Big Five Personality Traits and Insomnia Severity in Adults: The Conditional Role of Anxiety and Depression

Personality Dispositions as Diatheses for Sleep Disturbance

The Five Factor Model (FFM) of personality posits five broad trait domains—Neuroticism, Extraversion, Openness, Agreeableness and Conscientiousness—which together capture most of the systematic variance in adult personality. Large prospective cohorts and two recent meta-analyses converge on a robust bivariate pattern: high Neuroticism and low Conscientiousness reliably forecast shorter, poorer and more variable sleep (Duggan et al., 2014; Krizan & Hisler, 2019). Neuroticism is marked by emotional lability, rumination, and exaggerated threat appraisal—features commonly observed in individuals suffering from insomnia (Ellis et al., 2021) and conducive to nocturnal hyperarousal (Harvey, 2002). In contrast, conscientious individuals tend to maintain regular routines and exercise behavioural self-control, both protective for sleep regulation (Akram et al., 2023). Extraversion and low Openness appear to confer modest protective effects, while the role of Agreeableness remains equivocal (Conway et al., 2023; Emert et al., 2017; Fabbri et al., 2022; Kim et al., 2015).

Experimental studies demonstrate that neurotic individuals exhibit difficulties in emotion regulation, greater propensity for negative affect (e.g., pronounced worry about sleep, anxiety), and heightened physiological and cognitive hyperarousal, which, in turn, contribute to poorer subjective sleep quality (Zhang et al., 2023). Conscientiousness, by contrast, reflects self-discipline, planning and rule adherence. Epidemiological studies show that conscientious adults tend to follow earlier and more regular bedtimes, avoid evening stimulants, and adhere more consistently to stimulus-control guidelines, promoting healthier and more efficient sleep (Duggan et al., 2014; Emert et al., 2017). Extraversion’s association with positive affect and social connectedness may buffer stress reactions (Fabbri et al., 2022), whereas Agreeableness could foster interpersonal harmony and perceived bedtime safety (Hintsanen et al., 2014; Kim et al., 2015). Nonetheless, the modest size and contextual variability of these effects highlight the need to identify higher-order moderators that determine whether, when, and how personality traits influence sleep outcomes.

Affective States as Modulators of Dispositional Risk

Anxiety and depressive symptoms co-occur with insomnia at rates exceeding chance; population studies place the odds ratio for comorbid anxiety around 3.23 and for depression around 2.83 (Hertenstein et al., 2019). The dominant aetiological account—the hyperarousal model—posits that chronic insomnia reflects sustained cortical and autonomic activation driven by threat monitoring, rumination and dysfunctional beliefs (Dressle & Riemann, 2023). These processes are amplified by anxiety and depressive symptoms, which co-occur with insomnia at rates exceeding two-thirds in some populations. These psychopathological conditions are strongly associated with higher levels of Neuroticism and lower levels of Conscientiousness, Agreeableness, and Extraversion, which may exacerbate vulnerability to both sleep disturbances and affective dysregulation (Lester, 2021; Lyon et al., 2021; Nouri et al., 2019).

Until recently, empirical models treated distress primarily as a mediator explaining why neurotic or unconscientious individuals sleep poorly (Akram et al., 2019). However, accumulating evidence hints that anxiety and depression may instead—or also—moderate the strength of trait–insomnia couplings. Laboratory manipulations of pre-sleep worry show that once state anxiety surpasses a moderate threshold, Neuroticism no longer predicts polysomnographic sleep onset latency, implying a ceiling phenomenon in which intense affective arousal overrides the influence of stable personality traits (Horváth et al., 2016).

Building on vulnerability–stress theory, a two-tier perspective can be advanced. Under sub-clinical affective load, personality traits exert discernible influence: elevated Neuroticism heightens threat appraisal and perseverative cognition, whereas elevated Conscientiousness stabilises circadian and behavioural rhythms. As anxiety or depression escalates, global arousal reaches a ceiling where insomnia becomes widespread, attenuating between-person variance attributable to traits. Methodologically, this non-linear dynamic, demands moderation analyses able to detect trait × state interactions (Hayes, 2022). Individuals high in neuroticism are prone to maladaptive emotional and cognitive responses to sleep-related difficulties—such as persistent rumination and worry—that can intensify and prolong sleep disturbances (Krizan & Hisler, 2019). Moreover, neuroticism has been consistently associated with elevated levels of anxiety and depressive symptomatology (Lester, 2021), both of which are well-established correlates of insomnia (Comas et al., 2023). Taken together, these findings provide both theoretical and empirical support for examining whether anxiety and depression function as moderators in the relationship between neuroticism and insomnia.

The Present Study

Most trait-sleep studies have been conducted in Northern European or North American populations, while evidence from Mediterranean contexts remains sparse. Cultural differences in social rhythm regularity, dietary habits, and collectivistic coping strategies could modify both baseline sleep patterns and the salience of personality in sleep regulation. Moreover, extant research rarely examines personality, anxiety and depression simultaneously within the same moderation framework. By situating these factors within the mechanisms described above, the study offers a rigorous test of whether psychological distress amplifies or obscures the well-documented influence of personality on sleep. In doing so, it addresses a critical gap in the literature and informs precision-tailored interventions aimed at matching CBT-i to the individuals’ dispositional and affective profile.

Demographic factors such as sex and age also intersect with personality and distress in shaping sleep outcomes. Middle-aged women exhibit the highest chronic-insomnia prevalence, a pattern attributed to hormonal transitions, multi-role strain and heightened pre-sleep arousal (Morin & Jarrin, 2022; Tsou, 2022; Zeng et al., 2020). Younger adults report more sleep-onset difficulties (Morin & Jarrin, 2022) and greater evening physiological activation (Sidani et al., 2021), whereas older adults face circadian advance and difficulties maintaining sleep (Patel et al., 2018). These observations underscore the value of including demographic covariates when probing personality and distress effects.

The present study extends prior work in several ways. First, by analysing a large Greek sample it tests whether well-documented trait–insomnia associations generalise to a southern European cultural milieu characterised by later average bedtimes and strong familial networks. Second, the study explicitly models anxiety and depression as moderators rather than mere mediators, thereby adjudicating between additive-risk and ceiling accounts of dispositional vulnerability. Third, it incorporates sex and age as covariates to clarify whether previously reported demographic gradients persist after accounting for personality and distress. Fourth, it adds a facet-level perspective, moving beyond broad Big Five domains to identify the specific lower-order trait components most strongly associated with insomnia severity. This more fine-grained approach helps clarify whether insomnia risk is linked primarily to broad dispositional tendencies or to narrower affective, self-regulatory, and vitality-related personality features.

Drawing on the Five-Factor Model, the dynamic hyperarousal framework, and recent insights into psychological vulnerability, the literature review provides a conceptual foundation for examining a nuanced vulnerability–saturation model of insomnia. In doing so, it aims to move beyond static main-effect thinking and towards a more conditional, personalised understanding of why some individuals remain sleepless under stress while others sleep soundly despite similar challenges. Thus, drawing on the integrative hyperarousal model (Riemann et al., 2022), the study conceptualises insomnia risk as a two-tier system: a baseline predisposition (trait layer), where stable personality differences shape habitual cognitive-emotional reactivity, and affective activation (state layer) where situational anxiety or depression adds acute allostatic load. Whether state symptoms amplify or mask trait influences remains an open empirical question with direct clinical relevance. If an additive-risk model holds, interventions should address both personality-related cognitions and current emotional states. Alternatively, if saturation occurs, treatment should align with the individual’s current distress level.

Thus, the purpose of this study was to examine: (a) demographic covariates (age, sex) commonly implicated in sleep research; (b) the associations of each Big Five trait with insomnia severity; and (c) whether anxiety and depression condition the association between Neuroticism and insomnia severity. In supplementary exploratory analyses, we further examined whether specific Big Five facets uniquely contributed to insomnia severity and whether distress-related or demographic factors moderated selected facet-level associations. Building on prior findings, insomnia is expected to relate positively to Neuroticism and negatively to Conscientiousness. Regarding moderation, we tested two competing possibilities: an additive-risk pattern, in which trait–insomnia associations would be stronger under elevated distress, and an attenuation pattern, in which high psychological distress would reduce the relative contribution of stable personality traits.

Participants and Procedure

A convenience sample of Greek adults was recruited between January and March 2025 through social media, word of mouth, and university-affiliated networks. Of the 1,838 individuals who participated, 1,349 (63.2 % female) provided consent and completed the Limesurvey online questionnaire (completion rate = 82%), while 489 were excluded due to incomplete responses or invalid response patterns. Regarding age (18 to 79 years; M = 35.26, SD = 16.41), 37.4% were between 18 and 29 years old (n = 504), 30.1% were aged 30–44 (n = 406), 20.2% were 45–59 years old (n = 272), and 12.3% were 60+ years (n = 167). The institutional review board approved all procedures.

Measures

Statistical Analysis Plan

Data were screened for outliers and missing values (< 2% per variable) prior to analysis using SPSS 29.0 and the PROCESS macro v4.3 (Hayes, 2022). All variables met univariate normality assumptions (|skew| < 2, |kurtosis| < 7). Pearson correlations were conducted to examine zero-order associations among study variables. Hierarchical multiple regression analyses were performed to predict insomnia severity.

In addition to the primary domain-level analyses, supplementary models examined personality at the facet level to provide a more fine-grained understanding of the associations. Facet-level predictors were entered in the final step of the regression models after controlling for demographic variables and psychological distress.

Moderation analyses were conducted using PROCESS Models 1 and 2 to examine whether anxiety symptoms, depressive symptoms, sex, and age moderated the association between Neuroticism and insomnia severity. Exploratory facet-level moderation analyses were additionally conducted for the Neuroticism facets that demonstrated unique associations with insomnia severity in the hierarchical regression analyses. To reduce overparameterization and limit Type I error inflation, moderators were examined in separate models.

To further evaluate the attenuation pattern observed in the moderation analyses, additional exploratory analyses examined the variability of insomnia severity across low, moderate, and high levels of anxiety and depression. Mean scores, standard deviations, and variances of insomnia severity were compared across distress groups, and Levene’s tests were conducted to evaluate homogeneity of variances.

Significant interactions were further probed via simple-slope analysis and Johnson–Neyman regions of significance with 5,000 bias-corrected bootstrap samples. Effect sizes were interpreted according to Cohen (1988) guidelines.

Sex and Age Effects

Because Levene’s tests indicated heterogeneity of variances across demographic groups (p < .05), a two-factor Aligned-Rank-Transform ANOVA (ART-ANOVA) was conducted on ranked SCI scores, with sex (male vs female) and age group (18–25, 26–45, 46–65, 66+) entered as between-subject factors. The analysis revealed a significant main effect of sex, F (1, 1338) = 4.82, p = .028, whereby women reported higher insomnia severity than men. A robust main effect of age group also emerged, F (3, 1338) = 7.68, p < .001, indicating that younger participants (18–25 years) and older adults (66+ years) experienced more pronounced insomnia symptoms than the other cohorts. Tukey-adjusted pairwise contrasts confirmed that each of the two older groups differed significantly from the 18–25 and 26–45 groups (all adjusted ps < .01), whereas the two younger groups did not differ from each other. The sex × age interaction was non-significant F (3, 1338) = 0.25, p = .864, suggesting that the age-related rise in insomnia severity was comparable for men and women. The ART-ANOVA approach was preferred over traditional parametric alternatives (e.g., Welch ANOVA) as it allows for the examination of interaction effects in factorial designs under violations of homogeneity of variance.

Correlations

Pearson correlations (Table 1) indicated that insomnia severity correlated positively with Neuroticism (r = .38), Anxiety (r = .35), Depression (r = .39) and, weakly with Openness (r = .07), and negatively with Extraversion (r = −.10), Conscientiousness (r = −.18) and Agreeableness (r = −.15).

Facet-Level Analyses

To extend the domain-level findings, we conducted additional analyses at the facet level of the BFI-2. Correlational analyses indicated that insomnia severity was positively associated with all Neuroticism facets, particularly Depression (r = .38) and Anxiety (r = .32), and negatively associated with Conscientiousness facets, including Organization (r = −.17), Productiveness (r = −.16), and Responsibility (r = −.12). Among Extraversion facets, only Energy level was significantly related to insomnia (r = −.19), whereas Sociability and Assertiveness were not. Internal consistency estimates and zero-order correlations for all BFI-2 facets are presented in Supplemental Table S1.

A hierarchical regression analysis was then conducted to examine whether these facets uniquely predicted insomnia severity beyond demographic variables and psychological distress. In Step 1, age was a significant predictor, whereas sex was not. In Step 2, both anxiety (β = .128, p < .001) and depression (β = .172, p < .001) were positively associated with insomnia severity, explaining 17.4% of the variance.

In Step 3, the inclusion of personality facets led to a significant increase in explained variance (ΔR² = .041, p < .001), with the final model accounting for 21.5% of the variance. Among the facets, Neuroticism-related Anxiety (β = .129, p < .001) and Depression (β = .119, p = .004) emerged as significant predictors, whereas Emotional Volatility was not. From the Conscientiousness domain, only Organization was a significant negative predictor (β = −.100, p = .001). Additionally, Energy level, a facet of Extraversion, was negatively associated with insomnia severity (β = −.066, p = .019) (see Supplemental Table S2).

Moderation Analysis

A hierarchical multiple regression analysis was conducted to establish which covariates would enter the subsequent moderation analysis. The final model—including Neuroticism, BSI depressive symptoms, and BSI anxiety symptoms—was significant, F (5, 1343) = 63.33, p < .001, and explained 19% of the variance in self-reported insomnia (R² = .19, adj. R² = .19). Within this model, Neuroticism remained the most robust predictor (β = .20, t = 6.13, p < .001), followed by BSI depressive symptoms (β = .20, t = 5.62, p < .001) and BSI anxiety symptoms (β = .11, t = 3.08, p = .002). Sex and Openness to Experience were no longer significant once these variables were included, underscoring the primacy of the three focal predictors for the moderation stage.

Using these predictors, a PROCESS Model 2 moderation analysis tested whether BSI anxiety symptoms and BSI depressive symptoms jointly moderated the Neuroticism–insomnia association. The combined model accounted for 19.8% of the variance (R² = .198). Main effects remained significant for Neuroticism (B = 0.17, SE = 0.03, p < .001), Depression (B = 0.21, SE = 0.04, p < .001), and Anxiety (B = 0.12, SE = 0.04, p = .001).

A modest but statistically significant Neuroticism × Anxiety interaction emerged (B = −0.08, SE = 0.04, p = .045; ΔR² = .002), whereas the Neuroticism × Depression interaction was not (B = −0.04, p = .352; ΔR² = .001). Simple-slope analysis showed that the positive association between Neuroticism and insomnia was strongest at low Anxiety (−1 SD; b = 0.33, SE = 0.04, p < .001), moderate at the mean (b = 0.25, SE = 0.03, p < .001), and weakest at high Anxiety (+1 SD; b = 0.17, SE = 0.04, p < .001). Thus, Anxiety attenuates—but does not reverse—the Neuroticism–insomnia link, while Depression does not further moderate this relationship. The interaction pattern is shown in Figure 1.OPEN IN VIEWER

Exploratory facet-level moderation analyses were conducted for the Neuroticism facets that uniquely predicted insomnia severity. Because these analyses involved multiple interaction tests and conceptually overlapping constructs, findings should be interpreted cautiously. The results indicated that higher levels of depressive symptoms were associated with weaker facet–insomnia associations for both the Anxiety facet (b = −0.080, SE = 0.039, p = .042) and the Depression facet (b = −0.060, SE = 0.025, p = .015). Specifically, the associations between these facets and insomnia were weaker at higher levels of depressive symptoms.

In addition, age significantly moderated the association between the Anxiety facet and insomnia severity (b = 0.059, SE = 0.024, p = .014), such that the positive association between the Anxiety facet and insomnia severity was stronger among older participants. No significant moderation effects emerged for sex or for the interaction between the Depression facet and age. Full facet-level moderation results are presented in Supplemental Table S3.

Variance Check for the Ceiling-Effect Interpretation

To further examine whether the attenuation of the Neuroticism–insomnia association at higher levels of distress reflected a ceiling or variance-restriction effect, additional exploratory analyses compared the variability of insomnia scores across low, moderate, and high levels of anxiety and depression.

Results indicated that mean insomnia severity increased across distress groups; however, the variability of insomnia scores did not decrease at higher levels of anxiety or depression. In contrast, standard deviations and variance estimates remained relatively stable or slightly increased across groups. Levene’s tests further indicated significant heterogeneity of variances for both anxiety and depression groupings (ps < .01), suggesting that the attenuation pattern was unlikely to be explained solely by statistical variance restriction.

Accordingly, the findings do not provide strong support for a strict ceiling-effect interpretation and instead suggest that alternative mechanisms associated with elevated psychological distress may contribute to the weakened association between Neuroticism and insomnia severity.