Cross-cultural validation of the revised Temperament and Character Inventory: Serbian data

Available online at www.sciencedirect.com

Comprehensive Psychiatry 51 (2010) 649 – 655 www.elsevier.com/locate/comppsych

Cross-cultural validation of the revised Temperament and Character Inventory: Serbian data Tamara Dzamonja-Ignjatovic, Dragan M. Svrakic⁎, Nenad Svrakic, Mirjana Divac Jovanovic, Robert C. Cloninger Abstract Objective: In this work, we report data on construct validity and cross cultural applicability of the revised Temperament and Character Inventory (TCI R) (Cloninger, C.R., Przybeck, T.R., Svrakic, D.M., & Wetzel, R.D. (1999). The Temperament and Character Inventoryrevised, Washington University, St. Louis), a 5-point scale scoring formatrevision of the original, true-false version TCI (Cloninger, C.R., Przybeck, T.R., Svrakic, D.M., & Wetzel, R.D. (1994). The Temperament and Character Inventory-A guide to its development and use, Washington University, St. Louis). Both versions are based on the seven factor Psychobiological Model of Personality (Cloninger CR, Svrakic, DM & Przybeck TR (1993). A Psychobiological Model of temperament and Character, Archives of General Psychiatry, 50, 975-990). Methods: The sample consisted of 473 normal adult subjects representing a number of Serbian cities and towns, recruited consecutively while registering at the National Employment Center, located in Belgrade, Serbia. The sample was the designed to be highly representative of urban and suburban population in Serbia to match to TCI R sample in the US. The 240-item, 5-point scale scoring TCI R was used to assess temperament and character traits. In addition to a number of other revisions, the scoring format in the TCI R was changed into a 5-point Likert scale to increase its sensitivity to subtle variations in personality expression. The TCI R mean scores and standard deviations were compared between Serbian and US subjects, internal consistency of the TCI R scales was estimated using Cronbach's alpha coefficients, and principal component analysis was used separately for temperament and character (because of their non-linear relationship) to test the underlying factorial structure of the TCI R. Parallel analysis and randomized simulation data were used to determine the number of factors for temperament and character. Results: The results generally supported the construct validity and the cross cultural applicability of the TCI R in Serbia. With a few exceptions, the observed internal consistency for the TCI R scales was acceptable. For the most part, the US and Serbian subjects manifested comparable temperament traits, whereas US subjects had higher character scores. The observed differences are understood as partly reflective of local culture and partly of dramatic socio-economic change in Serbia over the last 20 years. Principal component analysis fully supported the four factor structure of temperament and the three factor structure of character, as postulated by theory. The inadequacy of using linear statistical methods in studying complex non-linear systems such as personality is discussed in some detail. Published by Elsevier Inc.

1. Introduction The Psychobiological Model of Personality [1] describes 7 personality dimensions (or traits) to account for individual differences in behavior. These 7 traits are conceptualized based on a number of complementary scientific perspectives, such as evolutionary, ethologic, epigenetic, neurobiologic, pharmacologic, cognitive, and so on. Personality develop⁎ Corresponding author. Department of Psychiatry, Washington University School of Medicine, Campus Box 8134, St Louis, Mo 63110. E-mail address: [email protected] (D.M. Svrakic). 0010-440X/$ – see front matter. Published by Elsevier Inc. doi:10.1016/j.comppsych.2009.09.009

ment is understood as a complex, self-organizing adaptive process reflective of person-environment interaction. Personality structure is conceptualized as consisting of 4 emotional dispositions (ie, temperament) and 3 adaptive traits (ie, character) and their lower order facets [2]. As described elsewhere [3], temperament and character are in constant bidirectional interaction to maximize adaptation to the changing environment. The Temperament and Character Inventory [2] is a family of tests, self-reports, and interviews, designed to measure temperament and character traits as defined by the 7-factor model.

650

T. Dzamonja-Ignjatovic et al. / Comprehensive Psychiatry 51 (2010) 649–655

In what follows, we briefly describe temperament and character traits as conceptualized by Cloninger et al [1] and recently reviewed by Cloninger et al [3]. Temperament (or the “biologic core” of personality) involves 4 broad, heritable dispositions to early emotions (eg, fear, aggression, attachment, and perseverance) and related behaviors (eg, inhibition, activation, continuation, and persistence of behavior) in response to specific environmental stimuli (eg, danger, novelty, reward, and frustrative nonreward, respectively). The 4 temperament traits, labeled harm avoidance, novelty seeking, reward dependence, and persistence (Table 1), are independently inherited, relatively stable over lifetime and crossculturally universal [1,3,4]. Of note, one of the most important findings in personality research for the past 30 years is that these 4 dimensions underlie not only normal

Table 1 Descriptors of individuals with high and low scores on temperament and character scales and subscales Temperament trait Harm avoidance (HA) HA1: worry and pessimism HA2: fear of uncertainty HA3: shyness HA4: fatigability Novelty seeking (NS) NS1: exploratory excitability NS2: impulsiveness NS3: extravagance NS4: disorderliness Reward dependence (RD) RD1: sentimentality RD2: openness to communication RD3: attachment RD4: dependence Persistence (PS) PS1: eagerness of effort PS2: work hardened PS3: ambitiousness PS4: perfectionism II. Character traits Self-directedness (SD) SD1: responsibility SD2: purposefulness SD3: resourcefulness SD4: self-acceptance SD5: congruent second nature Cooperativeness (CO) C1: social acceptance C2: empathy C3: helpfulness C4: compassion C5: pure hearted Self-transcendence (ST) ST1: self-forgetful ST2: transpersonal identification ST3: spiritual acceptance

High scorer

Low scorer

Pessimistic Fearful Shy Fatigable

Optimistic Daring Outgoing Energetic

Exploratory Impulsive Extravagant Irritable

Reserved Deliberate Thrifty Stoical

Sentimental Open Warm Appreciative

Detached Reserved Cold Independent

Industrious Determined Enthusiastic Perfectionistic

Inert Spoiled Underachiever Pragmatic

Responsible Purposeful Resourceful Self-accepted Disciplined

Blaming Aimless Inept Vain Undisciplined

Tenderhearted Empathic Helpful Compassionate Principled

Intolerant Insensitive Hostile Revengeful Opportunistic

Intuitive Acquiescent Spiritual

Contrived Controlling Materialistic

temperament but also represent dimensional structure of personality disorders [5]. The latter are diagnosed based on immaturity, that is, low character traits [6]. From a developmental perspective, these early and heritable behavior dispositions are, as it were, “building blocks” for a wider repertoire of adaptive behaviors to specific environmental stimuli. Recent advances in epigenetic studies [7] point to the critical significance of gene-environment (GxE) interaction in behavior development. Specifically, developmental branching of behaviors into more and more specialized traits results from GxE interaction, with numerous epigenetic modulations, such as environmentally induced silencing or activation of large numbers of gene networks and coordinated binding of transcription factors (so-called transcriptomes) in multiple gene regions (so-called regulons) in response to specific environmental stimuli —all without changes in the DNA sequence. This corresponds to our conceptualization of character development as an adaptive interface between temperament and the surrounding world that optimizes one's adaptation through person-environment interaction [8]. Specifically, character (or the “adaptive interface” of personality) involves individual differences in higher cognitive processes, such as logic, formal construction, symbolic interpretation, and so on. Character traits, such as self-directedness, cooperativeness, and self-transcendence (Table 1), facilitate one's adaptation to the society and are clinically used to assess the probability of personality disorder [6]. Character traits develop during childhood and adolescence and continue to change epigenetically, through GxE interaction, during lifetime, more so than temperament traits. As shown in prior work [8], character develops as a nonlinear function of antecedent temperament traits, sociocultural factors, and random life events unique to the individual. Character is an emerging property at the “interface” between internal and external environments to optimize one's adaptation to these environments. The Temperament and Character Inventory (TCI) main temperament and character scales, their facets, and descriptors for low and high scorers are illustrated in Table 1.

2. Aims of the study The primary goal of this study was to test the conceptual validity and cross-cultural applicability of the revised TCI (TCI R), developed in the United States, in a culturally, socially, and economically diverse setting of Serbia. This study is a continuation of our previous work with an earlier version of the TCI in Serbia in a number of clinical and nonclinical samples [9,10]. It is also a contribution to the current research studying psychometric properties of the TCI R in different societies, cultures, and social environments [11]. Pelissolo, et al, 2008.

T. Dzamonja-Ignjatovic et al. / Comprehensive Psychiatry 51 (2010) 649–655

651

3. Methods 3.1. Sample The sample consisted of 473 subjects, 174 males (36.8%) and 299 females (63.2%), between 18 and 50 years of age (average age, 27.33 years). The subjects were recruited while registering at the National Employment Center, Belgrade, Serbia. All subjects gave consent to participate in the study. All educational levels and socioeconomic classes (to represent urban and suburban population in Serbia) were included to match the US normative sample as closely as possible. A total of 158 (33.4%) of subjects were from Belgrade and 315 (66.6%) from other cities and towns, which is roughly paralleling the ratio of population distribution in the Republic of Serbia. 3.2. Personality measures The TCI R is a 240-item self-report developed by [12] to revise some scales and items, particularly reward dependence, persistence, and self-transcendence, and to change the old “true-false” format into a 5-point Likert scoring format. The Likert-type scales are widely accepted as reliable and valid for personal attitudes. The subject is asked to indicate whether he/she agrees or disagrees with each of the items (statements). In the TCI R, 5 options are provided: “strongly agree,” “agree,” “undecided,” “disagree,” and “strongly disagree,” and these responses are scored on a 1 to 5 scale (3 = “undecided”). We used the most recent version of the TCI R [12], with 7 main scales, 29 subscales, and 240 items. Although the total number of items was not changed from the original TCI (ie, both versions have 240 items), some items were added or removed based on their tested psychometric characteristics. The novelty seeking scale was reduced from 42 to 35 items and the harm avoidance scale from 36 to 33 items. Three items were added to the reward dependence scale that now consists of 30 items. One additional subscale (“openness to warm communication vs aloofness”) was added to the persistence scale, which now has 4 subscales and a total of 35 items (previously 11 items) (see Table 1 for details). All 3 of the TCI R character are reduced in the number of items, self-directedness from 44 to 40 items, cooperativeness

Fig. 2. Scree plot for character data (real + 2 random).

from 42 to 36 items, and self-transcendence from 33 to 26 items. In addition, a few self-transcendence items were conceptually revised to be more reflective of healthy, mature, and advanced spiritual components of personality, whereas some old items that were depicting psychotic symptoms were eliminated. 3.3. Statistical analyses A set of statistical analyses traditionally used in crosscultural studies was used in our study as well: 1. The US and Serbian subjects were compared with respect to their temperament and character scores (means, SDs), using t test comparisons, to examine differences in temperament and character as measured by the TCI R; 2. Internal consistency of the TCI R scales was tested using the Cronbach internal consistency α coefficients; 3. Principal component analysis of the TCI R scores, Guttman-Kaiser criteria, and Promax rotation of the axes were used to test the postulated 7-factor model of temperament and character. As an additional test, we performed parallel analysis of simulated data as follows. The set of 473 simulated responses, randomly generated from a uniform distribution, was produced, and these data were factor analyzed. This procedure is then repeated several times. For comparison, scree plots of eigenvalues of the correlation matrix of the real data and of simulated data are shown in Fig. 1 (for temperament traits) and Fig. 2 (for character traits). These calculations reveal that the appropriate number of factors is four (4) in the temperament subspace and three (3) in the character subspace, as expected on the basis of general arguments outlined above.

4. Results

Fig. 1. Scree plot for temperament data (real + 2 random).

First, the mean scores and SDs for the TCI R scales in the Serbian sample were compared to those observed in the United States (Table 2). With respect to temperament, Serbian subjects were somewhat higher in novelty seeking (United States = mean, 98.3; SD, 13.3; Serbia = mean, 100.9; SD, 13.7; t test, 4.25; P = .000) and lower in persistence

652

T. Dzamonja-Ignjatovic et al. / Comprehensive Psychiatry 51 (2010) 649–655

(United States = mean, 124.7; SD, 17.5; Serbia = mean, 120.6; SD, 17.4; t test, 5.03; P = .000). With respect to character, self-directedness and cooperativeness were significantly lower in Serbian subjects, much as found in previous work [10]. Other temperament and character dimensions were comparable between the 2 samples. Second, the internal consistency of the TCI R scales was tested as a general measure of its construct validity. As shown in Table 3, the Cronbach α coefficients ranged from .78 to .90 (persistence, .90; self-directedness, .89; harm avoidance, .87; cooperativeness, .82; novelty seeking, .78; reward dependence, .79; and self-transcendence, .78). The subscales had somewhat lower but still acceptable internal consistency, comparable to the US data (except for 3 subscales—C2 empathy, C3 helpfulness, and RD4 dependence, with α b .50). Third, a number of the TCI R temperament and character dimensions were intercorrelated as expected for functionally Table 2 TCI 5 mean values (SD), t test comparisons, and P values for US and Serbian subjects Scales Mean (SD) —US

Mean (SD)— Serbia

NS NS1 NS2 NS3 NS4 HA HA1 HA2 HA3 HA4 RD RD1 RD2 RD3 RD4 P PS1 PS2 PS3 PS4 SD SD1 SD2 SD3 SD4 SD5 C C1 C2 C3 C4 C5 ST ST1 ST2 ST3

100.90 (13.72) 32.52 (4.86) 21.97 ± 5.05 27.97 ± 6.62 18.64 ± 4.08 88.11 ± 16.13 28.95 ± 6.38 21.94 ± 5.06 17.92 ± 5.06 20.99 ± 5.03 102.96 ± 12.34 28.29 ± 4.34 35.66 ± 5.77 20.67 ± 4.60 18.43 ± 3.37 120.56 ± 17.39 29.39 ± 5.76 28.06 ± 4.82 36.32 ± 5.46 25.77 ± 4.94 141.49 ± 19.41 30.03 ± 5.23 23.22 ± 3.98 17.67 ± 3.54 31.22 ± 6.80 39.36 ± 6.76 132.89 ± 14.76 30.56 ± 4.17 18.76 ± 2.83 27.79 ± 3.29 26.16 ± 5.67 29.61 ± 4.65 75.59 ± 14.71 28.90 ± 6.74 23.65 ± 5.13 23.05 ± 6.17

98.3 (13.3) 30.5 (4.7) 23.1 (4.8) 26.2 ± 6.3 18.5 ± 4.2 89.6 ± 17.7 27.9 ± 6.3 22.2 ± 4.8 19.4 ± 6.0 20.1 ± 5.6 103.8 ± 13.8 28.4 ± 4.3 35.2 ± 6.0 19.7 ± 4.6 20.6 ± 3.3 124.7 ± 17.5 31.6 ± 5.3 29.3 ± 4.6 35.9 ± 5.7 27.9 ± 4.6 145.6 ± 19.3 30.4 ± 5.0 22.2 ± 4.1 19.0 ± 3.1 32.9 ± 7.1 41.1 ± 6.4 139.0 ± 14.3 30.1 ± 3.4 18.5 ± 2.6 31.3 ± 3.4 27.4 ± 5.1 31.7 ± 4.6 77.3 ± 13.9 27.7 ± 5.7 22.7 ± 4.9 26.9 ± 6.8

Refer to abbreviations in Table 1.

t test 4.25

df

P

472 .000

Mean difference 2.60

−1.96

.050 −1.49

−1.83

472 .064 −1.01

−5.932 472 .000 −4.65

−4.96

472 .000 −4.65

Table 3 TCI 5 internal consistency (Cronbach α) in United States and Serbia Scale

No. of items

Cronbach α—US

Cronbach α—Serbia

NS NS1 NS2 NS3 NS4 HA HA1 HA2 HA3 HA4 RD RD1 RD2 RD3 RD4 P PS1 PS2 PS3 PS4 SD SD1 SD2 SD3 SD4 SD5 C C1 C2 C3 C4 C5 ST ST1 ST2 ST3

35 10 9 9 7 33 11 7 7 8 30 8 10 6 6 35 9 8 10 8 40 8 6 5 10 11 36 8 5 8 7 8 26 10 8 8

.80 .61 .71 .78 .61 .91 .81 .73 .87 .82 .87 .68 .81 .79 .51 .93 .82 .79 .84 .75 .91 .78 .76 .72 .82 .81 .89 .72 .55 .62 .87 .62 .87 .72 .74 .84

.78 .54 .70 .77 .50 .87 .69 .69 .75 .70 .79 .60 .71 .73 .47 .90 .76 .70 .74 .67 .89 .75 .69 .66 .73 .79 .82 .67 .42 .30 .85 .57 .78 .75 .70 .78

Refer to abbreviations in Table 1.

related behavior traits (Table 4). The nature of this interrelatedness is addressed in the discussion section. Fourth, to test the postulated dimensional structure of the TCI R, principal component analysis was used (Tables 5 and 6). Following recent studies on the subject [11,13], the factorial structure of temperament and character was tested

Table 4 TCI R scale correlations NS

−9.167 472 .000 −6.03

−2.53

472 .012 −1.63

NS HA RD PS SD C ST

−.276a .111b −.169a −.160a −.151a .126a

HA

RD

PS

SD

C

ST

−.276a

.111 −.133a

−.169a −.479a .124a

−.160a −.609a .206a .481a

−.151a −.264a .485a .285a .508a

.126a −.005 .052 .216a −.250a .038

−.133a −.479a −.609a −.264a −.005

.124a .206a .485a .052

.481a .285a .216a

.508a −.250a

Refer to abbreviations in Table 1. a Correlation is significant at the 0.01 level (2-tailed). b Correlation is significant at the 0.05 level (2-tailed).

.038

T. Dzamonja-Ignjatovic et al. / Comprehensive Psychiatry 51 (2010) 649–655

The limitations of linear statistical methods in studying complex nonlinear systems such as personality are discussed below in some detail.

Table 5 Principal component analysis of the TCI R temperament scales Structure Matrix

NS1 NS2 NS3 NS4 HA1 HA2 HA3 HA4 RD1 RD2 RD3 RD4 PS1 PS2 PS3 PS4

1

2

3

4

.199 −.415 −.235 −.110 −.322 −.271 −.353 −.525 .358 .243 −.083 −.114 .761 .819 .823 .806

−.458 −.183 −.168 −.141 .787 .772 .681 .654 .478 −.146 −.165 .218 −.200 −.396 −.276 −.239

.480 .082 .410 .026 −.269 −.085 −.491 −.186 .431 .841 .810 .273 .102 .122 .169 .255

.366 .594 .599 .784 −.117 −.402 −.261 −.157 −.354 .176 .179 −.463 −.252 −.257 .108 −.119

Refer to abbreviations in Table 1. Extraction method used was principal component analysis and for rotation method, Promax with Kaiser normalization. Expected theoretical factors are underlined, and the highest loadings are in bold.

separately due to their nonlinear relationship. As noted earlier, simulated data were used to determine the number of real data factors to explain more variance than all combined factors with eigenvalues more than 1.0 extracted from randomized data (Figs. 1 and 2). Principal component analysis fully supported the 4-factor structure of temperament and the 3-factor structure of character, as postulated by theory (Tables 5 and 6). For temperament, the factors were solid, with the exception of 2 subscales: one subscale of novelty seeking (NS1—exploratory excitability) loaded with harm avoidance and reward dependence in addition to loading with other novelty seeking subscales and one subscale of reward dependence (RD4—dependence) loaded negatively with the novelty seeking subscales. For character, the factors were solid, except for the SD4 subscale (selfacceptance) that loaded with the cooperativeness subscales on the same factor much as seen in other studies [11]. Table 6 Principal component analysis of the TCI R character scales

SD1 SD2 SD3 SD4 SD5 C1 C2 C3 C4 C5 ST1 ST2 ST3

1

2

3

.465 .337 .270 .578 .423 .700 .512 .658 .800 .674 −.197 .114 .060

.755 .827 .846 .319 .782 .406 .267 .368 .292 .186 −.200 .094 −.173

−.320 .214 .115 −.427 −.191 −.141 .390 −.084 −.032 .070 .836 .807 .723

Refer to abbreviations in Table 1.

653

5. Discussion In accord with recent cross-cultural studies of the TCI R in Belgium and France [11,13], our results support the crosscultural applicability and the construct validity of the TCI R in Serbia. We discuss main points in turn below. The internal consistency of the TCI R scales was acceptable, except for a few subscales with low α coefficients, for example, C2—empathy and C3—helpfulness with α of .30 and .42, respectively. This is more likely reflective of internal consistency problems rather than of the (small) number of items in C2 and C3, as some other selfdirectedness and self-transcendence subscales with the similar number of items had α coefficients around .70. Of note, the same TCI R scales in the US and Serbian subjects had low or high internal consistency again indicating that features inherent in these scales rather than cultural factors influenced their internal consistency. The TCI R temperament and character scores in the Serbian sample were generally comparable to those obtained in the United States with some notable exceptions. The observed differences between US and Serbian subjects (Table 2) may not be surprising with respect to the character traits, which are expected to vary with culture but are unusual with respect to the temperament given the stability of temperament traits across cultures [4]. We interpret the observed higher novelty seeking in Serbian subjects as at least partly reflective of dramatic social and economic changes in Serbia for the last 20 years, socialism, and civil wars replaced by personal freedoms and entrepreneurship in particular. Note that in previous work published in 1991 [14], no difference in novelty seeking between Serbian and US subjects was found. In fact, in the same 1991 study, Serbian subjects had higher harm avoidance than the US subjects, but 20 years later, in the present study, the difference has disappeared. Instead, Serbs are now scoring higher on novelty seeking. Taken together, these variations (ie, increase in novelty seeking and decrease in harm avoidance over time) point to a speculative but plausible interpretation: environmental influences appear to be (epigenetically) suppressing harm avoidant behaviors and reinforcing behaviors related to novelty seeking. The observed higher US scores in persistence are not surprising because US subjects consistently have higher persistence than subjects from other societies [4], likely reflecting social or biologic factors specific to the United States. The observed lower self-directedness and cooperativeness scores in the Serbian sample do not necessarily mean an increased prevalence of personality disorder in Serbia, which has been shown to be comparable to the US population—cca

654

T. Dzamonja-Ignjatovic et al. / Comprehensive Psychiatry 51 (2010) 649–655

15% in the general population and 50% in psychiatric patients [9]. With this in mind and as 2 of the present authors are bicultural (DMS and NS), we have chosen to interpret these lower character scores in Serbian subjects as reflective of local Serbian norms about self-directed and cooperative behaviors rather than of increased prevalence of personality disorder pathologic condition. Our results support the factorial structure of the TCI R when temperament and character are analyzed separately. The factor loadings were generally lower for most of the TCI R scales than those reported by other authors [11], but the differentiation of the factors was satisfactory. A possible explanation for the relatively lower factor loadings in our sample is the new 5-point scale format with the “neutral” option for undecided subjects. In our sample, the “neutral” option was chosen very frequently, a total of 25% of all TCI R items was answered as “undecided” by about 40% of the subjects, which may have interfered with a more clear differentiation of the traits (something to watch for in future studies!). Quite similar to the Belgian study [11], the persistence dimension separated as a very robust factor justifying the inclusion of 3 additional persistence subscales in TCI R. A number of the TCI R temperament and character traits were interrelated (Table 4). For illustration, persistence was correlated with self-directedness and harm avoidance, and self-directedness was in addition correlated with cooperativeness. Although these interrelated traits reflect different personality processes, some more emotional and closer to their biologic roots, for example, harm avoidance and persistence, and some more rational and adaptive in nature, for example, self-directedness and cooperativeness, they functionally interact to shape expressed behaviors. For example, behaviors described as high self-directedness (eg, resourcefulness) are usually superimposed on behaviors described as low harm avoidance (eg, optimism)—as low harm avoidance sets a facilitating emotional stage for the development of adaptive behaviors such as high selfdirectedness. This functional relationship is detected as high correlation in linear statistics. We question here the adequacy of linear statistical methods (factor analysis included) to validate theoretical models of complex nonlinear dynamical systems such as personality. For example, the observed linear correlation between harm avoidance and self-directedness by no means implies their conceptual or functional equality. Each of these traits has its distinct etiology, distinct underlying learning processes, specific eliciting stimuli, unique motivational power, and variable level of activation (from dormant to active) before the graded severity of environmental circumstances [3]. In fact, internal consistency for harm avoidance and self-directedness was higher than for other TCI R scales in our study indicating that each is conceptually valid and homogenous. Linear factor analysis, as it were, takes only a snapshot of underlying interacting traits: it cannot discriminate between distinct and sequential developmental strata in

personality structure (eg, emotional vs adaptive) and cannot detect their dynamic interaction or their coordinated activation by specific environmental stimuli. Rather, linear approach flattens out all behaviors into an artificial 1dimensional plane and lumps all correlated behaviors (regardless of the source of correlation) into higher order “factors.” If linear factor analysis is used, it should be applied separately to temperament and character to reduce its inaccuracy [15]. In summary, more fundamental research (eg, epigenetic, functional neuroimaging, and others), nonlinear developmental models, and treatment outcome studies are needed to achieve a full understanding of naturally occurring complex dynamical systems such as personality and to validate theoretical models of such systems. Nonlinear factor analysis, although providing a more accurate fit for the observed variables, also does not fully reveal the underlying relationships among the observed traits. As shown in prior work [8,16], using a nonlinear developmental model, one temperament configuration can lead to several character outcomes and vice versa (the “multifinality” and “equfinality” of character development, respectively). The most likely outcomes, that is, those with the highest probability, are most frequently manifested, but other outcomes can also be activated by specific internal and external influences, for example, situational factors, physical health, random life events, developing intelligence, and so on. Personality traits, especially character traits, are not fixed and permanent but rather flexible and susceptible to epigenetic changes, although at times reach a point of adaptive stability, observed as equilibrium states in nonlinear dynamical systems. On the positive side, linear methods can be a useful first approximation, as when we sum scores to estimate individual traits in the TCI. In addition, factor analytically derived clusters of items are used as a starting point to identify symptomatic domains (eg, mood instability, impulsivity) or specific phenomenological experiences (eg, grandiosity, entitlement) as focus for psychotherapy [5]. References [1] Cloninger CR, Svrakic DM, Przybeck TR. A psychobiological model of temperament and character. Arch Gen Psychiatry 1993;50:975-90. [2] Cloninger CR, Przybeck TR, Svrakic DM, Wetzel RD. The Temperament and Character Inventory—a guide to its development and use. St. Louis: Washington University; 1994. [3] Cloninger CR, Svrakic DM. Personality disorder. In: Sadock V, Sadock B, editors. Comprehensive textbook of psychiatry. 9th ed. Williams and Wilkins; 2009. p. 2197-240. [4] Miettunen J, et al. International comparison of Cloninger's temperament dimensions. Pers Indiv Diff 2006;41(2006):1515-26. [5] Svrakic DM, Tosevski-Lecic D, Divac Jovanovic M. DSM II: personality or adaptation disorders. Curr Opin Psychiatry 2009;22: 111-7. [6] Svrakic DM, Whitehead C, Przybeck TR, Cloninger CR. Differential diagnosis of personality disorder by the seven factor personality inventory. Arch Gen Psychiatry 1993;50:991-9.

T. Dzamonja-Ignjatovic et al. / Comprehensive Psychiatry 51 (2010) 649–655 [7] Colvis C, Pollock J, Goodman R, et al. Epigenetic mechanisms and gene networks in the nervous system. J Neurosci 2005;25(45): 10379-89. [8] Svrakic N, Svrakic D, Cloninger CR. A general quantitative theory of personality: fundamentals of a self-organizing psychobiological complex. Deve Psychopathol 1996(8):247-72. [9] Džamonja Ignjatovic T, Svrakic D. Western personality models applied in Eastern Europe: Yugoslav data. Compr Psychiatry 2003; 44(1):51-9. [10] Džamonja Ignjatovic T. Psihodijagnosticka procena licnosti u okviru savremenih modela i DSM-IV klasifikacije poremecaja. Beograd: Filozofski fakultet; 1997. [11] Hansenne M, Delhez M, Cloninger CR. Psychometric properties of the Temperament and Character Inventory—Revised (TCI—R) in a Belgian sample. J Pers Assess 2005;85(1):40-9.

655

[12] Cloninger CR, Przybeck TR, Svrakic DM, Wetzel RD. The Temperament and Character Inventory—Revised. St. Louis: Washington University; 1999. [13] Pelissolo A, Ecochard P, Falissard B, et al. Psychometric characteristics of Cloninger's criteria for personality disorder in a population of French prisoners. Int J Methods Psychiatr Res 2008;17(1):30-4. [14] Svrakic D, Przybeck TR, Cloninger CR. Further contribution to the conceptual validity of the unified biosocial model of personality: US and Yugoslav data. Compr Psychiatry 1991;32(3):195-209. [15] Cloninger CR. The psychobiological theory of temperament and character: comment on Farmer and Goldberg. Psychol Assess 2008;20 (3):292-9. [16] Cloninger CR, Svrakic N, Svrakic D. Role of personality selforganization in development of mental order and disorder. Dev Psychopathol 1997(9):881-906.