Character Virtues in Romantic Relationships and Friendships During Emerging Adulthood: A Latent Profile Approach

Abstract

The present study aimed to examine the role of character virtues in the quality of romantic relationships and friendships during emerging adulthood. The sample included 505 individuals, comprising 269 single and 236 partnered emerging adults. Participants completed three self-report instruments: the Values in Action-114GR (VIA-114GR), Sternberg’s Triangular Love Scale (STLS), and the Friendship Network Satisfaction scale (FNS). Latent Profile Analysis (LPA) applying ipsatization identified three distinct virtue profiles, which were not significantly associated with participants’ relationship status. Concerning friendships, analyses of variance revealed that emerging adults characterized by a balanced virtue profile reported significantly greater perceived closeness than those classified within profiles defined by lower levels of restraint and transcendence virtues. However, no significant differences were observed in the frequency of socializing. Regarding romantic relationships, it was found that different virtue profiles support different love dynamics. Specifically, the results showed that profiles differing in endorsing the knowledge virtue were overrepresented in love profiles characterized by low passion or low commitment. These findings offer significant theoretical contributions and practical implications for strengths-based interventions designed to enhance emerging adults’ social and emotional well-being.

1. Introduction

Emerging adulthood, covering ages 18–29, is widely recognized as a critical period for developing relationships (Arnett, 2015; Camirand & Poulin, 2022). Spanning the late teens through the twenties, this developmental stage is often associated with forming deeper emotional bonds and greater complexity in relationship dynamics (Barry et al., 2017; Waworuntu et al., 2022). Higher education demands, job market instability, and financial pressures have significantly reshaped how young people perceive and engage in friendships and romantic relationships (Waters et al., 2018). Although the significance of close relationships for well-being and developmental outcomes is well-established (e.g., Kaufman et al., 2022), research remains limited on how personal characteristics such as character virtues contribute to relationship quality during this critical life stage.

At the same time, personality traits are well-established as influential in social relationships (Beeney et al., 2019; Kashdan et al., 2018). Among these, character strengths and virtues, defined as stable, morally valued traits that promote human flourishing (Peterson & Seligman, 2004), have been linked to positive interpersonal outcomes, including greater satisfaction, trust, and emotional intimacy in relationships (Boiman-Meshita & Littman-Ovadia, 2022; de Matos Fernandes et al., 2025; Wagner et al., 2021). Despite these promising findings, much of the existing research examines romantic relationships and friendships separately, often employing variable-centered approaches that may fail to capture the nuanced ways in which individuals combine multiple strengths and virtues simultaneously. In addition, empirical work on virtues in relationship quality or non-marital adult relationships remains limited, as do comparative studies across single and partnered individuals.

To address these gaps, the present study aimed to examine how distinct profiles of character virtues relate to relationship quality during emerging adulthood. Specifically, a person-centered approach was adopted using Latent Profile Analysis (LPA) to identify meaningful configurations of virtues and assess their associations with both friendship quality and components of romantic love. In addition, single and partnered individuals were compared to explore whether virtue profiles differ according to relationship status.

1.1. Virtues and Character Strengths

Virtues are universally valued moral qualities that transcend cultural boundaries (Peterson & Seligman, 2004; Pezirkianidis et al., 2020a). While the expression of virtues may vary based on cultural, historical, and social factors (McGrath, 2025), research suggests that their underlying essence is universal. Character strengths, in contrast, are the specific, positive traits manifested in thoughts, feelings, and behaviors that enable individuals to enact these virtues in daily life (Peterson & Seligman, 2004), serving as internal capacities that facilitate personal development, resilience, and well-being (Pezirkianidis et al., 2020b).

Peterson and Seligman’s (2004) Values in Action (VIA) Classification of Character Strengths and Virtues ranks among the most prominent and well-established frameworks. The VIA Institute on Character identifies 24 character strengths: fairness, kindness, teamwork, forgiveness, modesty, leadership, appreciation of beauty and excellence, love, creativity, bravery, perspective, open-mindedness, curiosity, humor, social intelligence, prudence, persistence, self-regulation, honesty, hope, spirituality, zest, gratitude, and love of learning. These strengths are categorized under six broader virtues: wisdom, courage, humanity, justice, temperance, and transcendence (Peterson & Seligman, 2004). The VIA model also proposes that virtues are universal, morally valued traits manifesting through specific character strengths. These traits are stable over time and across situations, guiding consistent patterns of thought, feeling, and behavior (Peterson & Seligman, 2004). However, differences exist between types of virtues. For instance, cognitive virtues exhibit greater individual variability due to factors such as intellectual engagement and access to learning resources, whereas interpersonal virtues are more socially shaped and stable, reflecting common cultural and developmental influences (Luria et al., 2015). Despite their established relevance to well-being, further research is needed to understand how configurations of character strengths relate to key life domains, including close relationships.

1.2. Romantic Relationships During Emerging Adulthood

Romantic relationships are intimate and deeply personal connections, characterized by distinct cognitive, affective, and behavioral aspects (for a review see Gómez-López et al., 2019). This definition aligns with Sternberg’s Triangular Theory of Love (1986), which identifies love as comprising three core components: intimacy, passion, and commitment. Specifically, passion refers to the physical attraction one feels toward another person. Intimacy involves feelings of warmth, closeness, and sharing within a relationship. Commitment reflects a deliberate choice to maintain the partnership over time, highlighting the importance of dedication and loyalty (Nabila & Gunawan, 2023).

Romantic relationships in young adulthood play a key role in personal growth and the development of interpersonal skills (Shulman & Connolly, 2013; Stephanou, 2012). Compared to previous generations, contemporary young adults show shifting patterns, expectations, and motivations in romantic involvement (Tillman et al., 2019). This stage often involves balancing emotional commitment with the need for autonomy and personal goal pursuit (Watkins et al., 2024). However, romantic relationships during young adulthood often emphasize exploration and personal growth rather than formal long-term commitments (Arnett, 2015). Young adults are more likely to experience multiple romantic relationships during this stage of development rather than just one (Boisvert & Poulin, 2016).

Romantic relationships share many similarities with friendships. In contemporary Western cultures, both are usually characterized as voluntary and committed (Camirand & Poulin, 2022). However, while friendship mainly fulfills social integration and self-esteem needs, romantic relationships offer intimacy and emotional support. According to the hierarchical-compensatory model (Cantor et al., 1987), individuals prioritize their relationships in a hierarchical order based on whom they turn to fulfill their needs. In this context, romantic relationships occupy the highest position in this hierarchy during emerging adulthood, relegating friendship to a compensatory role (Costello et al., 2023).

1.3. Friendships During Emerging Adulthood

Friendship is a complex and multidimensional relationship defined by emotional closeness, reciprocity, intimacy, and voluntary investment of time (Bukowski et al., 2009). The number of friendships tends to peak during the early years of young adulthood, gradually declining as individuals transition into intermediate adulthood (DeSousa & Cerqueira-Santos, 2012). During this period, friendships support young adults navigate significant life transitions, e.g., entering higher education, launching careers, or forming romantic relationships, and are linked to enhanced self-esteem, reduced stress, and better mental health (Chow & Ruhl, 2014). Moreover, strong friendships have been identified as a protective factor against loneliness, particularly in times of romantic or professional uncertainty (Barry et al., 2017).

Research has consistently shown that individuals in committed romantic relationships often reduce their engagement with broader friendship networks, increasingly relying on their partner as the primary source of emotional support (Seffrin et al., 2009). Empirical evidence further indicates that such individuals report fewer interactions with friends, diminished perceived closeness, and a gradual contraction of their social networks over time (Costello et al., 2023). Nevertheless, rather than abandoning friendships altogether, they tend to maintain a smaller, more selective group of close friends who offer emotional support and serve to complement the emotional functions of the romantic relationship (Hojjat & Moyer, 2017). However, it is important to distinguish between the quality of time spent with friends and the quantity of that time. Research suggests that friendships marked by emotionally rich and authentic connections contribute more significantly to well-being than merely spending much time with others (Langheit & Poulin, 2022). These dynamics underscore the importance of concurrently examining romantic and friendship relationships to better understand their interrelated nature.

1.4. Character Virtues and Strengths and Their Role in Shaping Relationships

Research shows that cultivating character virtues and strengths enhances life satisfaction, relationship quality, and psychological well-being (Habenicht & Schutte, 2023; Peterson & Seligman, 2004). Interpersonal strengths such as kindness, love, social intelligence, forgiveness, and humility foster empathy, trust, and conflict resolution, supporting stronger connections in both romantic and platonic relationships (Baumann et al., 2020; Niemiec, 2018). Gratitude and appreciation have also been linked to increased closeness and mutual recognition (Algoe et al., 2010). Despite these benefits, the contribution of character strengths to non-marital adult relationships has received limited empirical attention.

Recent studies suggest that virtues and character strengths influence romantic relationships in distinct ways. For example, self-transcendence virtues, such as benevolence and universalism, are associated with higher relationship quality by fostering empathy and concern for a partner’s well-being (Van der Wal et al., 2024). Complementing these, interpersonal strengths facilitate the development and maintenance of emotionally supportive and trusting partnerships (Weber & Ruch, 2012). In addition, cognitive virtues, including curiosity and the pursuit of understanding, guide individuals to engage in relationships through mutual exploration, perspective-taking, and open communication, enhancing relational depth rather than emotional intensity or physical attraction (Karremans & Van Lange, 2008). Such patterns may reflect a stronger focus on autonomy and self-development, particularly during emerging adulthood, when identity exploration is a central developmental task.

Regarding friendships, prior studies have shown that strengths such as kindness, love, honesty, and social intelligence predict greater friendship satisfaction and quality (e.g., Gander & Wagner, 2022). Dyadic analyses have shown that leadership and humor predict peer acceptance and friendship quality, whereas honesty and gratitude contribute more specifically to the perceived quality of close friendships (Wagner, 2019). Notably, close relationships amplify the benefits of strengths-based behaviors, underscoring the value of character strengths in fostering meaningful social connections (Lavy et al., 2014). In addition, emerging adults display meaningful differences in their psychological profiles, particularly in domains relevant to well-being and relationships. For instance, Walsh et al. (2024) conducted an LPA among single emerging adults and identified five distinct happiness profiles based on indicators such as friendship satisfaction, neuroticism, and self-esteem. They found that friendship satisfaction remained a key factor distinguishing the most positive profiles, regardless of the participants’ romantic relationship status. This suggests that internal psychological traits provoke well-being and satisfaction independently of whether an individual is single or in a romantic partnership.

2. Materials and Methods

2.1. Aims and Hypotheses of the Study

Taking the above into consideration, the present study aimed to: (a) identify distinct latent profiles of character virtues among emerging adults using LPA, (b) investigate whether the character virtue profiles are related to the participants’ relationship status, (c) explore how character virtue profiles relate to friendship quality, (d) identify distinct latent profiles of love components among emerging adults as conceptualized by Sternberg’s triangular theory of love (intimacy, passion, commitment), and (e) examine how character virtue profiles relate to love profiles.

Based on the review of the literature, the following hypotheses were proposed:

H1. 

Distinct latent profiles of character virtues will emerge among emerging adults, reflecting meaningful differences in the configuration of these psychological traits.

H2. 

Character virtue profiles will not significantly associate with participants’ relationship status (single vs. in a relationship).

H3. 

Character virtue profiles will significantly differ in terms of friendship quality, with certain profiles (e.g., those high in interpersonal strengths, such as kindness and social intelligence) reporting higher friendship quality.

H4. 

Character virtue profiles will significantly associate with love profiles. It is expected that virtue profiles encompassing interpersonal and emotional strengths (e.g., interpersonal, transcendence) will be associated with profiles reporting higher levels of intimacy and commitment, reflecting stronger emotional bonds and relational stability (Hypothesis 4a). Conversely, virtue profiles emphasizing cognitive or achievement-related strengths (e.g., intellect, love of learning) may be associated with profiles reporting lower levels of intimacy or commitment, but possibly higher levels of passion (Hypothesis 4b).

2.2. Participants

The participants were 505 Greek emerging adults aged 18 to 29 (M = 22.42, SD = 3.25). The sample consisted of 162 men (32.1%) and 342 women (67.9%). One participant (0.2%) did not wish to provide an answer about their gender. Regarding marital status, 269 (53.3%) were single and 236 (46.7%) were in a relationship. Concerning their education, 42 adults (8.3%) had just finished high school, 332 adults (65.7%) were university students, 82 adults (16.2%) held a Bachelor’s degree, 46 (9.1%) held a Master’s degree, and 3 (0.6%) held a Doctoral degree. Finally, in terms of occupational status, 254 adults (50.3%) were exclusively engaged in their studies, 168 (33.3%) combined work with their studies, 78 (15.4%) held either temporary or permanent employment, and 5 (1%) were unemployed.

2.3. Measures

2.3.1. Values in Action-114GR (VIA-114GR)

The Values in Action-120 (VIA-120) instrument, created by Peterson and Seligman (2004) and translated into Greek by Dimitriadou and Stalikas (2012), assesses positive personality traits considered to be morally valued and universally recognized across cultures. In its original form, it includes 120 items that assess six virtues (wisdom, courage, humanity, justice, transcendence, and moderation) and 24 character strengths (Curiosity, Love of learning, Judgment, Creativity, Perspective, Zest, Bravery, Persistence, Honesty, Social intelligence, Kindness, Love, Teamwork, Fairness, Leadership, Self-regulation, Prudence, Forgiveness, Humility, Appreciation of beauty and excellence, Gratitude, Hope, Spirituality, and Humor). Participants provide their responses on a 5-point Likert scale ranging from 1 (totally disagree) to 5 (totally agree).

The internal consistency, test–retest reliability, and validity of the VIA-IS scales have been well-established through numerous empirical studies conducted over the past several years (e.g., Peterson & Seligman, 2004; Ruch et al., 2010). Pezirkianidis et al. (2020a) adapted the scale into Greek and validated the structure of the 24-character strengths but proposed a model comprising 114 items (VIA-114GR) and five virtues (CFI = 0.93, RMSEA = 0.05): interpersonal (fairness, kindness, teamwork, forgiveness, modesty, leadership, appreciation of beauty and excellence, and love), intellect (creativity, bravery, perspective, open-mindedness, curiosity, humor, and social intelligence), restraint (prudence, persistence, self-regulation, and honesty), transcendence (hope, spirituality, zest, and gratitude), and knowledge (love of learning). They reported good internal consistency across 20 strengths (α = 0.70 to 0.85) and marginal reliability for the 4 remaining strengths (α = 0.60 to 0.66). The internal consistency of the 5 virtues was satisfactory (α = 0.81 to 0.96). Convergent validity was demonstrated through significant positive correlations with PERMA profiler factors (r = 0.13–0.59). Discriminant validity was supported by the results, showing negative or zero correlations of all character strengths and virtues with the variables of negative emotions and loneliness. In the present study, the reliability of 16 strengths was found satisfactory (Cronbach’s α ranged from 0.70 to 0.85). The other 8 strengths showed marginal reliability coefficient values (Cronbach’s α ranged from 0.62 to 0.69). The reliability of the virtues ranged from α = 0.79 to α = 0.91.

2.3.2. Sternberg’s Triangular Love Scale (STLS)

Sternberg’s Triangular Love Scale (Sternberg, 1986, 1997) was developed based on the triangular theory of love and was used to evaluate the nature and quality of romantic relationships. The scale has been adapted in Greek by Kordoutis and Manesi (2012). It includes 45 items that assess three love components: (a) intimacy (15 items, e.g., I share deeply personal information with my partner), (b) passion (15 items, e.g., I am physically attracted to my partner), and (c) decision/commitment (15 items, e.g., I am committed to maintaining my relationship with my partner). Participants were asked to rate their agreement with each statement on a 9-point Likert scale, ranging from 1 (not at all) to 9 (extremely). Participants who were not in a romantic relationship were asked to recall their most important past relationship and respond to all the items with that relationship in mind. However, individuals who have recently been in a relationship are particularly vulnerable to recall bias, as they may reinterpret their memories of past relationships influenced by their present emotional condition or the time elapsed since the relationship ended. This distinction holds theoretical significance, as the end of a relationship may lead to cognitive and motivational distortions when recalling events, evaluations, and feelings related to that relationship (Fletcher, 2002). Therefore, although the findings could have offered insights into the relationship between personality strengths, virtues, and the components of love, we chose not to include single adults in the analyses related to the final aim and the last hypothesis.

Sternberg (1997) reported high internal consistency for each subscale (α = 0.93–0.95) and provided strong support for the three-factor structure through confirmatory factor analysis. Sorokowski et al. (2021) examined the configural, metric, and scalar invariance across 25 countries, and all tests supported the cross-cultural validity of the love construct. The Greek adaptation by Kordoutis and Manesi (2012) confirmed the original structure (CFI = 0.95, RMSEA = 0.06) and the good reliability of its subscales (α = 0.96–0.98). Convergent validity was supported by correlations with relationship duration (r = 0.28–0.48). In the present study, the three-factor solution was confirmed, and the reliability of the subscales was found to be good (Cronbach’s α_intimacy = 0.97, Cronbach’s α_passion = 0.90, Cronbach’s α_commitment = 0.96).

2.3.3. Friendship Network Satisfaction (FNS)

Satisfaction with friendships in adulthood was assessed using the Friendship Network Satisfaction scale (Kaufman et al., 2022). As this was the first use of the scale with Greek samples, it was translated from English into Greek and subsequently back-translated by bilingual researchers. It comprises 14 items, of which 8 assess closeness (e.g., It is hard to imagine my life without my close/best friends) and 6 assess socializing (e.g., My friends and I go out and do things together). Participants provided their answers on a 5-point Likert scale ranging from 1 (not at all agree) to 6 (completely agree).

Kaufman et al. (2022) introduced the FNS and suggested a bi-factor model (OmegaH of 0.79 for general factor, OmegaH of 0.23 for Closeness, and 0.21 for Socializing). Convergent and discriminant validity were supported through correlations with related constructs, such as Miller’s Social Intimacy Scale, ranging from r = 0.51 to 0.69, and loneliness (r = −0.22), indicating the scale’s appropriate alignment with theoretically associated variables. However, as the scale is newly developed, evidence regarding its test–retest reliability is not yet available. Given that the FNS was used with a Greek sample for the first time, a confirmatory factor analysis was performed. The analysis revealed two factors, and the model provided an adequate fit to the data (χ²(64) = 3.621, p < 0.005, CFI = 0.933, IFI = 0.934, SRMR = 0.062, 90% CI [0.063, 0.089], RMSEA = 0.071). The reliability of the two factors that emerged was found α_closeness = 0.91 and α_socializing = 0.89.

2.4. Procedure

The initial step involved obtaining approval from the Research Ethics Committee at the first author’s affiliated university (Approval No. 6/30-10-2024) and ensuring compliance with the ethical standards outlined by the American Psychological Association (APA) and the European Union’s General Data Protection Regulation (GDPR) for managing sensitive personal data. In the subsequent step, a convenience sample was obtained, primarily through the dissemination of the questionnaire on social media platforms. During data collection, strict measures were implemented to safeguard the anonymity and confidentiality of the respondents. All participants were adults who provided informed consent to participate in the study and explicitly agreed to the publication of the findings. Data collection was conducted over a five-month period, from November 2024 to March 2025. A total of 527 questionnaires were collected, of which 22 were identified based on boxplot inspections and standardized z-scores exceeding ±3 and excluded to ensure normality and reduce the influence of extreme values, resulting in a final dataset of 505 responses.

2.5. Data Analyses

Preliminary analyses were conducted with IBM SPSS (Version 27) to explore the associations among character virtues, components of love, and friendship satisfaction, as well as to evaluate the suitability of parametric analyses. Given that some character strengths exhibited marginal reliability, it was decided to conduct the analyses using only the overarching virtues, which demonstrated high reliability, to ensure that the observed variables contributed meaningfully to the identification of latent profiles. The analyses included descriptive statistics, such as means, standard deviations, skewness, and kurtosis.

Concerning Hypothesis 1, a series of Fisher’s Z tests was initially performed to explore whether the associations between character virtues differed between individuals in romantic relationships and those who were single. Correlation coefficients were first computed separately for each group and then compared using Fisher’s r-to-z transformation to test for significant differences between the two independent samples. Analyses focused on each virtue in relation to the remaining strengths. This procedure allowed for the identification of potential group-specific relational patterns prior to conducting LPA. To control for Type I error across multiple comparisons, both conventional (|Z| ≥ 1.96) and adjusted thresholds (e.g., Bonferroni correction) were considered. No statistically significant differences were found, suggesting that we could proceed with LPAs in the total sample.

In the next step, ipsatization was applied prior to LPA to control for individual differences in overall response tendencies and to ensure that profiles reflected relative, not absolute, levels of virtues. Given that virtue scores were based on self-report measures, participants may have varied in their interpretation and use of the response scale, potentially introducing individual differences in response styles. By centering each participant’s subscale scores around their own mean, ipsatization highlighted intra-individual patterns, allowing for the identification of meaningful virtue configurations (Kim, 2024). This procedure enhanced the interpretability of the profiles by reducing global response bias.

Following ipsatization, we performed the LPA in the total sample. LPA was performed using the mclust package in R (R Core Team, 2022; Scrucca et al., 2023), within the RStudio environment (version 2024.12.1 Build 563). In alignment with established LPA guidelines, models with two to five profiles were explored to determine the optimal class solution. Model fit indices such as AIC, BIC, AWE, CLC, and KIC were used to identify profiles. However, these indices often produce divergent results, thereby introducing subjectivity into model selection (Akogul & Erisoglu, 2017). The Analytic Hierarchy Process (AHP; Saaty, 2013) was implemented to address this limitation. AHP synthesizes multiple fit indices into a single Composite Relative Importance Vector (C-RIV), offering a more systematic and data-driven approach to identifying the best-fitting solution. In the present study, AHP weighting was determined algorithmically rather than by expert judgment to minimize subjectivity. Specifically, the relative importance of each fit index (AIC, BIC, AWE, CLC, and KIC) was computed based on the normalized inverse of their values across the competing models, ensuring that lower index values received proportionally higher weights. These normalized scores were then used to generate a pairwise comparison matrix for AHP computation. Consistency ratios (CRs) were calculated for each AHP iteration and were all below the conventional threshold of 0.10 (mean CR = 0.04), indicating satisfactory internal consistency and stability of the weighting structure. The resulting C-RIV scores were subsequently used to identify the model demonstrating the optimal balance among fit quality, parsimony, and classification precision. The model associated with the highest C-RIV was considered the most optimal.

To examine Hypothesis 2, cross-tabulation analyses were conducted in SPSS. Chi-square tests of independence were employed to evaluate the statistical significance of associations between character virtues and relationship status, while Cramer’s V was calculated to assess the strength of these associations.

To test Hypothesis 3 concerning whether the derived profiles would significantly differ in their reported experiences of friendships, a series of Multivariate Analyses of Variance (MANOVAs) and Analyses of Variance (ANOVAs) were conducted.

Finally, to test Hypotheses 4a and 4b, two separate LPAs were conducted exclusively among partnered participants, as romantic love components (e.g., intimacy, passion, commitment) are conceptually and contextually relevant primarily within ongoing romantic relationships. The first analysis included the five virtues, and the second the three love components. For analyses involving sample sizes smaller than 300, researchers advise employing Monte Carlo simulations to assess the adequacy of statistical power (Muthén & Muthén, 2002). The Monte Carlo simulation was conducted to evaluate the optimal number of latent profiles in a sample with ipsatized indicators of virtue-related constructs. The simulation consisted of 100 iterations, each based on bootstrap resampling of the original dataset. Concerning virtue profiles, the frequency of the selected best number of profiles across simulations indicated that the three-profile solution was most frequently preferred (84% of simulations), followed by the five-profile solution (12%) and the two-profile solution (4%). Regarding love profiles, results were mixed. The frequency of the selected best number of profiles across simulations indicated that the five-profile solution was most frequently preferred (32% of simulations), followed by the three-profile solution (26%), the two-profile solution (22%), and the four-profile solution (20%). This pattern indicated some ambiguity in identifying the optimal number of latent classes. Although the 5-profile model was slightly more common in the simulations, it produced very small subgroups and notable conceptual overlap between neighboring profiles, making theoretical interpretation less coherent. In contrast, the 3-profile model provided a clearer and more stable differentiation between distinct relational orientations, offering an interpretable and parsimonious solution consistent with the broader goal of identifying meaningful, generalizable patterns rather than highly specific micro-clusters. Accordingly, parsimony, theoretical clarity, and model stability were prioritized over the marginally higher complexity of the 5-profile solution.

Following LPAs, cross-tabulation analysis was conducted in SPSS. Chi-square tests of independence were employed to evaluate the statistical significance of associations between character virtues and love profiles, and Cramer’s V was calculated to assess the strength of these associations. To evaluate the adequacy of statistical power for this test, a post hoc power analysis was conducted using G*Power 3.1 (Faul et al., 2007). Based on the observed effect size (Cramer’s V = 0.243), α = 0.05, df = 4, and total sample size of N = 236, the achieved power (1 − β) was estimated at 0.95. This indicates that the chi-square test had adequate sensitivity to detect an association of the observed magnitude. Moreover, inspection of the contingency table confirmed that all expected cell frequencies were well above the minimum threshold of 5, thereby supporting the validity of the chi-square approximation.

3. Results

3.1. Means, Standard Deviations, Skewness, and Kurtosis of Character Virtues and Strengths, Love Components, and Satisfaction from Friendships

Initially, preliminary analyses assessed the variables’ means, standard deviations, skewness, and kurtosis.

Table 1. Means, standard deviations, skewness, and kurtosis of character virtues, love components, and satisfaction from friendships.

	Single			In a Relationship
	Mean	SD	Skewness, Kurtosis	Mean	SD	Skewness, Kurtosis
Love components a
Intimacy	-	-	-	8.18	0.68	−1.13, −1.68
Commitment	-	-	-	7.67	1.12	−0.98, −0.37
Passion	-	-	-	7.80	1.04	−0.56, −0.74
Friendships b
Closeness	5.28	0.72	−1.22, 0.93	5.02	0.87	−1.07, 0.78
Socializing	4.61	1.05	−0.76, 0.31	4.14	1.12	−0.18, −0.80
Virtues c
Knowledge	3.64	0.81	−0.32, −0.62	3.72	0.81	−0.36, −0.86
Interpersonal	4.01	0.39	−0.22, −0.25	4.03	0.43	−0.31, −0.57
Intellect	3.83	0.42	−0.10, −0.46	3.85	0.45	−0.16, −0.35
Transcendence	3.71	0.64	−0.08, −0.97	3.74	0.59	−0.00, −0.69
Restraint	3.73	0.53	−0.18, −0.84	3.83	0.51	−0.50, −0.35

Note. ^a 1–9 Likert scale, ^b 1–6 Likert scale, ^c 1–5 Likert scale.

illustrates that adults who were in a relationship reported moderate to high levels across all love components. Regarding their evaluations of friendships, both single and partnered individuals rated the closeness subscale highly, while the socializing subscale received moderate to high ratings. Finally, both groups demonstrated higher levels of the interpersonal virtue and lower levels of the knowledge virtue.

3.2. Virtue Profiles

To identify distinct virtue profiles, LPA applying ipsatization on the total sample (N = 505) was performed, testing models with two to five latent profiles while utilizing the Analytical Hierarchy Process (AHP) approach as outlined by Akogul and Erisoglu (2017). The fit criteria and C-RIV for the estimated profile solutions are presented in

Table 2. Fit indices of virtue profiles in the total sample.

N Profiles	AIC	AWE	BIC	CLC	KIC	Prob Min	Prob Max	Entropy	C-RIV
2	1743	1957	1811	1713	1760	0.900	0.949	0.776	0.870
3	1589	1883	1682	1546	1612	0.901	0.933	0.834	0.809
4	1481	1856	1600	1427	1511	0.692	0.934	0.799	0.800
5	1410	1865	1553	1343	1445	0.803	0.911	0.788	0.803

. Model 3 offered the most suitable solution, as it struck an optimal balance among statistical fit, classification precision, model parsimony, and theoretical interpretability. While models with four or five profiles demonstrated marginally improved fit indices, these gains were minimal and accompanied by increased model complexity and a slight decline in entropy. In contrast, Model 3 maintained high entropy and robust classification quality and demonstrated a high C-RIV score. The elevated posterior probabilities (ranging from 0.90 to 0.93) further supported the adequacy of this solution, indicating that individuals were assigned to profiles with high certainty. Thus, three subgroups were identified, each representing a unique pattern of mean scores across the five core virtue domains: Intellect, Interpersonal, Knowledge, Restraint, and Transcendence. Thus, Hypothesis 1 was confirmed.

Figure 1 displays the 3-class personality profile identified. Class 1 (n = 176; 34.9%) was designated as “Low Restraint and Transcendence” due to its members exhibiting higher than their own average across intellect, interpersonal, and knowledge virtues but significantly lower scores in restraint and transcendence virtues. Class 2, the largest subgroup in the sample, was labeled “Balanced” (n = 269; 53.3%). This group appeared to reflect individuals with similar average scores in all virtues. Class 3 was labeled “Low Knowledge” (n = 60; 11.9%) because its members scored higher than their average across all virtues, compared to the knowledge virtue, which was rated significantly lower. Importantly, because ipsatized scores were used, these profiles reflect each participant’s relative emphasis among virtues rather than their absolute levels compared to others. Thus, the labels describe intra-individual configurations (i.e., lower relative to one’s own average virtue expression) and should not be interpreted as indicating absolute deficiencies.

3.3. The Association Between Character Virtue Profiles and Relationship Status

The second aim of the study was to examine whether character virtue profiles were associated with the participants’ relationship status. To investigate this, a cross-tabulation and Chi-square analysis were conducted to test the relationship between the three latent profiles of virtues and relationship status (single vs. in a relationship). The results of the Pearson Chi-square test indicated that the association between virtue profiles and relationship status was not statistically significant, χ²(2, 505) = 1012, p = 0.603. Additionally, the effect size, as measured by Cramer’s V, was small (V = 0.015), suggesting a minimal and non-meaningful overlap between the participants’ character virtues profiles and their relational status. Thus, Hypothesis 2 was confirmed.

3.4. The Association of Virtue Profiles with Satisfaction from Friendships

The next aim of the study was to examine whether the virtue profiles identified with the LPA analysis were associated with friendship quality. The results presented below partially confirmed Hypothesis 3. A multivariate analysis of variance (MANOVA) was conducted to examine whether virtue profile membership was associated with the two dimensions of friendship (closeness and socializing) in a significant manner. The assumption of equality of covariance matrices was met (Box’s M = 8.12, p = 0.156). Levene’s test for equality of error variances was also non-significant for both closeness (p = 0.789) and socializing (p = 0.919), indicating that the homogeneity of variance assumption was satisfied.

The MANOVA revealed statistically significant associations of virtue profiles with the combined dependent variables [Wilks’ Λ = 0.134, F(2, 502) = 1.96, p = 0.037, partial η² = 0.013], suggesting that social relationships differ by virtue profile. Follow-up univariate ANOVAs indicated a marginally significant effect of virtue profile on closeness [F(2, 502) = 3.29, p = 0.037, partial η² = 0.017], but not on socializing [F(2, 502) = 1.25, p = 0.256, partial η² = 0.005]. Post hoc comparisons using the Games–Howell test revealed that the “Balanced” profile assessed closeness higher (M = 5.24, SD = 0.79) compared to the “Low restraint and transcendence” profile (M = 5.06, SD = 0.73). No other pairwise comparisons were statistically significant. These results suggest that virtue profiles are associated with differences in perceived closeness with friends, but not necessarily with the frequency of socializing.

3.5. The Association of Virtue Profiles with Love Components

To address the study’s final aim and hypotheses, two new LPAs applying ipsatization were performed exclusively among partnered participants. Two to five latent profile models were tested using the Analytical Hierarchy Process (AHP) approach.

Concerning virtues, following the recommendations of the Monte Carlo simulation, the 3-class model was selected due to its strong statistical performance, high classification quality, and conceptual clarity. It achieved high entropy, indicating excellent class separation, and demonstrated high certainty in class assignments. Classes were well-represented, avoiding the small, unstable groups found in the 4- and 5-class models, where some classes comprised less than 5% of the total sample. The fit criteria and C-RIV for the estimated profile solutions are presented in

Table 3. Fit indices of virtue and love profiles in partnered adults.

N Profiles	AIC	AWE	BIC	CLC	KIC	Prob Min	Prob Max	Entropy	C-RIV
Virtue profiles
2	3286	3511	3352	3250	3308	0.967	0.978	0.913	0.870
3	3209	3517	3299	3238	3238	0.893	0.975	0.881	0.856
4	3111	3502	3225	3147	3147	0.847	0.982	0.890	0.809
5	3036	3511	3174	2079	3047	0.840	0.979	0.895	0.802
Love profiles
2	1111	1249	1166	1112	1144	0.885	0.976	0.839	0.957
3	999	1165	1048	973	1016	0.912	0.964	0.895	0.952
4	938	1152	1000	903	903	0.770	0.972	0.889	0.938
5	944	1206	1020	901	901	0.477	0.955	0.708	0.775

.

Regarding the love components, the Monte Carlo simulation recommendations were mixed. The 3-class model was chosen based on a balance of model fit and parsimony. It presented the lowest AIC and CLC among the models tested, indicating a good fit without overfitting. While the 4- and 5-class models showed slightly lower BIC or KIC, their entropy decreased, and minimum class probabilities suggested weaker class separation and potential overextraction. The 3-class solution maintained high entropy and strong classification probabilities, supporting statistical adequacy and interpretability, essential for meaningful latent class analysis. In addition, classes were well-represented, avoiding the small, unstable groups found in the 4- and 5-class models.

Figure 2 displays the 3-class virtue profile identified, which was similar to the one identified in the LPA performed on the total sample. Class 1 (N = 83; 34.16%) was labeled “Balanced with High Knowledge” because its members exhibited higher scores on knowledge than their average across the other virtues. Class 2 was labeled “Balanced” (N = 65; 26.75%). This group appeared to reflect individuals with relatively similar average scores in all virtues. Class 3 (N = 95; 39.09%) was labeled “Balanced with Low Knowledge” because its members scored balanced across all virtues, except for the knowledge virtue, which was rated significantly lower. Since ipsatized scores were employed, these profiles represent each participant’s relative prioritization of virtues rather than their absolute standing compared to others. Therefore, the classes capture within-person patterns and should not be taken to imply absolute shortcomings. Thus, Hypothesis 3 was confirmed.

Figure 3 displays the 3-class love profile identified. Class 1 (N = 38; 16.64%) was labeled “Low Passion” because its members exhibited higher scores on intimacy and commitment but lower scores in passion than their average. Class 2 was labeled “Stable” (N = 162; 66.67%). This group comprised individuals who demonstrated relatively similar average scores across all love components. Class 3 (N = 43; 17.70%) was labeled “Low Commitment” because its members exhibited balanced average across intimacy and passion, but the commitment component was rated significantly lower. Because ipsatized scores were used, these profiles show each person’s relative emphasis on love components rather than absolute levels.

Finally, a cross-tabulation analysis was conducted to examine the relationship between the virtue and love profiles. The chi-square test indicated a statistically significant association between these variables (χ²(4) = 28.794, p < 0.001) with a moderate strength of association (Cramer’s V = 0.243). Specifically, within the “Low Passion” group, 57.9% of cases belonged to the “Balanced with High Knowledge” profile, 39.5% to the “Balanced” profile, and 2.6% to the “Balanced with Low Knowledge” profile. In contrast, the “Stable” group was distributed as 33.3% in “Balanced with High Knowledge”, 25.9% in “Balanced”, and 40.7% in “Balanced with Low Knowledge”. Lastly, within the “Low Commitment” group, 16.3% belonged in “Balanced with High Knowledge”, 51.2% in “Balanced”, and 32.6% in “Balanced with Low Knowledge”. Overall, the association suggests meaningful differences between the virtue and love profiles, partially confirming Hypotheses 4a and 4b.

4. Discussion

This study used a person-centered approach to investigate the intricate relationships between adults’ character virtues and their assessments of romantic relationships and friendships. In this context, some significant findings were identified.

4.1. Virtue Profiles

To begin with, Fisher’s Z tests indicated no significant differences in the interrelations among character virtues between single individuals and those in romantic partnerships. This suggests the presence of a shared, underlying typology of virtue-related dispositions that operates independently of relational status. In other words, the stability of virtue profiles across groups implies that individuals tend to cluster into broad dispositional patterns of character strengths, regardless of whether they are currently in a relationship. This finding is consistent with the Values in Action (VIA) framework (Peterson & Seligman, 2004), which conceptualizes virtues as stable, cross-situational traits that shape behavior over time. From a personality theory standpoint, character strengths and virtues are considered trait-like attributes, often reflecting broader dimensions of personality such as agreeableness or conscientiousness (Noronha & Campos, 2018). Thus, individuals in both single and partnered groups may exhibit similar character strengths and virtue profiles if their experiences of connection, autonomy, and support do not differ meaningfully. This finding reinforces the conception that virtues are universal psychological structures, while the context of a relationship may influence the intensity of relational dynamics more so than the categorical distinction of relationship status.

LPAs conducted on the full sample yielded a three-profile solution, indicating distinct patterns of virtue expression across latent classes. It is essential to interpret these profiles within the context of ipsatized data. Ipsatization centers each individual’s virtue scores around their own mean, thereby removing between-person differences in overall virtue endorsement and emphasizing within-person relative patterns (Kim, 2024). Consequently, the identified profiles illustrate how participants prioritize or de-emphasize specific virtues relative to their overall tendency, not their absolute standing compared to others. The following labels should therefore be understood as reflecting intra-individual relative patterns rather than absolute deficits.

Notably, the virtue of Knowledge displayed the highest degree of variability, highlighting its central role in differentiating individual profiles. These results suggest that cognitive virtues may be particularly influential in shaping distinct dispositional profiles, whereas interpersonal virtues and strengths appear more consistently expressed across individuals. This pattern aligns with theoretical accounts asserting that cognitive virtues, such as love of learning, exhibit greater individual variability, often due to differences in intellectual engagement, epistemic curiosity, and access to educational resources (Pezirkianidis et al., 2020a; Shimai et al., 2006). Conversely, interpersonal virtues, including strengths like fairness, kindness, teamwork, forgiveness, and humility, are typically socially cultivated from early childhood. These virtues are frequently emphasized within family systems, educational settings, and broader sociocultural environments that promote prosocial behavior (Luria et al., 2015). Thus, interpersonal virtues may demonstrate reduced variability, shaped more by normative social expectations than by individual disposition or contextual factors.

In addition, the restraint virtue, including strengths such as prudence, persistence, self-regulation, and honesty, and the transcendence virtue, including hope, spirituality, zest, and gratitude, demonstrated some variability between profiles. According to Peterson and Seligman’s (2004) framework, all virtues are universally valued across cultures; however, their expression may differ depending on contextual factors. Specific character strengths, such as honesty or gratitude, may be more sensitive to influences like moral identity salience or spiritual beliefs, whereas others, such as curiosity or teamwork, tend to be more consistently reinforced across diverse societal systems and institutions. Additionally, research suggests that emerging adulthood is characterized by divergent developmental pathways, particularly in areas related to meaning, purpose, and self-discipline (Lapsley et al., 2023). Transcendence virtues, such as spirituality, hope, and a sense of purpose, are intimately connected to the existential exploration that characterizes emerging adulthood (Galanaki et al., 2023). During this developmental stage, individuals often engage with questions related to personal belief systems and life meaning. While some may establish clear spiritual or future-oriented frameworks, reflecting high levels of transcendence, others may experience ambiguity or detachment in these areas, indicating lower levels. This range of expression illustrates the diverse pathways individuals take in constructing meaning. Similarly, restraint-related virtues such as self-regulation and prudence are still undergoing cognitive and behavioral maturation during emerging adulthood, contributing to notable individual differences in these capacities.

4.2. The Association of Virtue Profiles with Relationship Status

Research in positive psychology consistently highlights the role of character strengths in fostering positive interpersonal relationships, including romantic ones (Wagner et al., 2021). From a positive psychology and personality theory perspective, character strengths are considered to be relatively stable, trait-like attributes that reflect core aspects of a person’s identity, moral functioning, and psychological well-being (Peterson & Seligman, 2004). In contrast, relational status is a more situational or contextual variable as it can change over time, may be influenced by social, cultural, or developmental factors, and may not reflect the depth or quality of one’s social connections (Chen & Liao, 2021).

The findings of the present study showed that the profiles of character virtues identified among participants via LPA did not significantly associate with the relationship status of emerging adults, suggesting that it is not the character virtues or strengths that determine one’s relational status, but rather the quality and nature of interpersonal experiences. This conclusion aligns with Attachment Theory (Bowlby, 1979), which emphasizes the quality of emotional bonds over the mere presence or absence of a relationship. Attachment security is conceptualized in terms of the quality of the emotional bond, specifically, the extent to which the relationship is experienced as safe, trusting, and emotionally supportive, rather than by the mere presence or absence of a romantic partner. This finding highlights the need to move beyond simplistic, categorical representations of relational variables and instead consider the nuanced, qualitative dimensions of relationships, which may interact more meaningfully with individual psychological traits and processes.

4.3. The Association of Virtue Profiles with Friendships

The next aim of the study was to explore the association between variations in character virtues and aspects of friendship quality. The findings revealed that virtue profile membership was significantly related to perceived closeness with friends, but not to the frequency of social interactions. Specifically, individuals classified within the “Balanced” profile reported significantly higher levels of perceived closeness than those in the “Low Restraint and Transcendence” profile.

Research on adult friendships has emphasized the positive influence of character strengths and virtues in fostering close relationships (e.g., Gander & Wagner, 2022; Pezirkianidis et al., 2020b; Wagner et al., 2021). Interpersonal and transcendence virtues, encompassing strengths such as humor, gratitude, forgiveness, and love, promote trust, empathy, emotional safety, and mutual understanding (Peterson & Seligman, 2004), which are central to feeling close to others. Individuals endorsing a variety of character strengths often possess secure attachment styles, higher emotional intelligence, and stronger communication skills, all of which contribute to enhanced relational depth (Kerns et al., 2023; Ros-Morente et al., 2018).

However, virtue profiles were not associated with socializing. From a Self-Determination Theory perspective (Ryan & Deci, 2000), the need for relatedness is possibly fulfilled not through the number of interactions, but through their authenticity and emotional resonance. Studies have highlighted developmental changes in friendship quality during emerging adulthood (Langheit & Poulin, 2022), suggesting a tendency to prioritize a smaller circle of close friends who offer emotional support and complement romantic relationships. Additionally, the frequency of interaction may depend more on external factors (e.g., work demands, social availability) and not necessarily reflect the strength of the bond (Oswald, 2017). Overall, the absence of a significant association between virtue profiles and socializing frequency suggests that virtues may shape the depth, authenticity, and emotional tone of interactions, but not necessarily their frequency.

4.4. The Association of Virtue Profiles with Love Components

The final aim of the study was to investigate whether the virtue profiles of partnered adults were associated with their love profiles. The results revealed a significant association between the love and virtue profiles, indicating that different virtue profiles support different love dynamics. Given the use of ipsatized scores, the profiles represent the individuals’ relative prioritization of love components rather than absolute levels. Accordingly, the labels denote intra-individual patterns, not absolute deficits.

In relationships marked by low passion, the overrepresentation of individuals in the “Balanced with High Knowledge” group suggests that people who especially value cognitive growth and learning, as reflected in higher-than-average endorsement of the knowledge virtue, may approach romantic relationships in a more thoughtful, curious, and reflective way. Despite endorsing interpersonal, intellectual, restraint, and transcendence virtues at similarly balanced levels, these individuals emphasize acquiring knowledge and understanding, which may foster relational depth through exploration and meaning-making rather than emotional intensity or physical attraction. As a result, their love style may feel more intellectually intimate than passionately charged. This aligns with theoretical accounts suggesting that passion often relies on emotional arousal, creativity, and physiological engagement (Carswell et al., 2019; Carswell & Impett, 2021), which may be less central in relationships where learning, reflection, or meaning take precedence.

In contrast, the “Low Commitment” group included a larger proportion of individuals in the “Balanced with Low Knowledge” profile, suggesting that love of learning was endorsed less than the person’s average across all virtues. While their overall character profile may still be well-rounded, the lower relative focus on knowledge acquisition and reflective thinking might impact how they approach long-term romantic commitments. Sustaining commitment during emerging adulthood is challenging and often requires a future-oriented mindset, deliberation, and the ability to reflect on one’s needs, goals, and values over time (Konstam et al., 2019; Monk et al., 2014). In this context, when the virtue of knowledge is somewhat de-emphasized, individuals may be more inclined to make relational decisions based on emotion, situational cues, or short-term gratification, rather than careful reflection or a desire to grow within the relationship.

The relatively equal representation of the stable love group (those with similar scores across all love components) across all virtue groups suggests that balanced romantic functioning is not strongly tied to any single virtue orientation. Instead, it may reflect a broader capacity for emotional regulation and psychological adaptability that enables individuals to integrate passion, intimacy, and commitment cohesively, independent of their dominant value system. Depending on the context, these individuals may draw on a range of virtues, such as interpersonal warmth, self-restraint, intellect, or transcendence, which can lead to more stable relationship dynamics.

5. Conclusions

The findings of the present study offer several significant theoretical contributions to the literature on character virtues, interpersonal relationships, and emerging adulthood. First, the identification of distinct virtue profiles among emerging adults supports the utility of a person-centered approach to virtue research, extending the Values in Action (VIA) framework (Peterson & Seligman, 2004) beyond traditional variable-centered analyses. By demonstrating that individuals cluster into meaningful latent profiles, this study advances the understanding of how virtues are internally organized and expressed during a key developmental stage.

Furthermore, the association between virtue profiles and different relational outcomes (e.g., friendship closeness, love components) underscores the relevance of virtues concerning different qualitative aspects of close relationships. Specifically, the finding that virtue profiles were unrelated to relationship status but associated with perceived closeness and love dynamics supports interactionist personality and relationship development models. These models emphasize how dispositional characteristics influence relational processes over time, rather than momentary status distinctions. In addition, the observed patterns among character virtues offer important insights into how endorsing specific virtues is associated with distinct relational profiles. These findings contribute to the growing body of literature suggesting that character strengths are differentially linked to relational experiences, highlighting the need for more nuanced theoretical frameworks that examine how particular configurations of virtues shape romantic and interpersonal dynamics during emerging adulthood.

The findings of this study offer several important practical applications for interventions designed to enhance the relational well-being of emerging adults. First, the identification of distinct virtue profiles provides a valuable foundation for designing personalized strengths-based interventions. For example, individuals classified within the “Low Restraint and Transcendence” profile, who reported lower closeness in friendships, may benefit from targeted programs that foster self-regulation, gratitude, or hope. Such interventions can be delivered through psychoeducational workshops, coaching, or even digital well-being applications that focus on cultivating specific character strengths.

Second, the study’s results suggest that virtues are more closely linked to the quality of relationships than to relationship status. This insight has implications for counseling and therapy with emerging adults, as it shifts the focus away from relationship acquisition toward enhancing relational depth, emotional intimacy, and long-term satisfaction. Counselors and relationship educators might incorporate character strengths assessments into their practice to help clients reflect on how their virtues shape the emotional tone and sustainability of their romantic or platonic connections.

Additionally, the observed link between virtue profiles and various love dynamics, such as low passion or commitment, could inform relationship education programs. For instance, individuals with high endorsement of knowledge virtues may benefit from strategies that promote emotional expressiveness and physical intimacy. At the same time, those low in reflective traits may need support in developing long-term commitment through deeper relational understanding and meaning-making. Relationship skills training programs within university settings could integrate these findings to offer more differentiated and psychologically informed content tailored to emerging adults’ diverse character profiles and relational patterns.

Finally, these findings have implications for curriculum and policy development in higher education. Universities and colleges can support the socio-emotional development of emerging adults by embedding character education, positive psychology, and relational skills training into student well-being curricula. Such initiatives could enhance students’ interpersonal competence and contribute to broader institutional goals of promoting mental health, academic engagement, and personal growth.

While the study offers valuable insights, it is subject to several limitations. First, as character strengths and virtues are not culturally invariant, the study’s focus on a single cultural context may constrain the generalizability of its findings to other settings. Additionally, the small sample size, especially within partnered emerging adults, limits representativeness. Future research with larger, more diverse samples is essential for replicating and validating the results of the present study. Moreover, relationship length was only assessed in a basic form, without considering potential qualitative differences across stages of relational development. This may restrict the ability to fully capture how relationship duration relates to variations in virtue profiles and components of love. Another limitation is the reliance on self-reported measures, which may be influenced by social desirability bias (Brutus et al., 2013). Employing alternative methodologies, such as behavioral assessments or third-party evaluations, could enhance the validity of future findings. Finally, the cross-sectional design precludes the establishment of causal relationships between character strengths, components of love, and friendships. To address this limitation, future studies should adopt longitudinal designs with repeated assessments to explore potential causal links and deepen the understanding of these associations.