For decades, research in personality has predominantly followed a variable-centered approach. This methodology has substantially advanced the understanding of the structural organization of constructs and the relationships among variables, operating under the assumption that these associations apply uniformly across individuals (Howard and Hoffman, 2018). However, this assumption has increasingly been questioned, with some scholars arguing that variable-centered models may obscure critical insights into how personality characteristics are configured and interact within individuals (Asendorpf, 2002; Herzberg and Roth, 2006). In contrast, a person-centered approach enables researchers to identify subpopulations based on shared patterns of variable expression (Howard and Hoffman, 2018), offering a more nuanced perspective on individual personality functioning. Among person-centered approaches, latent profile analysis (LPA) has emerged as a particularly powerful method, as it probabilistically identifies subgroups based on multivariate response patterns and offers a flexible framework for evaluating their distinctiveness. While both approaches are valuable and complementary (Mervielde and Asendorpf, 2000; Roth and Herzberg, 2017), the person-centered framework, and LPA in particular, provides a robust basis for typological research aimed at capturing coherent constellations of enduring characteristics rather than isolated dimensions (Williams and Kibowski, 2016).

Within this broader methodological context, temperament research has followed a similar trajectory. Although temperament is conceptually distinct from personality traits, typically referring to biologically rooted dispositions evident early in life (Teglasi et al., 2023), many temperament models have been developed using variable-centered, factor-analytic techniques. These include foundational frameworks developed by Rothbart and colleagues (Rothbart et al., 2000), see also Evans and Rothbart (Evans and Rothbart, 2007; Evans and Rothbart, 2009), Buss and Plomin (1975; 1984), Bates and associates (Bates et al., 1979), and Strelau (1996), all of which have been influential in structuring the understanding of temperament across the lifespan. However, longitudinal studies indicate that developmental continuity in temperament is rarely attributable to isolated traits; instead, it reflects stable configurations of interrelated characteristics (Caspi and Silva, 1995; Korn, 1984; Pesonen et al., 2003; Thomas and Chess, 1986). These findings underscore the value of person-centered methodologies for capturing the complexity and stability of temperament over time.

Among temperament models, the New York Longitudinal Study (NYLS) stands out as a landmark investigation that integrated both variable- and person-centered perspectives. Conducted by Thomas, Chess, and colleagues (Chess and Thomas, 1999; Thomas et al., 1963; Thomas et al., 1968), the NYLS followed a cohort from infancy into adulthood to examine the hypothesis that individual behavioral styles emerge early in life, remain relatively stable, and shape responses to environmental contexts. The NYLS was particularly focused in identifying patterns of temperament that either predisposed individuals to, or protected them from, behavioral disorders (Thomas et al., 1968). Through systematic observations and caregiver interviews, nine core temperament dimensions were identified: activity level (the proportion of active versus inactive periods), regularity/rhythmicity (predictability in timing of functions), adaptability (responses to new or altered situations), approach or withdrawal (initial reactions to novel stimuli), intensity of reaction (energy level of responses), quality of mood (emotional tone, either positive or negative), persistence (continuation of activities despite obstacles), distractibility (susceptibility to disruption by unrelated stimuli), and threshold of responsiveness (sensitivity to stimuli and intensity needed to elicit a response). These dimensions became the foundation for understanding individual differences in temperament across development.

Through a combination of clinical judgment and early statistical methods, the NYLS identified three temperament types: Easy, Difficult, and Slow-to-warm-up (Thomas et al., 1963; Thomas and Chess, 1977). The Easy type was characterized by regularity, positive responses to novelty, quick adaptation to change, mild emotional reactivity, and an overall positive mood. In contrast, the Difficult type was characterized by irregularity, negative reactions to novelty, low adaptability, intense emotional responses, and a predominantly negative mood. The Slow-to-warm-up type exhibited initial withdrawal, gradual adaptation, and milder expressions of intensity and regularity compared to the Difficult type. Prevalence estimates from the NYLS assigned 40% of participants to the Easy group, 10% to the Difficult group, and 15% to the Slow-to-warm-up group. However, 35% of participants did not fit these categories, suggesting the presence of additional or intermediate profiles and highlighting the need for a more differentiated typology (Carey, 1970; Chess and Thomas, 1986).

Despite its age, the NYLS typology remains highly influential in both research and practice. It continues to inform theoretical models (Caspi, 2000; Maine et al., 2025), cross-cultural studies (Super et al., 2008; Super et al., 2020), and research on specialized populations (Cassiano et al., 2020). It has also demonstrated predictive validity for adult mental health outcomes (Wu et al., 2022). Beyond research, the NYLS model underpins clinical guidelines (Carey and McDevitt, 2016; Delgado et al., 2022) and is widely cited in parenting literature (Kurcinka, 2020; Turecki and Tonner, 2000), further affirming its practical relevance.

Although temperament is generally considered relatively stable across the lifespan, the application of person-centered methods to adult temperament has been limited. Nevertheless, existing studies based on frameworks such as Cloninger’s psychobiological model (Cloninger et al., 1993; Strelau, 1996) and Strelau’s regulative theory of temperament (Strelau, 1996) have demonstrated the utility of person-centered analyses in adult populations. For example, adult temperament profiles have been associated with diverse outcomes, including health status (Keltikangas-Järvinen et al., 1999; Wessman et al., 2012), clinical symptoms (Müller et al., 2014; Rzeszutek and Oniszczenko, 2022), physical activity (Karppanen et al., 2023), criminal behavior (Moreira et al., 2022b), humor potential (Moreira et al., 2022a), and cognitive-emotional regulation (Wytykowska et al., 2022). These findings underscore the potential value of identifying temperament profiles, defined as biologically based configurations of behavioral tendencies, using person-centered methods in adult populations.

While the NYLS provided foundational insights, its typology was limited by a small sample size (N = 133), the analytic tools available at the time, and its exclusive focus on childhood data. Notably, over one-third of participants could not be categorized into any of the three original types, and the model was never formally updated. This substantial proportion of unclassifiable cases raised questions about the comprehensiveness of the three-type system, limited its capacity to capture the full range of individual differences, and reduced its practical utility. In applied and clinical settings, the absence of a comprehensive classification complicates the tailoring of interventions to individual needs. These limitations highlight the need for a more differentiated typology and point to the value of modern person-centered methods, such as LPA, which can identify additional subgroups based on multivariate response patterns and provide a more systematic framework for classification. Furthermore, it remains unclear whether comparable temperament profiles emerge in adulthood, given developmental shifts in temperament expression. To address these gaps, the present study revisits and extends the NYLS typology using LPA applied to a large, cross-national adult sample from the USA and Lithuania. Including participants from both countries allowed for analysis of the pooled sample while preserving statistical power and treating country as a distinct demographic variable.

The current study had two primary objectives. First, we aimed to reassess the viability of the NYLS typology by identifying adult temperament profiles using LPA, based on data from a standardized instrument derived from the original NYLS framework. This instrument assesses biologically based characteristics while clearly distinguishing temperament from broader trait- or state-like personality constructs. LPA, a probabilistic, person-centered method, provides greater nuance and flexibility than traditional clustering techniques (Spurk et al., 2020; Weller et al., 2020). It is particularly well-suited to typological research and aligns with professional practice, where categorical classifications are often more informative than dimensional scores (Mervielde and Asendorpf, 2000; Williams and Kibowski, 2016). Although LPA has been widely applied in temperament research with infants and toddlers (Augustine et al., 2022; Gartstein et al., 2017), as well as with children and adolescents (Dalimonte‐Merckling and Brophy‐Herb, 2019; Guerin et al., 2023; Hawes et al., 2022; Martin and Lease, 2022; Moran et al., 2017), its application to adult samples remains limited.

Second, we examined whether the resulting profiles varied by age, gender, or country of residence. Prior studies have reported gender-based differences in temperament patterns (Korn, 1984; Wessman et al., 2012; Wytykowska et al., 2022), although findings at the profile level remain mixed or ambiguous. While cultural influences on temperament are well documented in children (DeVries, 1984; Putnam et al., 2024; Super et al., 2020), few cross-national studies have investigated adult populations. Although both the USA and Lithuania are considered Western societies, they represent distinct cultural contexts that may shape behavioral tendencies in different directions (Allik et al., 2017; Schmitt et al., 2007). Age is also a relevant factor, as the expression of specific temperament characteristics may shift across the lifespan; for example, behaviors associated with difficult temperament in childhood may become more adaptive in later life (Carey and McDevitt, 1995). Exploring how temperament profiles relate to demographic variables is essential for both theoretical understanding and applied interpretation.

In summary, this study addressed two key objectives: (1) to revisit and refine the NYLS temperament typology using modern person-centered methodology in an adult sample, and (2) to examine how sociodemographic factors may influence or interact with temperament profiles. The findings aim to contribute to both theoretical advancement and practical application, helping to bridge the gap between research and real-world practice.

This study identified five qualitatively distinct adult temperament profiles—Easy, Difficult, Slow-to-warm-up, Curious, and Shy—based on a large cross-national sample assessed with standardized self-report instruments grounded in the NYLS framework. The first three profiles closely resembled the original NYLS types (Chess and Thomas, 1999; Thomas et al., 1963), reinforcing the framework’s enduring relevance. However, the Easy profile was less prevalent in this adult sample, and no “unclassified” or “intermediate” individuals were identified. Instead, participants not captured by the original typology were meaningfully grouped into two additional profiles: Curious and Shy. This suggests that the classic three-type structure, while foundational, may have overlooked important temperament configurations that become detectable when using modern person-centered methods.

The Curious and Shy profiles represent conceptually coherent extensions of the NYLS model, distinguished primarily by approach-withdrawal tendencies and activity level—dimensions central to many temperament theories (Zuckerman, 2012). The Curious profile was characterized by high activity and strong approach tendencies, whereas the Shy profile reflected low activity and marked withdrawal. These opposing behavioral styles help address a theoretical gap left by the original NYLS structure. It is plausible that these patterns were not previously detected due to their limited relevance to symptomatic behaviors.

The five-profile model also aligns conceptually with other large-scale typologies. For instance, the Dunedin study proposed five temperament types: Undercontrolled, Inhibited, Confident, Reserved, and Well-adjusted (Caspi and Silva, 1995; Caspi et al., 2003). Several of these correspond to NYLS categories: Undercontrolled with Difficult, Inhibited with Slow-to-warm-up, and Well-adjusted with Easy (Caspi, 2000). The Curious profile resembles the Confident type, defined by high approachability in childhood and impulsivity in adulthood, whereas the Shy profile parallels the Reserved type, characterized by withdrawal and sluggishness in childhood and a less decisive disposition in adulthood. Additional support comes from person-centered models grounded in broader personality frameworks. For example, Herzberg and Roth (2006) used Big Five data to identify five clusters—Resilient, Overcontrolled, Undercontrolled, Reserved, and Confident—that closely parallel both the Dunedin and current model. Although based on different theoretical foundations, this structure shows organizational similarities, suggesting common patterns of personality and temperament across frameworks. This is consistent with Asendorpf’s (2008) work, which directly linked NYLS temperament concepts to adult personality prototypes, highlighting continuity from early behavioral tendencies into adulthood. Taken together, these parallels position the present five-profile solution within a broader family of five-type models, underscoring its theoretical significance and its convergence with established temperament and personality typologies.

The present typology also helps clarify previously overlapping constructs in the literature. For instance, the Slow-to-warm-up profile identified here, characterized by withdrawal, persistence, and low distractibility, more closely aligns with the Reserved type in the Dunedin study (Caspi and Silva, 1995) than with the Inhibited type. In contrast, the Shy profile, which also exhibits withdrawal but is characterized by low persistence and high distractibility, aligns more closely with the Inhibited type. This distinction addresses a long-standing conflation between temperamental shyness and Slow-to-warm-up temperament. Asendorpf (2008) described a “Slow-to-warm-up inhibited temperament” (p. 113) as a developmental precursor to adult introversion and neuroticism. Our findings suggest that this description more accurately characterizes the Shy profile, while the Slow-to-warm-up profile appears more self-regulated and emotionally stable in terms of behavioral reactivity. These refinements enhance conceptual precision and may inform both developmental theory and clinical assessment.

Beyond temperament, person-centered models based on broader personality inventories further highlight the relevance of the current findings. For instance, studies by Isler et al. (2016) and Daljeet et al. (2017), using the Big Five and HEXACO models, identified three- to five-profile structures through latent modeling approaches. Similarly, Fisher and Robie (2019), using an exceptionally large global dataset, proposed a three-profile system—Maladaptive, Adaptive, and Highly-adaptive—based on overall psychological adjustment. On the one hand, these models, though grounded in different traditions, show similar patterns of high and low trait expression. On the other hand, they typically do not capture qualitative distinctions in behavioral reactivity, such as approach-withdrawal or emotional intensity, which are central to temperament theory. This comparison highlights the complementary value of temperament-based typologies in capturing early-emerging affective and regulatory styles that may precede or inform broader personality constructs.

Demographic factors, including age, gender, and country, were not meaningfully associated with temperament profile membership. Although some effects reached statistical significance, their effect sizes were small, indicating limited practical relevance. This finding aligns with previous research suggesting that demographic effects may be weaker in adulthood, particularly in older samples (Brancart et al., 2024). Prior studies have reported mixed findings regarding gender-based differences (Wessman et al., 2012; Wytykowska et al., 2022), but these often relied on broader constructs or included non-temperamental variables, making direct comparisons difficult. The similarity in profile distributions across the USA and Lithuanian samples further supports the potential cross-cultural generalizability of the five-profile model. Nonetheless, future research should examine whether more nuanced patterns were obscured by sample characteristics, cohort effects, or broader sociocultural factors.

From an applied perspective, the current findings are potentially relevant for temperament-based assessment and for identifying individual differences in behavioral style, emotional responsiveness, and person-environment fit in both clinical and everyday contexts. The clear differentiation among the five profiles may also facilitate more personalized approaches in psychological assessment, counseling, and workplace or interpersonal settings by linking characteristic patterns of activity and approach-withdrawal tendencies to adaptive functioning and social behavior. The cross-national similarity of profiles further indicates that these configurations reflect broad, culturally generalizable dimensions of temperament rather than sample-specific artifacts. Regarding developmental continuity, the present typology could facilitate integration of the NYLS framework across the lifespan if replicated in younger samples. Establishing whether a similar structure emerges in childhood will require empirical testing using existing caregiver-based assessment instruments rather than self-report measures (Carey and McDevitt, 2016). Such research would clarify whether the five-profile solution reflects a stable organizational pattern of temperament from early life into adulthood.

The current study re-evaluated the classic NYLS temperament typology using modern person-centered statistical methods in a large adult sample. Five temperament profiles—Easy, Difficult, Slow-to-warm-up, Curious, and Shy—were identified. Three profiles closely aligned with the original NYLS categories, while two extended the model in theoretically meaningful ways and demonstrated conceptual connections to other temperament and personality typologies. Notably, profile membership was largely independent of age, gender, and country, suggesting broad applicability across demographic groups.

By offering greater structural clarity and conceptual refinement, the updated typology advances efforts to understand individual differences in temperament across the lifespan. Future research should aim to replicate this model in diverse populations, evaluate its predictive validity, and explore its relevance across developmental, clinical, and applied contexts.

The data that support the findings of this study are available from the authors upon reasonable request.

SM played a lead role in supervision and an equal role in conceptualization, data curation, investigation, methodology, writing—original draft, writing—review and editing. TL played a lead role in formal analysis and visualization and an equal role in conceptualization, data curation, investigation, methodology, writing—original draft, writing—review and editing. Both authors contributed to editorial changes in the manuscript. Both authors read and approved the final manuscript. Both authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The study was conducted in accordance with the Declaration of Helsinki. The research protocol adhered to herein was approved by the Vilnius University Committee on Research Ethics in Psychology (No. 31/(1.3 E) 250000-KP-50). Authorization for the use of anonymous server data was granted by the Project Manager at Behavioral–Developmental Initiatives. Informed consent was obtained from all participants included in the study.

The authors wish to express their gratitude to Roy P. Martin, Professor Emeritus at the University of Georgia, for his valuable contribution to this paper. His thoughtful review of an earlier version of the manuscript and helpful comments greatly enhanced the quality and clarity of this paper.

This research received no external funding.

The authors declare no conflict of interest.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/BP45740.