The Cross-Sectional and Longitudinal Association of Grandiose Narcissism With the Quality of the Parent-Child Relationship

The concept of grandiose narcissism, henceforth referred to simply as narcissism, describes a dimensional personality characteristic, marked by intense feelings of grandiosity, a sense of being unique, and a strong entitlement attitude (Bosson et al., 2008; Morf & Rhodewalt, 2001). Individuals high in narcissism tend to exhibit an extreme self-focus, to overrate their own abilities and achievements, and to undervalue communal attributes like empathy and trustworthiness (Back et al., 2013; Brummelman et al., 2016). While initially perceived as a singular personality trait, recent empirical and theoretical work has shown that narcissism comprises various dimensions (e.g., Ackerman et al., 2011; Back et al., 2013; Krizan & Herlache, 2018). The Narcissistic Admiration Rivalry Concept (NARC; Back et al., 2013), for example, explicitly takes account of the agentic and antagonistic qualities of narcissism. Agentic and antagonistic narcissism are argued to include different sets of intra- and interpersonal strategies which aim to maintain the grandiose self-concept. Agentic narcissism involves agentic, self-enhancing strategies aimed at acquiring social adulation by outgoing, charming, and assertive behaviors. By contrast, antagonistic narcissism is associated with a more antagonistic orientation and includes strategies aimed at defending the self from perceived threats through aggressive and derogatory behaviors.

In the interpersonal domain, narcissism has been shown to be both a source of social potency and social conflict (Back et al., 2013). In the context of short-term relationships, where interaction partners are mainly focused on getting to know each other, particularly agentic narcissism has been found to be related to various positive outcomes, such as higher likability, attractiveness, or popularity ratings (Wurst et al., 2017). In long-term relationships, however, such as in romantic relationships or friendships, which demand for more emotional intimacy, selflessness, and warmth, the antagonistic qualities of narcissism are becoming salient. Antagonistic narcissism has been shown to be associated with lower relationship satisfaction and a higher occurrence of conflicts and transgressions (Wehner & Ziegler, 2023). Furthermore, within long-term relationships, individuals scoring high on antagonistic narcissism are often found to be perceived as emotionally detached, cold, lacking empathy, and as self-focused (Leckelt et al., 2020; e.g., Wurst et al., 2017).

The parent-child relationship, inherently long-term and demanding for a high degree of perspective taking, readiness to compromise, and willingness to subordinate own motives and needs for the sake of the interaction partner’s needs, may be particularly susceptible to the influences of narcissism. Although studies that explicitly investigated the effect of narcissism on the parent-child relationship are still lacking, cross-sectional evidence indicates that narcissism can undermine functional parenting behavior, which is a crucial aspect of the quality of the parent-child relationship (Vignando & Bizumic, 2023). In a retrospective study, adult children of parents with higher levels of narcissism recalled their parents as less caring and at the same time more excessively controlling in their childhood compared to those whose parents had lower levels of narcissism (Dentale, 2015). Additionally, children of parents with higher narcissism reported experiencing greater humiliation by their parents and more instances of parental favoritism, where parents showed preferential treatment toward siblings to the detriment of the subject. Similarly, more narcissistic fathers tend to report to be less acceptant of their children’s needs and behaviors than less narcissistic fathers (Finzi-Dottan & Cohen, 2019).

While the research reviewed above did not distinguish between the agentic and antagonistic dimensions of narcissism, there is some evidence suggesting that particularly antagonistic narcissism is related to dysfunctional parenting. For example, Hart et al. (2017) found that only antagonistic narcissism, as measured by the entitlement, exploitativeness, and exhibitionism items of the Narcissistic Personality Inventory (NPI; Raskin & Terry, 1988), was cross-sectionally and positively related to an authoritarian parenting style characterized by excessively high levels of parental control, such as the use of threats, guilt induction, love withdrawal or punishments (Baumrind, 1991). Furthermore, antagonistic narcissism, but not agentic narcissism, was negatively associated with authoritative parenting, a more adaptive parenting style where parents monitor rather than control their children, encourage autonomy, and promote social competence. In a recent cross-sectional study, parents scoring higher on antagonistic narcissism reported more dysfunctional parenting behavior, such as very punitive or permissive discipline, and less functional parenting behavior, such as higher levels of emotional warmth or autonomy support. Agentic narcissism, by contrast, was unrelated to dysfunctional parenting and even positively associated with functional parenting behavior (Rawn et al., 2023).

Literature on the personality correlates of child abuse indicate a higher potential for physical child abuse for parents scoring higher on antagonistic narcissism (Crouch et al., 2015). Relatedly, Wiehe et al. (2003) observed substantial differences in the distinct facets of narcissism when comparing physically and emotionally abusive parents with non-abusive parents. Whereas non-abusive parents scored higher on the more agentic subscales of the NPI, abusive parents had higher scores on the more antagonistic subscales.

The evidence outlined so far suggests that narcissism has the propensity to undermine the establishment of a benevolent parent-child relationship. However, the focus of most research has been more on parenting behavior rather than the nuances of the parent-child relationship itself. Furthermore, prior studies were either based on cross-sectional or retrospective data, single informant paradigms that did not explicitly include the child’s perspective, or convenient and small samples that are not necessarily representative of the broader population. Additionally, up to this date, longitudinal studies on the differential effects of the agentic and antagonistic qualities of narcissism on the parent-child relationship are lacking. Hence, it is still unknown how agentic and antagonistic narcissism are related to relative change in the quality of the parent-child relationship across time.

Using longitudinal data of a large nationwide clustered sample in Germany, this preregistered study aimed to provide a systematic investigation of the differential links between parental agentic and antagonistic narcissism and multiple indicators of the quality of parent-child relationship (i.e., intimacy of the parent-child relationship, parental admiration of the child, level of conflicts within the parent-child relationship, children’s fear of parent’s love withdrawal, parental warmth, parental negative communication, parental monitoring). Furthermore, we sought to investigate whether the cross-sectional and prospective relationships generalize across distinct informants rating the parent-child relationship (i.e., parent and child reports) and distinct time frames (i.e., one to four years). To capture the unique effects of agentic vs. antagonistic narcissism on the parent-child relationship, we aimed to statistically control for the respective other facet of narcissism. We expected parent’s antagonistic, but not necessarily agentic, narcissism to be associated with lower levels of intimacy and admiration and higher levels of parent-child conflicts and children’s fear of love withdrawal. Similarly, we expected parent’s antagonistic, but not necessarily agentic, narcissism to be related to lower levels of parental warmth and parental monitoring as well as higher levels of parental negative communication.

This study was preregistered, with all hypotheses, methods, hypothesis-testing analyses as well as the selection of the study variables specified prior to data analysis (see https://osf.io/6n8w7).

We used data from the from the Panel Analysis of Intimate Relationships and Family Dynamics (pairfam), release 12.0 (Brüderl et al., 2021), which focuses on partnership and family dynamics in Germany. Pairfam was launched in 2008/2009 with a nationwide random sample of anchor persons from three age cohorts (aged 15-17, 25-27, and 35- 37 years). Anchor persons are contacted at one-year intervals, with data being obtained using either face-to-face, telephone, or self-administered interviews. The self-reports encompass a wide array of subjects, including personality measures and anchor persons’ ratings of the parent-child relationship for children aged 6 to 15 years. Additionally, as a multi-actor-study, interviews are not only conducted with the anchor persons but also, among others and if available, with the anchor persons’ children in the age of 8 to 15 years. Further details on the study design are found in Huinink et al. (2011).

As a measure agentic and antagonistic narcissism was first introduced in 2016/2017, we included four waves of data in our analyses to which in the following we refer to T1 (measurement time: 2016/2017), T2 (measurement time: 2017/2018), T3 (measurement time: 2018/2019), and T4 (measurement time: 2019/2020). Additionally, we included only anchor persons who were parents and either provided data on the parent-child relationship or gave their agreement for the child interview.

Of the 2,192 parents who participated in the study, 1,743 (62.1% female) with a mean age of 39.8 years (SD = 5.3; range = 23.0 – 46.0) provided data on narcissism and the parent-child relationship at T1, 1,132 (65% female) with a mean age of 41.1 years (SD = 5.1; range = 25.0 – 47.0) provided data on the parent-child relationship at T2, 1,268 (65.7% female) with a mean age of 40.5 years (SD = 5.4; range = 25.0 – 48.0) provided data on the parent-child relationship at T3, and 1,203 (66.5% female) with a mean age of 41.1 years (SD = 5.5; range = 25.0 – 49.0) reported data on the parent-child relationship at T4.

At T1 parents had a total of 2,008 eligible children for whom data was available (48.5% female) with a mean age of 10.5 years (SD = 2.8; range = 5.0 – 16.0), at T2 parents had 1,896 eligible children (48.9% female) with a mean age of 10.6 years (SD = 2.8; range = 5.0 – 16.0), at T3 parents had 2,049 eligible children (48.3% female) with a mean age of 10.5 years (SD = 2.7; range = 5.0 – 16.0), and at T4 parents had 1,945 eligible children (48.4% female) with a mean age of 10.5 years (SD = 2.7; range = 5.0 – 16.0).

Narcissism was measured using a short form of the Narcissistic Admiration Rivalry Questionnaire (NARQ-S; Back et al., 2013; Leckelt et al., 2018) at T1 by parent reports. The NARQ-S is a 6-item self-report instrument that distinguishes between agentic (admiration) and antagonistic (rivalry) aspects of grandiose narcissism, with three items for each subscale (e.g., agentic narcissism: “I deserve to be seen as a great personality.”; antagonistic narcissism: “I react annoyed if another person steals the show from me.”). For reasons of consistency with other scales of pairfam, the original 6-point response format was reduced to a 5-point response format, ranging from 1 (not at all) to 5 (absolutely). Internal consistencies for the agentic and antagonistic subscales were $α$ = .80 and $α$ = .67, respectively.

All indicators of the parent-child relationship were measured at T1, T2, T3, and T4. If not otherwise stated, they were measured on a 5-point scale ranging from 1 (never) to 5 (very often/always) using parent as well as child reports. In the case parents had multiple eligible children, a separate parent questionnaire was administered for each child.

Level of Intimacy. The level of intimacy within the parent-child relationship was measured using two items of the Network of Relationships Inventory (NRI; Furman & Buhrmester, 1985) (e.g., parent report: “Your child shares with you his/her feelings and thoughts”; child report: “You share your secrets and private feelings with [name parent].”). As Cronbach’coefficent alpha is biased for two-item scales, Spearman-Brown formula has been used to estimate scale reliability (Eisinga et al., 2013). The calculated scale reliabilities¹ for the four measurement points ranged between $ρ$ = .86 and $ρ$ = .88 for parent reports and between $ρ$ = .80 and $ρ$ = .82 for child reports. Correlations between child and parent reports within each measurement wave ranged from r = .42 to r = .47.

Level of Conflicts. The level of conflicts within the parent-child relationship was measured by two items of the NRI (e.g., parent report: “You and your child disagree and quarrel”; child report: “You are annoyed or angry with each other.”). Reliability coefficients ranged between $ρ$ = .77 and $ρ$ = .79 for parent reports and between $ρ$ = .69 and $ρ$ = .72 for child reports. Correlations between child and parent reports ranged from r = .39 to r = .41.

Parental Admiration. Parent’s recognition for the things their children do was measured using two items of the NRI (e.g., parent report: “You show your child that you respect and like him/her”; child report: “[name parent] shows you that he/she likes you.”). Reliability coefficients were $ρ$ = .76 at all measurement points for parent reports and ranged between $ρ$ = .62 and $ρ$ = .63 for child reports. Correlations between child and parent reports ranged from r = .29 to r = .38.

Children’s Fear of Love Withdrawal. Children’s fear of love withdrawal was measured by three items using an adapted version of the Munich Individuation Test of Adolescence (Walper, 1997) using a 5-point scale ranging from 1 (not at all correct) to 5 (completely correct). Only children reported their fear that their own bad behavior could lead to a loss of love and affection (e.g., “When you disappoint [name parent] you are afraid that he/she will love you less.”). Internal consistencies ranged between $α$ = .79 and $α$ = .83.

Parental Warmth. Parents’ emotional warmth was measured by three items adapted from a scale by Jaursch et al. (2003). Parents and children indicated the degree of affirmative attention and care in parenting behavior (e.g., parent report: “You show your child with words and gestures that you like him/her”; child report: “[name parent] shows you that he/she likes you.”). Internal consistencies ranged between $α$ = .79 and $α$ = .83 for parent reports and between $α$ = .74 and $α$ = .76 for child reports. Correlations between child and parent reports ranged from r = .43 to r = .48.

Parental Negative Communication. Parents’ level of negative communication was measured by three items based on a scale by Schwarz et al. (1997). Parents and children indicated the degree of negative behavior of parents to their children (e.g., parent report: “You yell at your child because he/she did something wrong”; child report: “[name parent] yells at you because you did something wrong.”). Internal consistencies ranged between $α$ = .72 and $α$ = .75 for parent reports and between $α$ = .62 and $α$ = .64 for child reports. Correlations between child and parent reports ranged from r = .32 to r = .36.

Parental Monitoring. The scale monitoring measures the degree to which parents are interested in and informed about the activities and social contacts of their child. Parents and children rated parent’s monitoring using four items (e.g., parent report: “When your child makes new friends, you talk to her/him about them”; child report: “[name parent] knows with whom you spend your time when you go out.”), which were based on the Expanded German Version of the Alabama Parenting Questionnaire (Reichle & Franiek, 2005). Internal consistencies ranged between $α$ = .69 and $α$ = .71 for parent reports and between $α$ = .63 and $α$ = .84 for child reports. Correlations between child and parent reports ranged from r = .32 to r = .39.

All analyses were run using multilevel structural equation modelling, which takes account of the hierarchical data structure with children nested within parents. We specified two separate random-intercept models, one for the parent reports and one for the child reports of the outcome variables. Each model included agentic and antagonistic narcissism at T1 and the indicators of the parent-child relationship at T1, T2, T3, and T4. Agentic and antagonistic narcissism were specified as manifest level-2 predictor variables, and all indicators of the parent-child relationship were modelled as manifest level-1 variables. To investigate the differential links between agentic and antagonistic narcissism and the parent-child relationship, we specified direct pathways between both dimensions of narcissism and the outcome variables at T2, T3, and T4. Thus, the effects of agentic narcissism on the parent-child relationship were controlled for antagonistic narcissism, and vice versa. All variables measured at T1 were allowed to covary as were all residuals of the dependent variables. Additionally, all outcomes were controlled for temporal stability by specifying direct pathways between the respective outcome variables at T1 and the outcome variables at T2, T3, and T4. Figure 1 provides an illustrative example of the specified multilevel models².

Models were estimated using Mplus Version 8.00 (Muthén & Muthén, 2017), employing the robust maximum-likelihood estimator (MLR). Evaluation of model fit was based on the comparative fit index (CFI), the root-mean-square error of approximation (RMSEA), and the standardized root mean residual (SRMR), with a good fit being indicated by a CFI above .90, a RMSEA coefficient of less than .08, and a SRMR of equal or less than .11 (Schermelleh-Engel et al., 2003).

Attrition was handled using the full information maximum likelihood procedure (FIML), which is a model based statistical approach for dealing with missing data (Enders, 2010; Schafer & Graham, 2002).

All Mplus output files are provided online in the OSF repository (at https://osf.io/fh2qa).

Descriptive statistics including the manifest indicators’ means, standard deviations, ranges, intraclass correlations, and intercorrelations are provided online on the OSF repository separately for the parent reports (Table S1) and child reports (Table S2).

The parent model provided an acceptable fit with the data ($χ2$(180) = 726.761, p < .001, RMSEA = .031, CFI = .978, SRMR = .037). The correlations and standardized path coefficients as well as their respective confidence intervals for the between-level part of the model are provided in Table 1. As depicted, all indicators of the parent-child relationship were highly stable across the study period, with the standardized coefficients ranging between β = .673 and β = .833.

Level of intimacy. Antagonistic narcissism at T1 was negatively related to intimacy only at T1 (r = -.135, p = .001), but not at T2 (β = -.025, p = .587), T3 (β = -.098, p = .052), or T4 (β = -.059, p = .264). Agentic narcissism did not exhibit any significant associations with intimacy (T1: r = -.061, p = .113, T2: β = .034, p = .448, T3: β = .035, p = .470, T4: β = -.004, p = .930).

Level of conflicts. Both, antagonistic and agentic narcissism were neither cross-sectionally nor prospectively associated with the level of conflicts within the parent-child relationship (antagonistic narcissism: T1: r = .031, p = .409, T2: β = .059, p = .174, T3: β = .074, p = .088, T4: β = .030, p = .562; agentic narcissism: r = -.017, p = .636, T2: β = .014, p = .733, T3: β = .005, p = .899, T4: β = .037, p = .501).

Parental admiration. For admiration, there was a negative association with antagonistic narcissism only at T1 (r = -.161, p < .001), but not at T2 (β = -.073, p = .174), T3 (β = -.047, p = .293), or T4 (β = -.026, p = .561). Agentic narcissism was unrelated to admiration across all measurement points (T1: r = -.055, p = .097, T2: β = .049, p = 154, T3: β = -.009, p = .817, T4: β = .005, p = .908).

Parental warmth. Antagonistic narcissism at T1 was associated with lower levels of parental warmth at T1 (r = -.140, p < .001), and T2 (β = -.064, p = .043). This effect, however, decreased in magnitude across T3 (β = -.035, p = .350) and T4 (β = -.024, p = .540). Agentic narcissism showed a significant negative correlation with parental warmth only at T1 (r = -.086, p = .008), but neither at T2 (β = .058, p = .076), T3 (β = .036, p = .314), nor T4 (β = .044, p = .245).

Parental negative communication. Regarding negative communication, we observed a significant positive association with antagonistic narcissism only at T1 (r = .103, p = .001), but not at T2 (β = .060, p = .109), T3 (β = .026, p = .489), or T4 (β = .059, p = .181). Agentic narcissism showed exclusively non-significant relationships with negative communication across all time points (T1: r = .023, p = .462, T2: β = .006, p = .850, T3: β = .014, p = .696, T4: β = .004, p = .930).

Parental monitoring. For parental monitoring, antagonistic at T1 showed only a significant negative relationship at T1 (r = -.122, p < .001) and T3 (β = -.090, p = .014) but not at T2 (β = -.048, p = .134) and T4 (β = -.034, p = .390). Similarly, agentic narcissism at T1 was only significantly related to monitoring at T1 (r = -.082, p = .012) but not at T2 (β = -.025, p = .444), T3 (β = .011, p = .748), or T4 (β = -.017, p = .635).

The child model provided an acceptable fit with the data ($χ2$(252) = 975.733, p < .001, RMSEA = .031, CFI = .961, SRMR = .073). The correlations and standardized path coefficients as well as their respective confidence intervals for the between-level part of the model are provided in Table 2. As in the model for parent reports, all indicators of the parent-child relationship were highly stable across the study period, with the standardized coefficients ranging between β = .691 and β = .968.

Level of intimacy. For antagonistic narcissism, a significant negative association was observed at T1 (r = -.187, p = .001). Whereas antagonistic narcissism was unrelated to intimacy at T2 (β = .059, p = .322) and T4 (β = .013, p = .854), it was surprisingly positively associated with intimacy at T3 (β = .166, p = .010). Agentic narcissism turned out to be unrelated to intimacy both cross-sectionally and prospectively (T1: r = -.041, p = .408, T2: β = -.044, p = .433, T3: β = -.105, p = .089, T4: β = -.029, p = .676).

Parental admiration. A significant negative relationship was observed for antagonistic narcissism only at T1 (r = -.172, p = .010), but not at T2 (β = -.012, p = .871), T3 (β = .057, p = .478), or T4 (β = .040, p = .669). For agentic narcissism no significant effects were found (T1: r = -.085, p = .146, T2: β = .085, p = .187, T3: β = -.033, p = .636, T4: β = -.100, p = .237).

Level of conflict. Both antagonistic narcissism and agentic narcissism were unrelated to conflicts at any time points (antagonistic narcissism: T1: r = .094, p = .124, T2: β = -.048, p = .492, T3: β = -.050, p = .529, T4: β = -.062, p = .463; agentic narcissism: T1: r = .053, p = .094, T2: β = .074, p = .250, T3: β = .033, p = .625, T4: β = .017, p = .814).

Children’s fear of love withdrawal. Both antagonistic narcissism and agentic narcissism were unrelated to children’s fear of love withdrawal at any time points (antagonistic narcissism: T1: r = .086, p = .087, T2: β = .009, p = .862, T3: β = .003, p = .966, T4: β = .026, p = .719; agentic narcissism: T1: r = .028, p = .600, T2: β = -.024, p = .660, T3: β = .022, p = .721, T4: β = .042, p = .554).

Parental warmth. Antagonistic narcissism was significantly and negatively related to parental warmth only at T1 (r = -.199, p < .001). The effect decreased throughout T2 (β = .089, p = .074), T3 (β = .015, p = .808), and T4 (β = .014, p = .838). For agentic narcissism, no significant effects were found across all measurement points (T1: r = -.076, p = .097, T2: β = -.016, p = .741, T3: β = .056, p = .241, T4: β = -.045, p = .456).

Parental negative communication. Both, antagonistic and agentic narcissism were completely unrelated to parents’ negative communication (antagonistic narcissism: T1: r = .086, p = .064, T2: β = -.054, p = .359, T3: β = -.077, p = .215, T4: β = -.031, p = .652; agentic narcissism: T1 r = .040, p = .400, T2: β = .083, p = .157, T3: β = .050, p = 349, T4: β = .126, p = .061).

Parental monitoring. For antagonistic narcissism and monitoring a significant negative relationship was found at T1 (r = -.214, p < .001). However, no prospective links were observed (T2: β = -.028, p = .619, T3: β = .004, p = .951, T4: β = .005, p = .943). For agentic narcissism, a significant negative effect was also observed at T1 (r = -.196, p < .001) and even a positive prospective effect at T2 (β = .118, p = .036). The prospective association between agentic narcissism and monitoring turned out to be non-significant for T3 (β = .074, p = .244), and T4 (β = .071, p = .320).

The objective of this study was to provide a thorough investigation of both the cross-sectional and longitudinal relationships between parental agentic and antagonistic narcissism and the quality of the parent-child relationship. By considering distinct time frames and parent reports as well as child reports, we further aimed to assess the stability and generalizability of our findings.

Previous research suggested that narcissism has the propensity to undermine effective parenting practices and, hence, hinder the establishment of a positive parent-child relationship (Dentale, 2015; Finzi-Dottan & Cohen, 2019). In our study and in line with our hypotheses, particularly antagonistic narcissism was cross-sectionally associated with a dysfunctional parent-child relationship across multiple dimensions, including decreased intimacy and admiration in the parent-child relationship as well as reduced parental warmth, increased negative communication, and diminished monitoring. These effects were found for both parent reports and, although to a lesser degree, child reports, thereby highlighting the validity of our results.

Various mechanisms may account for the negative effect of antagonistic narcissism on the parent-child relationship. A positive parent-child relationship requires a high level of parent’s empathy, the ability to compromise, and the willingness to prioritize the child’s needs over one’s own (e.g., Bornstein, 2016). Hence, a positive parent-child relationship demands for behaviors that are not necessarily characteristic of narcissism. Antagonistic narcissism, in particular, has been shown to be related to lower levels of empathy and perspective taking (Back et al., 2013; Hart et al., 2017). Accordingly, individuals high in antagonistic narcissism may show deficits in the understanding of their children’s motives, needs, and wishes and, hence, may be less able to react sensitively and warmly. Moreover, children, in their formative years often engage in behaviors that are exploratory, boundary-testing, and, sometimes, challenging. Failing to view these actions as expressions of unmet needs or as part of normative developmental stages may lead parents high in antagonistic narcissism to interpret such behaviors more readily as personal attacks, affronts, or deliberate provocations. Individuals high in antagonistic narcissism are particularly inclined to react aggressively to ego threats (Du et al., 2022), which may prompt behaviors that are more punitive and hostile rather than nurturing or corrective in nature. Exacerbating to this issue is that particularly antagonistic narcissism has been shown to be related to lower levels of self-control (Back et al., 2013; Vazire & Funder, 2006), which may prompt inconsistent, unpredictable, and overreactive parental behavior and thereby further destabilize the parent-child relationship. Additionally, particularly parents high in antagonistic narcissism may be less capable or less willing to subordinate their own needs at the expense of their children’s needs. Consequently, they may be less interested in and less tolerant of the child’s behavior and generally show a lower involvement in child rearing.

Importantly, our findings suggest that the cross-sectional correlation between antagonistic narcissism and the quality of the parent-child relationship does not imply a prospective association. After controlling for the temporal stability of the outcome variables, contrary to our hypotheses, the analyses yielded that both antagonistic and agentic narcissism were largely unrelated to relative increases or decreases of the quality of the parent-child relationship. However, it is important to note that all indicators of the parent-child relationship were highly stable across the study period, which may have substantially limited the predictive effect of parental narcissism (Adachi & Willoughby, 2015). Therefore, the bivariate associations observed between narcissism at T1 and the indicators of the parent-child relationship at T2, T3, and T4 may be largely explained by the shared variance between narcissism and the indicators of the parent-child relationship already present at T1. Due to the high temporal stability of the outcome variables, we also estimated several multilevel latent growth curve models (MLGCM) as additional exploratory analyses. These models replicated the main findings obtained with the preregistered models using autoregressive pathways for the outcome variables in that particularly antagonistic narcissism was related to the intercept but not the slope of the parent-child relationship indicators. The Mplus output files of all MLGCMs may be retrieved online on the OSF repository. Taken together, although cross-sectionally associated, narcissism did not predict how the parent-child relationship evolved across time, indicating a more static rather than dynamic relationship between the constructs.

This study’s results must be interpreted while considering several limitations. First, all constructs were measured using very brief scales with a limited number of items. Despite the widespread use of these instruments in prior studies, brief scales often face criticism for their lower reliability, validity, and capacity to capture the nuances of personality variance compared to more comprehensive scales (Sleep et al., 2021). In fact, some of our scales included as few as two items, leading to reliability coefficients that were slightly below the recommended levels. We recommend that future research explore the relationship between grandiose narcissism and the parent-child relationship with the use of more detailed scales.

Second, to maintain consistency with other scales in the pairfam study, the original 6-point response format of the NARQ was modified to a 5-point scale. This reduction in response categories may have reduced the variability of responses and, consequently, altered the original psychometric properties of the NARQ, including its reliability and validity (Preston & Colman, 2000). Therefore, future studies should aim to replicate our findings using the original 6-point response format.

Third, both the parent and child models involved a substantial number of tested associations, which increased the risk of Type I errors. While multiple-testing corrections like the Bonferroni adjustment are often employed to address this issue, they can be overly conservative and may actually increase the risk of Type II error rate when a large number of tests are conducted (Maxwell, 2004). For this reason, we chose not to apply such corrections in the present study.

Fourth, indicators of the parent-child relationship were highly stable across the study period which limited our ability to predict their longitudinal development by variables such as agentic and antagonistic narcissism. Future studies might want to focus on outcome variables with lower stabilities or specifically focus on samples or developmental phases (such as entering puberty or leaving the parental home for higher education) during which the parent-child relationship can be expected to be less stable.

Fifth, the absence of prospective associations between mean-level narcissism and relative increases or decreases in the indicators of the parent-child relationship does not preclude that change in narcissism might be related to change in the quality of the parent-child relationship (although narcissism often tends to be rather stable across time, e.g. Jung et al., 2024). Nevertheless, we encourage future studies using repeated measurements of narcissism to investigate the correlated change of narcissism and the quality of the parent-child relationship.

Sixth, although we argued that deficits in empathy, perspective taking, and self-control as well as higher levels of impulsivity, distrust, and entitlement typical for individuals high in antagonistic narcissism may be crucial for explaining the negative effects on the quality of the parent-child relationship, we did not actually measure those potentially mediating variables. Future research should delve into the mechanisms that mediate the relationship between the facets of narcissism and the quality of the parent-child relationship.

Seventh, while our study focused on the cross-sectional and longitudinal links between parental narcissism and the parent-child relationship, it is important to note that the parent-child relationship is also a significant determinant in the development of narcissism, especially during childhood and adolescence (e.g., Brummelman et al., 2015; Cramer, 2011). In this context, it is possible that parental narcissism affects the dynamics of the relationship in such a way that it fosters narcissism in the child. Therefore, an interesting future research project could involve examining the parent-child relationship as a key link in the intergenerational transmission of narcissism.

Finally, our analyses were solely focused on grandiose narcissism, which refers to a dimensional personality construct in the general population and its agentic and antagonistic qualities. We deliberately excluded vulnerable narcissism, which may be considered as a more pathological form of narcissism (Miller et al., 2021). Although there is a conceptual overlap between vulnerable narcissism and antagonistic narcissism (Wink, 1991), and existing cross-sectional evidence hints a negative association with parenting (Crouch et al., 2015; Wiehe, 2003), investigating the longitudinal effects of vulnerable narcissism on the quality of the parent-child relationship remains an avenue for future research.

Despite these limitations, this study revealed profound insights into the complex dynamics of narcissism within the family environment, highlighting the particularly problematic effect of antagonistic narcissism on the parent-child relationship.

This paper uses data from the German Family Panel pairfam, coordinated by Josef Brüderl, Sonja Drobnič, Karsten Hank, Franz J. Neyer, and Sabine Walper. pairfam is funded as long-term project by the German Research Foundation (DFG).

The Mplus output files for all specified models are available online at https://osf.io/fh2qa. We do not have permission to share the data publicly. However, access to the data can be requested at https://www.pairfam.de/.

Janis Jung played a leading role in the conceptualization of the study, the analysis and interpretation of the data, as well as in writing the original draft. Michela Schröder-Abé made significant contributions to the study’s concept and interpretation of the data and played a leading role in supervising, revising, and editing the manuscript.

Both authors have no competing interests to declare.