.
9. Sava F. (2004): "Analiza datelor in cercetarea psihologica. Metode statistice complementare", Editura ASCR, Cluj-Napoca.
10. Svrakic, D.M., Whitehead C., Przybeck T.R. si Cloninger C.R. (1993): "Differential diagnosis of personality disorders by the seven factor model of temperament and character", Arch. Gen. Psychiatry 50, p. 991-999.