Introduction

Social anxiety disorder (SAD) is characterized by a marked and persistent fear of social situations in which embarrassment or scrutiny from others may occur [1]. Individuals with SAD fear that their way of acting will be negatively evaluated by others, and they often fear that their anxiety symptoms (e.g., sweating, trembling, stumbling over one’s words) will be noticed. Consequently, individuals with SAD often avoid social situations. To be diagnosed with SAD, the fear must be disproportionate to the triggering event, and must cause significant distress or impairment in important areas of functioning, such as in social, academic or occupational domains [1]. SAD is one of the most common occurring mental health disorders in the general population [2], and the onset of this disorder is usually in early adolescence [3]. Hence, it is essential to study SAD in adolescent samples to examine its origins, as well as and risk and protective factors.

Personality may have a large impact on psychosocial functioning, and certain personality traits may thus represent important risk factors in the development of mental disorders [4]. Furthermore, personality and psychopathology are closely related conceptually [5]. Personality refers to our characteristic ways of thinking, feeling and behaving. Similarly, all mental disorders are characterized by disturbances in cognition, emotion regulation and/or behavior [1]. Indeed, numerous studies have examined the relationship between personality and psychopathology among adults [6–8]. Adult personality research has largely focused on the Big Five traits, which include Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness [9]. Those who score high on neuroticism tend to experience negative emotions, worry a lot, and are vulnerable to stress. Individuals who score high on extraversion tend to be sociable, talkative, experience positive emotions, and gain energy in social situations. Individuals scoring high on openness tend to be imaginative, intellectually curious, behaviorally flexible, and like to try new things. Furthermore, agreeable individuals tend to be trusting, altruistic, empathic, and care about others. Lastly, those who score high on conscientiousness tend to be thoughtful, well-organized, and goal directed.

Regarding SAD, this diagnosis has been particularly associated with high neuroticism and low extraversion, but also low conscientiousness in adult samples [7, 10]. Of note, this pattern of personality traits is not unique for SAD, but it also found for other common mental disorders such as depression and substance use disorders [7, 11]. Studies on the relationship between personality and anxiety disorders in child and adolescent samples indicate that a similar pattern of personality traits is associated with anxiety in childhood and adolescence (i.e., a pattern of primarily high neuroticism, but also low extraversion and low conscientiousness). Although most studies are cross-sectional, prospective studies of adolescent samples have demonstrated that these personality traits predict subsequent anxiety and depression diagnoses and symptoms [12–16]. Of note, most of these studies have focused on neuroticism and extraversion, while fewer studies have examined conscientiousness.

Beyond the Big Five traits, other personality related characteristics have also been associated with SAD. Loneliness is commonly defined as a negative emotional response to a perceived discrepancy between the desired and actual quantity or quality of social relationships [17]. Of note, empirical studies have found the rank-order of loneliness to be as stable as the rank-order of other personality characteristics, supporting a trait-like nature [18]. It has been hypothesized that loneliness leads to increased sensitivity to signs of rejection and more negative interpretations of others’ behavior, thus making loneliness a potentially important predictor of SAD [19, 20]. However, although numerous cross-sectional studies of child and adolescent samples have reported associations between loneliness and SAD, little is known about how these constructs are related longitudinally [21]. Another concept assumed to play a role in SAD, is self-efficacy. According to social cognitive theory, self-efficacy (i.e., an individual’s belief that he/she will be able to execute the actions necessary to achieve a desired outcome) plays an important role in anxiety arousal [22]. Thus, individuals with lower self-efficacy are believed to develop and maintain higher levels of anxiety as compared to individuals with higher self-efficacy. Although the direction of the effect is unclear, cross-sectional studies of adolescent samples have reported associations between lower self-efficacy and higher levels of anxiety, supporting a proposed relationship between self-efficacy and SAD [23, 24].

Regarding protective factors for SAD (and for psychopathology in general), the concept of resilience has long been a topic of interest. Resilience may be conceptualized as a personality trait that increases the ability to cope with adversity [25]. A meta-analysis of cross-sectional studies examining the relationship between trait resilience and mental health found that trait resilience was negatively associated with mental illness, including anxiety [26]. Although some studies in this meta-analysis included adolescents, most study samples consisted of adults. Related to trait resilience is the concept of sense of coherence. The theory of sense of coherence aims to explain the relationship between stressors, coping and health [27]. Conceptually, sense of coherence may be considered as a personality trait characterized by a relatively stable tendency to view the world more or less predictable, manageable and meaningful [28]. Supporting its trait-like nature, empirical studies have found that sense of coherence shows similar rank-order stability as the Big Five traits [29–31]. Lower sense of coherence is shown to be strongly associated with mental health problems, including anxiety, in both adult and adolescent populations [32–35].

To summarize, there is a growing consensus about the importance of personality in predicting psychopathology [36]. However, most studies on the relationship between personality and anxiety disorders have used adult samples, ‘personality’ is often narrowed down to the Big Five traits in the research literature, and the majority of studies are cross-sectional. Given that the onset of SAD is usually in adolescence, it is essential to study SAD in this age period. Furthermore, longitudinal studies are needed in order to better understand the temporal associations between personality characteristics and SAD. Being able to identify youth at risk of developing SAD before the symptoms have become so severe that they have developed into a clinical diagnosis, may improve the chance of effectively preventing development of the diagnosis and the associated functional impairment. Furthermore, not much is known about the nature of the associations between personality traits and SAD beyond neuroticism. More specifically, genetically informative studies have consistently reported that shared genetic factors account for a substantial amount of the positive association between neuroticism and internalizing disorders, including SAD [37–41]. Studies have also found substantial negative genetic correlations between SAD and extraversion [39, 40], while genetic research on other personality traits and SAD are scarce. Since both SAD [42] and personality traits [43] show substantial heritability, it is important to use research designs that are able to examine to what degree SAD and personality are related due to shared genetic and environmental influences, respectively. For example, associations that are driven to a greater extent by environmental influences are assumed to be more easily modifiable by preventive efforts as compared to associations that are mainly due to shared genetic influences.

By using a longitudinal twin design, this study sought to contribute to a better understanding of the longitudinal relationship between a broad range of personality-related variables throughout adolescence, and SAD in early adulthood. More specifically, we examined the temporal relationship between SAD and several personality-related characteristics, including the Big Five traits, loneliness, self-efficacy, resilience, and sense of coherence. We also examined whether concepts more related to externalizing behavioral tendencies (i.e., impulsivity, conduct problems, and delinquency) were related to SAD. Although SAD is typically characterized by shyness, introversion and behavioral inhibition, a subgroup of SAD seems to be characterized by high impulsivity and disinhibited behavior tendencies [44, 45]. Furthermore, a significant number of adolescents who engage in delinquent behavior also have internalizing mental health problems, such as anxiety symptoms [46]. Hypotheses have been made both of a positive and a negative relationship between delinquency and SAD, and results from empirical studies have yielded mixed results [47, 48]. Impulsivity is often considered a personality trait [49], and both the level of delinquent behavior and conduct problems behavior is shown to be relatively stable in childhood [50], which supports viewing these concepts as trait-like constructs.

Throughout the paper, we will use the word ‘personality’ as a collective term when referring to all independent variables. All these concepts tap into people’s tendencies to think, feel, and behave, and may thus be considered as personality characteristics in a broad sense. In addition, as described throughout in the paragraphs above, empirical studies have found these constructs to have trait-like features. Some concepts are more related to our tendencies to feel (e.g., loneliness), while others tap more into cognition (e.g., self-efficacy) or behavior (e.g., impulsivity).

We had the following specific aims; first, to estimate the phenotypic temporal associations between personality throughout adolescence and SAD in early adulthood; and second, to estimate to what degree the associations are accounted for by genetic and environmental influences.

Method

Sample and procedure

Data from the Oslo University Adolescent and Young Adult Twin Project were used for the present study [51, 52]. All twins born in Norway between 1988 and 1994 were invited to participate. This longitudinal project includes three waves of questionnaire data collected throughout adolescence and young adulthood with two years between measurements (12 to 18 years at Wave-1), and data from one face-to-face interview when the twins were around 19 years old (M = 19.1, SD = 1.2). The study was approved by the Norwegian Data Inspectorate and the Regional Committees for Medical and Health Research Ethics, ref. 2015/4 (19661). American Psychological Association ethical standards were followed in the conduct of the study. Written informed consent was obtained from both the twins and their parents. The process of obtaining informed consents began on March 21st 2006 and ended in mid-August 2006.

In the present study, the self-report questionnaire data from the three waves (in which each wave included data from seven birth cohorts) were rearranged into data from the age of 12–13 years, 14–15 years, 16–17 years, and 18 years and older (i.e., until the time of the interview assessment of SAD). The whole sample consisted of 3394 twins (56% females) from 1716 twin pairs. All twins from both complete and incomplete twin pairs were included in the study. Sample characteristics are presented in Table 1. The majority of those who responded to the questionnaires also participated in the interview (i.e., 76%, 80%, 82% and 87% at age 12–13, 14–15, 16–17 and 18, respectively).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Twin sample characteristics.

https://doi.org/10.1371/journal.pone.0299766.t001

Results

Phenotypic associations

For SAD, measured with a diagnostic interview around the age of 19 (M = 19.1, SD = 1.2), the mean was 0.18 (SD = 0.52). More specifically, 148 (5.3%) and 174 (6.2%) participants were classified as having subthreshold SAD and SAD, respectively. Means and standard deviations for the scales from the questionnaires are presented in S3 Table. We also provide inter-scale correlations and correlation with sex in S4 Table. Overall, the correlations between sex and the study variables were weak (i.e., below .20).

Table 2 presents polyserial bivariate correlations and results from regression analyses, predicting SAD from various personality-related variables measured at four different ages throughout adolescence. Looking at the bivariate associations across all age groups between the Big Five traits and SAD, the results showed that especially higher neuroticism and lower extraversion were associated with SAD. Beyond the Big Five traits, also higher levels of loneliness, and lower levels of resilience, lower self-efficacy, and lower sense of coherence were associated with SAD. Furthermore, it is noteworthy that the bivariate associations were so similar in magnitude, whether personality was measured at age 12–13 years (i.e., 6–7 years prior to the assessment of SAD) or relatively concurrent in time as the assessment of SAD. The beta values in Table 2 are probably underestimated using linear regression when the dependent variable is ordinal. However, as described in the method section, these analyses were conducted to be able to create composite scores of personality in which the variables are given weight depending on their importance for SAD.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Bivariate correlations between personality and SAD, and results from linear regression analyses, predicting SAD.

https://doi.org/10.1371/journal.pone.0299766.t002

Biometric analyses

We fitted univariate twin models to SAD and the composite scores of personality in order to partition the variances into genetic and environmental influences. SAD showed high genetic effects, with additive genetic influences explaining 59% of the variance. The remaining variance was accounted for by non-shared environmental influences. For the composite scores, heritability estimates ranged from 46% to 61%. Twin correlations, univariate parameter estimates, and fit statistics are presented in S5 Table.

To quantify the extent to which personality in adolescence are related to SAD in young adulthood due to genetic and environmental influences, we fitted a series of bivariate Cholesky decomposition models. More specifically, we partitioned the phenotypic correlations between the composite scores of personality and SAD into genetic and environmental influences. Table 3 presents polyserial cross-trait correlations between the composite scores and SAD. The phenotypic correlations reflect how well SAD can be predicted from personality. As expected, the phenotypic correlations increased somewhat in magnitude as the time interval between the measurements decreased. The patterns of twin correlations indicate that the associations between personality and SAD were mainly due to genetic influences. More specifically, the DZ correlations were not greater than half the size of the MZ correlations, indicating no influence of shared environmental factors. Furthermore, the extent to which the MZ correlations are lower than the phenotypic correlations is indicative of shared influences in the non-shared environment between the predictor variables and SAD. Thus, shared influence of non-shared environmental factors in the association between personality and SAD seem to be negligible when personality was measured at 12–13 years and 14–15 years. From 16–17 years, on the other hand, non-shared environmental influences seem to make a significant contribution to the association between personality and SAD.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Cross-trait correlations.

https://doi.org/10.1371/journal.pone.0299766.t003

Given that none of the DZ correlations were greater than half the size of the MZ correlations, we fitted ADE models to data and compared them to reduced AE models. According to the AIC values, an AE model was the best fitting model for all Cholesky decompositions except the model including the composite score of personality measured at age 16–17 years. The proportions of the phenotypic correlations between personality and SAD due to genetic and environmental influences are presented in Fig 1 (standardized parameter estimates are given in S6 Table). For ease of presentation, results from the AE model are presented for all age groups in Fig 1. However, results from the full ADE models yielded similar results regarding the proportions due to genetic and environmental influences (i.e., the proportions due to non-shared environmental influences were estimated to 0%, 6%, 23% and 39% at age 12–13, 14–15, 16–17 and 18 years, respectively). Thus, the proportions due to additive genetic influences in Fig 1 could be interpreted as a broad/total percentage due to genetic influences (i.e., both additive and non-additive). In line with the pattern of cross-trait correlations, the phenotypic associations between personality and SAD were almost exclusively due to shared genetic influences when personality was measured at 12–13 years and 14–15 years. From 16–17 years, non-shared environmental influences also contributed to the association between personality and SAD.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Genetic and environmental influences on the association between personality and SAD.

Note. SAD = social anxiety disorder; A = additive genetic influences; E = non-shared environmental influences. The percentages represent the proportions of the phenotypic correlations between composite scores of personality and SAD due to genetic and environmental influences.

https://doi.org/10.1371/journal.pone.0299766.g001

Genetic and environmental correlations between personality and SAD are presented in Table 4. These correlations provided further support for the results presented in Fig 1. That is, all genetic correlations were high, whereas the environmental correlations were negligible when personality was measured at 12–13 years and 14–15 years. From 16–17 years of age, however, the non-shared environmental correlations between personality and SAD were statistically significant and moderate in magnitude. Genetic and environmental correlations between SAD and all individual traits are given in S7 Table.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Genetic and environmental correlations.

https://doi.org/10.1371/journal.pone.0299766.t004

In Fig 2, we present the genetic and environmental variance unique to SAD and the portions shared with personality at different ages throughout adolescence. The percentage of the total variance in SAD that were shared with personality due to common genetic influences was 33%–41% when personality was measured at the age of 12–15 years and 24%–27% when personality was measured at 16 years and older. If we look at only the genetic variance in SAD (i.e., ignoring the environmental variance), 57%–72% was shared with personality measured at 12–15 years (i.e., [33/33+25] and [41/41+16]). When personality was measured from 16 years and older, the percentage of the genetic variance in SAD that where shared with personality was lower, i.e., 41%–46%. Similarly, if we look at only the environmental variance in SAD (i.e., ignoring the genetic variance), SAD did not share any non-shared environmental influences with personality measured at 12–15 years. However, from 16 years of age, a small amount of the environmental variance in SAD was shared with personality (i.e., 10% and 19% when personality was measured at 16–17 years and 18 years, respectively).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Proportion of variance in SAD that are unique to SAD and shared with personality.

Note. SAD = social anxiety disorder; Common A = genetic factors common to both personality and SAD, Specific A = genetic factors unique to SAD; Common E = non-shared environmental influences common to both personality and SAD; Specific E = non-shared environmental influences unique to SAD.

https://doi.org/10.1371/journal.pone.0299766.g002

Discussion

This study examined the longitudinal associations between personality throughout adolescence and SAD in young adulthood. Despite the fact that SAD usually has its onset in early adolescence [3], the majority of studies on the relationship between personality and SAD have used adult samples. Furthermore, the overwhelming body of research has focused on the Big Five traits, and most studies are cross-sectional. The present study sought to add to the literature by examine the temporal relationship between a range of personality related variables measured at different ages throughout adolescence, and SAD in young adulthood. In line with findings from previous studies of both adult [7, 71] and adolescent [13, 16] samples, results from the present study showed that higher levels of neuroticism, lower levels of conscientiousness, and lower levels of extraversion were associated with SAD.

Of note, ‘personality’ has traditionally been called ‘temperament’ when studying children and adolescents [16]. Studies have consistently reported associations between the temperamental trait ‘behavioral inhibition’ (the tendency to be shy, quiet, and restrained in unfamiliar situations) and anxiety disorders in childhood and adolescence, especially SAD [72, 73]. In addition, ‘negative emotionality’ (the tendency to feel sadness, fear and frustration after high arousal) and ‘effortful control’ (the ability to sustain attention and inhibit behavior when appropriate) have also been linked to anxiety symptoms in childhood and adolescence [16, 74]. With respect to personality, behavioral inhibition is found to be associated with low extraversion and high neuroticism, negative emotionality shows high overlap with neuroticism, and effortful control is closely linked to conscientiousness [16]. Thus, a similar pattern of traits is found to be important for anxiety within the temperament tradition as in the personality literature, and this pattern of associations was also observed in the present study. Beyond the Big Five traits, the results showed that SAD was also associated with higher levels of loneliness, lower levels of resilience, lower self-efficacy, and lower sense of coherence throughout adolescence. Associations between SAD and these variables have been reported in previous cross-sectional studies of adolescent samples [21, 23, 26, 35].

The present study adds to the literature by showing that personality measured already in early adolescence were associated with SAD in young adulthood. More specifically, the phenotypic correlation between the composite score of personality and SAD was moderate already when personality was measured at age 12–13 years. The correlations ranged from 0.42 when personality was measured at 12–13 years of age, to 0.52 when personality and SAD was measured relatively concurrent in time. Given our finding that measures of personality may represent vulnerabilities to later development of SAD, this highlights the potential value of early assessment of personality related characteristics. This is in line with an influential meta-analysis of the relationship between the Big Five traits and common mental disorders which emphasized that more focus on personality dimensions is needed in clinical psychology [7]. The authors concluded that personality “may be helpful in directing prevention efforts, developing case conceptualizations, and making clinical prognoses” [7, p. 810]. Indeed, personality assessment may be useful for identifying adolescents at risk of developing psychopathology in general since e.g., high neuroticism and low extraversion seem to represent general psychological vulnerabilities [6, 75, 76]. In addition, personality itself should perhaps be the focus of intervention before the onset of psychopathology. For example, results from a meta-analysis suggest that personality traits seem amenable to psychological intervention, although it should be acknowledged that one does not know whether the observed changes in personality reflect ‘real’ change or reflect change in which people return to their prior baseline [77]. Nevertheless, assessment of personality may be useful to determine the focus of treatment. For example, a personality assessment may inform the clinician of a person’s strengths and vulnerabilities which may be used to guide treatment.

By using a longitudinal twin design, we were also able to examine the nature of the phenotypic associations between personality and SAD. Overall, the results showed that personality were associated with SAD mainly for genetic reasons. More specifically, the covariance between personality and SAD was exclusively due to shared genetic influences when personality was measured at 12–13 years and 14–15 years. When personality was measured later in adolescence (and closer in time to the assessment of SAD), also non-shared environmental influences contributed weakly to the covariance. That is, shared genetic influences still accounted for 2/3 of the covariance. We are not aware of any prior study that have examined the genetic and environmental contributions to the associations between personality and SAD beyond the Big Five traits. However, in line with our results, prior studies have found that shared genetic factors account for a substantial amount of the association between SAD and both neuroticism and extraversion [37–41]. Of note, genetic links have been found between personality and psychopathology in general. For example, results from a recent genetically informative study suggest that the phenotypic associations between the Big Five traits and internalizing disorders (including SAD) mainly stem from shared genetic influences [78]. More specifically, higher neuroticism and several facets from the domain of extraversion and conscientiousness seem to represent genetic susceptibility to internalizing disorders, rather than these traits having direct environmental effects on internalizing disorders.

The results in the present study contradict commonsense reasoning that would suggest that the reason why a person is the way he/she is, is because of what happened to the person in childhood or adolescence. More specifically, the results showed that environmental factors influencing both personality and SAD was negligible when personality was measured 4–5 years or more prior to the assessment of SAD. Consequently, one should be careful jumping to conclusions based on results from cross-sectional studies. Of course, this shared environmental variance will only capture environmental factors creating variance in both personality and SAD. However, it is reasonable to think that significant events that could potentially influence SAD, also contribute to variance in personality. For example, if problems at school is contributing to SAD, it is likely that this also will affect measures of personality.

It is important to have in mind that the data for the present study was based on a naturalistic study and not an intervention design. Thus, even if personality is related to SAD mainly for genetic reasons, this does not necessarily mean that interventions seeking to for example increase a person’s self-efficacy or decrease feelings of loneliness would not work. It is also important to emphasize that genetically influenced conditions may also potentially be modified, perhaps particularly through individualized/personalized therapy.

In the present study, the heritability of SAD was estimated to 59%, which is in the higher range compared to heritability estimates that have typically been reported for anxiety disorders and anxiety problems [43, 79–83]. Of note, a substantial amount of the heritability in SAD was not shared with personality, indicating that SAD is something more than personality. Surprisingly, the proportion of the genetic variance in SAD that was shared with personality was somewhat higher when personality was measured in early adolescence compared to when personality was measured at a later age and also shorter in time until the assessment of SAD. A reason may be that personality in later adolescence is influenced by “new” genes that is more unrelated to SAD.

Conclusion

The present study demonstrated that personality measured already at the age of 12 is associated with SAD in early adulthood. More specifically, we found that higher neuroticism, lower extraversion, lower conscientiousness, higher levels of loneliness, lower levels of resilience, lower self-efficacy, and lower sense of coherence are relevant traits for the development of SAD. These traits may thus be important to pay attention to in order to identify adolescents with risk of developing SAD later in life. Furthermore, results from the present study indicate that personality is associated with SAD mainly due to shared genetic influences, rather than personality having a direct effect on SAD. Associations that are accounted for by environmental influences of a size that are practically meaningful are assumed to be more effective for prevention efforts compared to highly genetically driven associations. However, although the associations were primarily genetically mediated, this does not mean that interventions aimed to e.g., strengthen a person’s self-efficacy or reduce loneliness do not help.

Supporting information

Acknowledgments

RBUP, Eastern, and Southern Norway contributed to data collection.