Abstract
Although situations themselves are objective in nature, each individual could experience same situation differently. Within the current manuscript, in an ecological momentary study during which N = 502 participants provided responses seven times per day for seven consecutive days (k = 19 108 observations), we investigate the mutual relationship between momentary narcissism and situation perception in naturalistic setting. Results of the tested dynamic structural equation models revealed a considerable amount of bi-directional interplay in which momentary increases in narcissism were related to subsequent increases in how participants perceived situations. Furthermore, using the dynamic latent class analysis we explored how these situations are perceived during specific narcissism modes. We distinguished four types of narcissism modes: (1) non-narcissistic, (2) agentic, (3) neurotic, (4) antagonistic, and demonstrate how they differ in terms of situation perception. All narcissism modes were characterized of seeing situation as more deceptive when compared to non-narcissistic mode. We also found specific patterns of perceiving situations for each mode. These results, on the one hand support claims on the default nature of agentic, and on the other, neurotic narcissism and highlight the reactive nature of antagonistic narcissism.
Plain Language Summary
People do not always feel equally confident, insecure, or defensive. These moment-to-moment shifts are especially important for understanding narcissism, which is usually seen as a stable personality trait but actually changes across situations in everyday life. In our study, over 500 people reported how they felt and how they saw their current situation seven times a day for a week. This allowed us to track narcissistic feelings in real time. We found that narcissism takes three momentary “modes”: Agentic, feeling confident, powerful, and positive; neurotic mode, feeling ignored, insecure, and negative, and antagonistic mode, feeling aggressive, defensive, and distrustful. Most of the time, people were in a non-narcissistic state. But when narcissistic modes appeared, they were closely tied to how people interpreted their situations. For example, in agentic mode, people saw situations as more positive while in neurotic mode as more negative. The antagonistic mode stood out: people in this state consistently viewed situations as more deceptive and hostile. Overall, our study shows that narcissism is not just a trait people “have” but a dynamic process shaped by how they interpret what is happening around them. Understanding these fluctuations may help explain why people with narcissistic tendencies sometimes react confidently, sometimes defensively, and sometimes aggressively.
Narcissism, which can be broadly understood as an entitled sense of self-importance (Krizan & Herlache, 2018). Most of the existing research conceptualize and study narcissism as a stable personality trait that people from the general population possess to a varying extent (Wetzel et al., 2016). However, narcissism is not only conceptualized as such stable personality trait but also as a more situational dependent state (Fleeson & Jayawickreme, 2015; Kandler & Rauthmann, 2021). Existing research indeed indicate that narcissism states change throughout time from a sense of superiority to being helpless (Back et al., 2013; Edershile & Wright, 2021; Rogoza et al., 2025). These changes are hypothesized to be initiated in response to specific situational stimuli. Assessing the effects of situation on thoughts, feelings, and behavior is a challenging task as even if situations may be considered objective (in the sense that every condition can be controlled), two participants may perceive the very same situation differently, which highlights the importance of subjective perception of situation characteristic (Leikas et al., 2012; Mischel & Shoda, 1995). Thus, the goal of the current work is to assess narcissism states in the context of the experienced perception of situations characteristics.
The complex structure of narcissism
Narcissism is a multidimensional construct full of apparent paradoxes, which have hindered attempts to grasp its complexity into a single model (Back et al., 2013; Morf & Rhodewalt, 2001). Narcissism can manifest in two distinct phenotypes: grandiose and vulnerable (Krizan & Herlache, 2018). The grandiose form is characterized by excessively positive self-perceptions and a belief in being superior to others and more deserving. In contrast, the vulnerable form involves hypersensitivity, paranoid vulnerability, self-referential thinking, and broader psychological maladjustment. These two phenotypes are usually unrelated in empirical research and exhibit distinct nomological networks (Jauk et al., 2022; Rogoza, Cieciuch et al., 2022; Wright, 2016). Despite such differences, these phenotypical manifestations are both related to interpersonal hostility and beliefs of deserving more than the others (Maciantowicz & Zajenkowski, 2020). These are considered as an antagonistic core, connecting grandiose and vulnerable narcissism, which led to the development of the three-factor model of narcissism (Krizan & Herlache, 2018; Miller et al., 2021), which is illustrated in Figure 1. The three-factor model of narcissism.
The model distinguishes three facets of narcissism: agentic (marked by self-enhancement and self-promotion, typical for grandiose narcissism), neurotic (characterized by hypersensitivity and social withdrawal; distinctive for vulnerable narcissism), and antagonistic (corresponding to hostile protection of self and entitlement; common to both forms; Back et al., 2013; Rogoza, Cieciuch et al., 2022). According to this model, antagonistic narcissism is the most central feature in the structural and functional organization of narcissistic personality (Rogoza, Crowe, et al., 2022; Rogoza, Krammer et al., 2024). Each facet correlates with distinct basic personality traits, have their own nomological network, and theoretical characteristics (Rogoza et al., 2019). Importantly, in processual models of narcissism, agentic and neurotic facets are viewed as default modes associated with grandiose and vulnerable narcissism, respectively. Such mode is interpreted as a phase during which the most basic underlying goals of maintaining exaggerated self-view (in grandiose narcissism) or protecting the fragile self from being harmed (in vulnerable narcissism) are successfully realized (Back et al., 2013; Rogoza, Cieciuch et al., 2022). That is, if one receives admiration from others, which reinforces grandiose self-view or if one avoids feelings of shame through isolation, these goals are realized and there’s no need of changing them. However, these modes change over time as these goals—despite phenotypical differences—they are both unrealistic in nature, as it is impossible to maintain a stable, overly positive evaluation of self (Geukes et al., 2017) nor avoid any potential threats due the increased tendency towards self-referencing (Holtzman et al., 2025). This inevitable shift is hypothesized to initiate the antagonistic mode, which goal is to deal with the existing inter- or intrapersonal threats. As such, the antagonistic facet is considered a reactive mode that is triggered under specific situations (Back, 2018; Grapsas et al., 2020; Rogoza, Cieciuch et al., 2022). Such changes between narcissism modes, which are operationalized as successive changes between narcissism states are labeled as fluctuations, which posits that individuals with high levels of narcissism may display not only agentic states, but also antagonistic and neurotic states over time (Edershile & Wright, 2021). Of note, recent advancements in assessing state narcissism emphasize that the three-factor structure found in trait-narcissism studies is perfectly mirrored at the state level (Bauditz et al., 2025; Rogoza et al., 2025; Rogoza, Krammer et al., 2024). Thus, when studying the process of change in narcissistic states, it is essential to take into account all of their facets. This perspective emphasizes that narcissism should not be studied solely as a trait, but to understand its underlying processes, it is necessary to conceptualize narcissism as a state, which is more responsive to specific situations (Fleeson & Jayawickreme, 2015; Kandler & Rauthmann, 2021).
Narcissism in situations
The way people perceive their momentary environment, that is, the situation, can be described in terms of situational cues (e.g., the presence of the partner), situational characteristics (e.g., the opportunity to receive support from the partner), and situational classes (e.g., relationship-related situations; Rauthmann et al., 2015). In regard to narcissism, Maaß et al. (2018) argue that situational interactions depends on both situation-invariant variables (e.g., trait narcissism) as well as situation-varying variables (e.g., state narcissism), which are sensitive to behavioral cues (e.g., visual attention; Grapsas et al., 2024), momentary feelings (e.g., affectivity; Scharbert et al., 2024), and cognition (e.g., situation perception; Bauditz et al., 2025). As a result, narcissism states are considered highly responsive to situations and change over time (Edershile & Wright, 2021). Back (2018) proposed a theoretical model, which explained why narcissism may fluctuate (e.g., from agentic to antagonistic), highlighting the role of the situation perception. For instance, when a situation is perceived as ego-threatening, one may cease agentic self-promotion and shift toward more defensive, antagonistic behavior. In a similar vein, Grapsas et al. (2020) argue that changes in narcissistic behavior, from the default agentic self-promoting behaviors to antagonistic reactive hostile derogation of others, can occur in response to hindrance of status pursuit. Indeed, experimental studies have provided support that objectively manipulated perception of status (e.g., attack and neglect of status) is differently associated with narcissism facets and their fluctuation over time (Edershile et al., 2024).
It is established that giving more attention to cues regarding status in social situations is important in explaining narcissism (Mahadevan & Jordon, 2022; Zeigler-Hill et al., 2019). However, in everyday life, such situations are not experienced objectively, but are instead filtered through the individual’s subjective lens. For example, neutral feedback might be perceived as constructive and helpful or might be interpreted as a personal attack. Therefore, it is not the objective nature of the situation, but its subjective interpretation, that is, situation perception that determines how narcissistic states are expressed in the given moment. Thus, situation perception may serve not only as a contextual background but also as a central regulatory mechanism influencing the expression, variability, and responsiveness of narcissistic states over time.
The taxonomy of situation characteristics—The situational DIAMONDS
The way people perceive their momentary environment, that is, the situation, can be described in terms of situational cues (e.g., the presence of the partner), situational characteristics (e.g., the opportunity to receive support from the partner), and situational classes (e.g., relationship-related situations; Rauthmann et al., 2015). Available situation taxonomies focus mostly on its subjective characteristics given that they offer the most information about the psychological meaning of the situations (e.g., Oreg et al., 2020; Parrigon et al., 2017; Rauthmann et al., 2014). Among the frameworks that systematically capture psychologically meaningful situation characteristics shaping individual perceptions are the empirically derived situational DIAMONDS or lexically derived CAPTION framework (Rauthmann & Sherman, 2020). The DIAMONDS taxonomy provides a structured understanding of how individuals perceive the psychological characteristics of situations across eight core dimensions, which are relevant for understanding the dynamic expression of narcissism. The dimensions are defined as follows: Duty (evaluates whether there is a perceived obligation in the situation; asks if something needs to be done), Intellect (evaluates whether deep cognitive processing is required; asks if the situation demands intellectual engagement), Adversity (evaluates whether the individual faces adversity in the form of negative external forces; asks if external threats or challenges are present), Mating (evaluates whether there is the potential for social and sexual attraction; asks if there is an opportunity to attract potential mates), Positivity (evaluates whether the situation is perceived as pleasant; asks if about overall mood or emotional quality of the situation), Negativity (evaluates whether the situation has the potential to evoke negative emotions; asks if the situation can lead to feelings of sadness, frustration, or other negative responses), Deception (evaluates the level of trust in the situation; asks if others are deceptive or trustworthy), and Sociality (evaluates whether social interaction is possible or expected; asks if individuals will engage in social behaviors or isolate themselves).
These eight dimensions provide a nuanced framework for capturing how individuals subjectively interpret situational meaning, which plays a central role in shaping their psychological and behavioral responses. For example, research shows that dimensions of DIAMONDS predicted compliance with restrictions during the COVID-19 pandemic (Zajenkowski et al., 2020). Moreover, they are also specifically relevant for narcissism-related research as they, mediated relationships between narcissism facets and relationship satisfaction (Rentzsch et al., 2021). Moreover, all eight situational perceptions were associated with individual differences in personality (Jonason & Sherman, 2020), showing specific patterns of relationships between stable traits and how people view the world around them. Furthermore, the DIAMONDS framework may offer a systematic approach for identifying situational triggers underlying narcissistic variability. By capturing subjectively perceived features such as adversity, mating opportunities, or deception, it allows researchers to explore the possibility that distinct situational dimensions elicit agentic, antagonistic, or neurotic narcissistic states in everyday life.
The perception of situations in narcissism
Does narcissism represent a specific cognitive bias leading to distortions in perceiving situations as favorable to the self? Answering this question appears more complex and dependent on the conceptualization of narcissism. For instance, agentic narcissism leads to self-promotion regardless of the situation strength (Maaß & Ziegler, 2017) and is associated with a self-perpetuating state of feeling overly positive (Scharbert et al., 2024). In contrast, neurotic narcissism is related to intense and painful feelings of shame, helplessness, and overwhelming negative emotions (Back, 2018; Di Sarno et al., 2020). As such, agentic and neurotic narcissism may reflect perceiving the world in pink vs. gray-colored glasses, as shown in the recent study by (Rogoza et al. in press), in which agentic and neurotic narcissism were related to increased inertia in positive and negative affect, respectively (but see Scharbert et al., 2024 for non-significant associations). These affective biases may, in turn, influence how individuals interpret the situations around them. Existing studies suggest that agentic and neurotic narcissism are indeed related to more positively (e.g., intellect, mating, positivity, sociality) and negatively (negativity, deception) valenced situational perceptions, respectively (Bauditz et al., 2025; Jonason & Sherman, 2020; Rentzsch et al., 2021). While agentic and neurotic narcissism are sometimes considered as the “default” modes of grandiose and vulnerable narcissism, antagonistic narcissism is considered as a reactive mode used to diminish (e.g., status) threat or to deal with the unaccepted feelings (Back, 2018; Edershile et al., 2024; Rogoza, Cieciuch et al., 2022). It is not surprising then, that antagonistic narcissism is associated with increased propensity to perceive situations as adverse (Bauditz et al., 2025).
While Rauthmann et al. (2016) provided evidence that there are limited bi-directional associations between personality traits and situation perceptions, Bauditz et al. (2025) found moderate effects in both directions. Specifically, perceiving a situation as higher than usual in regard to intellect and positivity leads to increases in agentic narcissism. Perceiving the situation as more dutiful, intellectual, and negative was associated with increases in antagonistic narcissism, while greater negativity predicted increases in neurotic narcissism. In contrast, increases from a person’s average level of agentic narcissism were related to an increase perception of mating, deception, and sociality. However, neither antagonistic nor neurotic narcissism were related to increased propensity to perceive situation in subsequent measurement occasions (Bauditz et al., 2025). While pioneering in nature, this study had several limitations. First, it was a longitudinal study carried out in an experimental setting—while this allows to objectify parameters of the study, it also constrains a large portion of situations which may happen in everyday life. Second, the sentence-based measures used to assess narcissistic states likely limited within-person variability in data, undermining the ecological validity of momentary assessments (Rogoza et al., 2025). Third, the relatively low number of measurements (18) might have been insufficient to capture the complex interplay between situation perception and narcissism (Schultzberg & Muthén, 2017). We attempt to address all these limitations within the current work.
Current study
The current study is the first to assess all three facets of momentary narcissism in their relations to perception of situations in naturalistic settings. While experimental studies offer a unique opportunity to manipulate the situation (e.g., Bauditz et al., 2025), people tend to behave differently in naturalistic settings, thus, especially when assessing perception of situations, it is important to study these perceptions outside of the laboratory (Mehl et al., 2006). Expanding these prior findings, we focus on addressing two research questions: (1) what are the bi-directional relationships between momentary narcissism and situation perception in everyday life and (2) how do people perceive situations when they are in a specific narcissism mode? According to the literature, which emphasizes the role of affect and stresses the positive vs. negative perception of self in agentic and neurotic narcissism (Di Sarno et al., 2020; Scharbert et al., 2024; (Rogoza et al. in press), we expect that this would also be reflected in cognition, that is, how one perceives situations. Specifically, we expect that agentic narcissism should be related to perceiving situations as positive, intellectual, and as an opportunity for mating, while neurotic narcissism should be related to negativity and deception (Bauditz et al., 2025; Rentzsch et al., 2021). In regard to antagonistic narcissism, we expect it will be related to adversity, which reflects hostile situational self-protection (Jonason & Sherman, 2020; Rauthmann et al., 2016).
Although the bi-directional relationship between personality states and situation perception is modest at most (Rauthmann et al., 2016), narcissism states are expected to change in response to specific situations, which may amplify the bi-directional relations with situation perception (Back, 2018; Edershile & Wright, 2021; Rogoza et al., 2025). The only study that assessed such bi-directional associations suggests that it might be limited; however, the experimental nature of this study might hinder the experience of some situations (Bauditz et al., 2025). Using ecological momentary assessment and a high-frequency sampling strategy, the present study captures within-person variability in narcissism states and perceived situational meaning with greater ecological validity and temporal precision. Thus, we adopt a more exploratory approach in evaluating how narcissism states and momentary situation perception change together over time. Finally, in an exploratory fashion, we also assess whether it is possible to empirically distinguish different narcissism modes in everyday life and if so, how they are related to psychological perceptions of situations. While we expect these to reflect the dynamics of narcissistic fluctuation (Back, 2018; Rogoza et al. in press), we do not make any a priori assumptions about either the frequency of narcissistic vs. non-narcissistic modes. However, given the theoretical characteristics of different narcissism modes, mirroring our expectations regarding the between-person relations between narcissism and perceptions of situational characteristics, we expected that these modes will meaningfully differ in contemporaneous perceptions of situational characteristics. That is, the agentic narcissism mode is associated with perceiving oneself as better than the average in overly positive light (Balcerowska et al., 2023; Zajenkowski, Czarna, et al., 2020), thus—during this mode individuals should perceive situations as more positive and intellectual and mating. The neurotic narcissism mode, associated with negative affect, paranoidal thoughts, and social withdrawal (Di Sarno et al., 2020; Rogoza, Cieciuch et al., 2022) should be associated with perceiving situations as more negative and deceptive. Finally, the reactive antagonistic narcissism mode—which is used to cope with the emerging threats (Back, 2018; Grapsas et al., 2020), should be characterized by perceiving situations as highly adverse.
Method
Participants and procedure
The sample consisted of 502 participants aged between 18 and 67 (M = 27.58, SD = 8.72, 75.1% females, 23.9% males, 1.0% other gender), recruited among university students and individuals from the general population (invited to participate by students for course credit). Higher education was reported by 53.0% of the sample, secondary education by 43.2%, and primary education by 3.8%. Initially, participants were invited to complete a set of baseline measures, after which they were invited to participate in a follow-up intensive longitudinal study. Participants were informed that they could withdraw at any time without providing a reason. The study has been programmed in the SEMA3® app and was conducted on participants’ smartphones. For seven consecutive days, participants received seven push notifications per day with questions regarding how they felt in their present situation. The surveys were conducted according to a fixed probing schedule with extended time intervals. Specifically, prompts were sent at fixed time (9:00 and 10:45 a.m., and 12:30, 2:15, 4:00, 5:45, 7:30 p.m.) and were available for 45 min to be completed. Participants received a push notification at the scheduled time and reminder 15 min before the survey expired. Those who completed at least 70% of the surveys received online vouchers worth approximately €12.50. Additionally, participants who completed at least 80% were drawn for one of the extra vouchers valued at around €50 each.
Measures
Narcissism
To assess narcissism, we used a 12-item measure, which is a combination of brief forms of three adjective-based measures of narcissism: Narcissistic Grandiosity Scale assessing agentic narcissism (Crowe et al., 2016; Rosenthal et al., 2020: items: brilliant, glorious, powerful, prestigious), Narcissistic Antagonism Scale assessing antagonistic narcissism (Rogoza et al., 2025; items: abusive, depreciating, exploitative, nasty), and Narcissistic Vulnerability Scale assessing neurotic narcissism (Crowe et al., 2018; items: ignored, resentful, misunderstood, underappreciated). These measures align with the three-factor model of narcissism at both the within- and the between-person level and are considered as valid measures of narcissism (cf., Rogoza et al., 2025). Participants were presented with all 12 adjectives, and asked to evaluate the extent to which each of them describes them right now using a visual analogue slider bar ranging from 0 (not at all) to 100 (completely). The scales were characterized by good internal consistency across both, within- (αagentic = .80; αantagonistic = .79; αneurotic = .86) and between-person level (αagentic = .96; αantagonistic = .99; αneurotic = .97).
Psychological perception of situations
To assess how participants viewed the ongoing situation, we used the ultra-brief Situational Eight DIAMONDS measure (Rauthmann & Sherman, 2016), specifically designed to be included in ambulatory assessment designs. Although given the format of the scale (i.e., one item per domain) we were not able to assess the internal consistency, it has been found as valid as the original version (cf., Rauthmann & Sherman, 2016). The scale comprises eight items, each corresponding to the specific domains of: Duty, Intellect, Adversity, Mating, pOsitivity, Negativity, Deception, and Sociality. Similar to the narcissism measures, we also asked participants to assess the extent to which each statement describes how they feel right now in a given situation using a visual analogue slider bar ranging from 0 (not at all) to 100 (completely). Full content of the items can be found at the OSF project site.
Statistical analyses
Reciprocal relations between narcissism and situations perception—The dynamic structural equation modeling approach
The Dynamic Structural Equation Modeling (DSEM; Asparouhov et al., 2018) is a statistical technique specifically designed to handle intensive longitudinal data. It combines features of three different approaches to data modeling, such as time-series modeling (used for testing lagged associations across time), multilevel modeling (accounting for nested data structure), and structural equation modeling (which provides a general framework; Hamaker et al., 2021). DSEM is effective in dealing with missing observations, which are iteratively estimated through the Bayesian Markov Chain Monte Carlo algorithm, so that they are sampled from their conditional posterior and as result, all possible information can be used even if a substantial amount of data is missing. In DSEM, the effect is interpreted as significant when the confidence interval does not contain zero within its’ boundaries, which are designated by 2.5 and 97.5 percentiles of the posterior distribution (in case of the reported 95% credible interval; McNeish & Hamaker, 2020). To assess convergence of the estimated Monte Carlo chains, we assumed the potential scale reduction (PSR) value at < 1.05. We first run each model using 10,000 iterations and then, we assessed the stability of this convergence by examining if the PSR value remained stable after increasing the number of iterations to 20,000. Results of this validity check are included at the OSF. Regarding the evaluation of statistical power in DSEM models, only one simulation study has examined this issue. It states that, for complex models where all estimates are regressed on between-person variables, a minimum of 200 participants assessed at least 25 times each is required to achieve satisfactory statistical power (Schultzberg & Muthén, 2017).
DSEM decomposes data into within- (i.e., time-varying) and between-person models (i.e., time-invariant) in which estimates from the between-person model reflect a latent variable similar to the mean across all measurements, while estimates from the within-person model reflect temporal deviations from this mean (Hamaker et al., 2015; McNeish & Hamaker, 2020). When decomposing DSEM into specific effects, one could encounter following effects: (1) inertia (the carry-over effect), which is the autoregressive parameter, reflecting the time needed to return to individual equilibrium after experiencing a change from personal mean; (2) the spill-over effect, which is the cross-regressive parameter, reflecting the cascade effects where deviations in one state are related to subsequent changes in other states; (3) innovation variance (i.e., the random factor), reflecting the random, within-person variability not explained by the model. All of these parameters accounts for individual differences, assuming that each individual has its own unique pattern of within-person relations. Also, in DSEM it is possible to differentiate between within-person model, reflecting contemporaneous momentary relations at given time-point; and between-person model reflecting average relations across all time-points (Hamaker et al., 2018, 2021). In the current study, we were specifically interested in assessing the fluctuation effects between narcissism states and situation perceptions to test whether perceiving a situation as higher than usual in X may be related to a successive increase in narcissism (and vice versa). To avoid partialling of the shared variance between narcissism facets and situation perception, we tested each model (which comprised only one narcissism facet and only one situation perception) separately, which resulted in a total of 24 tested DSEM models.
Differentiating narcissism modes
Within the literature, usually two different approaches to data assessment are contrasted: the variable-oriented approach, which focuses on describing associations between variables, vs. the person-oriented approach, which focuses on describing intra-individual differences between people (von Eye & Bogat, 2006; Muthén & Muthén, 2000). For instance, while a variable-oriented approach assesses the relationship between narcissism and self-esteem (Mota et al., 2019), a person-oriented approach assesses whether narcissism may predict membership to a latent profile typical for people with different levels of self-esteem (Rogoza et al., 2018). Person-oriented approach has been widely used in personality research (e.g., Gorgol et al., 2024) and narcissism specifically (Wetzel et al., 2016), providing its robust utility and complementarity in explaining complex psychological phenomena (Morin et al., 2017). In the current study, instead of focusing on the differentiation of subgroups of people, we propose to differentiate subgroups based on narcissism states.
For this purpose, we used the time-intensive longitudinal latent class modeling framework—the dynamic latent class analysis (DLCA; Asparouhov et al., 2017). It integrates features of the previously described DSEM and general multilevel mixture framework (Asparouhov & Muthén, 2008) by combining standard latent class modeling techniques with the time-series analysis. The goal of this approach is to explain inter-situational differences in item response patterns using a fixed number of subgroups, which are referred to as latent classes nested within participants. That is, the DLCA seeks subgroups of narcissism modes in which responses are characterized by homogeneity within latent class (i.e., similar levels of narcissism states) and heterogeneity between these classes (i.e., each class should represent different patterns of narcissism states; Wetzel et al., 2016). While there are no clear procedures on how to evaluate the results of the DLCA, we acknowledge criteria proposed by Wetzel and colleagues (2016), which regards evaluation of: (1) Bayesian Information Criterion (BIC)—lower value suggest best fitting solution; (2) interpretability of classes—the defining characteristics of a latent class should be clearly identifiable; (3) size of classes—a class to be distinguished requires a sensible number of observations; and (4) overall certainty, which should exceed 85% of the probabilities for the most likely latent class membership. The analysis is repeated, starting with two classes and continues until all these criteria are met (or unmet). The result of the analysis would assign each situation into one of the distinguished latent classes with constant within-person variance and varying between-person means.
Furthermore, we extended the DLCA with latent regression techniques (Wetzel et al., 2016), which allows to include external variables within the best-fitting latent class model used as predictors of latent class membership. In the current study, we used situational DIAMONDS as predictors of the probability of membership in selected latent classes. In this analysis, subgroups are compared to a selected group of reference by examining odds ratios. The odd ratio higher than one means that for every one-unit increase in a situational characteristics, the log-odds of belonging to one group vs. the group of reference increase proportionally. For instance, an odd ratio of 1.024 means that for every one-unit increase in a given situational characteristics, the odds of belonging to given group (vs. group of reference) increases by approximately 2.4%. The analysis was also carried out in Mplus (v. 8.3; Muthén & Muthén, 2017) using a two-level mixture type with maximum likelihood estimator.
Results
Within- and between-person associations between momentary narcissism states and situation perception.
Note. Given almost 20, 000 observations at the within-person level, all estimates were significant, although the actual effect size was spurious and close to zero. At the between-person level, coefficients ≥.10 were significant.
Bi-directional relations between narcissism and situation perception
Spill-over effects (cross-lagged associations) between narcissism states and momentary situation perception.
Note. Associations which did not contained zero were bolded. All of the presented results are standardized. Due to convergence issues, we fixed innovation variance of adversity (neurotic, antagonistic), mating (agentic), negativity (neurotic), and deception (agentic, neurotic) to equal (marked with *). For deception and mating in antagonistic narcissism, both innovation variances were set to equal (marked with **).
Clustering narcissistic modes
Summary of fit indices of dynamic latent class analysis.
Distinguished latent profiles representing different narcissism modes.
Results of dynamic latent class regressions.
Compared to the antagonistic narcissism mode, both agentic and neurotic modes were characterized by lower scores in adversity and deception. Despite this similarity, the agentic and neurotic modes differed in terms of affect—perceiving situations as more positive and less negative increased probability of being assigned to the agentic mode, whereas perceiving situations as more negative and less positive increased probability of being assigned to the neurotic mode. The neurotic mode was also characterized by lesser mating, while agentic mode by higher intellect. Finally, when directly comparing the agentic and neurotic modes, the latter was characterized by greater negativity and adversity, while perceiving situations as positive, opportunity for mating, or more intellectual increased probability of being assigned to agentic mode. In regard to duty nor sociality, we did not find any notable differences between compared modes.
Discussion
Within the current study, we examined the bi-directional relationship between three facets of narcissism (agentic, antagonistic, and neurotic) and psychological perceptions of situations characteristics in an ambulatory naturalistic setting. Prior research, which assessed such relations, was conducted in laboratory settings (cf. Bauditz et al., 2025), which potentially may influence obtained results, especially as the experimental situation in itself may lead to specific perceptions (Mehl et al., 2006). Moreover, we extended this approach by testing how individuals in different narcissism modes perceive the situation at given moment. At the cross-sectional level, we have found support for our expectations at the within- but not on the between-person level. Specifically, agentic narcissism was related to more positively valenced situational perceptions, such as greater positivity, increased opportunity for mating, or enhanced sociality. In contrast, both antagonistic and neurotic narcissism were related to perceiving the situation as more deceptive, negative, or adverse. At the between-person level, we have found larger effect sizes suggesting similar patterns of association. However, most of the relations were significant except for perceiving situation as either negative or positive for agentic and neurotic narcissism, respectively. Such a confounding pattern of relationship at the between-person level is not surprising, given that the personal means are usually artificially interrelated (Baird et al., 2006). Thus, when assessing the bi-directional relationship between narcissism and psychological perception of situations, we used the DSEM, which not only takes this issue into consideration when estimating the results but also allows for separation of the between-person effects so the observed effects correspond to actual momentary change and not the relations between individual means (Hamaker et al., 2018).
The results of the tested DSEM models revealed a number of bi- and one-directional relationships between momentary narcissism and perception of situations. Momentary increases in agentic narcissism as well as in perceiving the situation as more positive or as an opportunity for mating—were mutually related one to another. However, increases in agentic narcissism beyond personal mean were related to successive increase in seeing the situation as more socializing, intellectual, and deceptive. In turn, seeing situation as less negative than usual was associated with subsequent increase in agentic narcissism. In regard to antagonistic narcissism, we found only one bi-directional relation—seeing situation as more deceptive was related to increases in antagonistic narcissism and vice versa. Momentary increase in antagonistic narcissism were also related to increase in propensity to perceive the situation as adverse and negative at the next time point. Finally, momentary increases in neurotic narcissism or seeing the situation as more negative or deceptive—were interrelated over time one to another. While increases in neurotic narcissism were related to seeing the next situation as more adverse, seeing the situation as less positive than usual was related to subsequent increases in neurotic narcissism.
The spectrum of narcissistic traits is set on a temperamentally based continuum delineating the valence of the narcissistic experiences across its facets (Krizan & Herlache, 2018; Rogoza et al., 2019). As such, narcissistic expression ranges from overly positive self-evaluations, typical for agentic narcissism, to overly negative self-perceptions, typical for the neurotic facet (Balcerowska et al., 2023; Zajenkowski, Czarna, et al., 2020). Existing empirical works largely support such claims, consistently reporting associations between distinct patterns of affect across the spectrum of narcissism (Blasco-Belled et al., 2023; Edershile & Wright, 2021; Fatfouta & Rogoza, 2024; Scharbert et al., 2024). Psychological perceptions of situations can also differ in more general tones as some impressions are considered as more positive (i.e., mating, positivity, and sociality), while others are related to negativity and threats (i.e., adversity and deception; Moon & Ahn, 2024; Rauthmann & Sherman, 2020). As such, we extend findings reported by Bauditz et al. (2025), who reported limited bi-directional relationships between narcissism and situation perception in a laboratory-based experimental setting. In contrast, our ambulatory assessments reveal that the interplay between momentary narcissism and situation perception is much more complex in naturalistic contexts. Yet, these perceptions are forming a theoretically meaningful pattern of relations, emphasizing overly positive perceptions for agentic narcissism, overly negative for neurotic narcissism, and focused on paranoidal threat-detection related to antagonistic narcissism (Back, 2018; Rogoza et al., 2025). To better understand these intricate relations between narcissism and situation perceptions, we changed our approach and investigated how different characteristics of situation are seen during the theoretically relevant modes of narcissism.
The result of the DLCA suggests that it is possible to differentiate four profiles of narcissistic modes. In more than two-thirds of all studied situations (which equals over 13,000 observations), participants reported low levels of all narcissism facets. This finding suggests that people from the general population do not constantly experience heightened levels of narcissism in their daily lives. Apart from these, we also observed three qualitatively different profiles of narcissism modes, which we labeled agentic, neurotic, and antagonistic. These profiles correspond well with existing theoretical and empirical claims regarding the process of fluctuations in narcissism (Back, 2018; Edershile et al., 2024; Rogoza et al., 2025). That is, the agentic profile represents the default mode of narcissism in which one craves receiving admiration to boost the ego (Back et al., 2013; Wetzel et al., 2016). The neurotic profile corresponds to the state of vulnerability during which everything is perceived against the odds and the restoration of status is impossible (Grapsas et al., 2020; Rogoza, Cieciuch, & Strus, 2022). Interestingly, antagonistic narcissism was conceptualized as a reactive strategy occurring on the same level as agentic and neurotic narcissism states (Back, 2018). However, the results of the current study somehow challenge this claim. Specifically, we did not identify a distinct antagonistic profile; instead, antagonistic narcissism appeared only in combination with elevated levels of both agentic and neurotic narcissism. These results suggest that antagonistic narcissism is mutually related to both agentic and neurotic narcissism, interconnecting these into a joint network and preceding changes in them (Rogoza, Crowe, et al., 2022; Rogoza et al., 2025). However, such an antagonistic orientation is initiated together with agentic and neurotic reactions, which structurally changes the perspective on the process of fluctuation in narcissistic states. It is possible that when the ego or status is threatened, one does not simply forget about their own grandiosity, but these feelings may be fueled by the arousing sense of insecurity, together resulting in antagonistic reactions.
We scrutinized this issue in greater detail by assessing how momentary perceptions of situations predict the increased probability of being classified as narcissistic (i.e., agentic, neurotic, or antagonistic mode) and non-narcissistic. When compared to the non-narcissistic cluster, all narcissistic modes seen situations as much more deceptive. That is, an increase of ten points (out of 100) at seeing the situation as deceptive, increased probability to be classified to narcissistic mode by 36% (agentic) or even 65% (antagonistic). The remaining results largely corroborated the general findings as neurotic narcissism mode was related to perceiving situation as more negative and adverse, agentic narcissism mode was related to higher propensity of seeing situation as more positive and as an opportunity for mating, while being in an antagonistic narcissism mode was mostly related to increased propensity to seeing such situations as adverse. Perhaps the most interesting findings regarded the comparison to the antagonistic profile. Such situations, in contrast to both agentic and neurotic modes, were perceived as most adverse and deceptive, which corresponds to the perceptions of situational threat (Rauthmann & Sherman, 2020) directly posited as a central feature underlying the process of fluctuations in narcissism (Back, 2018; Grapsas et al., 2020; Kroencke et al., 2023; Rogoza et al., 2025; Scharbert et al., 2024). The antagonistic mode differed also from the remaining profiles, however in different directions—it was more positive than neurotic, but less than agentic and more negative than agentic but less than neurotic. It was also less intellectual as compared to agentic, but more as an opportunity for mating when compared to the neurotic profile. Finally, we also observed the differences in hedonic tone between the agentic and neurotic modes, which largely differed between themselves in terms of positivity vs. negativity (cf., Czarna et al., 2018, 2024), which may potentially also reflect the differences in temperamental traits associated with these two narcissistic phenotypes (Krizan & Herlache, 2018).
Limitations and future directions
Within the current experience sampling study, we have analyzed data from almost 20,000 different situations nested in over 500 individuals in an attempt to differentiate situations characterized by the increased levels of narcissism. Although the sample was large enough to differentiate meaningful subgroups of situations which may be considered as narcissistic, there are several limitations that need to be taken into account when interpreting our results. First, the studied sample was mostly female and comprised mostly students from a single European country, which should constrain generalizability of these findings to other populations. Future studies may attempt to gather a more robust and balanced sample in terms of gender identification and age. Furthermore, all person-oriented analyses, such as latent profile analysis, are heavily data-dependent (Muthén & Muthén, 2000; von Eye & Bogat, 2006). The dynamic approach adopted in the current work is no exception to this rule, so our findings shouldn’t be treated as assessing the prevalence of narcissistic behaviors in the general population. Although it is reasonable that in the vast majority of situations (i.e., almost 70%) people report very low levels of narcissism, evidence from a single study is insufficient to conclude that the remaining 30% of situations may be considered as somehow narcissistic. Thus, future studies adopting such an approach are needed to address the prevalence of narcissistic situations in general.
In the tested DSEM models, we opted to keep random effect for each parameter (e.g., mean, inertia, spillover, innovation variance) so for every single individual, we kept these individual differences in how these parameters were estimated. Nevertheless, for some of the tested DSEM models, we did not accounted for individual differences in variance in situational characteristics. It might be that the variance of this random parameter was very low, which resulted in problems with convergence of the estimated chains. In fact, most problematic models involved adversity and deception, both of which showed very low mean levels (i.e., M < 10), reducing variability across participants. Although this is understandable as these perceptions are most likely to be uncommon, we suggest interpreting the results of these models with increased caution.
Although we adopted a longitudinal approach to data collection, we ultimately relied on self-reported perceptions of momentary feelings regarding the psychological perception of situations (Rauthmann et al., 2015). While this approach is common in intensive longitudinal studies in general (e.g., Di Sarno et al., 2020; Edershile & Wright, 2021; Rogoza et al., 2025), it is also possible to standardize situations in experimental design (Bauditz et al., 2025; Szücs et al., 2023). Although people may behave differently in laboratory and naturalistic settings (Mehl et al., 2006), these two perspectives could be intertwined to provide a more robust understanding of the ongoing processes. Although naturalistic designs offer several advantages, they are also biased by their tendency to conflate subjective perceptions with the objective nature of the situation itself, which is the advantage of the experimental approach. Also, the current design is limited by the fact that we did not include any additional external validation of the actual situations itself. This limits the possibility to generalize our findings as depending on the context, individuals may behave differently even if the situation remains the same (Kandler & Rauthmann, 2021). Thus, future studies should attempt not only to address the types of situations itself but also to include information about the context. Furthermore, both methodological designs could include biological correlates such as momentary heart rate or assessment of levels of hormones (Behnke et al., 2023; Zajenkowski et al., 2023). Thus, future intensive longitudinal studies should ideally combine experimental and naturalistic settings (cf. Edershile et al., 2024) and put more emphasis on assessing biological correlates of narcissism (cf. Jauk & Kanske, 2021).
Within the current approach, we relied on the fixed probing schedule with prolonged time intervals. That is, participants received seven prompts during specifically set hours and had 45 min to report on the current situation. Adjusting the temporal spacing to the nature of the study is an important step as this may influence the observed spillover effects (Hecht et al., 2023). Although daily diary designs proved to effectively capture dynamic processes related to narcissism such as changed in seeing one’s own intelligence or experiences shame (Di Sarno et al., 2020; Rogoza, Zajenkowski et al., 2024), the perception of situational characteristics is more momentary in nature. Thus, although the temporal spacing—given the short-term nature of the situation itself, could be potentially even more dense, the current design seems to balance between the bandwidth and fidelity. Another limitation regards the lack of validation of measures in experience sampling studies. While there is a considerable amount of psychometric evaluation of the used narcissism scales in terms of successful verification of the underlying three-factor structure, validity at the within- and between-person level, cross-level invariance, and reliability of measurement (Rogoza et al., 2025), that is not the case for the ultra-brief measure of situational diamonds (Rauthmann & Sherman, 2016). Although it has been successfully tested against the already existing longer version (cf. Rauthmann & Sherman, 2016), we were unable to assess the internal consistency or test its factorial structure. Although it might be debatable whether inclusion of a longer measure of situational diamonds would bring more benefits (in terms of psychometric evaluation) or harm (in terms of increasing the fatigue of participants), future research may attempt to assess the extent to which choice of a specific version influence results in intensive longitudinal studies.
Although the used adjective-based measures of narcissism have been extensively validated as a measure of state narcissism, including tests of factorial structure, validity of measurement, and cross-level invariance (Rogoza et al., 2025), the content of the items may represent different modalities—those concerning agentic narcissism are mostly cognitive (e.g., brilliant), those regarding antagonistic narcissism are mostly behavioral (e.g., depreciating), while those measuring neurotic narcissism are emotional in nature (e.g., ignored). Although this may be concerned as a limiting factor related to the interpretation of the results of dynamic latent class analysis, these modalities may be inherently related to the process of fluctuation in narcissism states (Back, 2018). For instance, thinking about oneself as more intelligent is strongly associated with agentic narcissism and reflects the general tendency to maintain an overly positive self-view (Zajenkowski, Czarna, et al., 2020). Aggressive, reactive, and entitled behaviors enacted in self-defense are embedded within the nature of antagonistic narcissism (Back et al., 2013), while feelings of intense shame, hopelessness, and negativity—are among the core descriptors of neurotic narcissism (Di Sarno et al., 2020; Rogoza, Cieciuch, & Strus, 2022). Although it is possible to distinguish different modalities for each facet (e.g., charming behaviors in agentic, feelings of unforgiveness in antagonistic, or self-referencing thoughts in neurotic narcissism; Back et al., 2013; Fatfouta et al., 2017; Holtzman et al., 2025), future studies should both attempt to mix these modalities to test whether the findings from the current work, especially from the dynamic latent classes would replicate as well as considering formalizing if these modalities indeed might be best reflective for each narcissism facet.
Finally, although longitudinal design goes beyond studying mere covariations and may be seen as a hint to causality, they are insufficient to unambiguously address the directionality of these links (Costantini & Perugini, 2018; Di Sarno et al., 2023). For instance, while seeing situation as deceptive is related to higher probability of being classified into any of the differentiated narcissism modes, the current design is insufficient to explain whether increasing feeling of being deceived increases (any) narcissism causally, or whether it increases narcissism, which could be assessed in future experimental studies. Longitudinal designs are still limited by the potential influence of confounding variables. Within the current study, the three-factor model of narcissism (Krizan & Herlache, 2018; Miller et al., 2021) delineated the theoretical boundaries allowing us to explore the full spectrum of narcissistic states, including the narcissistic antagonism, which is sometimes missing in similar designs (Edershile et al., 2024). Although the situational DIAMONDS provide an integrative framework for assessing situations (Rauthmann et al., 2014), within the current study, we did not assess whether there are any factors that may shape these perceptions in the first place. For instance, individuals who may be more prone to ruminations and experience self-perpetuating negative affectivity may perceive situations as more negative as well. Thus, we believe that the ongoing processes regarding situational perception are more complex and future studies may systematically and empirically uncover this complexity.
Conclusion
To date, the current study is the most comprehensive ambulatory investigation of the interplay between narcissism and psychological perceptions of situations in a naturalistic setting. We provided robust evidence that each of the facets of narcissism is characterized by qualitatively different proneness towards experiencing situations as more positive (agentic), negative (neurotic) or threatening (antagonistic). Our results also shed new light on general process of fluctuations in narcissism. That is, while antagonistic reactions were treated as independent process, our situation-based approach suggests they are underpinned by the elevated level of not only antagonistic narcissism itself but also simultaneously elevated levels of agentic and neurotic narcissism. Such situations differ from others by being focused on the experienced threats. Possibly, the effective vs. ineffective removal of such a threat may potentially lead to further switching to an agentic or neurotic narcissism mode. Our findings challenge traditional trait-based models by emphasizing the importance of incorporating dynamic, context-dependent processes into understanding of narcissistic functioning. Summarizing, by putting situations into the center of interest, we were able to describe the process of fluctuations in narcissism from a new, unexplored perspective.
Footnotes
Author contributions
RR: conceptualization, formal analysis, methodology, writing—original draft, writing-review & editing. ML: writing-review & editing. LB: writing-review & editing. JGW: writing-review & editing. ZB: methodology, validation. KW: methodology, validation.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by National Science Centre, Poland (2023/49/B/HS6/00067).
