Political cynicism is usually positively related to conspiracy beliefs (Swami, 2012; Swami et al., 2011). People who hostilely distrust politicians with regard to their motives and actions tend to question official explanations of events, which might result in endorsing conspiracy theories postulating the involvement of politicians in “covering up” evidence related to real-life events.

Life satisfaction might be negatively related to conspiracy beliefs (Swami et al., 2011); poor life satisfaction might result from feeling disadvantaged, and it has been speculated that underprivileged people are more likely to endorse conspiracy theories. However, the findings are mixed (Swami, 2012; Swami et al., 2011). Examining the intricate relationships at the item level in the present study could shed light on this relationship; it cannot be ruled out that only certain aspects of life satisfaction are related to conspiracy beliefs.

Several different constructs included in the present study are concerned with the self-appraisal of a person (Judge et al., 2002). High neuroticism, low self-esteem, low self-efficacy, high external locus of control, and low internal locus of control are generally thought to reflect a negative self-view. On the surface level, it might appear unnecessary to include several self-appraisal constructs. However, as shown in the following paragraphs, the strength of the relationship with conspiracy beliefs differs between the individual personality traits.

According to meta-analytic findings, a small negative relationship between self-esteem and conspiracy beliefs is usually reported. For example, Bowes and colleagues (2023) reported an average correlation of r = -.09, 95% CI [-.14, -.04]. Biddlestone and colleagues (2025) observed a correlation of r = -.10, 95% CI [ -.20, .01], and Stasielowicz (2022b) found a correlation of r = -.06, 95% CR [-.11, <.00]. Theoretically, this relationship could be bidirectional. On the one hand, people with relatively low self-esteem may be vulnerable to conspiracy theories because they provide potential explanations or the opportunity to blame other entities, including politicians and institutions, for their own problems and failures. On the other hand, believing in conspiracy theories could lead to a negative self-evaluation of one's ability to deal with problems. Exposure to conspiracy theories postulating the existence of uncontrollable forces or actors with bad intentions could lead to low self-esteem with a pessimistic worldview that assumes that external forces make it impossible for ordinary citizens to change the environment.

For self-efficacy and neuroticism, negligible average correlations were found in meta-analyses. Specifically, the average correlations with neuroticism reported across meta-analyses are r = .05, 95% CI [.03, .07], r = .03, 95% CI [-.02, .09], and r = .04, 95% CR [.01, .07] (Bowes et al., 2023; Goreis & Voracek, 2019; Stasielowicz, 2022b). For self-efficacy, Bowes and colleagues (2023) reported a correlation of r = -.07, 95% CI [-.22, .09] with substantial heterogeneity, indicating that self-efficacy might be relevant under certain circumstances.

Finally, people with low perceived control tend to believe in conspiracy theories (Biddlestone et al., 2025; Bowes et al., 2023). Whereas Bowes and colleagues found an average correlation of r = -.17, 95% CI [-.21, -.12], Biddlestone and colleagues reported a meta-analytic correlation of r = -.20, 95% CI [-.25, -.15] with external control and a correlation of r = -.04, 95% CI [-.11, .03] with personal control. The present study also distinguishes between control dimensions to enable the investigation of differential relationships with conspiracy mentality. Similar to self-esteem, a bidirectional relationship between perceived control and conspiracy mentality is plausible. People consuming conspiracy theories might conclude that politicians, health institutions, and other entities control many areas of everyday life (i.e., high external locus of control) and that their own actions do not matter (i.e., low internal locus of control). However, it is also possible that people who already have a low sense of control are attracted to conspiracy theories, as they offer the opportunity to blame others for the perceived lack of control.

Conspiracy mentality might be positively related to the tendency to go to bed later than intended (Kroese et al., 2014). Such bedtime delays can partially result from consuming online content in the evening. Ensuing sleep deficiency and associated symptoms, such as concentration difficulties, might make people more vulnerable to conspiracy theories disseminated on the Internet that bedtime procrastinators are likely to encounter while searching for exciting online content at night.

Overall, the present study seeks to:

Given the present research’s goal of identifying potential intervention targets, a threshold had to be set to decide what constitutes a promising intervention target. As various undergraduate projects were conducted as part of the current study, a simple correlation criterion was used. The correlation with conspiracy mentality had to be greater than a typical correlation between personality variables and conspiracy beliefs. Since the average correlation between conspiracy beliefs and previously examined personality variables is close to r = . 15 (Biddlestone et al., 2025), the threshold was set to r = .20. To achieve a power of 80% at this correlation threshold, a sample size of approximately 200 participants is required (R package pwr2ppl 0.5.0). Therefore, it was planned to collect data from at least 200 participants, with no upper limit to enable more complex analyses.

Users of internet forums and social media groups discussing conspiracy theories were asked to participate in a German online survey. In addition, other groups (primarily students) were approached to increase variability in the conspiracy mentality. Data were collected in two waves, before and during the COVID-19 pandemic (2019 and 2020, respectively), providing the unique opportunity to investigate network stability across different circumstances. To mitigate potential data dependency issues, different groups (e.g., student groups and internet forums) were contacted at each wave. In total, 673 people answered at least one question (1^st wave – 438 people, 2^nd wave – 235 people).

The project was carried out in accordance with the ethical guidelines of the American Psychological Association. After providing informed consent, participants received demographic questions. Only participants older than 18 received further questions. Specifically, these participants completed a few questionnaires for different undergraduate and graduate projects: Bedtime procrastination, neuroticism, self-esteem, life satisfaction, self-efficacy, political cynicism, locus of control (only during the second recruitment wave), conspiracy beliefs, and conspiracy mentality. Finally, participants could choose the compensation for completing the survey: (1) Draw (eight vouchers each worth 20 €), (2) Course credit, or (3) No compensation.

Descriptive statistics of the relevant questionnaires are presented in Table 1. All relevant instruments had good or very good reliability across all samples.

Bedtime procrastination was assessed with the Bedtime Procrastination Scale (Kroese et al., 2014). Participants completed all nine items (e.g., “Often I am still doing other things when it is time to go to bed”) using a five-point scale (1 – Almost never, 5 – Almost always).

Conspiracy mentality was measured with the Conspiracy Mentality Questionnaire (Bruder et al., 2013). The scale consists of five items (e.g., “I think that there are secret organisations that greatly influence political decisions”), and the responses are provided on an 11-point scale (1 = Certainly not, 11 = Certain).

Life satisfaction was assessed using the Satisfaction with Life Scale (Glaesmer et al., 2011). Participants responded to five items (e.g., “In most ways my life is close to my ideal”) using a seven-point scale (1 = Does not apply at all, 7 = Applies completely).

Locus of control was assessed only in the second sample. To this end, the two-ways-questionnaire for the assessment of control beliefs was used (Maes & Schmitt, 1999). Specifically, the internal locus of control (e.g., “Everyone has their life in their own hands”) and external locus of control (e.g., “Others have a decisive influence on your own life”) were assessed by four items using a five-point scale (1 = Completely disagree, 5 = Completely agree).

Neuroticism was measured using one subscale of the NEO−Five−Factor Inventory (Körner et al., 2008). Participants completed the relevant six items (e.g., “Sometimes I feel completely worthless”) using a five-point scale (1 = Completely disagree, 5 = Completely agree).

Political cynicism was assessed with the Political Cynicism Scale (Pattyn et al., 2012). Responses to the eight items (e.g., “Our political leaders are prepared to lie to us whenever it suits their purposes”) were provided using a five-point scale (1 = Completely disagree, 5 = Completely agree).

Self-efficacy was measured using the General Self-Efficacy Scale (Schwarzer & Jerusalem, 1995), which consists of ten items (e.g., “Thanks to my resourcefulness, I know how to handle unforeseen situations”). The responses were provided using a four-point scale (1 = Not true at all, 4 = Exactly true).

Self-esteem was assessed using the Rosenberg Self-Esteem Scale (von Collani & Herzberg, 2003). Participants completed the ten items (e.g., “I certainly feel useless at times”) using a four-point scale (1 = Does not apply at all, 4 = Applies completely).

Seventy-seven out of 673 participants did not complete the survey or provided low-quality data (see Stasielowicz, 2024b for all exclusion criteria). Overall, before the COVID-19 pandemic, 403 participants could be included. During the pandemic, data from 193 participants were included. Since skipping questions was not possible in the online survey, there were no missing data in the final sample.

Most participants were female (Sample 1: 277 women, 123 men, and three intersex/diverse people; Sample 2: 140 women, 52 men, and one intersex/diverse person). The age varied between 18 and 81 (Sample 1: Mdn = 23, M = 28.43, SD = 12.24; Sample 2: Mdn = 23, M = 25.91, SD = 9.03).

The descriptive statistics of the main questionnaires are reported separately for both measurement waves and the combined sample in Table 1. Bayesian bivariate correlations between demographic variables and questionnaire scores are presented in Figure 1.

Despite the different external circumstances, both the pre-pandemic and pandemic samples showed similar patterns. For example, there were strong bivariate relationships among the three traditional self-appraisal constructs (i.e., neuroticism, self-efficacy, and self-esteem). Notably, only political cynicism was strongly related to conspiracy mentality across both measurement waves. Interestingly, during the pandemic, conspiracy mentality was more strongly related to life satisfaction and neuroticism than before the pandemic. Specifically, people with relatively low life satisfaction or high neuroticism reported a higher conspiracy mentality. In addition, people with a high external locus of control, which was only assessed in the second sample, tended to have a high conspiracy mentality.

To examine whether relationships hold up when simultaneously accounting for multiple variables, network analyses with mean questionnaire scores were conducted in the next step.

Since focusing on the relationships between aggregate scores might mask intricate patterns, network models consisting of individual questionnaire items were built in the next step. Because of model complexity (e.g., the number of items, ordinal data), the uncertainty around the estimates is large for these models. The uncertainty is additionally increased by the reduced willingness to participate in surveys early during the COVID-19 pandemic (e.g., taking care of children during lockdowns had higher priority). Therefore, unless otherwise stated, rather than directly comparing the pre-pandemic and pandemic data, pre-pandemic data were compared to the pooled data to determine whether the relationships identified in the pre-pandemic data changed considerably after including data collected during the pandemic.

Although most of the identified item-level relationships were plausible (e.g., the sign of the relationship), and the respective estimates might help inform future hypothesis-based research, it would be too speculative to provide recommendations for designing interventions at this stage. Therefore, a cautious interpretation of the item-level patterns is provided for interested readers in the Open Repository at Stasielowicz (2025).

Overall, the network models based on aggregate scores paint a slightly different picture than the bivariate correlations (see Figure 1 and Figure 2). While both bivariate and partial correlations confirmed that political cynicism tends to go along with a high conspiracy mentality, the relationships with life satisfaction, self-esteem, and external locus of control were different. In the pre-pandemic data, the negligible bivariate correlation between conspiracy mentality and life satisfaction became positive after accounting for various personality variables in the network model. Furthermore, in the network model based on the pandemic data, a negligible negative relationship with self-esteem turned into a small positive relationship, and the positive relationship with an external locus of control became negligible.

Because of the differences between bivariate correlations and partial correlations from network models, one should not assume that the magnitude or even sign of the causal relationship with antecedents and consequences of conspiracy beliefs can be derived from the meta-analytic estimates of bivariate correlations (Biddlestone et al., 2025; Bowes et al., 2023; Hornsey et al., 2023; Stasielowicz, 2022b). At the same time, it is worth noting that cross-sectional network models do not satisfy all three typical social science causality criteria (Preacher, 2015). Although cross-sectional network models can help, (a) establish a relationship between two variables, and (b) rule out alternative explanations by adjusting for further variables, the (c) temporal causality condition is not met. Therefore, it is recommended to conduct longitudinal network analyses to ensure that the presumed cause is assessed before the consequence, increasing the plausibility of a causal relationship.

The unexpected start of a pandemic before the second data collection enabled the investigation of the stability of psychometric networks. Despite the radically different external circumstances (e.g., pandemic measures such as lockdowns and social distancing), many network relationships were stable across the samples, including the positive relationship between political cynicism and conspiracy mentality and the negligible relationship between self-efficacy and conspiracy mentality.

Still, a test systematically comparing the networks revealed that, despite the high correlation, the networks cannot be regarded as equal. For example, before the COVID-19 pandemic, life satisfaction was positively related to conspiracy mentality, whereas during the pandemic, there was a negative relationship. The data collection strategies and sample characteristics cannot explain this pattern because they were similar across both samples (see Table 1). However, Germany implemented strict restrictions, such as mask-wearing mandates and social distancing protocols, in the first years of the pandemic. It is possible that during the pandemic, people with a high conspiracy mentality had relatively low life satisfaction because they were frustrated by the wide-ranging preventive measures they attributed to the purported ongoing conspiracies. In contrast, before the pandemic, a high conspiracy mentality corresponded to higher life satisfaction, which could mean that conspiracy theories offer consolation by providing explanations for events in relatively peaceful times.

The observed pattern of relationships changing across different circumstances may have implications for the development of interventions. Amenable intervention targets may be different during crises. Paying attention to citizens’ life satisfaction, especially during crises, could reduce vulnerability to implausible conspiracy beliefs. However, these novel insights need to be confirmed in future studies comparing the stability of networks before, during, and after political, societal, and economic crises.

Item-based networks were somewhat less stable than networks based on aggregate scores. On the one hand, it could mean that some item-level relationships are spurious. On the other hand, it cannot be ruled out that model complexity (e.g., number of estimated relationships, ordinal data) contributed to the uncertainty of the estimates. Furthermore, similar to the aggregated data, some sample differences might stem from the varying external circumstances.

Notably, various additional comparisons indicated that networks were relatively stable. For example, aggregate score networks did not change considerably after the locus of control variables were included. Finally, the rank order and estimated magnitude of the centrality indices (i.e., betweenness, closeness, strength, and expected influence) were often similar across samples.

Identifying novel intervention routes is important as available interventions usually have a small impact on conspiracy beliefs (O’Mahony et al., 2023; Stasielowicz, 2026a). Since bivariate analyses and mean-score-based network models across all samples confirmed a positive relationship between political cynicism and conspiracy mentality (Swami, 2012; Swami et al., 2011), it may be worth examining whether increasing the politicians’ transparency and accountability (e.g., compulsory lobby registers) could help reduce the conspiracy mentality. Future studies could use databases with transparency rankings to investigate whether countries with higher transparency and accountability tend to have a lower conspiracy mentality.

Finally, centrality measures in both samples indicated that conspiracy mentality had the highest expected influence in the networks based on mean scores. Thus, reducing conspiracy mentality might impact various personality variables. Such hints are important as the causal effects of conspiracy beliefs are not well-established; intervention studies focus on reducing conspiracy beliefs without examining downstream consequences (Stasielowicz, 2026a).

Although sample composition was similar before and during the COVID-19 pandemic, which enabled assessing the robustness of the findings in similar groups, most of the 600 participants were relatively young (Mdn = 23, Min = 18, Max = 81) compared to the median age in Germany - 47 (Ritchie & Roser, 2019). Furthermore, twice as many women as men participated in the current study, which differs from the balanced ratio in the general population.

As indicated by the uncertainty of some network estimates, especially at the item level, the power for some analyses was not optimal. Sample size planning was primarily conducted with bivariate correlation analyses in mind for undergraduate projects. While the target sample size for those analyses could be reached, more participants are required for more complex analyses. To reduce the uncertainty of the estimates, researchers are encouraged to conduct large-scale studies, preferably as multi-lab projects, to ensure the project's feasibility.

For practical reasons, the present study could not consider all correlates of conspiracy beliefs examined in meta-analyses (Biddlestone et al., 2025; Bowes et al., 2023; Hornsey et al., 2023; Stasielowicz, 2022b). Including constructs such as narcissism, paranoia, schizotypy, and religiosity could lead to other fine-grained implications in future studies.

Finally, the current study's cross-sectional design and unrestricted time frame in the questionnaires’ instructions cannot be used to derive causal conclusions. It is plausible that personality acts both as a cause and an outcome. Furthermore, similar to the bidirectional relationship between conspiracy beliefs and COVID-19-related preventive behavior (Stasielowicz, 2022a), it cannot be ruled out that the strength of the relationships examined in the present study depends on the time lag between assessing the respective constructs. Furthermore, although mostly stable personality chararcteristics were assessed in the present study, which were thought to occupy a similar place in the causal chain, it cannot be ruled out that there are collider-type relationships which may bias correlation estimates (Cinelli et al., 2024; Rohrer, 2018) or that relevant variables were not included in the networks. Future studies will benefit from incorporating theoretical deliberations about the causal order of the variables and the meta-analytic evidence about the strength of the relationships, using longitudinal designs accounting for variables regarded as relevant predictors of conspiracy beliefs (Biddlestone et al., 2025; Bowes et al., 2023; Hornsey et al., 2023; Stasielowicz, 2022b) and using time lags between three and six months. Following these recommendations may help establish whether the relationship between conspiracy mentality and personality is bidirectional and how it evolves over time.