Personality Traits and Types in Relation to Career Success: An Empirical Comparison Using the Big Five

The purpose of this study is twofold: First, it discusses and derives personality types based on Big Five traits. Second, it compares their associations with career success. After deriving both a statistical and content-wise meaningful two-type solution referring to a resilient and a distressed profile, the explanatory value for both objective (i.e., promotions and income) and subjective career success (i.e., self-reported career success and career satisfaction) is tested for both traits and types. For objective career success, only traits appeared to be relevant predictors. For subjective career success, types appeared to have explanatory value as well, next to traits. This study concludes with a short dis- cussion of its implications and possible further research avenues.


INTRODUCTION
Personality significantly determines individual behaviour in the workplace (Penney, David, & Witt, 2011), and has been reported to be an important predictor of work and career success in both cross-sectional and longitudinal studies (see, e.g., Seibert & Kraimer, 2001;Wille, De Fruyt, & Feys, 2013). Two different operationalisations of the personality construct have been used in previous empirical research. First, the so-called trait approach typically focused on personality traits that were assumed to have predictive power, and-at the same time-convincingly attested to the importance of examine the unique (separate) contribution of traits for explaining relevant key outcome variables related to job and career success. Moreover, earlier empirical studies showed that combinations of Big Five traits (i.e., interactive effects) are instrumental in predicting key outcome variables (see, e.g., Ilies, Scott, & Judge, 2006;Jensen & Patel, 2011).
Personality types (profiles), on the other hand, can represent two-or k-way (k = 3 or more) interactions of certain trait levels, enabling all distinguished Big Five traits to simultaneously shape the interaction pattern (Asendorpf & Denissen, 2006). Most profile solutions (using the same sample and clustering method) vary from two to five types (solutions with an increasing number of types often giving more nuances on some types, leading to a modified type "label", but not on other types, preserving their label). Also across samples and/or clustering methods, some types may be labelled differently whereas other types are labelled identically. Table 1 summarizes previous scholarly work and portrays exemplary two to five type personality configurations. In empirical work examining job or career outcomes to date, mainly two profiles have been identified in research by De Fruyt (2002) and by Van der Wal et al. (2016).
As can be seen from Table 1, in all studies a so-called resilient (or well-adjusted) type is prevalent, accompanied by one or more different profiles having diverse labels. In terms of the Big Five, Resilients are generally characterised by relatively low scores on Neuroticism, as well as by high scores on all other traits. The relative size of the Resilient profile varies considerably across studies, and this variation cannot solely be attributed to a different number of profiles in the personality type configuration. As Herzberg and Roth (2006) argued, several factors may determine the size of a personality profile, including the number of profiles, sample size, sample composition, and the method by which respondents are assigned to profiles.

Procedure and Sample
This study was carried out among a variety of employees working for various Dutch plants of a worldwide multinational (> 50 countries) in the habitat and construction markets. Employees working in diverse middle up to higher-level positions were asked to fill in an e-survey. Using company-owned lists of e-mail addresses, an independent market research agency took care of all electronic communication as well as data storage handling and guaranteed anonymity and confidentiality. To prevent Common-Method Bias (CMB), several remedies have been applied (see, e.g., Podsakoff, MacKenzie, Lee, & Podsakoff, 2012    and scale anchors were different across the e-survey measures, and reversed items were included. Moreover, one of our main outcome variables, objective career success (see for more details Part 2), represents "factual data that are, in principle, verifiable from other sources" (Podsakoff & Organ, 1986, p. 532), comprising a data type which reduces CMB. Our final sample consisted of 293 employees (response rate was 91.8%) and included 242 males (82.6%) and 51 females (17.4%). Their mean age was 41 years (SD = 9.2), and their organisational tenure was, on average, 10.7 years (SD = 9.7). The employees' educational level comprised the following categories: (1) primary school (1.0%); (2) high school or equivalent (45.1%); (3) lower technical and vocational degree, typically earned before 18 years old (34.1%); (4) higher technical and vocational degree, typically earned after 18 years old (17.1%); and (5) academic degree, such as a bachelor's or a master's degree (2.7%). Commonly encountered job titles are: "adjunct director" (N = 11), "head of department" (N = 24), "plant manager" (N = 27), "head of product group" (N = 24), "commercial collaborator" (N = 50), "administrative collaborator" (N = 14), "collaborator in finance and accounting" (N = 25), "project leader" (N = 12), "show room manager" (N = 11), and "system (IT) manager" (N = 6).
measures Personality was measured using the 60-item short version of the validated Dutch translation (Hoekstra, Ormel, & De Fruyt, 1996) of the NEO Five-Factor instrument (Costa & McCrae, 1992). All items were scored using a five-point rating scale ranging from: (1) strongly disagree to (5) strongly agree. Cronbach's alpha values for the (12-item) subscales were .69 for Neuroticism, .67 for Extraversion, .64 for Openness to experience, .58 for Agreeableness and .72 for Conscientiousness, respectively. A recalculation of the reliabilities excluding observations with neutral (i.e. midrange) scores provided (overall) slightly better scores, with the final lowest value being .63, more specifically, for Agreeableness. Given these slightly improved Cronbach's alpha values and following the argumentation by Kruyen et al. (2012), we accepted the somewhat lower alpha levels (and, obviously, included the neutral response category).

analytical Strategy
Using respondents' mean item scores for the NEO subscales, a special form of "mixture modelling" namely "latent profile analysis" (see for details Oberski, 2011) was relied on. Unlike the traditional "hard clustering" techniques (e.g., k-means clustering), in which an individual is entirely assigned to one cluster (here: latent profile), latent profile analysis derives a probabilistic © 2018 The Authors. Applied Psychology published by John Wiley & Sons Ltd on behalf of International Association of Applied Psychology cluster solution (here: latent profile configuration). Obviously, all latent profile probabilities calculated for the same individual sum up to 1.00.
Latent profile analyses were run in MPlus version 7.11 (Muthén & Muthén, 1998 to derive alternative latent profile configurations, each configuration containing a different number (i.e., two, three, four, or five) of latent profiles. Statistically speaking, the choice for one of the alternative latent profile configurations, each of them representing an increasing number of personality types, is guided by a series of statistical comparisons of latent profile configurations containing k versus k -1 latent profiles (k = 2 first, then k = 3, k = 4, k = 5, etc.). Such a statistical comparison is enabled through the Vuong-Lo-Mendell-Rubin Likelihood Ratio Test (LRT) and its adjusted variant (i.e., the Adjusted LRT). In addition to statistical comparisons of alternative latent profile configurations, content-wise examination of each latent profile configuration guided our final choice. In line with personality (profile) descriptions as found in other studies using latent profile analysis (see Ferguson & Hull, 2018), each individual (employee) may be assigned to the most likely latent (personality) profile.

latent Profile Configurations of employees' Personalities
Using our data, successive statistical comparisons of latent profile configurations using the Vuong-Lo-Mendell-Rubin Likelihood Ratio Test (LRT) showed that: (a) two latent profiles are preferable to one latent profile (p = .01); (b) three latent profiles are not preferable to two latent profiles (p = .28); and (c) four latent profiles are not preferable to three latent profiles (p = .09). For more detailed results, we refer to the online supplement for this study.
The two resulting personality types can be recognized as the Resilients (with relatively low scores on neuroticism and relatively high scores on the other traits) and the Internalizer/Externalizer type (with relatively high scores on neuroticism and relatively low scores on the other traits, especially extraversion and agreeableness) of De Fruyt (2002). These types are also comparable with the two types identified by Van der Wal et al. (2016), who labelled the Resilients type Well-adjusted and the Internalizer/Externalizer type Distressed. For more detailed (statistical and content-wise) information of alternative typologies including three, four and five profiles, we refer to the online supplement.
For the purpose of cross-validation, we repeated our latent profile analyses using a second data set (e-survey data from Dutch teachers in higher education; the same personality measures were used). This cross-validation attested to a latent profile configuration with two latent profiles (for further details see the online supplement).
All in all, our (two) derived latent profiles are very similar to the personality typology as identified by De Fruyt (2002) and by Van der Wal et al. (2016). Therefore, we decided to base all subsequent analyses on the latent profiles identified. Objective and Subjective Career Success Objective career success was measured using three single items (Gattiker & Larwood, 1988). Objective hierarchical success was measured as the number of promotions, which was defined as "any increase in hierarchical level and/or any significant increase in job responsibilities or job scope" employees have experienced "since joining their current organization" [organization-specific objective hierarchical success (first item)] and "in their entire career" [overall objective hierarchical success (second item)]. Objective financial success was measured as the logarithm of current gross income (per month) (excluding bonuses, share options, etc.) (third item). Subjective career success was measured using five multi-item scales from Gattiker and Larwood (1986) comprising job success (8 items; Cronbach's alpha was .68, example item: "I am fully backed by management in the work I do"), interpersonal success (4 items; Cronbach's alpha was .62, example item: "I am respected by my peers"), hierarchical success (4 items; Cronbach's alpha was .61, example item: "I am pleased with the promotions I have received so far"), financial success (3 items; Cronbach's alpha was .70, example item: "I am receiving fair compensation compared to my peers") and a non-organisational component, so-called life success (4 items; Cronbach's alpha was .66, example item: "I am satisfied with my life overall"). All items were scored on a five-point rating scale ranging from: (1) disagree completely to (5) agree completely.
Career satisfaction was measured by means of the frequently used and thoroughly validated five-item Career Satisfaction Scale (Greenhaus, Parasuraman, & Wormley, 1990). All items were scored on a five-point scale ranging from: (5) strongly disagree to (1) strongly agree. An example item is: "I am satisfied with the progress I have made toward meeting my overall career goals". Numerous studies have attested to the high internal consistency of this scale; Cronbach's alpha values systematically exceeded .80 (Judge, Kammeyer-Mueller, & Bretz, 2004;Seibert & Kraimer, 2001). In our sample, Cronbach's alpha was .79.
Additionally, given their previously found effects on career success (see also Ng, Eby, Sorensen, & Feldman, 2005), the following socio-demographic control variables were included: age, gender, and highest educational qualification. Moreover, tenure with current employer (in years), being an important career-related variable, was included as well.
To analyse the impact of personality (both types and traits) on career success outcomes we relied on regression analyses.

Descriptives and Correlations for all Study Variables
First, for the entire sample, descriptive results for, and correlations between, all study variables are presented in Table 2.
Both the correlational data (i.e., correlations are all below .60) and Variance Inflation Factor (VIF) calculations (highest VIF value: 1.78) did not reveal a multi-collinearity issue. The (significant) correlations as found in Table 2 are in line with previous and meta-analytic findings on traits, except for the negative correlations between Extraversion and Openness on the one hand, and perceived financial success on the other hand. All control variables (age, gender, educational qualification, and organisational tenure) were found to have substantial associations with one another and at least some of the outcome variables. Therefore, we included all control variables in all our regression analyses.

Personality traits and types explaining Objective Career Success
Linear regression analyses were performed to examine the effects of traits and types on objective career success. For the results we refer to Table 3.
As shown in Table 3, traits appeared to have no significant association with the number of promotions within the employee's current organization. However, Agreeableness appeared to be significantly negatively associated with the number of promotions made during one's entire career. Furthermore, Neuroticism showed a significant negative association with (the logarithm of) income. For all other traits, no associations with objective career success outcomes were found. When examining our two-type personality configuration, the Resilient personality type did not show any significant association with any objective career success outcomes. Differences in explained variance (adjusted R 2 ) of the regression models for traits versus types, appeared to be small. Across all objective career success outcomes, the largest difference in the amount of explained variance between traits and types amounted up to 0.031, meaning 3.1 percentage points.

Personality traits and types explaining Subjective Career Success
In a next step, linear regression analyses were performed to examine the effects of traits and types on subjective career success outcomes. The results are shown in Table 4.
The results in Table 4 (as opposed to Table 3) show that the difference in adjusted R 2 between models based on traits versus types was more pronounced As regards the trait models, significant associations involved Conscientiousness (positive association) and Neuroticism (negative   association) on the one hand, and subjective career success outcomes, on the other hand. Furthermore, Extraversion showed a significantly positive association with three out of the six subjective career success outcomes (job, hierarchical, and life success). In addition, Extraversion and Openness showed one significantly negative association with financial success. In line with earlier studies, Agreeableness appeared not to be (significantly) associated with any of the subjective career success outcomes. As regards the so-called type models, we found that the Resilient type had significant positive associations with four out of the five subjective career success outcomes, excluding financial success. Additionally, no significant association was found between profiles and career satisfaction.

DiSCUSSiOn anD COnClUSiOn
The purpose of this study was two-fold: First, to derive a meaningful (Big Five) personality type configuration (for the sample at hand) and, second, to systematically assess the predictive value of the set of five Big Five traits versus personality types with respect to career success outcomes. Related to our first purpose, our Latent Profile Analysis using Big Five data led to a convincing configuration including two personality types (profiles) which resembled the personality types as previously identified by De Fruyt (2002) and by Van der Wal et al. (2016). Our sample does not support configurations including three personality types, which in some studies (see, e.g., Alessandri et al., 2014;Asendorpf et al., 2001), seem to be considered as being established both statistically and content-wise, that is, as regards the meaning of the types. It may well be that the most adequate configuration of personality types using Big Five data is dependent on "study context" (e.g., sample characteristics) as well as on "study method" (e.g., the analytical procedure used to identify personality types). In other words, a personality configuration, which seems most adequate for the study at hand, may not be fully comparable to personality configurations as identified in previous research as each and every profile that is found using different data sets is in fact unique. In this respect, trait activation theory (Tett & Burnett, 2003) suggests that situational factors cue the expression of traits at work and, as a result, the same trait may be expressed in different ways in different contexts, and/or across different jobs.
As regards our second purpose, our data supported previous notions in the literature that being a Resilient personality type can have positive consequences for one's career. Our regression analyses (predicting career outcomes based on Big Five traits) produced trait-outcome associations that are congruent with results as obtained in earlier (meta) analyses. All identified © 2018 The Authors. Applied Psychology published by John Wiley & Sons Ltd on behalf of International Association of Applied Psychology significant associations between traits and career success outcomes are congruent, except for the negative association between Openness and Extraversion on the one hand and perceived financial success on the other hand. It might be that especially people who score high on Openness may not be that easily satisfied with the financial rewards for their labour. In this respect, Ganzach and Pazy (2015) found that Openness is well associated with actual income of both men and women. Dissatisfaction with one's current income status might therefore help in being focused on obtaining a higher income later in time. An interpretation of the negative association between Extraversion and perceived financial success might be that people scoring high on this trait might be more inclined to actually report their dissatisfaction with their income through their score on this scale. The same results could account for being a Resilient type of personality [i.e. also implying high(er) scores for Openness and Extraversion] and (perceived) financial success in our study. However, we did not find such an effect.
In line with earlier findings (see, e.g., Asendorpf & Denissen, 2006), we identified a loss of explanatory power when using a type-based representation of personality compared to a trait-based representation, in particular in explaining subjective career success. Taking into account as well that the explanatory power of personality is known to be larger for subjective career outcomes than for objective career outcomes (see, e.g., Ng & Feldman, 2014), one may convincingly argue for the use of traits over the use of types in a (predictive) statistical analysis. In comparison to types, traits have both higher explanatory power and they reveal more explicit effects. That being said, when considering the practical relevance of personality at work, considering types may still be worthwhile. After all, the Resilient type is consistently returning in the literature as the preferable type in terms of, for example, social, health-related and work-related outcomes. In addition, personality types were found to be useful in different contexts in which interaction with people happens on a daily basis (see, e.g., Altmann et al., 2013;Roth & Von Collani, 2007). Taken all together, as personality is not rigid, yet rather changeable to a certain extent (see also Wille et al., 2013), it might be worth taking personality into account in career counselling or coaching, in order to stimulate employee behavioural patterns that are known for their positive effects on career success.
The present study has limitations as well. First, our (male-dominated) data have all been collected using self-reports for both personality and career success outcomes, herewith risking some common-method bias (see, e.g., Podsakoff et al., 2012). In future scholarly work, studies should therefore include both males and females in a more balanced representation, using multi-method measurements in a longitudinal design, to overcome © 2018 The Authors. Applied Psychology published by John Wiley & Sons Ltd on behalf of International Association of Applied Psychology the cross-sectional nature of this study as well (De Lange, Taris, Kompier, Houtman, & Bongers, 2004). Moreover, personality may not only predict career outcomes, yet also reciprocal effects might be possible (Sutin, Costa Jr, Miech, & Eaton, 2009). It would therefore be worthwhile to further study possible reversed causality issues, and developments in both personality and career outcomes over time (see also Ganzach & Pazy, 2015;Wille et al., 2013).
The current study did not include situational factors that may also moderate the personality-career success relationship. For example, high-quality leader-subordinate relationships (LMX; Leader-Member eXchange) and perceived organisational support are known to moderate the work value fit with career success (see, e.g., Erdogan et al., 2004). Similarly, these factors might be relevant for the personality-career success relationship.
Despite these limitations, this research adds to the literature in two important ways: the present study is the first that systematically compares a two-type solution for personality types, as regards their comparative and complementary value, to traits in explaining both objective and subjective career success. Second, this study acknowledges the complementary value of types based on the Big Five scores, in addition to the Big Five traits. Although from a statistical point of view, traits have the advantage of revealing the most (trait) specific information on what personality characteristics explain different aspects of career outcomes, in day-to-day reality, using profiles makes more sense. Moreover, as up until now, in the context of work and organisations, a more integrated indication of personality (types) is often applied by using unreliable and invalid methods (Vermeren, 2013), we call for more empirical work using latent profile analyses that is aimed at enhancing our insights into how personality and career success are interrelated.