Abstract
National narcissism and national identification, two distinct types of national commitment, differ in terms of their psychological concomitants. Therefore, in the current article, we hypothesized that they would also relate to different adult attachment styles. Namely, we proposed that national narcissism would be positively associated with higher attachment anxiety, while national identification would be associated with lower attachment anxiety and avoidance. These hypotheses were tested in three cross-sectional surveys (Study 1 N = 570; Study 3 N = 558; Study 4 N = 649) and one longitudinal survey (Study 2 N = 808). In all studies, we found a consistent positive relationship between attachment anxiety and national narcissism, and a negative relationship between attachment avoidance and national identification. Finally, we also demonstrated indirect effects of attachment anxiety (via national narcissism) on maladaptive group-related outcomes: conspiracy beliefs, non-normative collective action, and willingness to conspire.
Keywords
Investigations of social identity-related processes have often focused on individual factors shaping different types of in-group commitment (e.g., Adorno et al., 1950; Fromm, 1973; Turner, 1982). In line with classic psychological theorizing (Tajfel & Turner, 1986), social identity may serve as a compensation for the frustration of different psychological needs. Indeed, previous research showed that people identified with their in-groups, for example, to manage feelings of uncertainty (Mullin & Hogg, 1998), existential threats (Castano et al., 2002), or lack of personal control (Agroskin & Jonas, 2013). In such cases, the group is perceived as offering opportunities for self-fulfillment and, thus, being at the service of the self. Those who are related to their in-group in this way are less motivated to invest the individual effort to benefit the group. Instead, they search for a possibility to satisfy their individual needs by becoming part of a strong collective (Cichocka, 2016). This type of in-group commitment is often defensive and destructive, both from the perspective of intra- and intergroup processes (Golec de Zavala et al., 2009). It is related to in-group disloyalty and hostility toward out-group members (Marchlewska et al., 2020). Not all types of in-group commitment, however, are based on mechanisms of psychological compensation. In fact, some individuals are more focused on group- than individual-level benefits and commit to their in-group in a constructive way.
In this article, we discuss the concomitants of these two distinct forms of in-group commitment and assume that individual frustration and satisfaction should translate into collective defensiveness and security, respectively. We hypothesize that the way people form personal relationships should be related to the way they commit to their in-groups as well. Specifically, we are the first to explore the links between adult attachment and defensive versus secure commitment related to one’s national in-group. We focus on adult attachment styles to better understand the role of basic individual-level factors in adopting different forms of national commitment—phenomena shaping attitudes and behaviors that have a significant impact on entire societies.
National Narcissism Versus Secure National Identification
We operationalize defensive national commitment as national narcissism (Cichocka & Cislak, 2020)—a grandiose image of one’s national group that is contingent on the external recognition of its worth (Golec de Zavala et al., 2009). In turn, secure (i.e., non-narcissistic) national commitment is defined as an unpretentious investment in the national in-group, independent of the recognition of the group in the eyes of others (Golec de Zavala et al., 2013). This distinction is inspired by research on self-evaluation, which differentiates between individual narcissism (i.e., an inflated self-evaluation associated with the need for external validation and defensiveness in response to ego threats; Horvath & Morf, 2009) and secure self-esteem (i.e., a realistic pride people take in their strengths, which serves as a buffer against psychological threats; Kernis, 2005; Marchlewska & Cichocka, 2017). In our methodological approach, we follow the previously used procedure (e.g., Cichocka et al., 2018; Golec de Zavala et al., 2009; Marchlewska, Cichocka, et al., 2022; Paulhus et al., 2004), where researchers distinguish the unique effects of individual narcissism versus self-esteem by accounting for their shared variance. Similarly, to observe the unique effects of narcissistic versus secure national commitment, researchers co-vary out the variance shared between national narcissism and national identification. Such an approach gives the possibility to observe the distinctive effects of defensive national commitment, which is national narcissism minus its overlap with national identification, and the unique effects of secure national commitment, which is national identification minus its overlap with national narcissism (e.g., Cichocka & Cislak, 2020). Controlling for national narcissism while measuring national identification makes it possible to obtain an index of secure national identification (i.e., an unpretentious investment in the in-group, independent of the recognition of the group in the eyes of others; Golec de Zavala et al., 2013; Marchlewska, Cichocka, et al., 2022), previously linked to psychological security (e.g., higher personal control; Cichocka et al., 2018) and positive inter- and intra group outcomes (Cichocka, 2016).
Previous research has shown that national narcissism is a type of in-group commitment that is built on the foundations of a threatened ego (Cichocka et al., 2018; Golec de Zavala et al., 2020 see also Fromm, 1973). For example, it was shown to increase in response to low levels of personal control (Cichocka et al., 2018; see also Marchlewska et al., 2020) and observed among those individuals who scored low (vs. high) on self-esteem (Golec de Zavala et al., 2020). Still, it is worth stressing that although national narcissism results from the frustration of individual needs, it does not necessarily reinforce feelings of personal control or boost self-esteem (Cichocka et al., 2018). In contrast, due to its defensive nature, it leads to maladaptive psychological outcomes (Marchlewska et al., 2020; for a review, see Cichocka & Cislak, 2020).
In fact, individuals scoring high on collective narcissism are focused on defending their group from real or imagined enemies, rather than on investing the individual effort to actually benefit the group (Cichocka, 2016; Marchlewska, Cichocka, et al., 2019, 2022; Marchlewska et al., 2020). This is probably due to the exaggerated feelings of insecurity, which underlie this particular in-group commitment (Cichocka, 2016). Prior studies demonstrated that collective narcissism was positively related to perceiving even ambiguous intergroup situations as threatening (Golec de Zavala et al., 2009). This may explain its positive relationships with conspiracy beliefs (i.e., beliefs in secret plots by powerful and malevolent groups; Douglas & Sutton, 2008) and out-group hostility (Marchlewska, Cichocka, et al., 2019; Marchlewska, Górska, et al., 2022). As previously mentioned, national narcissism is not only related to negative perceptions of out-groups but also to lack of trust toward other in-group members. Recent research revealed that in some cases, those scoring high on collective narcissism are even ready to act contrary to the interests of their own group members, by, for example, showing a stronger willingness to conspire against them (e.g., Biddlestone et al., 2022; Molenda et al., in press), which may result in negative relations within the group itself (Cichocka & Cislak, 2020; Marchlewska et al., 2020, in press).
This, however, is not the case among people who commit to their nation in a secure way. Secure (i.e., non-narcissistic) national commitment stems from satisfied psychological needs (e.g., Marchlewska et al., 2020) and is based on the foundation of a stable and secure self (Cichocka et al., 2018). For example, it was previously linked to high levels of personal control (e.g., Cichocka et al., 2018) and positive emotionality (Golec de Zavala, 2019). Those who score high (vs. low) on secure national commitment do not react negatively toward out-group members (Górska, Stefaniak, et al., 2022). They are also less defensive and do not respond to criticism with retaliatory hostility (Cichocka, 2016). They also engage in constructive intragroup behaviors. For example, in their recent research Marchlewska, Hamer, et al. (2022) found that willingness to vaccinate against COVID-19 was positively related to secure national commitment and negatively to national narcissism.
These results suggest that collective narcissism and secure identification often have opposite relationships with many psychological variables. Collective narcissism (but not secure identification) is related to maladaptive traits and states. Given such findings, we expected that these two types of national commitment (i.e., narcissistic vs. secure) would be also related differently to insecure and secure interpersonal attachment.
Adult Attachment and Group Processes
Attachment in adulthood can be conceptualized as an orthogonal relationship between two dimensions; attachment anxiety, which reflects feelings of low self-worth and fear of abandonment and rejection, and attachment avoidance, which reflects discomfort with closeness, excessive self-reliance, and lack of confidence in depending on others (Bartholomew & Horowitz, 1991; for a review of attachment measurement, see Mikulincer & Shaver, 2016). Scoring low on both of these dimensions would constitute a secure attachment. Although adult attachment research has primarily been concerned with individual and interpersonal processes, a growing body of research has demonstrated that it is also able to explain people’s maladaptive and adaptive group processes.
For example, Rom and Mikulincer (2003) demonstrated how attachment anxiety and avoidance affect group-related regulatory strategies. Specifically, attachment-anxious people reported negative self-concepts as group members and demonstrated poor instrumental support (i.e., contributing to a successful completion of tasks) toward group interactions. On the other hand, attachment-avoidant people had negative views of other in-group members and group interactions (but did not find them threatening), sought self-reliance goals, and demonstrated poor instrumental and socioemotional support (i.e., contributing to morale and cohesion) toward groups. Rom and Mikulincer (2003) argued that their findings could be described in terms of hyperactivating and deactivating regulatory strategies, respectively (e.g., Cassidy & Kobak, 1988). Attachment anxiety is characterized by a heightened sensitivity toward threats, coupled with continual elicitation of attention, support, and care from others, while attachment avoidance is characterized by psychological avoidance of negative effects and denial of attachment needs.
The studies described above are theoretically important for understanding how attachment affects group processes but did not explore the different ways in which people may commit to social groups. In this vein, Milanov and colleagues (2013) examined such relationships and found unique associations between attachment and different types of in-group attitudes. Secure attachment was associated with a higher social (e.g., higher similarity to in-group members) and communal (e.g., able to empathize with in-group members) identification, compared with attachment avoidance only. Alternatively, attachment avoidance was associated with a higher interdependent identification (e.g., instrumental, exchange-orientated), compared with secure attachment only. This study shows that secure attachment is associated with constructive forms of in-group commitment, but does not appear to measure insecure, defensive types of in-group commitment (e.g., collective narcissism). Furthermore, like other studies described in this section, this study did not account for the context of the group (i.e., national) either. Moreover, no research has explored the relationships between attachment and insecure (particularly, narcissistic) versus secure national in-group commitment. The current research aims to delineate these relationships.
Adult Attachment and Different Types of In-Group Commitment
Thus far, one study measured maladaptive and adaptive aspects of national commitment (i.e., blind nationalism vs. constructive patriotism, respectively; Rothí et al., 2005) alongside attachment anxiety and avoidance (see Marsh & Brown, 2011). Specifically, nationalism was found to positively correlate with attachment anxiety only. Unfortunately, however, the relationships between patriotism and attachment were not the main focus of their study and were therefore not reported. 1 This study provides initial support for the notion that insecure attachment may be associated with insecure forms of national commitment; however, it falls short on simultaneously taking into account insecure and secure forms of national commitment, which would provide a more nuanced understanding of how attachment differences may affect these types of identification. Notwithstanding the dearth of literature on this approach, there is a lot of indirect evidence that suggests these variables are associated in disparate ways.
To begin, we know that collective narcissism is borne out of a weak and threatened ego (Golec de Zavala et al., 2020) and is associated with a heightened sensitivity toward threats (Cichocka, 2016). Similarly, attachment anxiety is characterized by negative models of the self as unworthy, vulnerable, and helpless (Bartholomew & Horowitz, 1991), and is associated with regulatory strategies that intensify feelings of distress (i.e., hyperactivating; Mikulincer & Shaver, 2003); all of which appear to be consistent at the group level (e.g., Rom & Mikulincer, 2003). Furthermore, like collective narcissism (Golec de Zavala & Cichocka, 2012; Marchlewska, Cichocka, et al., 2019), attachment anxiety has also been associated with an increased belief in conspiracy theories, which was argued to stem from the tendency to appraise life’s problems in catastrophic terms (Green & Douglas, 2018). Therefore, it seems reasonable to assume that attachment anxiety will be a concomitant of collective narcissism, due to their similarities with a threatened sense of self and the propensity to exaggerate threats.
Secure national commitment, on the other hand, is likely to be associated with both low attachment avoidance and anxiety—that is, secure attachment. For example, secure national identification is associated with satisfied psychological needs and a secure sense of self (Cichocka et al., 2018; Marchlewska et al., 2020), high personal control (Cichocka et al., 2018), and positive emotionality (Golec de Zavala, 2019); each of which is indicative of low attachment anxiety (i.e., a positive model of the self). Furthermore, secure national commitment is also associated with positive attitudes toward in-group and out-group members alike (Cichocka, 2016), which is indicative of low attachment avoidance (i.e., a positive model of others). In fact, secure attachment has been associated with a number of adaptive group-related processes, which further indicate its positive relationship with secure national commitment. As already mentioned, secure attachment is associated with constructive types of group identification (Milanov et al., 2013). Interestingly, other research has shown that both the majority and minority (i.e., immigrant) group members with secure attachment have positive attitudes toward integration, with the opposite only being true for attachment avoidance (Van Oudenhoven & Hofstra, 2006; see also Hofstra et al., 2005). In addition, experimental studies have demonstrated positive effects of situationally inducing a sense of secure attachment (Mikulincer & Shaver, 2001). Specifically, by priming a sense of attachment-related security—compared with a control group—participants were found to derogate out-group members less, even when their self-esteem and cultural worldviews were threatened.
These studies suggest that feeling secure with oneself and other people should lead to more secure forms of national commitment. Taken together, we assumed that high attachment anxiety underscores narcissistic national commitment, while low attachment avoidance and anxiety underscore secure national in-group commitment.
Overview of the Current Research
As stated before, national narcissism, in contrast to secure national identification, stems from unsatisfied individual needs and a threatened ego (e.g., Cichocka, 2016; Fromm, 1973; Golec de Zavala et al., 2020). However, the empirical evidence identifying those exact unsatisfied needs or psychological difficulties underpinning narcissistic in-group commitment is scarce. In turn, comprehending the psychological roots of national narcissism would be beneficial for future interventions—as targetting them at groups with specific difficulties could be more efficient. Therefore, our research aims to fill this gap by investigating another possible foundation of the types of in-group commitment: adult attachment styles.
We tested our hypotheses in a series of four studies. We assumed that high attachment anxiety should be linked to a defensive form of in-group commitment, that is national narcissism when controlled for national identification (H1). Moreover, we hypothesized that low attachment avoidance (H2) and anxiety (H3) should be linked to a secure form of in-group commitment, which is national identification when controlled for national narcissism. We expected the relationships between attachment and different types of in-group commitment to be especially pronounced when controlled for the shared variance between national narcissism and national identification. For this reason, we first report zero-order correlations and then, we report the effects of attachment anxiety and avoidance on each type of in-group commitment while considering both national narcissism and national identification in one model (i.e., controlling for their shared variance). We investigate these effects in three cross-sectional surveys (Study 1, Study 3, and Study 4) and one two-wave survey (Study 2), in which we focus on the reciprocal paths between attachment anxiety and avoidance, and both forms of in-group commitment. In Study 3 and Study 4, we additionally check whether both types of in-group commitment may differentially mediate the relationships between attachment and group-related outcomes: non-normative collective action (Studies 3), conspiracy beliefs (Studies 3 and 4), and willingness to conspire against one’s in-group (Study 4). Our studies were not preregistered. In all studies, we aimed to include at least 400 participants, which gave us a power of .80 for detecting even small associations between variables (for r = .14; Cohen, 1988; G*Power yields a target of 395 participants). Data (with labeled variables) for all studies are available at the Open Science Framework: https://osf.io/t7k4g/?view_only=9da20b2635bd460d928ddef3b988d511.
Study 1
In Study 1, we checked for the relationships between attachment anxiety and avoidance, and two forms of national identity. We assumed that high attachment anxiety should be associated with national narcissism (when controlled for national identification), whereas low attachment anxiety and avoidance should be linked to national identification (when controlled for national narcissism).
Method
Participants and procedure
The data for Study 1 were obtained from a 2019 online survey, conducted by a local research company, and also used in other academic studies (e.g., Marchlewska, Cichocka, et al., 2019). Five hundred seventy Polish participants were recruited for this study (284 women, 286 men), aged between 18 and 25 (M = 22.14, SD = 2.24). The sample was nonprobability, quota-based, and representative of young Polish adults in terms of gender, education, and size of the place of residence. Quotas were selected based on the last National Census of Population and Housing, carried out by the Central Statistical Office (Główny Urząd Statystyczny; GUS).
Measures
The scales employed in the present research were part of a larger questionnaire that included the measures of various personality and social psychology constructs (e.g., need for cognitive closure; Kossowska et al., 2012). 2 Unless otherwise noted, all measures used a response scale ranging from 1 (strongly disagree) to 5 (strongly agree).
Attachment
The Revised Adult Attachment Scale (RAAS; Adamczyk & Pilarska, 2012; Collins, 1996) was employed to measure adult attachment. While the initial version of the RAAS consists of three subscales (i.e., anxiety [e.g., “I often worry that romantic partners don’t really love me.”], close [e.g., “I find it relatively easy to get close to people.”], and depend [e.g., “I find it difficult to allow myself to depend on others.”]), for this study, we formed the attachment avoidance index by combining close and depend subscales (see Collins, 2008). Within the avoidance subscale, five items were recorded (three items of depend subscale and two items of close subscale) according to alternative scoring proposed by Collins (2008). At the same time, the original anxiety subscale was used as a measure of attachment anxiety. Both attachment avoidance (α = .70) and anxiety (α = .86) were internally consistent. 3
National identification
To assess national identification, we used the full, 12-item Social Identification Scale (Cameron, 2004), where Poles served as the reference group (e.g., “I have a lot in common with other Poles”; α = .86).
National narcissism
Collective narcissism with respect to the national in-group (national narcissism) was measured with the short, five-item version of the Collective narcissism Scale (e.g., “If the Polish nation had a major say in the world, the world would be a much better place”; α = 85; Golec de Zavala et al., 2013).
Covariates
Political conservatism (1 = left, 7 = right), gender (0 = female, 1 = male), age, education (1 = primary degree or no degree, 2 = vocational degree, 3 = high school or postsecondary degree, 4 = university degree), and size of the place of residence (1 = rural area or village, 2 = town up to 20,000 residents, 3 = town 20,001–100,000 residents, 4 = town 100,001–200,000 residents, 5 = city 200,001–500,000 residents, 6 = city with more than 500,000 residents) served as covariates. 4
Results
Analytic strategy
Our analyses were divided into two parts—inspecting zero-order correlations and hypotheses testing. To verify our hypotheses, we tested two path models. First, we regressed national narcissism on attachment anxiety and avoidance and on national identification. Accounting for the national identification allowed us to check whether attachment anxiety and avoidance explained the variance in national narcissism when national identification was controlled for. 5 Likewise, to check whether attachment anxiety and avoidance predicted national identification (when controlled for national narcissism), we tested a separate model, in which national identification was regressed on attachment anxiety and avoidance, and on national narcissism. Finally, we checked whether (a) accounting for the covariates and (b) using robust estimation method changed our results in a theoretically meaningful way. Missing data (0.8%) were handled with the Full Information Maximum Likelihood (FIML) estimator (Enders & Bandalos, 2001). All models were estimated with the use of Mplus 8.0. Given that the analyzed models were saturated (i.e., without degrees of freedom), we do not report any fit indices.
Preliminary analyses
Table 1 presents descriptive statistics and intercorrelations for the variables measured in Study 1. In comparison with women, men reported more conservative political views, M = 4.08, SD = 1.50 versus M = 4.58, SD = 1.30, t(362.04) = −3.48, p < .001, d = 0.35. Gender did not differentiate participants in terms of attachment anxiety or avoidance, national narcissism, national identification, or demographics (all ps > .064). Attachment anxiety was associated positively with national narcissism and negatively with national identification, whereas avoidance displayed a negative link with national identification only.
Descriptive Statistics and Correlations (Study 1).
Note. N = 570.
p < .05. **p < .01. ***p < .001.
Hypotheses testing
As presented in Figure 1, attachment anxiety was positively related to national narcissism (when controlled for national identification), B = 0.25, SE = 0.04, 95% confidence interval (CI) = [0.17, 0.34], β = .25, p < .001, but attachment avoidance was unrelated to national narcissism, B = −0.04, SE = 0.07, 95% CI = [−0.19, 0.10], β = −.02, p = .585. At the same time, both attachment anxiety, B = −0.09, SE = 0.03, 95% CI = [−0.16, −0.02], β = −.11, p = .009, and avoidance, B = −0.21, SE = 0.06, 95% CI = [−0.32, −0.10], β = −.16, p < .001, were negatively related to the non-narcissistic national identification (i.e., national identification when controlled for national narcissism). Adding covariates into the model or using the robust estimation method did not alter our conclusions.

Results of path models testing the effects of attachment anxiety and attachment avoidance on national narcissism (when controlled for national identification) and national identification (when controlled for national narcissism), Study 1.
Discussion
Study 1 offered initial support for our theorizing, confirming our hypotheses. In a sample of young adults, national narcissism (when controlled for national identification) was positively related to attachment anxiety but not to avoidance. Thus, H1 received full support. In turn, national identification (when controlled for national narcissism) was negatively associated with both attachment anxiety and avoidance, which confirmed H2 and H3 in full as well.
Study 2
Although the results from Study 1 were encouraging, they were rather preliminary. Therefore, in Study 2, we tested our main hypotheses utilizing a two-wave study design, assuming that the relations identified in Study 1 would remain stable over time. Moreover, as the sample in Study 1 was drawn from the population of young adults, generalizability of our findings could be questioned. Thus, in Study 2, data were collected in a nationwide survey of adult Poles.
Method
Participants and procedure
Study 2 was administered as part of a two-wave in-house survey of adult Poles. To obtain a representative sample in the first wave, we employed random sampling based on the Polish identity number (PESEL) as the sampling frame. The two measurements were divided by a 6-month interval. Data were collected using computer-assisted personal interviews. Out of the 1,300 respondents who participated in the first measurement, 808 (62.2%) 6 took part in the second measurement. 7 Only individuals who participated in both waves of the survey comprised the sample of this study (429 females, 379 males, Mage = 46.62, SDage = 16.27).
Measures
Measures used in the analyses presented below were embedded in a larger questionnaire that assessed a range of psychology constructs (e.g., right-wing authoritarianism; Funke, 2005). Due to space constraints, short versions of the original measures were employed.
Attachment
To measure the participants’ level of attachment avoidance, we used four items from the RAAS depend and close subscales (Collins, 2008): “I find it relatively easy to get close to others,” “I am comfortable developing close relationships with others,” “I know that people will be there when I need them,” and “I am comfortable depending on others” (1 = strongly disagree, 7 = strongly agree). Prior to forming the attachment avoidance index, all four items were reverse-scored. The scale showed satisfactory reliability across both measurement occasions, αT1 = .85, αT2 = .85. At the same time, attachment anxiety was assessed with two items taken from the RAAS anxiety subscale: “When I show my feelings for others, I’m afraid they will not feel the same about me” and “I often worry that romantic partners won’t want to stay with me” (1 = strongly disagree, 7 = strongly agree). The correlation between these items was moderate, rT1(724) = .40, p < .001 rT2(776) = .49, p < .001.
National identification
National identification was assessed with three items borrowed from Cameron’s (2004) scale (see also Górska et al., 2020): “I feel strong ties to other Poles” (in-group ties), “In general, being a Pole is an important part of my self-image” (in-group centrality), and “In general, I’m glad to be a Pole” (in-group affect). The response scale ranged from 1 = strongly disagree to 5 = strongly agree. The scale was internally consistent, αT1 = .86, αT2 = .82
National narcissism
National narcissism was assessed as in Study 1 (Golec de Zavala et al., 2013), αT1 = .89, αT2 = .87.
Covariates
Similar to Study 1, political conservatism (1 = left, 7 = right), gender (0 = female, 1 = male), age, education, and size of the place of residence served as covariates.
Results
Analytic strategy
Again, our analyses involved two steps. First, we inspected intercorrelations for the variables of interest. Next, an autoregressive cross-lagged path model (Selig & Little, 2012) was estimated to verify our hypotheses. In this solution, each out of the four variables assessed at T2 (i.e., attachment avoidance, attachment anxiety, national narcissism, and national identification) was regressed on attachment avoidance, attachment anxiety, national narcissism, and national identification measured at T1. Similar to Study 1, we accounted for the variance shared by national narcissism and national identification at T2. 8 To compare the specific paths, we performed Wald tests. Robustness checks were the last part of our analyses—we checked whether our findings changed substantively after (a) accounting for covariates and (b) employing a robust estimation method. Missing data (3.64%) were handled with the FIML estimator. The model was estimated with the use of Mplus 8.0.
Preliminary analyses
Descriptive statistics and intercorrelations for the variables assessed in Study 2 are presented in Table 2. In comparison with men, women were slightly older, M = 44.80, SD = 16.34 versus M = 48.23, SD = 16.05, t(802) = 2.99, p = .003, d = 0.21, and better educated, M = 12.63, SD = 2.68 versus M = 13.17, SD = 3.54, t(719.90) = 2.37, p = .018, d = 0.17. Gender did not differentiate the participants’ national narcissism, national identification, attachment anxiety, attachment avoidance, or any of the remaining covariates (ps ≥ .314). Attachment avoidance and anxiety, national narcissism, and national identification were stable over time, rs ≥ .53. Across both measurements, attachment anxiety correlated positively with national narcissism and attachment avoidance was related to national narcissism negatively. Moreover, attachment avoidance and anxiety were associated negatively with national identification at T2.
Descriptive Statistics and Correlations (Study 2).
Note. Ns from 655 to 808.
p < .05. **p < .01. ***p < .001.
Hypotheses testing
Figure 2 shows the results for the autoregressive cross-lagged path model tested in Study 2. Attachment anxiety, B = 0.04, SE = 0.02, 95% CI = [0.01, 0.07], β = .06, p = .012, and avoidance, B = 0.05, SE = 0.02, 95% CI = [0.01, 0.09], β = .06, p = .019 measured at T1 were positively related to national narcissism (when controlled for national identification) assessed at T2. Moreover, national narcissism assessed at T1 was positively related to itself at T2, B = 0.67, SE = 0.03, 95% CI = [0.62, 0.72], β = .71, p < .001. National identification at T1 was negatively related to national narcissism at T2, B = −0.17, SE = 0.04, 95% CI = [−0.24, −0.10], β = −.16, p < .001.

Results of autoregressive cross-lagged panel models testing the relationships between attachment anxiety, attachment avoidance, national narcissism (when controlled for national identification), and national identification (when controlled for national narcissism), Study 2.
Moreover, both attachment anxiety, B = −0.05, SE = 0.01, 95% CI = [−0.08, −0.02], β = −.09, p < .001, and avoidance, B = −0.08, SE = 0.02, 95% CI = [−0.11, −0.04], β = −.10, p < .001 assessed at T1 were negatively related to non-narcissistic national identification measured at T2. Furthermore, national identification at T1 was positively related to itself at T2, B = 0.60, SE = 0.03, 95% CI = [0.55, 0.66], β = .64, p < .001. National narcissism assessed at T1 was negatively related to national identification at T2, B = −0.20, SE = 0.03, 95% CI = [−0.26, −0.14], β = −.23, p < .001.
Interestingly, national narcissism assessed at T1 was positively related to attachment anxiety measured at T2, B = 0.28, SE = 0.05, 95% CI = [0.18, 0.37], β = .19, p < .001, and this effect was significantly stronger than the reverse effect of T1 attachment anxiety on T2 national narcissism, χ2(1) = 22.10, p < .001. Moreover, there was a negative overtime effect of the T1 national identification on T2 attachment anxiety, B = −0.27, SE = 0.06, 95% CI = [−0.38, −0.16], β = −.17, p < .001, which again was significantly stronger than its reverse counterpart, χ2(1) = 13.83, p < .001. The effect of T1 attachment anxiety on itself at T2 was positive and significant, B = 0.45, SE = 0.03, 95% CI = [0.39, 0.50], β = .48, p < .001. The effect of T1 attachment avoidance did not reach significance, B = 0.07, SE = 0.04, 95% CI = [−0.01, 0.15], β = .06, p = .067.
National identification assessed at T1 was negatively related to attachment avoidance measured at T2, B = −0.13, SE = 0.04, 95% CI = [−0.20, −0.05], β = −.11, p = .001, and this effect did not differ from the reverse negative effect of T1 attachment avoidance on T2 national identification, χ2(1) = 1.33, p = .250. By contrast, national narcissism assessed at T1 was not associated with T2 attachment avoidance, B = 0.03, SE = 0.03, 95% CI = [−0.04, 0.09], β = .02, p = .423. However, this effect did not differ significantly from the positive effect of T1 attachment avoidance on T2 national narcissism, χ2(1) = 0.31, p = .577. Moreover, T1 attachment avoidance was positively related to itself at T2, B = 0.56, SE = 0.03, 95% CI = [0.51, 0.61], β = .62, p < .001. Attachment anxiety measured at T1 was unrelated to attachment avoidance at T2, B = 0.03, SE = .02, 95% CI = [−0.004, 0.07], β = .05, p = .083.
Adding covariates into the model slightly changed the results. Specifically, while attachment anxiety measured at T1 still was positively associated with national narcissism measured at T2, attachment avoidance assessed at T1 was no longer associated with T2 national narcissism. At the same time, the negative effects of T1 attachment anxiety and T1 attachment avoidance on national identification assessed at T2 remained significant. Using MLR estimation did not alter the results in a meaningful way.
Discussion
Study 2 provided further support for our hypotheses and overcame some of the limitations of Study 1. Attachment anxiety exerted a positive overtime effect on national narcissism (when controlled for national identification), thus confirming H1. Moreover, we also found an unexpected positive effect of attachment avoidance on national narcissism (when controlled for national identification). At the same time, the longitudinal effects of attachment anxiety and avoidance on national identification (when controlled for national narcissism) were negative, which again confirmed H2 and H3. Thus, except for the significant positive effect of attachment avoidance on national narcissism, the results from a different population were in high congruence with the young adult population reported in Study 1.
Moreover, within the current study, we found evidence in favor of some reverse effects. Specifically, national identification (when controlled for national narcissism) had negative overtime effects on attachment anxiety and avoidance. Moreover, national narcissism served as a positive longitudinal predictor of attachment anxiety. As such, the relationships between attachment anxiety and the two types of national in-group commitment, as well as attachment avoidance and national identification, turned out to be reciprocal.
Study 3
Results of Studies 1 and 2 provided evidence that attachment anxiety is associated with national narcissism (when controlled for national identification). We also found that both types of attachment (i.e., anxiety and avoidance) were negatively related to national identification (when controlled for national narcissism). However, our knowledge of how this association might explain other, socially undesirable phenomena was still limited. Previous research found both anxious (but not avoidant; Green & Douglas, 2018) attachment and national narcissism (but not national identification; Marchlewska et al., 2020) to predict negative intra- and intergroup outcomes (e.g., conspiracy beliefs, see Green & Douglas, 2018). These studies, however, did not analyze the link between attachment and the two types of in-group commitment in question. Thus, the purpose of Study 3 was to assess whether the two types of attachment facilitated conspiracy theory beliefs and radical collective action by affecting different types of in-group commitment. Specifically, we hypothesized that the positive effect of attachment anxiety (but not avoidance) on conspiracy beliefs would be mediated by increased national narcissism (when controlled for national identification) but not national identification (when controlled for national narcissism). Understanding the relationship between attachment and these two types of socially undesirable phenomena could enable future interventions to be more effective. As studies show, endorsement of conspiracy beliefs might pose a threat to public safety as it, for example, can discourage people from vaccinating themselves against COVID-19 (Marchlewska, Hamer, et al., 2022). Radical collective actions are also dangerous for public safety, as they may be directly related to violence against out-groups (e.g., attacks on LGBTQ demonstrators in Poland).
Method
Participants and procedure
Data for Study 3 was obtained in an online survey of Poles aged between 18 and 26 (M = 23.30, SD = 2.12). 9 The sample consisted of N = 558 individuals (272 females, 286 males), which constituted 53.24% of the sample employed in the first measurement. 10 Importantly, the sample gathered in the first wave of the survey was representative of young adults in the Polish society in terms of gender, age, and size of the place of residence. 11 As in Study 1, data collection was carried out by an external, local research company specializing in market research. Quotas used by the chosen company were based on the last National Census of Population and Housing, carried out by the Central Statistical Office (Główny Urząd Statystyczny; GUS). In exchange for participation in the survey, participants were given small material rewards. Study 3 was conducted during the COVID-19 pandemic.
Measures
Measures employed in Study 3 were part of a larger questionnaire that included scales of various personality and social-psychological constructs. 12 Unless otherwise noted, all measures utilized a 5-point response scale (1 = strongly disagree, 5 = strongly agree).
Attachment
To assess attachment, we used the same measure as in Study 1 (RAAS; Adamczyk & Pilarska, 2012; Collins, 1996). Both attachment anxiety (α = .88) and avoidance (α = .76) subscales showed satisfactory reliability.
National identification
National identification was measured as in Study 1 (Cameron, 2004; α = .85).
National narcissism
National narcissism was assessed as in Studies 1 and 2 (Golec de Zavala et al., 2013). The measure was internally consistent, α = .89.
COVID-19 conspiracy beliefs
Fourteen items were used to assess COVID-19 conspiracy beliefs (based on Kowalski et al., 2020). Sample items read: “The coronavirus does not really exist—it was invented to distract attention from behind-the-scenes political games” and “The coronavirus was made up to restrict people’s freedom and control them” (1 = strongly disagree, 7 = strongly agree). 13 The scale showed high reliability, α = .97. 14
Non-normative collective action
Four items were used to tap into non-normative collective action. Participants were presented with different forms of non-normative engagement (i.e., blocking the streets, destroying property, attending an illegal demonstration, and attending an illegal gathering) and asked to declare whether they had done it, whether they might do it in the future or would never engage in it. Responses were recorded on a 4-point scale (1 = would never do, 2 = might do it, 3 = have done it once or twice, 4 = do it on a regular basis). The scale demonstrated good reliability, α = .95.
Covariates
Again, the covariates involved political conservatism, gender (0 = female, 1 = male), age, education, and size of the place of residence.
Results
Analytic strategy
Similar to Studies 1 and 2, our analyses were divided into two parts. First, we inspected gender effects and intercorrelations between the variables of interest. Next, two path models were estimated to test our hypotheses. In both models, national narcissism and national identification were regressed on attachment anxiety and avoidance. At the same time, COVID-19 conspiracy beliefs and non-normative collective action were regressed on the two aspects of attachment and different types of national in-group commitment. Robustness checks involved (a) accounting for the covariates and (b) using MLR estimation. Bootstrapping with 10,000 resamples was used to obtain 95% CIs for the indirect effects. Using the FIML as an estimation method accounted for the missing data (0.10%). Models were estimated with the use of Mplus 8.0. Given the analyzed models were saturated (i.e., without degrees of freedom), we do not report any fit indices.
Preliminary analyses
Table 3 presents the descriptive statistics and intercorrelations for the variables assessed in Study 3. In comparison with women, men exhibited higher national narcissism, M = 2.78, SD = 0.88 versus M = 2.94, SD = 0.92, t(556) = −2.03, p = .043, d = 0.17; declared higher non-normative engagement, M = 1.57, SD = 0.81 versus M = 1.82, SD = 0.92, t(547.31) = −3.38, p < .001, d = 0.29; and more right-wing political orientation, M = 3.82, SD = 1.37 versus M = 4.07, SD = 1.49, t(556) = −2.36, p = .019, d = 0.20. Gender did not differentiate the remaining variables, all ps ≥ .056.
Descriptive Statistics and Correlations (Study 3).
Note. Ns ranging from 552 to 558.
p < .10. *p < .05. **p < .01. ***p < .001.
Attachment anxiety correlated positively with national narcissism, national identification, COVID-19 conspiracy beliefs, and non-normative collective action. The association between attachment avoidance and national identification was negative. In line with past research (e.g., Górska & Bilewicz, 2021; Górska et al., 2020; Górska, Marchlewska, et al.,2022; Marchlewska, Cichocka, et al., 2019), national narcissism was positively linked to national identification, COVID-19 conspiracy beliefs, and non-normative collective action. National identification correlated positively with COVID-19 conspiracy beliefs but remained unrelated to non-normative engagement. Non-normative collective action was positively associated with COVID-19 conspiracy beliefs, which confirmed recent results (Imhoff et al., 2021).
Hypotheses testing
Figure 3 presents the results of the tested path model. Attachment anxiety was positively related to national narcissism (when controlled for national identification), B = 0.20, SE = 0.04, 95% CI = [0.12, 0.28], β = .20, p < .001, but attachment avoidance was not, B = −0.08, SE = 0.07, 95% CI = [−0.21, 0.04], β = −.05, p = .201. In turn, attachment avoidance was negatively related to national identification, B = −0.13, SE = 0.07, 95% CI = [−0.26, −0.01], β = −.08, p = .038, but attachment anxiety was not, B = 0.00, SE = 0.04, 95% CI = [−0.08, 0.08], β = .00, p = .991. In line with our expectations, national narcissism was positively associated to COVID-19 conspiracy beliefs, B = 1.03, SE = 0.08, 95% CI = [0.88, 1.18], β = .61, p < .001, and non-normative collective action, B = 0.29, SE = 0.05, 95% CI = [0.19, 0.39], β = .30, p < .001. By contrast, national identification showed opposite relationships with these two DVs (B = −0.35, SE = 0.08, 95% CI = [−0.50, −0.20], β = −.21, p < .001 and B = −0.20, SE = 0.05, 95% CI = [−0.30, −0.10], β = −.20, p < .001, respectively). The direct effects of attachment avoidance on COVID-19 conspiracy beliefs (B = 0.06, SE = 0.12, 95% CI = [−0.16, 0.29], β = .02, p = .592) and non-normative collective action (B = −0.06, SE = 0.08, 95% CI = [−0.21, 0.09], β = −.04, p = .449) were both nonsignificant. At the same time, while the positive effect of attachment anxiety on non-normative collective action approached significance (B = 0.09, SE = 0.05, 95% CI = [−0.003, 0.19], β = .09, p = .058), attachment anxiety was unrelated to COVID-19 conspiracy beliefs, B = 0.06, SE = 0.07, 95% CI = [−0.08, 0.21], β = .04, p = .389.

Results of path models examining the indirect effects of attachment anxiety and attachment avoidance on COVID-19 conspiracy beliefs and non-normative collective action via national narcissism (when controlled for national identification) and national identification (when controlled for national narcissism), Study 3.
There was a positive indirect effect of attachment anxiety on COVID-19 conspiracy beliefs through increased national narcissism, IE = 0.21, SE = 0.05, 95% CI = [0.10, 0.33], Z = 4.71, p < .001. Likewise, attachment anxiety was positively associated with non-normative collective action through increased national collective narcissism, IE = 0.06, SE = 0.02, 95% CI = [0.03, 0.10], Z = 3.80, p < .001. However, national narcissism did not mediate the effect of attachment avoidance on COVID-19 conspiracy beliefs (IE = −0.09, SE = 0.07, 95% CI = [−0.27, 0.09], Z = −1.27, p = .203) or non-normative collective action (IE = −0.02, SE = 0.02, 95% CI = [−0.08, 0.03], Z = −1.25, p = .211).
By contrast, attachment avoidance did not exert a positive indirect effect neither on COVID-19 conspiracy beliefs nor non-normative collective action through decreased national identification (IE = 0.05, SE = 0.03, 95% CI = [−0.01, 0.12], Z = 1.89, p = .058, and IE = 0.03, SE = 0.02, 95% CI = [−0.01, 0.07], Z = 1.85, p = .065, respectively). Likewise, the indirect effects of attachment anxiety via decreased national identification did not reach significance (IE = 0.00, SE = 0.02, 95% CI = [−0.04, 0.04], Z = −0.01, p = .991 for COVID-19 conspiracy beliefs and IE = 0.00, SE = 0.01, 95% CI = [−0.02, 0.02], Z = −0.01, p = .991 for non-normative collective action).
Adding covariates into the model or using the robust estimation method did not alter our conclusions.
Discussion
The goal of Study 3 was to scrutinize the mediating effects of national narcissism on the relation between attachment anxiety and: (a) conspiracy beliefs, and (b) non-normative engagement. More specifically, we expected that among the two types of national in-group commitment, only national narcissism (when controlled for national identification) should catalyze these socially undesirable phenomena. Results confirmed these expectations. National narcissism mediated the relationship between attachment anxiety and COVID-19 conspiracy beliefs and between attachment anxiety and non-normative collective action. National identification (when controlled for national narcissism) was negatively related to COVID-19 conspiracy beliefs and non-normative collective action. However, the indirect effect of attachment avoidance via national identification on our dependent variables did not reach significance. This time, we did not find a significant relationship between attachment anxiety and national identification either.
Study 4
In Study 4, we aimed to replicate the pattern of results obtained in previous studies in a different socio-political context (i.e., among British participants). As in Study 3, we focused on the relationship between attachment styles, different types of in-group commitment, and COVID-19 conspiracy beliefs. This time, however, we also included an additional variable, strictly related to negative intragroup processes (i.e., willingness to conspire; Douglas & Sutton, 2011). Previous research found that people usually endorsed conspiracy theories when they thought they would be willing, personally, to participate in the alleged conspiracies (Douglas & Sutton, 2011). Thus, in Study 4, we hypothesized that the positive effect of attachment anxiety (but not avoidance) on COVID-19 conspiracy beliefs would be mediated by increased national narcissism and, in turn, increased willingness to conspire in the COVID-19 context. In such a way, we aimed to examine whether defensive in-group commitment may be related to another form of an intragroup hostility (i.e., willingness to engage in secret harmful plots against one’s own in-group members; see also Biddlestone et al., 2022; Molenda et al., 2022). One limitation of Studies 1–3 was that we measured the crucial variable (i.e., national narcissism) with the use of a short (five-item) scale with no reverse-coded items. Therefore, in Study 4, we also examined whether the pattern of results obtained in previous studies would conceptually replicate if we used a better measurement tool (a full, 9-item version of the Collective Narcissism Scale; Golec de Zavala et al., 2009).
Method
Participants and procedure
Data for Study 4 was obtained in an online survey of British participants aged between 18 and 83 (M = 40.74, SD = 12.86) via Prolific Academic (Peer et al., 2017). The sample consisted of N = 649 individuals (321 females, 326 males, 1 nonbinary, 1 other).
Measures
Unless otherwise noted, all measures utilized a 5-point response scale (1 = strongly disagree, 5 = strongly agree).
Attachment
We assessed attachment with the same measure as in Studies 1 and 3 (RAAS; Collins, 1996). Both attachment anxiety (α = .89) and attachment avoidance (α = .90) subscales were reliable.
National identification
National identification was measured in the same way as in Study 1, that is with the full 12-item (e.g., “I often think about being British”) version of Cameron’s (2004) scale, α = .89.
National narcissism
National narcissism was assessed with the full 9-item (e.g., “The true worth of British people is often misunderstood”) version of the Collective Narcissism Scale (Golec de Zavala et al., 2009), α = .89.
Willingness to conspire
Willingness to conspire was measured with a 5-item scale based on Douglas and Sutton (2011; e.g., “If I were in the position of governments, I would manipulate the information about the coronavirus to increase my influence”), α = .93.
COVID-19 conspiracy beliefs
To assess COVID-19 conspiracy beliefs, we used the same 16 items as in Study 3, α = .93.
Covariates
Again, the covariates involved political conservatism, gender (0 = female, 1 = male), age, education, and size of the place of residence.
Results
Analytic strategy
Following Studies 1–3, we started by examining gender differences and intercorrelations. In the next step, two path models were estimated. While national narcissism and national identification were regressed on attachment anxiety and avoidance, willingness to conspire was regressed on the two types of national in-group commitment and the two facets of attachment. COVID-19 conspiracy beliefs—the outcome variable—was regressed on all other variables. To test the robustness of our conclusions, we (a) accounted for the covariates and (b) used MLR estimation. The 95% CIs for the indirect effects were obtained with bootstrapping (10,000 resamples). As far as the psychological variables were concerned, there were no missing values in the present dataset. Models were estimated with the use of Mplus 8.0. Given the analyzed models were saturated (i.e., without degrees of freedom), we do not report any fit indices.
Preliminary analyses
Table 4 displays the descriptive statistics and intercorrelations for the variables measured in Study 4. In comparison with women, men exhibited lower attachment anxiety, M = 2.94, SD = 0.98 versus M = 2.77, SD = 0.90, t(645) = 2.24, p = .025, d = 0.18; showed higher collective narcissism, M = 2.07, SD = 0.75 versus M = 2.23, SD = 0.78, t(645) = −2.58, p = .010, d = 0.20; declared higher willingness to conspire, M = 1.33, SD = 0.77 versus M = 1.53, SD = 1.10, t(583.82) = −2.75, p = .006, d = 0.22; were less educated, M = 3.38, SD = 0.79 versus M = 3.24, SD = 0.95, t(627.72) = 2.14, p = .033, d = 0.17; as well as older, M = 38.76, SD = 11.78 versus M = 42.63, SD = 13.58, t(634.87) = −3.87, p < .001, d = 0.30, and more conservative, M = 4.78, SD = 1.99 versus M = 5.37, SD = 2.06, t(645) = −3.68, p < .001, d = 0.29. None of the remaining variables was differentiated by gender ps ≥.093.
Descriptive Statistics and Correlations (Study 4).
Note. N = 649.
p < .10. *p < .05. **p < .01. ***p < .001.
Attachment anxiety correlated positively with attachment avoidance and COVID-19 conspiracy beliefs, and negatively with national identification. Attachment avoidance showed a negative association with national identification and was related positively to COVID-19 conspiracy beliefs. National narcissism was associated positively with national identification, COVID-19 conspiracy beliefs, and willingness to conspire. There was a positive correlation between COVID-19 conspiracy beliefs and willingness to conspire.
Hypotheses testing
Figure 4 displays the results for the path model tested in Study 4. Attachment anxiety was positively related to national narcissism (when controlled for national identification), B = 0.11, SE = 0.04, 95% CI = [0.05, 0.18], β = .14, p = .001, but avoidance was not, B = 0.02, SE = 0.04, 95% CI = [−0.07, 0.10], β = .01, p = .722. At the same time, both attachment avoidance (B = −0.14, SE = 0.04, 95% CI = [−0.21, −0.07], β = −.15, p < .001) and attachment anxiety (B = −0.08, SE = 0.03, 95% CI = [−0.14, −0.02], β = −.11, p = .006) were negatively associated to national identification (when controlled for national narcissism). National narcissism was positively related to willingness to conspire (B = 0.27, SE = 0.06, 95% CI = [0.16, 0.39], β = .22, p < .001), but national identification (B = −0.10, SE = 0.06, 95% CI = [−0.23, 0.02], β = −.08, p = .101), attachment anxiety (B = 0.02, SE = 0.05, 95% CI = [−0.07, 0.12], β = .02, p = .632), and attachment avoidance (B = 0.04, SE = 0.06, 95% CI = [−0.08, 0.16], β = .03, p = .524) were not. In turn, willingness to conspire was positively related to COVID-19 conspiracy beliefs (B = 0.28, SE = 0.03, 95% CI = [0.22, 0.35], β = .32, p < .001). National narcissism was also positively related to COVID-19 conspiracy beliefs (B = 0.28, SE = 0.05, 95% CI = [0.18, 0.37], β = .25, p < .001), replicating the results of Study 3. By contrast, national identification was negatively related to COVID-19 conspiracy beliefs, B = −0.17, SE = 0.05, 95% CI = [−0.28, −0.07], β = −.14, p = .001. Neither attachment anxiety (B = 0.04, SE = 0.04, 95% CI = [−0.04, 0.12], β = .05, p = .281) nor attachment avoidance (B = 0.04, SE = 0.05, 95% CI = [−0.06, 0.14], β = .04, p = .395) were related to the outcome variable.

Results of path models examining the indirect effects of attachment anxiety and attachment avoidance on COVID-19 conspiracy beliefs via national narcissism (when controlled for national identification) and national identification (when controlled for national narcissism), and willingness to conspire, Study 4.
In accordance with our expectations, there was a positive indirect effect of attachment anxiety on COVID-19 conspiracy beliefs via increased national narcissism and willingness to conspire, IE = 0.01, SE = 0.003, 95% CI = [0.003, 0.02], Z = 2.65, p = .008. At the same time, attachment anxiety exerted a positive effect on COVID-19 conspiracy beliefs by increasing national narcissism, IE = 0.03, SE = 0.01, 95% CI = [0.01, 0.06], Z = 2.93, p = .003. By contrast, the indirect effect of attachment avoidance on COVID-19 conspiracy beliefs thorough national narcissism and willingness to conspire (IE = 0.001, SE = 0.003, 95% CI = [−0.01, 0.01], Z = 0.35, p = .723), as well as national narcissism, were nonsignificant, IE = 0.004, SE = 0.01, 95% CI = [−0.02, 0.03], Z = 0.35, p = .723.
National identification mediated the effects of attachment anxiety (IE = 0.01, SE = 0.01, 95% CI = [0.004, 0.03], Z = 2.10, p = .035) and attachment avoidance (IE = 0.02, SE = 0.01, 95% CI = [0.01, 0.05], Z = 2.49, p = .013) on COVID-19 conspiracy beliefs. Neither attachment anxiety (IE = 0.002, SE = 0.002, 95% CI = [−0.001, 0.01], Z = 1.39, p = .165) nor attachment avoidance (IE = 0.004, SE = 0.003, 95% CI = [−0.001, 0.01], Z = 1.49, p = .138) exerted a serial indirect effect on the outcome variable by affecting national identification and willingness to conspire.
Using the robust estimation method did not alter our conclusions in a meaningful way. However, when the covariates were accounted for, the positive effect of attachment anxiety on COVID-19 conspiracy beliefs lost significance.
Discussion
The goal of Study 4 was to check whether the pattern of results obtained in previous studies would replicate in a different socio-political context (i.e., among British participants). In line with our assumptions, we found that anxious attachment was positively linked to increased national narcissism, which was further linked to increased willingness to conspire, which then predicted COVID-19 conspiracy beliefs. In Study 4, we also found that both attachment anxiety and avoidance were negatively related to national identification, which was then negatively linked to COVID-19 conspiracy beliefs. Overall, the results of Study 4 showed that it seems at least plausible that individual insecurity may translate into collective defensiveness that further goes hand in hand with negative intragroup (i.e., willingness to conspire) and intergroup (i.e., conspiracy beliefs) phenomena.
General Discussion
Across a series of four studies, we have investigated the links between attachment anxiety and avoidance, and the two types of national in-group commitment: secure national commitment (i.e., national identification when controlled for national narcissism) and defensive national commitment (i.e., national narcissism when controlled for national identification). We found consistent evidence that national narcissism is positively related to attachment anxiety, while national identification is negatively related to avoidance, but also, in a more limited extent, to attachment anxiety (see Table 5 for the summary of key results). These relationships were especially pronounced once we accounted for the variance shared between national narcissism and national identification.
Summary of the Key Results.
Theoretical Implications
The present results provided systematic evidence that attachment anxiety is related to national narcissism. This goes in line with previous theorizing (Tajfel & Turner, 1986), suggesting that social identity may serve as a compensation for the frustration of different psychological needs. In this case, individual-level anxiety translates into collective defensiveness in the form of national narcissism, which is full of entitlement and concern about the external recognition of the in-group in the eyes of others (Golec de Zavala et al., 2013). Previous research showed that this type of defensive in-group commitment results in maladaptive intra group (e.g., in-group disloyalty; Marchlewska et al., 2020) and intergroup outcomes (e.g., out-group hostility; Marchlewska, Górska, et al., 2022). Our studies extend this work by showing positive relationships between: attachment anxiety, defensive national commitment, and, in turn, COVID-19 conspiracy beliefs, willingness to conspire, as well as non-normative collective action. These phenomena have negative impacts on society. Conspiracy beliefs, of which adherence to COVID-19 conspiracies is but one example, stem from willingness to conspire (Douglas & Sutton, 2011) and may have vast consequences both for the in-group, as well as the out-groups (Douglas, 2021a). For example, in the context of the pandemic, past research showed that higher endorsement of COVID-19 conspiracies was positively related to the refusal to vaccinate or wear masks to prevent the spread of the disease (Cislak et al., 2021; Marchlewska, Hamer, et al., 2022), which may potentially have a negative effect on the health of other in-group members. It was also related to blaming out-groups for the pandemic’s outbreak (Douglas, 2021b), which may lead to the reinforcement of prejudice. On the other hand, non-normative collective actions deriving from defensive national commitment are probably directly aimed at out-groups and, therefore, endanger their safety in the public sphere.
Moreover, we also found that lower attachment anxiety and avoidance were related to secure national identification. This may suggest that the feelings of comfort with closeness and confidence in depending on others (Bartholomew & Horowitz, 1991) are related to an increased commitment to the national in-group in a more constructive way. This commitment is based on an unpretentious investment in the in-group, which is independent of the recognition of the group in the eyes of others (Golec de Zavala et al., 2013). This form of national commitment does not lead to maladaptive intra- or intergroup outcomes (Cichocka & Cislak, 2020).
At the same time, the results of the longitudinal Study 2 showed that the relationships between attachment anxiety and avoidance, and both types of national commitment, were reciprocal. First, we not only found that high attachment anxiety at Time 1 predicted greater national narcissism at Time 2, but also that high national narcissism at Time 1 predicted greater attachment anxiety at Time 2. These results suggest that attachment anxiety strengthens defensive national commitment a few months later and that defensive national commitment also strengthens attachment anxiety. Second, we also found that both attachment anxiety and avoidance at Time 1 predicted lower secure national identification and that secure national identification at Time 1 predicted lower attachment anxiety and avoidance at Time 2.
Taken together, these findings point to a conclusion that psychological compensation in the form of a defensive identity only seems to make things worse. Those who score high (vs. low) on attachment anxiety are highly sensitive toward threats, obsessively search for attention, support, and care from others (Cassidy & Kobak, 1988). This seems to evoke a kind of a “vicious circle,” in which individuals with attachment anxiety commit to a particular group in a narcissistic way, which strengthens their insecurity and defensiveness in the long-term.
Furthermore, our research extends past findings on the link between attachment styles and group-related processes. Our results align with previous studies, which showed a positive relationship between attachment anxiety and nationalism (Marsh & Brown, 2011), which has a lot in common with national narcissism (Cichocka & Cislak, 2020). However, our findings provide a more nuanced understanding of the role of attachment styles in shaping different attitudes toward one’s own national in-group by simultaneously including insecure and secure forms of national in-group commitment and demonstrating opposite effects of attachment anxiety on these types of in-group commitment. Moreover, while previous studies showed that secure attachment was linked to desirable group-related outcomes (positive attitudes toward integration, out-group acceptance; Mikulincer & Shaver, 2001; Van Oudenhoven & Hofstra, 2006), we identified the other side of the coin. More specifically, we demonstrated that anxious attachment might have adverse social consequences (e.g., non-normative collective action, conspiracy beliefs), and this relationship is mediated by national narcissism.
Therefore, our research is the first to show that the compensatory mechanisms related to high anxious attachment may only strengthen certain psychological problems. This, however, is not the case among people low in attachment avoidance and anxiety (secure attachment), who commit to their national group in a secure manner and, furthermore, seem to become even more secure individually. Indeed, this finding echoes Mikulincer and Shaver’s (2020) “broaden and build” theory of secure attachment, which has not been demonstrated at the national in-group level until now.
Limitations and Future Directions
In this research, we aimed to assess the relationships between attachment and defensive versus secure national in-group commitment. We did so in a series of four independent studies. However, research presented here is not without limitations.
First, within the current research, we used only one measure of attachment. The reported results, therefore, might be somewhat specific for this measure. Future research might consider replicating the results reported in the current article using other validated measures of attachment (e.g., Experiences in Close Relationships, Fraley et al., 2000).
Second, future research would also do well to develop and validate a questionnaire that allows researchers to investigate and measure secure in-group commitment per se (i.e., without the necessity to control for national narcissism in the analysis). In this research, we followed an established tradition of capturing security by controlling for defensiveness (e.g., Cichocka et al., 2018; Locke, 2009; Marchlewska & Cichocka, 2017). This approach, though easy to implement, does not allow for an unbiased interpretation of the obtained results. In fact, there are reasons to believe that secure national commitment can be more than just national identification without the narcissistic component. Thus, even though similar procedures were implemented in many different contexts (e.g., self-esteem; Marchlewska, Castellanos, et al., 2019; Paulhus et al., 2004 or perfectionism; Stoeber et al., 2014), future studies should focus on developing a tool that captures secure national identification more directly.
Third, an important limitation of the current work is that the observed results were solely based on self-reports. While this is a typical approach studying the correlates of national in-group commitment (e.g., Cichocka et al., 2018), future research using a multi-methodological approach including observational data, informant-reports, and other sources of data are desired to assess the degree of robustness of our findings.
Fourth, national narcissism is related to other forms of national commitment that have been linked to maladaptive psychological states and traits, such as nationalism (understood as national dominance; Kosterman & Feshbach, 1989), “blind” patriotism (understood as an uncritical attachment to the nation; Schatz & Staub, 1997), or glorification (understood as national superiority and respect for national symbols; Roccas et al., 2006). According to Cichocka and Cislak (2020), national narcissism can be seen as an underlying construct that could lead to both aggrandizing (captured by glorification) and dominating (captured by nationalism) strategies in the struggle for group recognition (see also Gronfeldt et al., 2021). Investigating the role of adult attachment styles in forming all these different forms of national commitment would be an interesting avenue for future research.
Finally, the designs of our studies were correlational, thereby limiting causal inferences. Future research would do well to manipulate attachment styles in experimental studies (e.g., in the form of trainings or psychoeducation) or use three-wave longitudinal research design to better verify causality (e.g., Górska, Marchlewska, et al., 2022). Future studies might also consider assessments during longer periods of time, as attachment is subject to change (Fraley, 2019), or alternatively, using the Experience Sampling Methodology, focus on intensive assessments during short periods of time.
Conclusion
We believe that the role of attachment for national in-group commitment is an important issue that may help explain some intra- and intergroup processes. We provide empirical evidence that types of adult attachment are related differently to secure and to defensive national commitment. Specifically, our results shed new light on that defensive national commitment is primarily associated with heightened attachment anxiety, whereas secure national commitment is related negatively to both attachment anxiety and avoidance. We found these relationships to be reciprocal. Summing up, the current research emphasizes that the role of attachment could be an important factor explaining not only intraindividual, but also group-level processes.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by Polish Ministry of Science and Higher Education (DIALOG Grant No. 0013/2019; financing period: 2019–2021), by the National Science Center (grant number 2017/26/M/HS6/00689) and by the Norface Democratic Governance in a Turbulent Age Research Programme (the project Threat, identity, and dissent: Understanding and addressing political polarization in European democracies is financially supported by the NORFACE Joint Research Programme on Democratic Governance in a Turbulent Age and co-funded by National Science Centre, Poland and the European Commission through Horizon 2020 under grant agreement No 822166).
Data Availability
Supplemental material
Supplemental material is available online with this article.
