Temperament and character in an Australian sample: examining cross-sectional associations of personality with age, sex, and satisfaction with life

TCI and other contemporary personality measures

Although the study of personality originated in the 1800’s, contemporary personality trait theories could arguably have stemmed from the work of Hans and Sybil Eysenck in 1975 (1916–1997). Eysenck proposed a two-axis measure of personality using extraversion-introversion and neuroticism-emotional stability. He later added psychoticism and hence developed the PEN model of personality, i.e., psychoticism, extraversion, and neuroticism (Eysenck, Eysenck & Barrett, 1985). Germaine to this article, and controversial at the time, is that he believed that biological factors, in particular the hereditary influence of genes played a vital role in one’s behavior.

The most recognized today are models that analyze differences in terms individual attitudes and behaviors across broad measures of personality. The Five-Factor personality model (FFM) (Costa & McRae, 1992) is a popular measure. The FFM was derived empirically from factor analysis to describe five main measures of personality, which are, Openness to Experience, Conscientiousness, Extraversion, Agreeableness and Neuroticism. There are other models measuring more dimensions such as the HEXACO, which although outwardly similar to the FFM includes three substantially different dimensions (Lee & Ashton, 2004). Another example is Cattell’s 16 PF (16 Personality Factors Questionnaire) comprising 16 distinct dimensions, clearly in disagreement with Eysenck’s theory of personality measurement (Cattell, Cattell & Cattell, 1995). Congruent with Eysenck’s model, the TCI was designed to measure the dynamic, non-linear configuration of personality by taking into account the complex architecture of the psychobiological mechanisms that contribute to individual adaptive processes

There is substantial overlap statistically in the content of the TCI and other popular personality inventories. The TCI accounts for all components of alternative tests of personality, such as the Five-Factor Model, and predicts clinical indicators of mental health as well as or better than other personality inventories (Grucza & Goldberg, 2007). There are significant correlations between the TCI and the NEO-PIR, for example, Harm Avoidance is correlated with Neuroticism (r2 = 0.55, p < 0.01), and Extraversion (r2 =  − 0.63, p < 0.01). Self-Directedness and Persistence are both correlated with Conscientiousness (r2 = 0.51, p < 0.01), (r2 = 0.50, p < 0.01) respectively. Cooperativeness and Agreeableness are positively related (r2 = 0.64, p < 0.01) (Cloninger, Svrakic & Svrakic, 1997; Cloninger, 2006; Grucza & Goldberg, 2007).

Cloninger’s psychobiological model of personality stresses the contributory interaction of biological mechanisms and learning within the context of one’s environment. Within an iterative epigenic process, the dimensions are distinct, yet each interacts with the other in influencing behavior (Cloninger, 2003). The TCI was designed to measure the dynamic structure of personality and therefore accounts for normal and abnormal personality, including mature, healthy patterns and personality disorders and a wide range of related psychopathology (Cloninger, 2006). For example, research has demonstrated that the TCI traits show relationships between perceptions of well-being, and happiness (Cloninger & Zohar, 2011; Josefsson et al., 2011) and resilience (Eley et al., 2013; Eley et al., 2013; Eley et al., 2017). The personality profile showing a temperament that is low in Harm Avoidance and high in Reward Dependence, and a character that is high in Self-Directedness and Cooperativeness is consistently associated with well-being. Self-Transcendence is also associated with well-being when combined with high Self-Directedness and Cooperativeness but can also increase the risk of ill-being under some circumstances (Josefsson et al., 2011).

The TCI and demographic characteristics

A consistent finding regarding sex differences is that females are higher in levels of Harm Avoidance, Reward Dependence and Cooperativeness (Ablin et al., 2016; Brooks, Eley & Zink, 2014; Zohar & Cloninger, 2011; Eley, Young & Przybeck, 2008; Eley et al., 2013; Goncalves & Cloninger, 2010; Gutierrez-Zotes et al., 2015; Hansenne, Delhez & Cloninger, 2005). Cross-sectional studies have shown that Novelty Seeking tends to decrease with age, (Cloninger, 2003; Trouillet & Gana, 2008) and in general character traits show changes over time. While Self-Directedness and Cooperativeness normally increase with age, Self-Transcendence decreases until middle age and then increases again as people face ultimate situations like suffering and mortality. Temperament is moderately stable throughout life except that Novelty Seeking decreases and Persistence may increase with age (Cloninger, 2003; Josefsson et al., 2013).

Rational and aims of this study

This study focused on the association between personality and subjective well-being among a general population sample of Australians. The increase in globalisation and ease of communication internationally suggests we are becoming more alike than different. Studies have shown the economic, social and geographical influences on personality and often find consistent differences between countries and cultures (Kajonius & Mac Giolla, 2017). However, as mentioned above, most of these studies measure personality with the Five-Factor Model. Despite the extant literature using the Temperament and Character Inventory in over 30 countries, as well as several studies of specific participant samples in Australia, there are few using an Australian general population. Using an Australian general population sample, we examined the psychometric properties of the Temperament and Character Inventory (TCIR140) and investigated the associations between TCI-traits with both sociodemographic variables and measures of well-being. In addition, we investigated differences in temperament and character between our Australian general population sample and published results of similar studies from other countries.

Study design and setting

The University of Queensland (UQ) Human Research Ethics Committee (HREC) (2013001083) granted ethical approval for the study. [Approval reference number: UQ HREC 2013001083] All participants provided consent documented on the survey. Participants were recruited from 2016-2017 within the general population via community newsletters and national organizations e.g., national seniors, toastmasters, sporting organizations and two universities. In each instance, the organization distributed an announcement about the study through online newsletters or regular publications. The announcement contained a short description of the study, an invitation to participate, and a direct link to the survey for more information. The survey preamble contained the ethical clearance declaration, all participant information and the consent statement. There were no conditions for participation. Participants were offered the option to request their temperament and character personality profile by providing their email address on the survey. All data were collected online via Survey Monkey^®.

Participants

There were 1772 respondents. Data cleaning excluded those cases that did not provide valid data on the TCIR-140 (the main outcome measure). Invalid responses did not answer at least 80% of the items (n = 112; 6%), and/or the four validity items (n = 150; 8%). If an individual had answered at least 80% of the items within a scale, the missing data was replaced based on the individual’s average responses on other items within the same scale. The final sample comprised 1510 participants with valid data on TCIR-140. The number of respondents with valid data on PANAS and SWLS were N = 1424 and N = 1419, respectively. Missing data were handled by pairwise inclusion to retain available information for the corresponding analyses.

Measures

Statistical analysis

We examined the descriptive statistics of the TCIR-140, PANAS and SWLS. To assess the reliability of the TCIR-140 descriptive statistics were estimated with Cronbach’s alphas to determine internal consistency.

Confirmatory factor analyses (CFA) were conducted to test the fit of each item into the Temperament and Character models of the TCIR-140. Our data meets the assumptions of a CFA based on assessments of multivariate normality, a sample size of n > 300, a-priori model specification, and sampling methods. The model structures were pre-determined based on the design of the TCI with four temperament traits in one model, and three character traits in another model. Paths were entered between the latent constructs of the traits based on known significant correlations. These included NS and RD, NS and HA, HA and PS, and HA and RD in the Temperament model, and SD and CO, SD and ST, and CO and ST in the Character model. Better model fit was determined by higher values of comparative fit index (CFI), normed fit index (NFI), Tucker-Lewis coefficient (TLI), and lower values of root mean square error of approximation (RMSEA).

Modification indices were used to inform model building for better model fit using a step-wise approach. Additional paths between the items were put into the model using the step-wise approach based on paths for model fit improvements with the highest values for the purpose of informing the model fit. The results for the a-priori paths of interests were presented because they were pre-determined and relevant to how each item loads onto the construct. The models were specified for paths from the latent dimensions to each observed item. Covariance between the latent subscales were informed by significant correlations between the TCI dimension scores. The CFA and the covariances to be analyzed were conducted using the maximum likelihood method.

The TCIR-140, PANAS and SWLS trait scores were compared by sex and age group using independent sample t-tests and ANOVAs with Bonferroni post hoc correction tests respectively. Partial correlation analyzes, controlling for sex and age, were conducted to examine the association between the TCIR-140 traits with the PANAS and the SWLS.

Lastly, we compared our TCIR-140 trait scores with those from studies of other populations using the same measures. The original data from those studies was not available, therefore, separate one sample t-tests were used to compare our findings with each of the published estimates from different populations. These included samples from an Israel community (Zohar & Cloninger, 2011), a Spanish community (Gutierrez-Zotes et al., 2015) and a Brazilian general population sample (Goncalves & Cloninger, 2010). All analyzes were conducted using SPSS. A significance level of p < 0.05 was used in this study.

Demographics

Table 1 presents the sample’s demographic characteristics. Females made up 78% of the sample. Age (mean = 44.8 years) ranged from 19 to 90 years and collapsed into three groups, young-adult 19-30 years, mid-adult 31-55 years, and older-adult 56-90 years.

The majority of the sample were partnered, lived longest in an urban/metro location, and lived the majority of their life in Australia. An undergraduate education comprised 38% of the sample. Overall, the participants reflected the demographic characteristics of the Australian general population as described by the Australian Bureau of Statistics report on People and Communities (Australian Bureau of Statistics, 2021). As noted above, our sample was recruited from a broad variety of community organizations on a national scale.

Descriptive statistics and psychometrics

Descriptive statistics of the TCIR-140 traits showed that the sample as a whole scored high (i.e., mean Likert values above 3.5) in levels of Cooperativeness, Self-Directedness and Persistence and average (i.e., mean Likert values of 2.51–3.5) on all other temperament and character traits (see Table 2). There was no strong skewness (skewness ranged from −.66 to .18), in addition, and due to our large sample size and the central limit theorem, the normally assumption of the data were met. Reliability scores (Cronbach alpha) in this sample ranged from α = 0.78 for Novelty Seeking to α = 0.92 for Harm Avoidance. For the PANAS, high levels of Positive affect (i.e., mean Likert value of 3.5) were found. The whole sample had low levels of Negative affect, with scores skewed to the right (i.e., more participants reporting lower values). The PANAS showed strong reliability in both the Positive affect (α = 0.88) and Negative affect (α = 0.89) traits. SWLS reliability in this sample was also strong (α = 0.88), with more participants having moderately high scores overall, but the skew was not strong (skewness −.73).

CFA results for the Temperament dimensions of the TCIR-140 are presented in Fig. 1. Overall, the items loaded moderately well onto their corresponding dimensions, and the model achieved acceptable model fit, CFI = 0.89, NFI = 0.84, TLI = 0.87, RMSEA = 0.039, chi-square = 8182.58, p < 0.001. In general, estimates were consistently stronger in the Harm Avoidance and Persistence dimensions, with more variability in Novelty Seeking and Reward Dependence. Only one item did not significantly load onto its designated dimension, but its removal did not significantly improve internal consistency.

CFA results for the Character dimensions are presented in Fig. 2. Similar to the Temperament model, the Character model results showed that overall, the items loaded moderately onto their corresponding scales, and the model achieved acceptable model fit, CFI = 0.88, NFI = 0.84, TLI = 0.87, RMSEA = 0.041, chi-square = 4298.53, p < 0.001. One item with the lowest standardised estimate loaded weakly on its scale, but its removal would only improve internal consistency by 0.01.

Correlation between the TCIR-140 with PANAS and SWLS

Table 3 provides partial correlations among the TCIR-140 traits PANAS and SWLS, controlling for sex. Significant moderate to strong correlations (r > 0.3) were found between Harm Avoidance with Persistence (r =  − 0.36, p < 0.001) and Self-Directedness (r =  − 0.61, p < 0.001), and between Cooperativeness with Reward Dependence (r = 0.44, p < 0.001), and Self-Directedness (r = 0.40, p < 0.001). Greater Positive Affect scores were correlated significantly with lower Harm Avoidance (r =  − 0.57, p < 0.001), higher Persistence (r = 0.54, p < 0.001) and higher Self-Directedness (r = 0.54, p < 0.001). Greater Negative Affect scores were correlated with higher Harm Avoidance (r = 0.56, p < 0.001) and lower Self-Directedness (r =  − 0.60, p < 0.001). Greater SWLS scores were correlated with lower Harm Avoidance (r =  − 0.38, p < 0.001) and higher Self-Directedness (r = 0.58, p < 0.001). Because the majority of the sample is female Appendix 2 shows individual correlations by male and by female. Here we see the strength of several correlations greater in males than females. See File S2 for more detail.

Association of demographics with the TCIR-140, PANAS and SWLS

The mean scores on the 5-point Likert scale for the TCIR-140 centred near three and the whole sample scored high (above 3.5) in levels of Persistence, Cooperativeness and Self-Directedness. Scores were average on all other traits. Comparison of variable mean scores for TCIR-140, PANAS and SWLS by sex and age are presented in Table 4.

Comparisons with other countries and cultures

The mean TCIR-140 trait scores from this study were compared with scores from similar general population/community studies and are presented in Table 5. Brief demographics of the comparator country samples are also listed on Table 5. Compared to the Spanish sample, Australians are significantly higher in Novelty Seeking, Persistence and Self-Transcendence and lower in every other trait. The Australians differ from the Israeli sample being significantly higher in Harm Avoidance, Persistence, Cooperativeness and Self-Transcendence. Compared to Brazilians, our sample is higher in Harm Avoidance and Cooperativeness, and lower in Reward Dependence and Self-Transcendence.

Australian temperament and character profile and associations with demographic characteristics

The temperament and character associations with the sample demographics were consistent with previous literature that shows females are higher in levels of Harm Avoidance, Reward Dependence and Cooperativeness compared to males (Hansenne, Delhez & Cloninger, 2005; Preiss et al., 2007; Goncalves & Cloninger, 2010; Cloninger & Zohar, 2011; Gutierrez-Zotes et al., 2015). Men were higher than females only in Self-Directedness but levels were not different in Persistence. This is contrary to studies in Israel and Belgium which found men to be higher than women in both traits (Zohar & Cloninger, 2011; Hansenne, Delhez & Cloninger, 2005), and with a French study showing men to be higher in Reward Dependence, Cooperativeness and Self-Directedness (Pelissolo et al., 2005). Schmitt et al. (2017) has proposed that counties with greater egalitarian cultures show greater differences in psychological constructs by gender. Australia is considered a democratic society yet differences in certain traits, common in other societies are not seen. A suggestion might be related to the gender-equality paradox which has been observed in relation to females entering STEM (science technology engineering and maths) degrees (Fors, Goossen & Hjerm, 2020). Nevertheless in keeping with the literature, significantly higher Harm Avoidance in females is an almost universal finding (Mendlowicz & Girardin, 2000; Miettunen et al., 2006), and is likely to be related to levels of anxiety shown to be more prevalent among females (Cloninger, 2006).

Our findings related to age are also largely consistent with the international literature using the TCIR-140. Novelty Seeking decreased with age (Aluja et al., 2010; Cloninger & Zohar, 2011; Fresán et al., 2011; Gutierrez-Zotes et al., 2015; Josefsson et al., 2013; Preiss et al., 2007). However, contrary to Gutierrez-Zotes et al. (2015), Harm Avoidance also decreased with age in this Australian sample. Furthermore, in our sample, levels of Self-Directedness and Cooperativeness were highest in older adults compared to the two younger groups. Reward Dependence was not significant between age groups but levels of Self-Transcendence, although significantly different between each group, were highest in the older adults.

In comparison to partnered respondents, those who were single scored higher on Harm Avoidance and lower on Self-Directedness, which fit with lower scores on SWL and higher Negative Affect. Partnered individuals may feel more settled and fulfilled with a steady partner and therefore are less anxious and more positive for their future. Respondents with a postgraduate degree had the highest levels of Self-Directedness and Reward Dependence, which suggests they are conscientious and highly sociable which may be characteristics valuable to earning a higher education degree. In contrast those with a secondary education had the lowest levels of Persistence and highest levels of Harm Avoidance and Negative Affect.

Temperament and character associations with measures of well-being

A unique aspect of our study is the inclusion of life satisfaction, and positive and negative affect, as proxy measures of subjective well-being, and to measure their associations with temperament and character traits. Correlations were calculated between all TCIR-140 traits with the PANAS and SWLS. Harm Avoidance and Self-Directedness had a contrasting pattern of association with all the others. Harm Avoidance was positively correlated with Negative Affect, and negatively to Positive Affect and SWLS. The opposite was seen with Self-Directedness, negatively correlated with Negative Affect but positive with Positive Affect and SWLS. Only Persistence was also strongly correlated with Positive Affect, but all other associations were weak. Zohar & Cloninger (2011) showed this same pattern in their Israeli sample. The pattern of correlations between Harm Avoidance and Persistence show temperament closely related to well-being, as is the character trait Self-Directedness and Cooperativeness. A profile of temperament that is low in levels of Harm Avoidance, and high in Persistence, alongside, a character of high Self-Directedness and Cooperativeness has been shown to be strongly associated with high levels of resilience (Eley et al., 2013; Eley et al., 2017) and well-being (Cloninger & Zohar, 2011; Josefsson et al., 2013).

According to Diener (2009), personality is strongly associated with well-being and there is substantial stability in subjective well-being over time. This is due in part to the heritable dimensions of personality being relatively stable throughout life. Our sample’s mean levels of each temperament and character trait follow this pattern, distinguished by average Harm Avoidance, indicating confidence and the ability to tolerate a level of uncertainty, plus high Persistence demonstrating industriousness and tenacity. This temperament will be relatively stable throughout life and complimented by a character prone to being highly Self Directed, responsible and conscientious, as well as highly Cooperative which speaks to the overall character of Australians.

Therefore, it is the combination and levels of each trait; stable temperament and character influenced by life experience, that helps us perceive our well-being and explain personality over time. In our sample, young adults had the highest levels of Harm Avoidance, and the lowest levels of Persistence. Self-Directedness and Cooperativeness. Essentially the opposite of the profile most conducive to well-being and perhaps unsurprisingly as they also scored lowest on Satisfaction with Life and Positive Affect and highest on Negative Affect. Sadly, this finding may reflect the rising concern over the poor mental health in our young people.

Comparison with similar temperament and character studies in other countries

Looking at the inter-correlation between the temperament and character traits across studies showed that although the trend of all associations is in the expected direction, this sample showed some moderate to strong relationships that differ from the others. For example, the negative correlation between Harm Avoidance with Persistence and Self-Directedness was higher, and the positive relationship between Cooperativeness with Reward Dependence and Self Directedness was also higher compared to those reported in the Spanish sample by Gutierrez-Zotes et al. (2015).

While an in-depth comparison of the temperament and character trait levels across the three countries was not within the scope of the study, it is of interest to note how differences may relate to the cultural values in each country. According to the data from the World Values Survey, the longitudinal database providing worldwide data on social, cultural and political change (Inglehart & Baker, 2000), the dimension of Traditional/Secular-rational values can be an indicator of a country’s values which is likely to be reflected broadly in the personality of their population. This dimension distinguishes between societies where religion and traditional values are very important and may feature in their laws and societal norms i.e., traditional, and those that are not i.e., secular. Australia is around the halfway point but leaning toward secular values along with much of Europe, including Spain and also Israel, while Brazil is more traditional.

Comparison of TCIR-140 trait levels with similar studies in other populations showed this Australian sample to be significantly different in mean levels of all seven traits to a Spanish sample (Gutierrez-Zotes et al., 2015). The Spanish sample in general is higher in levels of all the temperament and character traits across the three countries. The Australian sample is higher in Harm Avoidance, Persistence, Cooperativeness and Self-Directedness compared to the Israeli study (Zohar & Cloninger, 2011). In comparison to the Brazilian sample (Goncalves & Cloninger, 2010), Australians are higher in Harm Avoidance, and Cooperativeness, and lower in Reward Dependence and Self-Transcendence. When compared with the Brazilian and Israeli samples, Australians in this sample have a cautious and independent nature with a character that is cooperative, industrious, and self-reliant.