In earlier communication the authors have disclosed the funding sources that supported their research. One of them being a commercial source, Systembolaget, the authors would like to add that this does not alter their adherence to all the PLOS ONE policies on sharing data and materials.
Conceived and designed the experiments: DG SL TN NK HA CRC. Analyzed the data: SL SB DG. Contributed reagents/materials/analysis tools: DG SL SB HA. Wrote the paper: DG SL MR CRC NK TN HA.
The Child and Adolescent Twin Study in Sweden (CATSS) is an on-going, large population-based longitudinal twin study. We aimed (1) to investigate the reliability of two different versions (125-items and 238-items) of Cloninger's Temperament and Character Inventory (TCI) used in the CATSS and the validity of extracting the short version from the long version, (2) to compare these personality dimensions between twins and adolescents from the general population, and (3) to investigate the genetic structure of Cloninger's model.
Reliability and correlation analyses were conducted for both TCI versions, 2,714 CATSS-twins were compared to 631 adolescents from the general population, and the genetic structure was investigated through univariate genetic analyses, using a model-fitting approach with structural equation-modeling techniques based on same-sex twin pairs from the CATSS (423 monozygotic and 408 dizygotic pairs).
The TCI scores from the short and long versions showed comparable reliability coefficients and were strongly correlated. Twins scored about half a standard deviation higher in the character scales. Three of the four temperament dimensions (Novelty Seeking, Harm Avoidance, and Persistence) had strong genetic and non-shared environmental effects, while Reward Dependence and the three character dimensions had moderate genetic effects, and both shared and non-shared environmental effects.
Twins showed higher scores in character dimensions compared to adolescents from the general population. At least among adolescents there is a shared environmental influence for all of the character dimensions, but only for one of the temperament dimensions (i.e., Reward Dependence). This specific finding regarding the existence of shared environmental factors behind the character dimensions in adolescence, together with earlier findings showing a small shared environmental effects on character among young adults and no shared environmental effects on character among adults, suggest that there is a shift in type of environmental influence from adolescence to adulthood regarding character.
Cloninger's psychobiological model of personality
Different versions of the TCI (e.g., 125-items, 238-items) have been validated against other measures of personality
Twin studies suggest at least equally (if not more) important roles of genetic influence on the character dimensions as on the temperament dimensions. Ando and colleagues
Cloninger's psychobiological theory of personality provides a model that can be empirically verified by behavioral genetics methodology because it proposes that humans are an integrated hierarchy of biological, psychological, and social systems that adapt to changes
The present study was conducted using self-reported temperament and character measures from the Child and Adolescent Twin Study in Sweden (CATSS), which is an on-going, large population-based longitudinal twin study targeting all twins born in Sweden since July 1, 1992. By January 2010, the CATSS was comprised of around 17,000 twins and a response rate of roughly 80% (for a detailed description of the CATSS, see
The aims of the present study were:
to provide distributions (means and standard deviations) and to investigate the convergent reliability (
to compare the TCI dimensions of the CATSS-twins to previously published data of adolescents from the Swedish general population.
to investigate the genetic structure of Cloninger's model in adolescence by twin modeling.
Subjects are protected by informed consent process – they are informed of what is being collected and repeatedly given the option to withdraw their consent and discontinue their participation. All adolescents in the study had written consent from parents, caretakers, or guardians to participate in the study. The study has ethical approval from the Karolinska Institute Ethical Review Board: CATSS-15 Dnr: 2009/1599-32/5, DOGSS Dnr: 03-672 and 2010/1356/31/1.
Since 2004, parents of all twins in Sweden have been interviewed via telephone by lay interviewers who, after a brief introduction to child and adolescent psychiatry and twin research, used a computerized version of the Autism – Tics, AD/HD and other Comorbidities inventory (A-TAC)
The twins were contacted again when they were 15 years of age (CATSS-15) – the sample used in the present study. The CATSS-15 study consists of questionnaires that were sent home to the twins by mail (overall response rate 48%), including the short version (125 items) of the TCI. A thorough clinical follow-up study was also carried out of a selected part of the 15-year-old twins, creating a second sample that was included in the present study. In this clinical follow-up study, children who had been screened positive for different neurodevelopmental problems in the A-TAC interview at age 9/12, their co-twins, and matched controls were enrolled in a study called the Developmental Outcomes in a Genetic Twin Study in Sweden (DOGSS). The overall response rate in the DOGSS was over 60% with a tendency towards a higher response rate in controls and somewhat lower in the screened positive children. Participants in the DOGSS were asked to complete the longer version of the TCI (238 items) from which the shorter 125-items version could be extracted. This extracting procedure enabled the amalgamation of the CATSS-15- and the DOGSS-twins, which allowed us to achieve about the same response rate among children with psychosocial problems – a subgroup that is often under-sampled in population-based self-report studies – as what is found for normal children. Only twins from the CATSS-15 and the DOGSS who had no more than 5% missed TCI-items were included in the final sample. This procedure left a total of 2,714 twins (369 DOGSS-twins and 2,345 CATSS-15-twins) for individual comparisons (878 monozygotic, 885 same-sex dizygotic, 638 different-sex dizygotic, and 313 of unknown zygosity) and 423 monozygotic pairs and 408 same-sex dizygotic pairs for twin comparisons. See
We used data collected in two different published studies, Brändström
Both versions of the TCI used in the present study measure the seven dimensions of the psychobiological model of personality. The dimensions are computed by summarizing binary answers (
Zygosity was determined by a validated algorithm based on five questions on twin similarity, derived from 571 pairs of twins with known zygosity. Only twins with more than a 95% probability of being correctly classified, compared to DNA-testing, were assigned zygosity by this method
Using histograms, all TCI data (CATSS-15, DOGSS, Brändström, and Garcia) were considered to be normally distributed, thus all statistical tests regarding the first two aims were conducted using parametric methods in SPSS version 19. Regarding the first aim:
The third aim was analyzed using twin methodology, which is basically a comparison of monozygotic-twins, who are genetically identical, and dizygotic-twins who, on average, share 50% of their segregating alleles. As a first step, intraclass correlation (ICC) coefficients for the seven dimensions in the TCI were calculated separately for monozygotic-twins and same-sex dizygotic-twins. As a second step, we performed univariate genetic analyses, using a model-fitting approach with structural equation-modeling techniques, using Mx
The reliability coefficients (
Novelty Seeking | Harm Avoidance | Reward Dependence | Persistence | Self-directedness | Cooperativeness | Self-transcendence | |||
DOGSS | 369 | .77 | .84 | .68 | .61 | .82 | .83 | .83 | .77 |
Brändström | 332 | .73 | .82 | .64 | .60 | .80 | .84 | .79 | .74 |
Garcia | 399 | .72 | .85 | .66 | .62 | .81 | .82 | .82 | .75 |
CATSS-15 | 2345 | .66 | .81 | .46 | .66 | .81 | .72 | .78 | .70 |
DOGSS▾ | 369 | .70 | .81 | .57 | .59 | .79 | .76 | .79 | .71 |
CATSS-15 & DOGSS▾ | 2714 | .67 | .81 | .48 | .66 | .80 | .72 | .78 | .70 |
Brändström▾ | 332 | .65 | .81 | .52 | .60 | .77 | .79 | .76 | .70 |
Garcia▾ | 399 | .61 | .82 | .55 | .61 | .80 | .75 | .79 | .70 |
Novelty Seeking | Harm Avoidance | Reward Dependence | Persistence | Self-directedness | Cooperativeness | Self-transcendence | ||
DOGSS | 369 | 22.50±5.97 | 4.04±2.00 | |||||
Brändström | 332 | 22.84±5.62 | 14.88±6.53 | 13.75±3.73 | 3.84±1.90 | 24.85±6.78 | 27.38±7.00 | |
Garcia | 399 | 4.05±2.08 | 24.52±7.09 | |||||
CATSS-15 | 2345 | |||||||
DOGSS▾ | 369 | 10.85±3.69 | ||||||
CATSS-15 & DOGSS▾ | 2714 | 10.88±3.48 | 8.83±4.43 | 8.62±2.41 | 2.54±1.60 | 17.41±4.63 | 18.37±3.77 | 3.50±2.94 |
Brändström▾ | 332 | 8.25±2.58 | 2.23±1.54 | 14.60±4.72 | 16.36±4.65 | |||
Garcia▾ | 399 | 2.43±1.55 | 14.26±5.09 | 16.79±4.38♀ |
Strong correlations (ranging between 0.90 and 0.96) between the dimensions of the long and the extracted short version of the TCI were found (
NS | HA | RD | PS | SD | CO | ST | |
Novelty Seeking (NS) | −.18*** | .11* | −.19*** | −.16** | −.08 | .12* | |
Harm Avoidance (HA) | −.17** | .05 | −.09 | −.41*** | −.01 | −.12* | |
Reward Dependence (RD) | .00 | .11* | −.01 | .09 | .39*** | .06 | |
Persistence (PS) | −.36*** | −.18*** | .12* | .09 | .06 | .21*** | |
Self-directedness (SD) | −.31*** | −.42*** | .14** | .30*** | .33*** | −.13* | |
Cooperativeness (CO) | −.21*** | −.10* | .52*** | .27*** | .46*** | −.03 | |
Self-transcend ence (SD) | .02 | .04 | .19*** | .14** | −.17*** | .18*** |
Twins in the DOGSS showed higher levels of Harm Avoidance (
The TCI dimension showing the lowest heritability was the character dimension of Self-directedness (.29), while the highest heritability was found for two of the temperament dimensions: Harm Avoidance (.50) and Persistence (.52). More importantly, the results showed a shared environmental influence for all of the character dimensions (Self-directedness = .22; Cooperativeness = .21; Self-transcendence = .11) and for one of the temperament dimensions (i.e., Reward Dependence = .11). See
Dimension | Trait | MZ | DZ | A | C | E |
( |
( |
Additive Genetics | Common Environment | Unique Environment | ||
TEMPERAMENT | Novelty Seeking | 0.45 | 0.14 | .44 | .00 | .56 |
[.36–.52] | [.05–.24] | [.34–.51] | [.00–.07] | [.49–.64] | ||
Harm Avoidance | 0.51 | 0.19 | .50 | .00 | .50 | |
[.43–.57] | [.09–.28] | [.38–.56] | [.00–.09] | [.44–.57] | ||
Reward Dependence | 0.45 | 0.30 | .36 | .11 | .53 | |
[.38–.53] | [.21–.39] | [.14–.53] | [.00–.29] | [.46–.61] | ||
Persistence | 0.56 | 0.11 | .52 | .00 | .48 | |
[.49–.62] | [.01–.21] | [.45–.59] | [.00–.05] | [.41–.54] | ||
CHARACTER | Self-directedness | 0.52 | 0.36 | .29 | .22 | .49 |
[.44–.58] | [.27–.44] | [.09–.50] | [.04–.38] | [.43–.56] | ||
Cooperativeness | 0.59 | 0.40 | .38 | .21 | .41 | |
[.52–.65] | [.32–.48] | [.19–.57] | [.04–.37] | [.36–.47] | ||
Self-transcendence | 0.51 | 0.31 | .40 | .11 | .49 | |
[.43–.58] | [.22–.39] | [.19–.56] | [.00–.28] | [.43–.56] |
Note: MZ = Monozygotic; DZ = Dizygotic (same-sex).
In the present study, testing the reliability of both TCI versions included in the CATSS-15 and DOGSS follow up, as well as validation of the extraction procedure was of special interest. The first important finding of the present study is that the extraction procedure generated dimensions that were as reliable as those obtained by the longer version, both from the DOGSS and from the Brändström and Garcia studies as well. Our results also show that the TCI dimensions obtained by this procedure were highly correlated to those obtained with the longer version, suggesting that both the original version as well as the ”extracted short version” are equally reliable and could function as measures of personality dimensions in adolescents.
The second key-question of the present study was how similar/dissimilar Swedish twin adolescents are, compared to Swedish adolescents from the general Swedish population. Twins had scores about half a standard deviation higher in the Self-directedness and Cooperativeness character dimensions, while adolescents from the general Swedish population had higher scores in the Self-transcendence dimensions. Longitudinal studies in Finland have found that people increase in Self-directedness and Coooperativeness while decreasing in Self-transcendence as they mature from ages 20 to 45 years of age
The twins also reported higher scores, compared to the Brändström study group, in the temperament dimensions of Reward Dependence and Persistence. It is known that among children, Reward Dependence, Persistence, Cooperativeness, and Self-directedness are positively related to intelligence and academic achievement
The third and final aim was to investigate the contributions of genes and environment on the seven dimensions of Cloninger's psychobiological model of personality using same-sex twins from the CATSS-15 and DOGSS. The results showed that at least among adolescents there is a shared environmental influence on all of the character dimensions, but only on one of the temperament dimensions (i.e., Reward Dependence). Previous research about the etiology of personality dimensions in young adults
In the present study we had no information regarding the adolescents' birth weights, or parents' ages, educations, or socioeconomic statuses. Although this information is accessible for the CATSS data, it was not obtainable for the general population groups, making this analysis impossible. These variables might be important to examine before we can actually suggest that twins might have advantages, with regard to character maturity, compared to the general population. Moreover, the sample used here was composed of data from the DOGSS, in which adolescents were selected by one or both of the twins being screen positive for neurodevelopmental problems. This suggests that the results presented here should be interpreted with caution. Nevertheless, we suggest that the inclusion of the DOGSS-twins allowed a more robust comparison because this is a subgroup that is often under-sampled in population-based self-report studies.
It is possible that our findings regarding the genetic structure of Cloninger's model of personality differ from those of earlier research because of differences in measurement. Most research has been done using the longer version of the TCI. Nevertheless, similar results to those obtained with the long version have been found using shorter versions (e.g., see
Research on the development of twins has been described as a “natural experiment” and is commonly used to test “hypotheses about health, development, and behavior in general” (
Together with recent prospective studies showing increases in Self-directedness and Cooperativeness (which is an indicator of increasing responsibility and relatedness) with age (from 20 to 45
The character and temperament dimensions are higher order dimensions composed of lower order facets. There are both advantages and disadvantages when personality is investigated in terms of broad dimensions. One advantage is that each TCI dimension represents wide-ranging descriptions of personality, allowing the prediction of many outcomes (e.g., personality disorders). On the other hand, the aggregation of the lower order facets in one higher order dimension results in a loss of information – information that might be useful for psychological description, prediction, and explanation
The extraction of common items in the long and short TCI versions did not alter the psychometric properties of the seven personality dimensions, which were strongly correlated to their corresponding dimensions derived by the long TCI version, hence suggesting high reliability and correlation for the short version dimensions created by this procedure. Compared to adolescents from the general population, twins may have advantages pertaining character maturity. A more mature character might explain why twins, despite adversities related to prenatal and birth problems, sometimes show equal health outcomes or sometimes slightly better cognitive performance than singletons. Even if the confidence intervals are large and partly overlapping, our study supported the theoretical notion behind Cloninger's model and studies of longitudinal development showing that temperament traits are under strong genetic influence and are different from character scales that describe the development of regulatory cognitive-emotional strategies with a more complex etiology, including significant common environmental effects.
(DOCX)