Five-factor model-based personality profiles in South Korea

Abstract

This study aimed to (1) identify personality profiles based on the Five-Factor Model using a large South Korean sample (N = 813,611) and (2) examine their associations with well-being and related variables (e.g., self-esteem, loneliness, subjective social status). Using a Gaussian Mixture Model (GMM), we identified five personality types: Average, with mid-range scores on all traits; Introverted-Open, with high Openness and low Extraversion and Agreeableness; Introverted-Reactive, with low Conscientiousness and Extraversion and high Neuroticism; Expressive, with high Openness and Extraversion; and Conventional, with low Openness and, to a lesser extent, low Agreeableness and Neuroticism. Membership probabilities for the five personality profiles showed distinct patterns of association with the aforementioned variables, providing further insight into their characteristics. Expressive and Average type membership was associated with only beneficial outcomes, though the magnitude of these associations differed, while the Introverted-Reactive type was linked to lower well-being and various psychological vulnerabilities. The Introverted-Open and Conventional types displayed a nuanced profile, characterized by a combination of both adaptive and maladaptive attributes. These findings build on previous personality typology research by identifying empirically derived profiles in a Korean sample and offer insights for developing targeted, person-centered well-being interventions.

Introduction

“None of us is so exquisitely different as to defy a useful categorization”¹.

What “types” of people exist? Tiger moms, night owls, romantics, curmudgeons, and the list goes on. This basic human inclination to categorize and label serves the adaptive function of simplifying complex information, thereby facilitating comprehension, learning, and the discovery of patterns^2,3,4,5. To answer the opening question, researchers have explored the presence of common personality “types” or profiles within populations⁶, often derived from the Five Factor Model of personality (FFM)⁷. However, although different personality types may exist across different cultures, studies involving sizable South Korean samples remain relatively scarce. Accordingly, the current study employs a data-driven approach to examine the Big-Five-based personality types as well as their well-being-related correlates, in a large (N = 813,611) South Korean sample.

Identifying personality types in South Korea

The Five-Factor Model (FFM) is a widely accepted psychological framework that defines personality in terms of five key dimensions: Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism⁷. Much of the research on this model has assumed a variable-centered (i.e., dimensional) approach which examines individual differences along each personality dimension independently⁸. While this approach is helpful for dividing a complex construct into manageable components, it suffers from being too reductionistic in that it does not capture the role of a dimension in the context of other dimensions^8,9. The person-centered (i.e., typological) approach, however, addresses this limitation by uncovering configurations of traits that define subtypes of people and allowing researchers to “identify, organize, and systematically describe naturally occurring behavioral patterns of people in such a way that the wholeness of the people is retained”¹⁰.

Earlier studies taking the person-centered approach primarily highlighted 3 personality prototypes (hereafter ROU): Resilient, characterized by low levels of Neuroticism and high levels on other dimensions, Overcontrolled, characterized by high levels of Neuroticism (and at times Agreeableness) and low levels of Extraversion, and Undercontrolled, with low levels of Conscientiousness and Agreeableness^11,12,13,14. These types were labeled based on the constructs of ego-control (i.e., the tendency to inhibit and control impulses) and ego-resiliency (i.e., the ability to adapt to environmental changes by regulating one’s level of control), which Block and Block¹⁵ proposed reflected the underlying regulatory processes that give rise to personality. Since then, a growing number of studies have sought to replicate and/or extend these findings^{6,16,17,18,19}. Notably, although many of these studies at least in part adopted the labels for the ROU types, there were significant variations across studies in the identified profiles as well as the proposed number of profiles^6,18,20,21. Substantial discrepancies can result from differences in the sample, measure, and clustering method used^11,22, but this variability also suggests that a bottom-up approach to explaining the empirically derived personality types (rather than a top-down approach in which the types are interpreted through the lens of an existing theoretical framework) may prove valuable for uncovering and validating the existence of other personality types. As an example, Gerlach and colleagues¹⁸, using a purely data-driven approach, confirmed the existence of two personality profiles that did not map onto the ROU types.

We propose that this inductive approach is particularly useful when examining personality types in a new culture or context that is likely to differ from those that have been examined (e.g., Western cultures)²¹ and South Korea represents one such culture. Research indicates that the values of South Koreans differ markedly from those of Westerners, with South Koreans (vs. Westerners) placing greater emphasis on social harmony and cohesion^23,24,25,26, and resultantly adherence to social norms^27,28. These divergent values are likely to create social contexts that reward a different set of behaviors than those rewarded in Western cultures^29,30 and may in turn produce different “types” of personalities. In addition, research has shown that levels and representations of the Big Five may differ across individuals of Korean and Western cultural backgrounds^31,32,33. Despite these known cultural differences and prior research indicating that culture is an important factor that can influence profile solutions³⁴, to date, there are few studies employing Korean samples³⁵. Moreover, the studies that do exist have used smaller samples (N < 500) or shorter measures of the FFM, which can limit the reliability of findings when conducting clustering analyses^18,36. Thus, in the present research, we analyze the responses of an extremely large sample of individuals (N = 813,611) to a 120-item measure of the Big Five to identify personality profiles that exist in the Korean population. To the best of our knowledge, this is one of the largest datasets utilized in studies of this nature. We use the Gaussian mixture model, which allows probabilistic quantification of cluster membership and captures uncertainty in clustering, particularly for borderline cases where individuals may show mixed characteristics across different profiles³⁷. This is a key advantage over traditional clustering methods.

Correlates of personality types

Because the person-centered approach offers a coherent account of personality functioning that captures how the person (vs. individual traits) interacts with their environment^13,17,38, an investigation of the correlates of personality types can provide unique information not captured by studies focusing on individual personality dimensions. There has been empirical work comparing the predictive utility of the dimension- and person-centered approaches for various outcomes, and results have been mixed. Some studies have found that dimensions are stronger predictors of some outcomes than types^20,39,40,41, but others have also revealed that personality types (vs. dimensions) are better predictors of longitudinally measured temperamental outcomes^39,42 and exhibit differential associations with health outcomes⁴³. Although these direct comparisons yield mixed findings, studies show that personality profiles account for significant variation in individuals’ behavior^44,45 and numerous other variables^{16,17,19,46,47}.

Hence, we argue that an investigation into the correlates of personality profiles, which helps establish their external validity, is meaningful and warranted. In the current study, we focus on correlates related to well-being because not only is it a widely prized outcome universally^48,49, but it is of particular interest in Korea, where happiness levels are consistently low⁵⁰, and the country records the highest suicide rate among the 38 OECD nations despite its economic prosperity⁵¹. Specifically, we test the associations between personality type membership and (1) well-being and (2) various individual difference variables known to be associated with well-being: self-esteem⁵², stress⁵³, loneliness⁵⁴, essentialist beliefs about happiness⁵⁵, tendencies for social comparison⁵⁶, gratitude⁵⁷, and optimism⁵⁸.

Methods

The current manuscript reports secondary analyses of data collected by Kakao Corporation in partnership with the Center for Happiness Studies at Seoul National University, with informed consent obtained from all participants. The use of this data was approved by the Institutional Review Board of Kangwon National University (IRB No. 201910009002). This work was conducted in accordance with institutional and international ethical guidelines, including the Declaration of Helsinki and its subsequent amendments. The present study was not preregistered.

Participants

The present study utilized data collected via an online survey platform operated by Kakao Corporation, one of the largest telecommunication companies in Korea, in collaboration with the Center for Happiness Studies at Seoul National University. The data collected through the Kakao-SNU platform has previously been used in a number of psychological studies^59,60,61,62. The survey was distributed through the company’s primary application, KakaoTalk, and its affiliated website (http://together.kakao.com/hello). Participants could voluntarily respond to the survey at any time and as often as desired. If participants provided multiple responses for key variables of interest, only their first response was included in the analysis. Due to Kakao Corporation’s privacy policy, demographic information was limited to participants’ birth year, gender, and region of residence.

From January 1, 2018 to December 31, 2024, 813,611 participants (586,684 female, 226,926 male, and 1 who preferred not to answer) responded to all five personality scales. On average, participants took 53.07 days (SD = 221.56) to complete all five scales. Some participants responded to the personality scales multiple times; for these participants, only the first response was retained and subsequent responses were discarded. Respondents’ ages ranged from 14 to 74 years (M_age = 27.68, SD = 8.91). A subset of participants specified their age group (e.g., teens, 70s) rather than providing their exact age (N = 45,175), or chose not to respond (N = 1,646). As a result, the mean and standard deviation were calculated only for those who reported their specific age. The age distribution was as follows: 162,619 participants were between the ages of 14 and 19 years, 396,653 were between 20 and 29 years, 172,623 were between 30 and 39 years, 54,407 were between 40 and 49 years, 21,059 were between 50 and 59 years, and 4,604 were aged 60 years or older. Participants were not paid but received personalized feedback consisting of their score on the scale, its interpretation, and normative comparisons to other South Korean respondents.

Measures

Unless otherwise noted, all measures listed below were administered in Korean using the translation and back-translation method, conducted by native speakers of English and Korean.

Personality

The Big Five personality traits were assessed using a Korean adaptation of the International Personality Item Pool NEO-120 (IPIP-NEO-120)⁶³. This Korean version was developed through cultural and linguistic adaptation of the original IPIP–NEO–120 items to enhance clarity and appropriateness for Korean respondents. The IPIP-NEO-120 itself is a 120-item abbreviated form derived from the original 300-item IPIP-NEO measure^64,65 and assesses (1) Openness, (2) Conscientiousness, (3) Extraversion, (4) Agreeableness, and (5) Neuroticism, each consisting of six facets (4 items each). Participants completed the measure across five separate sessions, rating the 24 items for a different Big Five domain in each session using a 5-point scale (0 = disagree strongly, 4 = strongly agree). The 24-item Big Five domain scales demonstrated acceptable reliability, with McDonald’s omega coefficients of 0.86 (Openness), 0.90 (Conscientiousness), 0.91 (Extraversion), 0.79 (Agreeableness), and 0.92 (Neuroticism). Across domains, mean facet-level omega coefficients ranged from 0.70 to 0.74, with individual facet omegas spanning 0.51 to 0.87. A full facet-level reliability table for all 30 facets is provided in the Supplemental Material. Furthermore, correlations among facets within the same domain were markedly higher (Mr ≈ 0.41) than those between facets from different domains (Mr ≈ –0.02), consistent with the expected Big Five structure (see the Supplemental Material for domain-specific averages).

Correlates: well-being and related variables

We used the following measures of well-being: life satisfaction⁶⁶, positive and negative affect⁶⁷, and meaning in life⁶⁸. Additional individual-difference variables assessed included self-esteem⁶⁹, perceived stress⁷⁰, loneliness⁷¹, essentialist beliefs about happiness⁵⁵, tendencies for social comparison⁷², gratitude⁵⁷, and optimism⁷³. Subjective socioeconomic status was also assessed using the 10-point MacArthur ladder scale⁷⁴. A more detailed discussion of each variable is provided in the Supplemental Material.

Analytic procedure

Clustering and enrichment analysis

First, we performed an Exploratory Factor Analysis (EFA) with oblimin rotation on the 120 items to extract factor scores for the Big Five dimensions: Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism. Oblimin rotation was selected to allow for correlations between factors. We then identified latent personality profiles using Gaussian Mixture Models (GMM) via the R package ‘mclust’. In large-scale datasets, standard fit indices such as the Bayesian Information Criterion (BIC) often monotonically increase or decrease with sample size. To address this issue, we utilized the Normalized Entropy Criterion (NEC) to determine the optimal number of clusters. Given that GMMs can be sensitive to initial values, leading to local optima, we implemented multiple random initializations. To ensure the stability of our final solution, we calculated the Normalized Variation of Information (NVI) among the top candidate models with high NEC values. We selected the specific solution (seed) that minimized the average NVI distance to all other candidates, thereby identifying the most representative and stable clustering result. Finally, to statistically validate the distinctiveness of the identified profiles, we conducted an enrichment analysis. This involved calculating an enrichment score for each cluster, computed as an average of density ratios of original to permuted data, to assess whether the observed data density significantly exceeded what would be expected under a null hypothesis of independence¹⁸.

Further details regarding these analyses, model fit indices (e.g., BIC, ICL), and the results of two robustness checks (a split-sample replication and an analysis using simple mean scores) that confirm the stability of the five-profile structure are provided in the Supplemental Material.

Regression analyses

To examine how the uncovered profiles relate to well-being and related variables, we conducted log-contrast regression analyses with GMM-derived cluster membership probabilities as independent variables, and each correlate (hereafter outcome variable) as the dependent variable. Age and gender were included as covariates. Because participants were free to select the scales they completed, the number of responses varied across outcome variables. We used listwise deletion for each analysis separately to maximize the sample size, which resulted in different sample sizes across analyses. Given the compositional constraint—membership probabilities summing to one for each individual—we log-transformed the probabilities and imposed a zero-sum constraint on the regression coefficients. This modeling approach estimates the relationship between relative changes in cluster membership probabilities and the outcome measures where the regression coefficients quantify the degree to which the outcome variable changes for every unit deviation from the geometric mean of the membership probabilities, taking into account the dependent structure of the compositional components. For statistical inference, the errors from the log-contrast model must follow a normal distribution, which allows the construction of confidence intervals and hypothesis tests on the regression coefficients. In our analysis, since the residuals violated normality assumptions, we utilized bootstrap confidence intervals and permutation tests following Lee et al.⁷⁵.

Covariate adjustment was performed using a two-stage residual inclusion procedure⁷⁶. In the first stage, each outcome variable was regressed on the covariates (age and gender) using ordinary least squares (OLS), and the residuals from this model were retained. In the second stage, these residuals served as the dependent variable in the log-contrast regression model, with the GMM posterior membership probabilities as the compositional predictors. This two-stage approach estimates the association between the cluster membership probabilities and the outcome while adjusting for the covariates.

Results

Identification of personality profiles

Based on the GMM results, a five-profile solution was identified as the most appropriate. The enrichment analysis further supported this solution, indicating that all five clusters had enrichment scores greater than 1 (Introverted-Open = 1.11, Conventional = 1.50, Average = 2.01, Expressive = 2.30, Introverted-Reactive = 2.56), meaning that each cluster exhibited greater data density than expected under random assignment. The corresponding p-values—calculated as the proportion of permutation iterations yielding ratios greater than 1—were all below 0.05, providing statistical evidence for the cohesiveness of each cluster.

Table 1 presents the mean scores and standard deviations for the FFM dimensions for each profile, and Fig. 1 provides a graphical representation of these scores. To aid comprehension, we provide descriptive labels for each personality profile. We note that these labels are intended as heuristic descriptors and do not substitute for the full empirical characterization based on the underlying Big Five domain scores. Specifically, the most common profile (n = 384,967) was characterized by factor scores near the grand mean across all personality dimensions. Accordingly, we designate this profile the “Average” type and consider a mean dimension score that falls within this range to reflect an average level on the respective personality dimension. This personality type is one that has been identified in a number of prior studies^18,34,45,77.

The second most common profile was characterized by high levels of Openness and moderately low levels of Extraversion and Agreeableness. This is a new type that has not been identified in prior research and exhibits overlap on only single dimensions with previously identified personality types (e.g., low level of Extraversion seen in the overcontrolled type, low level of Agreeableness typical of the undercontrolled type). The distinctive configuration of Openness, Extraversion, and Agreeableness present in this type is unique, and depicts an individual who is intellectually curious yet not socially inclined. We denote this profile the “Introverted-Open” type (n = 120,501).

The third profile featured moderately low levels of Conscientiousness and Extraversion and a high level of Neuroticism. This type is similar to the undercontrolled type, which is typically described as having relatively low levels of Agreeableness, Conscientiousness, and at times emotional stability. However, unlike the undercontrolled type, this type exhibits an average level of Agreeableness. We designate this profile the “Introverted-Reactive” type (n = 111,132).

The fourth profile was distinguished by high levels of Openness and Extraversion. This profile partially resembles the Confident type, which is an additional personality type identified by a number of researchers^{6,19,41,45,78,79} that is characterized by high levels of Openness, Extraversion, and Agreeableness, with considerable variability in Conscientiousness and Neuroticism across studies. Given that the defining characteristics of this personality type are Openness paired with a predisposition toward active engagement in external and social environments (i.e., Extraversion), we label this profile the “Expressive” type (n = 100,771).

The fifth and final profile was characterized by a particularly low level of Openness, and to a lesser degree, Agreeableness and Neuroticism. This profile aligns closely with the Reserved type–marked by low levels of Openness and Neuroticism—which is also an additional personality type identified by a number of researchers^{6,18,19,41,45,78,79,80}. Based on the distinctively low level of Openness for this profile, we refer to this profile as the “Conventional” type (n = 96,240).

Table 1 Means and standard deviations of study variables across the five personality profiles.

Fig. 1

The five personality profiles identified through the GMM analysis.

The correlates of profile membership

Given that well-being is a matter of significant concern in South Korea, we examined the relationship between personality type membership and (1) well-being and (2) various individual-difference variables known to predict well-being. Table 1 presents the mean scores and standard deviations of all correlates for each profile. We conducted log-contrast regression analyses, using the log-transformed probabilities of membership in the five types as independent variables and each (standardized) correlate as the dependent variable. These analyses reveal the extent to which the probability of being classified into a given type (vs. the average composition) predicts the outcome variables. For simplicity, we refer to individuals with higher membership probability scores for a given profile as stronger representatives of that type (e.g., stronger Average type) and those with lower membership probability scores for a given profile as weaker representatives of that type (e.g., weaker Average type). We summarize notable findings for each profile here; full results are available in the Supplemental Material.

Average

Average type membership probability was generally positively associated with well-being and variables known to predict well-being. Specifically, stronger (vs. weaker) Average types reported greater life satisfaction, positive affect, and meaning in life, as well as elevated socioeconomic status (SES), gratitude, optimism, and self-esteem. They were also more likely to exhibit low levels of essentialist beliefs about happiness, believing that happiness is malleable and improvable with effort and not fixed or biologically predetermined. Additionally, stronger Average types experienced less negative affect, stress, and loneliness, and were less prone to engaging in social comparisons than weaker Average types.

Introverted-Open

The likelihood of being categorized as the Introverted-Open type relative to the average composition was negatively associated with not only positive affect, but also negative affect and stress, suggesting that stronger Introverted-Open types may have more muted emotional experiences than weaker Introverted-Open types. Membership probability for this profile also predicted a range of negative outcomes, such as higher levels of loneliness and essentialist views of happiness and lower levels of gratitude, optimism, and socioeconomic status. On the positive side, stronger (vs. weaker) Introverted-Open types were markedly less inclined to engage in social comparisons and more inclined to search for meaning.

Introverted-Reactive

The Introverted-Reactive type was found to be the most psychologically vulnerable: stronger Introverted-Reactive types were more likely to report lower SES and consistently low levels across a range of positive indicators, including happiness, meaning in life, self-esteem, gratitude, and optimism, than weaker Introverted-Reactive types. Stronger (vs. weaker) Introverted-Reactive types were also more prone to heightened stress, loneliness, social comparisons, and viewing happiness as fixed and immutable.

Expressive

The psychosocial strengths of the Expressive type closely mirrored those of the Average type, with stronger (vs. weaker) Expressive types exhibiting higher levels across measures of well-being and traits conducive to well-being and lower levels across maladaptive ones. However, Expressive type membership was a stronger predictor of SES, self-esteem, and optimism than Average type membership, and did not exhibit any relationship with social comparison.

Conventional

Conventional type membership probability was positively associated with SES, positive affect, and self-esteem, and negatively associated with negative affect, stress, loneliness, and social comparison behaviors. Despite these favorable attributes, stronger Conventional types did not experience greater life satisfaction than weaker Conventional types, but instead reported less optimism, gratitude, and experience of and search for meaning in life, and stronger essentialist beliefs about happiness.

General discussion

In the present study, we identified five personality profiles in a large South Korean sample: (1) Average, with average scores on all dimensions; (2) Introverted-Open, with a high level of Openness and low level of Extraversion and Agreeableness; (3) Introverted-Reactive, with low levels of Conscientiousness and Extraversion and a high level of Neuroticism; (4) Expressive, with high levels of Openness and Extraversion; and (5) Conventional, with a particularly low level of Openness and to a lesser degree, Agreeableness and Neuroticism. These personality profiles echo patterns identified in previous research while also revealing novel trait configurations. For instance, the Introverted-Reactive type resembles the Undercontrolled profile but also shows an average (vs. low) level of Agreeableness, the expressive type exhibits parallels to the Confident profile, and the Conventional type shows similarities to the Reserved profile but features a low (vs. average) level of Agreeableness. Notably, the Confident and Reserved profiles were identified in prior research as additional personality types beyond the traditional Resilient, Overcontrolled, and Undercontrolled classifications, suggesting that some of our identified types align more closely with these expanded typologies. Finally, the Introverted-Open type represented a novel profile. Interestingly, we did not find the Resilient (or Role-Model) type found in numerous prior studies.

Many prior studies have adopted the labels used in earlier studies (i.e., Resilient, Overcontrolled, and Undercontrolled) to describe their uncovered profiles, despite there being substantial variations in levels of some dimensions, some even across the negative and positive range^18,21. It is possible that some of the profiles identified in the present research are variations of the previously identified personality types, with discrepancies stemming from differences in factors such as sample characteristics, measures, and/or clustering method^11,19,22. However, in keeping with our goal of conducting a fully data-driven examination of personality types in South Korea, we assigned new labels that we believed best embodied the characteristics of each personality type. Furthermore, we examined the relationship between probability of membership in each of these personality types and various well-being-related variables, providing a richer understanding of these profiles: (1) Average, characterized by general happiness and broadly positive psychological traits; (2) Introverted-Open, emotionally reserved, lonely, and insulated from social comparison, yet seeking meaning in life; (3) Introverted-Reactive, the most vulnerable profile, distinguished by emotional strain and limited psychological resources for well-being; (4) Expressive, similar to the Average type but showing higher levels of SES, self-esteem, and optimism; and (5) Conventional, emotionally composed and prone to experiencing positive affect, yet falling short on life meaning, optimism, and gratitude.

Contributions

Our research contributes to the expanding body of literature on personality types. Investigations of typologies have primarily been undertaken using Western samples (e.g., Americans, Europeans). Although a small subset has used East Asian samples (e.g., Japan, China), studies using Korean samples—especially those of a significant size—remain scarce. Through a large-scale, data-driven analysis in an understudied cultural context, we found that the profiles we identified exhibited a degree of consistency with, but also meaningful differences from, those uncovered in prior studies. These findings add to the understanding of variability in the constellations of personality traits across national cultures, showing that not only are there likely to be nuanced differences in common personality types, but certain personality types represented in one culture may be absent in another (e.g., Resilient type). Notably, apart from the Introverted-Reactive type, which shares some characteristics with the Undercontrolled type, our profiles did not map onto the ROU classification. This underscores the value of an extensive data-driven, bottom-up approach in capturing personality structures specific to understudied cultures.

Beyond identifying culturally specific personality profiles, our findings also demonstrate the value of a person-centered approach for uncovering how trait configurations—rather than individual trait levels—relate to well-being. For example, although Openness is often associated with positive outcomes⁸¹, individuals with a stronger likelihood of belonging to the Introverted-Open profile reported lower positive affect, higher loneliness, and generally more muted emotional experiences despite their high Openness. At the same time, they showed lower tendencies toward social comparison, revealing a psychological pattern not easily inferred from Openness or Extraversion alone. Similarly, although Agreeableness is typically a positive predictor of well-being⁸¹, individuals with a stronger likelihood of belonging to the Conventional profile—despite their low Agreeableness and Openness—reported higher levels of positive affect and self-esteem, and lower levels of negative affect, stress, loneliness, and social comparison. This illustrates an additional contribution of the present study: person-centered analyses offer complementary insights to traditional variable-centered approaches by detecting non-additive, configuration-specific associations between personality and well-being that may otherwise remain obscured. We propose that these two approaches be viewed as mutually informative—with personality profiles offering an additional lens through which to understand how combinations of traits jointly shape functioning in the South Korean context.

Moreover, our study identifies personality type as a novel factor that can predict well-being among South Koreans, suggesting its potential relevance for well-being interventions. Despite substantial economic growth and development, South Korea continues to struggle with persistently low happiness levels, with only 48% of South Koreans reporting being happy—significantly lower than the global average of 71% and the 72% reported in the United States⁸². This disparity underscores the importance of continued efforts to identify the underlying causes of unhappiness and develop effective interventions. Traditional large-scale well-being initiatives often fail to account for individual differences, limiting their effectiveness in addressing specific psychological needs. While personalized interventions offer more targeted solutions, their feasibility is often constrained by the necessity of individualized assessments, which may be inaccessible, costly, or logistically challenging for large-scale implementation.

We propose that our exploration of how each personality type relates to well-being and its predictors can inform the development of interventions that strike a balance between these two extremes, incorporating their respective advantages while mitigating drawbacks. Tailoring interventions to specific personality types can offer more personalized solutions than generic approaches, while remaining scalable for large populations¹⁰. For example, Introverted-Open type membership probability was positively associated with essentialist beliefs regarding happiness and negatively associated with positive emotions. Individuals with this personality profile may benefit from interventions that promote the belief that happiness is malleable and can be enhanced through intentional effort. In addition, while the Introverted-Reactive type was also characterized by low levels of positive affect, it was the one profile that exhibited a positive relationship with social comparison. This tendency is well-documented as one of the most detrimental to happiness, and thus encouraging a shift toward making less comparisons with others may present a potential entry point for intervention. By identifying the strengths and vulnerabilities associated with each personality type in relation to well-being, our study offers practical insights that can inform the development of well-being policies and initiatives in South Korea.

Limitations and future directions

While this study provides valuable insights into the distinct types of personality that exist in Korea as well as their corresponding attributes, we acknowledge certain limitations. First, the present study was conducted using a convenience sample of self-selected respondents, which may limit the generalizability of the findings to the broader population. In terms of the demographic distribution, our sample included a higher proportion of female and younger individuals than the general population. We note, however, that Kakao’s platform and messaging application is used by over 98% of individuals residing in Korea⁸³ and is the platform on which people spend the longest amount of time after YouTube⁸⁴, which allowed extensive reach into the general population and access to an exceptionally large sample to power our analyses. As such, utilizing this platform was a sensible point of departure for our investigation. Nonetheless, future research with a nationally representative sample of the Korean population is needed to further elucidate the personality types common in South Korea.

Second, the relationships among the personality types and the various outcome variables are correlational in nature, which limits assumptions regarding causality. To establish temporal precedence, we included only those participants who responded to the personality scales before the scales for the outcome variable. Given that personality is not a trait that can easily be manipulated, additional longitudinal studies, such as those utilizing latent change score models that assess whether changes in personality over the course of one period predict changes in the outcome variable over the course of a subsequent period may provide greater insight into the ways in which various configurations of personality traits influence psychological and behavioral outcomes.

A promising direction for future research is exploring the relationship between personality types and outcome variables across various life domains. For instance, in organizational contexts, given that companies frequently utilize personality and character assessments in recruitment⁸⁵, understanding how specific personality types influence job engagement and performance—along with potential moderators such as job type, task characteristics, and cultural factors—could yield valuable insights with practical applications. Likewise, personality also plays a pivotal role in attraction, relationship satisfaction, and long-term relationship stability⁸⁶. Investigating how different personality types contribute to relationship dynamics, conflict resolution, and emotional intimacy could be another productive avenue for future research.

Conclusion

The human tendency to categorize and distill complex information into manageable “types” is deeply rooted in our psyche, as reflected in the vast taxonomies that exist across countless domains—from living organisms⁸⁷ to love⁸⁸ and even scents⁸⁹. It is only natural that this impulse extends to understanding people and personalities, and here, we describe one such endeavor, building on previous research within a relatively under-examined cultural context. In doing so, we contribute to a deeper understanding of the varied manifestations of personality.