The influence of personality traits on engagement in lifelong learning

ABSTRACT Today, adult individuals must be able to continuously learn and adapt to the rapid changes occurring in society. However, little is known about the individual characteristics, particularly personality traits, that make adults more likely to engage in learning activities. Moreover, few studies have longitudinally and objectively investigated the influence of personality on engagement in lifelong learning throughout working age. This study therefore used longitudinal data (15 years) to examine which personality traits predicted level and long-term changes in learning activities among 1329 Swedish adults aged 30–60. The results from growth curve modelling showed that over the follow-up period, novelty seeking and self-transcendence were both positively related to overall level of engagement in learning activities, although not to rate of change. Regarding specific activities, novelty seeking was related to higher levels of engagement in attending courses, taking on new education, and making occupational changes, while harm avoidance was negatively related to the likelihood of changing occupation. The results of this study underscore the importance of considering personality in relation to engagement in lifelong learning activities. Insights from this study can potentially increase the likelihood of finding methods to promote lifelong learning, which can be beneficial for educators, policymakers, and companies.


Introduction
Given the demographic changes happening worldwide, along with the fast development of knowledge, new innovations, and technological changes, it is crucial for individuals to constantly adapt to their evolving societies.In addition, and partly as a consequence of the rapid development of society, the labour market continuously puts new demands on the individual (see e.g.Kaplan, 2016;Tuckett & Field, 2016).However, it is not given that everyone has the will or motivation to continuously learn and change throughout life.Therefore, an increased understanding is required of the factors that motivate individuals to acquire new knowledge over their lifespan and working life.It has been suggested that factors such as age, educational level, socioeconomic status, parents' educational background, expectations of family, development opportunities, employment prospects, higher income, social benefits, health issues, and expectations from the workplace may influence the likelihood of engaging in learning activities (see e.g.Boudard & Rubenson, 2003;Gorard & Selwyn, 2005;Maclean et al., 2013;Macleod & Lambe, 2007;White, 2012).
However, relatively little is known about how individual characteristics, and personality traits in particular, may influence an adult's engagement (level and change) in learning activities by following behavioural patterns across time.Furthermore, as noted by Laible et al. (2020), most studies in this area have focused on predictors of work-related training activities rather than nonwork-related learning activities.A deeper understanding of how personality can influence engagement in learning activities over a working life, and perhaps also help the individual to maintain occupational and intellectual development, may be an important clue in understanding why some individuals are more likely than others to be able to adapt to societal change.

The concept of lifelong learning
In the Hamburg declaration (UNESCO, 1997) it is claimed that adult education, conceived within the framework of lifelong learning, should be looked upon as a lifelong process that will enable both individuals and society to handle future challenges.The question then is how to understand the concept of lifelong learning.The Commission of the European Communities (2001) has suggested that lifelong learning can be seen from a personal, social, civic, and employment perspective, and that it includes all learning activity undertaken over life to reflect how individuals gain knowledge, skills, and competencies.Lifelong learning has also been regarded as purposeful and ongoing acquisition of skills and knowledge throughout life (Longworth, 2006).Another relatively broad description used in the literature is that lifelong learning includes both formal and informal education that may help the individual to acquire knowledge that was lacking from previous formal education (Hus, 2011).Moreover, and of relevance to the current study, lifelong learning can be considered as the voluntary and self-motivated pursuit of knowledge, and as a type of learning that can be executed for both personal and professional reasons (Ates & Alsal, 2012).When considering such aspects of engagement in lifelong learning activities, and noting that to a large extent learning is driven by the interest and motivation of the individual, it seems plausible that engagement in learning may be differently influenced by personality characteristics or traits as compared to when engagement in activities is not fully driven by the individual's own interest or motivation.
Although many actions can be related to the acquisition of knowledge, there is no consensus in the literature regarding what specific activities may fit within the framework of lifelong learning, and so the concept may include any activity 'organised with the intention to improve an individual's knowledge, skills and competences' (Eurostat, 2016, p. 10).However, learning activities are often divided into formal, non-formal, and informal learning, with the first two of these referring to activities that are often organised, such as in a classroom.Formal learning activities often include an accepted degree or certified education.Non-formal education, on the other hand, often happens outside the formal educational curriculum but still within an organisational framework.Informal learning is often less structured, spontaneous, and unintentional.Such learning takes place outside school with, for instance, family and friends, or when reading literature.It may also take place spontaneously within the context of work (Marsick & Volpe, 1999).It should be noted, however, that others have suggested that two criteria must be fulfilled for activity to be defined as a learning activity, as compared to a non-learning activity.One is that the learning activity must be intentional with a purpose (not random), and organised in some way; the other is that it should include the transfer of information through, for instance, messages, ideas, knowledge, or strategies.Moreover, the intention of learning must be expressed before the activity starts (Eurostat, 2016).

The concept of personality
As previously noted, relatively few studies have investigated how individual characteristics such as personality traits influence engagement in learning activities over working life.Personality traits are indicative of relatively stable patterns of behaviours, thoughts, and feelings in an individual (Roberts et al., 2008).Within the trait perspective of personality, there exist several conceptual models.Common to all is that personality is described as a collection of traits that can be more or less adaptive.One popular model is the five-factor model of personality, which describes personality according to five core dimensions or traits (Digman, 1990;Goldberg, 1993).Another example is Cloninger's psychobiological model of personality (Cloninger et al., 1993), which guided research in the current study.Although Cloninger's model and the five-factor model share similarities, considering that they are both trait-based and aim to provide a comprehensive overview of personality, they also have fundamental differences.According to Cloninger's model, personality consists of four temperament traits (conceived to be genetically based and relatively stable) and three character traits (conceived to be shaped by experience and social learning and thus more susceptible to change).The notion is that personality can be understood from a perspective of interactions between temperament and character.Thus, whereas Cloninger's model distinguishes between inherited temperament traits and character traits shaped by the environment, the five-factor model does not distinguish between inherited and learned traits.Cloninger's model, measured by the Temperament and Character Inventory (TCI, Cloninger et al., 1993), also has a more complex structure than the five-factor model because it has a double emphasis on temperament and character.Despite these differences, there appears to be an overlap between the traits within the five-factor model and Cloninger's model (e.g.Aluja & Blanch, 2011;Capanna et al., 2012;MacDonald & Holland, 2002).For example, both Capanna et al. (2012) and MacDonald and Holland (2002) found significant correlations between one or more of the traits in Cloninger's model and the traits within the five-factor model.Nonetheless, both instruments are commonly used in psychological research (Feher & Vernon, 2021).

Personality and lifelong learning
Today, it is well-established that personality traits can predict educational level (Prevo & Ter Weel, 2015), academic performance (Andersen et al., 2020), job search activities (Caliendo et al., 2010) and occupational success (Furnham, 2018).Regarding engagement in adult learning activities, particularly in further education and training, personality traits have been found to influence attitudes towards learning (Fouarge et al., 2012).However, it is important to distinguish between attitudes and actual engagement in learning activities.
Regarding research on personality and actual engagement in learning activities over working life, the literature is relatively sparse.In one of the studies that constitute the exception, Offerhaus (2013) used longitudinal data to investigate the relationship between personality and participation in employment-related further education and training.The study, which collected data at three points over a nine-year period from a large sample, utilised the five-factor model of personality and locus of control as personality indicators.The results showed that individuals with high scores on the personality factor of openness, which indicates a tendency to engage in new experiences, curiosity, and an interest in exploring their inner life (Costa & McCrae, 1992), as well as those with high internal control beliefs, meaning a belief that they can influence their lives and outcomes, are more inclined to participate in further education and training related to employment.The personality factor conscientiousness, which describes an individual's ability to organise, sense of responsibility, diligence, and goal orientation, was also found to have an impact, especially in its most extreme forms where it decreased the chances of participation in employment-related further education and training.However, the other factors in the five factor model, namely agreeableness, extraversion, and neuroticism, had no impact on participation in employment-related further education and training.The results from this study are intriguing and offer insight into how personality influences engagement in lifelong learning.What still needs to be explored, though, is how personality influences adult learning when learning activities are not explicitly employment related.
In another of the few studies that have addressed the relationship between personality and lifelong learning from a longitudinal perspective, Laible et al. (2020) used objective measures from the German National Educational Panel study to examine whether the personality factors of the five-factor model could be related to engagement in lifelong learning.The participants were aged between 25 and 65, and were followed over a period of approximately three years.Overall, the results indicated that two personality factors, 'extraversion' and 'openness to experience', were significantly and positively related to the probability of engaging in lifelong learning, which included both non-formal and informal learning activities.It therefore seems that individuals who are more outgoing and energetic (extraversion) or open to unusual ideas and adventures (openness to experience) are also more likely to engage in learning activities during working life.For instance, it was found that openness to experience could predict engagement in competence development.It should nevertheless be stressed that openness had a somewhat stronger association with informal activities than with non-formal activities.The results also revealed that the personality factor 'agreeableness' (the capacity to put others' needs before one's own) was related to non-formal learning activities.Overall, these findings indicate that personality traits may need to be considered when investigating what makes an individual more likely to engage in learning activities over time.
However, to the best of the present authors' knowledge, no studies have objectively investigated how individual characteristics such as personality traits may influence an individual's engagement in learning activities over a longer time frame.Correspondingly, there is also a lack of studies that have investigated personality in relation to lifelong learning by using indicators of actual engagement in learning activities over time, rather than evaluating an individual's subjective willingness or likelihood to engage in lifelong learning.

The present study
In the current study, we included three activities related to lifelong learning that were undertaken voluntarily and by an individual's own initiative (in contrast to, for example, mandatory skills development within a workplace): 'attending courses', 'education', and 'occupational changes'.Courses often include a series of fixed classes to promote learning on a topic, whereas education, according to a Swedish definition, can refer to systematic teaching and training that provides knowledge and skills for a particular occupation and often provides some formal competence for this (Svensk ordbok, 2021).In this study, occupational change should be understood from a perspective of changing occupational category, and not be confused with a change of workplace.
As already noted, the fact that activities are undertaken voluntarily by the individual has been deemed important in relation to the motivation to engage in lifelong learning (Ates & Alsal, 2012).The activities included in this present study may support the individual to gain knowledge lacking from previous formal education (Hus, 2011).They also fit into the common definition that lifelong learning reflects purposeful learning that is undertaken throughout life, that aims to improve knowledge, skills, and competencies, and that can be seen from a personal, social, civic, or employment perspective (Commission of the European Communities, 2000).Two of them, attending courses and education, can be considered organised forms of learning activities, which can be either formal or non-formal.They can both be defined as activities with the purpose of learning, they both include transfer of information, and in both cases the intention of learning will most likely have been expressed before the activities started.The third learning activity included in this study, occupational change, is not a formal learning activity in the sense of being an organised form of learning, and it may not totally fulfil the criterion of learning as an activity that needs to be communicated before the activity starts (Eurostat, 2016).However, as noted above, changing occupation fits within the definition of lifelong learning as suggested by, for instance, Tuckett and Field (2016) and the Commission of the European Communities (2000).Moreover, and again as noted previously, both formal and informal learning activities can constitute a part of lifelong learning.
More substantiated research is needed to help us understand how to respond to the needs and fast changes taking place in today's society, in terms of rapid transitions, engagement in further education, and taking on new professional roles.In this study, we used unique longitudinal data to examine whether individual differences in personality could influence motivation to continuously learn over a working life.

Aims
The overall aim of the present study was to investigate how personality traits predict an adult's level of and long-term change in learning activities over their working life.More specifically, we aimed to investigate how personality traits novelty seeking, harm avoidance, reward dependence, persistence, self-directedness, cooperativeness, and self-transcendence were predictive of level and change for engagement in learning activities (courses, education, change of occupation).The study included data from baseline and from follow-ups at 5, 10, and 15 years.The potential influence of factors such as age, gender, years of education, occupational complexity, and indicators of socioeconomic status (SES) were also considered in the analyses.Results from this study will expand our understanding of how personality influences engagement in lifelong learning.By including learning activities reflective of various aspects of lifelong learning this study will contribute to a broader understanding of how personality affects adult learning.The longitudinal design and the use of a population-based sample will also offer unique insights into how the influence of personality on engagement in learning activities manifests both in the long and short term in an adult population still in working life.The outcome from this study may be of value for political decision making, companies, and educators when aiming to enable lifelong learning.The results could also potentially help individuals to understand the underlying mechanisms behind their own motivation (or lack thereof) to engage in various learning activities.

Study population
The data used in this study were drawn from the Betula prospective cohort study (Nilsson et al., 1997(Nilsson et al., , 2004;;Nyberg et al., 2020), which is a Swedish longitudinal study of ageing, memory, and health.The participants, selected by stratified (age, gender) random sampling from the population registry, have been tested at 5-year intervals since 1988, covering six test waves: 1988-1990, 1993-1995, 1998-2000, 2003-2005, 2008-2010, and 2013-2014.The Betula study has been shown to be representative in terms of population validity, considering factors such as education level, gender, marital status, income, and household size (Nilsson et al., 1997).Participation in this study includes extensive health assessment and cognitive testing.These assessments are separated into two sessions, with one focusing on health and the other mostly on cognitive functioning.At each test wave, these sessions are separated by about one week.Due to the number of participants to be tested, and the large amount of data to be collected from each participant, it has taken approximately two years to complete all testing in each test wave.However, in the Betula study, it has been carefully managed to ensure that each participant had approximately a 5-year interval between their individual testing sessions across the test waves.The test wave that was conducted between years 1993-1995 was used as the study baseline, Time 1 (T1), in this present study to allow the inclusion of repeated measures of data that were relevant to this study and that were equally measured over time.Three of the samples (sample 1, sample 2, and sample 3), each including different randomly selected individuals from the population, contribute with longitudinal data and hence were used in the present study (other samples have been included only in one test wave, primarily to control for potential learning and cohort effects).Sample 1 contributed with data over four test waves (1993-1995, 1998-2000, 2003-2005, and 2008-2010), sample 2 over two test waves (1993-1995, and 1998-2000), and sample 3 over four test waves (1993-1995, 1998-2000, 2003-2005, and 2008-2010).Data collected between 2013 and 2014 were not included in the present study because the recruitment procedure was different, also resulting in less available data on the key variables of interest.Thus, the data used in the present study included information from four measurement points: T1 (1993-1995), T2 (1998-2000), T3 (2003-2005), and T4 (2008-2010), equivalent to a follow-up period of 15 years.We included participants who were aged between 35 and 60 at the study baseline and set the end age of follow-up at 65 years (the common retirement age in Sweden) in order to be able to follow participants when they were still of working age.

Participants
At baseline (T1), the study included 1600 participants aged 35-60 years from samples 1 (n = 500), 2 (n = 600), and 3 (n = 500).Of the 1343 who responded to the questionnaire (TCI) about their personality, 14 had to be excluded due to missing data on engagement in learning activities at baseline, and so the final baseline sample consisted of 1329 participants (sample 1 = 394, sample 2 = 486, sample 3 = 449).Data were available for 1081 participants (sample 1 = 363, sample 2 = 298, sample 3 = 420) at the 5-year follow-up, for 584 participants (sample 1 = 268, sample 3 = 316) at the 10-year follow-up, and for 362 participants (sample 1 = 167, sample 3 = 195) at the 15-year follow-up.At baseline, the study sample had a mean age of 49.29 years (SD = 7.68, range 35-60 years) and comprised 52.4% women.It should be noted that the significant differences in sample size over time were largely due to the study design (i.e.Sample 2 did not contribute data for the entire follow-up period) and exclusion criteria (such as retirement age).The main reason for including all samples with longitudinal data was to enhance statistical power and reliability of the outcomes.

Personality
The Temperament and Character Inventory (TCI; Cloninger et al., 1993) is the instrument used in the Betula study for personality assessment.This paper-and-pencil-test of TCI is a validated instrument that includes 238 statements (true/false), and we coded and treated the data according to standard procedures.In this inventory, personality is explored based on seven dimensions (subscales) which can further be summarised to represent one of two overarching factors: temperament (novelty seeking, harm avoidance, reward dependence, persistence) or character (self-directedness, cooperativeness, self-transcendence).Novelty seeking (based on 40 items) is a heritable temperament trait that reflects a bias towards activation or initiation of actions and towards frequently exploring new things as a response to novelty.This personality dimension also includes impulsive decision making, quick loss of temper, avoidance of frustration, and excessive bias to approach cues for potential rewards.Harm avoidance (based on 35 items) reflects the individual's bias towards inhibiting behaviours, pessimistic worry about future problems, doubtfulness, fear of uncertainty, shyness, and being easily fatigued.Reward dependence (based on 24 items) reflects the tendency to preserve and continue to engage in ongoing behaviours associated with reward or relief of punishment, to react strongly to reinforcement and the approval of others, and to depend on social attachment.Persistence (based on 8 items) is the propensity for perseverance even in the presence of fatigue or frustration.Among the character traits, self-directedness (based on 44 items) is related to how responsible, dutiful, and self-accepting the individual is.Cooperativeness (based on 42 items) concerns the extent of agreeableness in relations with others, and is related to the degree to which the individual identifies with and accepts others.Self-transcendence (based on 33 items) reflects personal boundaries; for instance, how the individual considers themself as a part of the universe.

Lifelong learning
Indicators of lifelong learning including attending courses, engagement in new education, and occupational change were measured with the Life Event Inventory (Perris, 1984), which was administered at each test wave (T2-T5) used in the present study.The participants indicated whether any of the events had occurred during the last five years (i.e. between test waves), and if so, whether they felt that this event was something that could be influenced by the individual.Thus, participants only received a score (yes = 1, no = 0) for engagement in an activity that they themselves had actively influenced.

Confounders
Potential confounders included in the analyses were age, gender (1 = female, 0 = male), years of education, and occupational complexity (for a detailed description, see Sörman et al., 2021).Two self-reported living conditions were included as indicators of SES: number of rooms (excluding the kitchen), and number of persons in the household.

Statistical analysis
Descriptive statistics (means, standard deviations, skewness, kurtosis) were calculated for all the variables included in this study, and correlational analyses were employed between variables.The percentage distribution for engagement in learning activities over the test waves was also calculated.Next, latent growth curve modelling was executed.Models were analysed with SPSS-28 and AMOS-28 using maximum likelihood estimation, and the significance level was set to 0.05.First, an unconditional model (i.e.without predictor variables) including four time points of lifelong learning data (baseline, 5 years, 10 years, and 15 years) was tested to gain information about mean changes, variance in level (i.e. the intercept) and change (i.e. the slope), and the extent to which intercept and slope were associated when predictor variables were not considered.Three measures (attending courses, new education, and occupational change) were used as indicators of a latent lifelong learning construct across time.After this, the conditional model was tested.In this second-order latent growth model, the intercept and slope were conditioned by predictor variables, and personality factors and confounders were thus included in the model.Both growth curve models (unconditional and conditional) were evaluated with the χ2 statistic divided by its degrees of freedom (df), the Tucker -Lewis index (TLI), the comparative fit index (CFI), and the root mean square error of approximation (RMSEA) with a 90% confidence interval.Suggested cut-off criteria according to Kline (2010) and Hooper et al. (2008) were used to estimate adequate model fit (TLI > 0.90, CFI > 0.90, RMSEA < 0.08) and good model fit (TLI >0.95, CFI > 0.95, RMSEA < 0.06).

Results
Participant characteristics, measures of probability distribution, and intercorrelations between study variables are given in Table 1, and the percentage distribution for engagement in learning activities over the test waves is given in Table 2.
Most of the variables included in this study were normally distributed.The only variables with a value that might be slightly above the upper threshold suggested in the literature were learning activity at T3 and learning activity at T4; however, this was extremely marginal, with skewness  .values of 2.03 and 2.37, respectively.One suggested upper threshold for skewness is 2.0 (see e.g.Finney & DiStefano, 2006).
Although participants got increasingly older and the sample size became significantly smaller over time, sample characteristics remained relatively stable over the test waves.There was, however, a small tendency that those that were more highly educated, had higher level of occupational complexity, and more persons in the household to a higher extent contributed with data over time.This is most likely reflective of cohort effects (i.e.participants younger at the start of the project had more years of education, higher occupational complexity, and more people lived in their household).Sample characteristics at each wave of the study can be found in the supplementary material, Table S1.
Results from the correlational analyses are presented in Table 1.Overall, age showed a significant correlation with learning activities over all test waves ( r range: −.22 to −.32), indicating that older individuals were less likely to engage in such factors.More years of education also had a significant positive correlation with learning activities in all test waves ( r range = .14to .28).Being female was positively related to learning activities in test waves one (r = .06)and three (r = .10).Occupational complexity was significantly associated with learning activities at the first (r = .14)and second test wave (r = .10).Regarding SES, number of rooms in household was related to learning activities at the first two test waves (r = .07and r = .11),whereas number of persons in household was significantly and positively related to learning activities over the first three test waves ( r range: .14 to .19)Among the personality factors, novelty seeking was significantly and positively associated with learning activities over the first three test waves ( r range: .14 to .21),harm avoidance was negatively associated with learning activities over the first two test waves (r = −.07 and r =−.08), and reward dependence was positively correlated with learning activity at the second test wave (r = .07).Persistence was associated with learning activity at the first (r = .09)and second (r = .08)test waves, and self-transcendence was positively related to learning activities at the first three test waves ( r range: .07 to .09).
Results from the conditional growth curve model revealed good model fit according to all fit indices: TLI (= .958),CFI (= .979),RMSEA (= .020;90% CI [.01-.02]), and chi-square/df ( = 1.51, p = <.001).Standardised and unstandardised estimates for the predictor variables on lifelong learning (intercept and slope) in the model are presented in Table 3 along with standard errors, critical ratios, and p values.Among the personality factors, novelty seeking was positively associated with the intercept for lifelong learning, but also had a borderline (p = .070)negative association with slope.These findings suggest that individuals with higher levels of novelty seeking are more likely to engage in learning activities, but the trend in the data may indicate that this effect does not hold longitudinally, as shown by less interest in lifelong learning over time.Self-transcendence was also positively and significantly related to the intercept, suggesting that higher levels of this personality trait were associated with higher levels of engagement in learning activities, but it was not related to change over time.None of the other personality traits were associated with intercept or slope in lifelong learning.
Among the other predictor variables, age was significantly and negatively associated with intercept, suggesting that higher age is related to lower levels of engagement in learning activities.However, higher age was not related to slope.Thus, even if there is a negative change in engagement in lifelong learning over time (as indicated by changes in mean levels), which consequently is part of the ageing trajectory, age does not predict differences in change between individuals over time.A higher number of years of education was positively related to level of engagement in learning activities, but was not related to change.Finally, none of the SES indicators, gender, or occupational complexity were associated with intercept or slope of lifelong learning.It should be noted that gender had a borderline relationship (p = .065)with intercept, indicating that women on average might be more likely to engage in learning activities.

Additional analyses
The number of individuals in the sample who had engaged in a specific activity (courses, education, occupational change) at each test wave was overall too low to be able to reliably investigate effects of personality on each activity at each measurement point.This issue also affected the possibility of investigating what can cause change in each learning activity over time.For this reason, it was a correct choice to examine the total amount of learning activities at each measurement point via a latent construct.However, additional linear regression analyses were executed to gain some clarity regarding the role of personality traits as predictors of engagement in different learning activities.In these analyses, indexes were constructed for each learning activity, based on the number of times that the individual had engaged in each learning activity over the entire follow-up period divided by the number of measurement points at which that the participant had contributed data.The use of indexes to represent the total follow-up period cannot provide information about the extent to which independent variables can predict change in learning activities over time.However, it does provide some information about the amount of engagement in specific learning activities when the total follow-up period is considered.The results from these additional analyses are presented in Table 4.
Novelty seeking was significantly and positively related to all learning activities, confirming the findings from the growth model.Harm avoidance was related to being less willing to change occupation.Self-transcendence, which was related to learning in the growth model, did not have enough impact to significantly influence any individual learning activity, although it should be noted that it had a borderline relationship with occupational change (p = .053).Among the confounders, age was negatively related and number of years of education was positively related to all learning activities, again confirming the findings from the growth model.However, analyses of gender revealed a somewhat different pattern; being female was related to being more engaged in courses, but also to being less willing to change occupation.Finally, higher levels of occupational complexity were related to higher levels of attending courses, and the SES factor 'number of rooms in household' was related to less engagement in new education.

Summary of key findings
This study examined level and long-term change in lifelong learning in relation to personality factors in a sample of working-age individuals.The results indicate that two personality dimensions in particular were related to level of engagement in learning activities: novelty seeking and self-transcendence.However, these factors were not significantly related to change in lifelong learning over the 15-year follow-up period, although it should be noted that novelty seeking was borderline significant for slope in the growth curve model.Additional analyses, investigating the influence of personality traits on level of involvement in specific activities taking the total follow-up period into account, revealed that novelty seeking was related to higher engagement in all activities included in this study (courses, education, occupational change).Harm avoidance was negatively related to the likelihood of changing occupation, and self-transcendence, which was positively related to learning activities in the latent growth model, was borderline significant for change of occupation.No other personality trait (reward dependence, persistence, self-directness, cooperativeness) was associated with lifelong learning as measured in this study.

Interpreting key findings of this study: personality and lifelong learning
The finding that novelty seeking was related to overall engagement in learning activities, as well as to participation in specific learning activities, seems reasonable.This trait is very much heritable, and refers to a tendency towards sensation seeking and the pursuit of new experiences (Cloninger et al., 1993).Results from previous studies have, for instance, shown that novelty seeking is associated with the activation of dopaminergic pathways (Wiesbeck et al., 1995).The motivation to engage in new activities may therefore be supported and modulated by the fact that novel stimuli can excite dopamine neurons and trigger brain regions with dopaminergic input (V.D. Costa et al., 2014).Although it has been found that individuals with high novelty seeking may be at higher risk of negative health behaviours such as drug and alcohol addiction (Grucza et al., 2006), the results from this study conversely suggest that this personality trait may be adaptive with regard to the tendency and willingness to engage in new learning activities and adapt to evolving societies.On the other hand, we cannot rule out the possibility that individuals with high novelty seeking will also have difficulty engaging in specific activities over longer time periods.It may be that such individuals constantly seek novelty, and thus tend to leave 'old' activities behind in the search for new activities that will feed their curiosity or help them avoid boredom (Liang et al., 2020).In the present study, we were not able to investigate change in any specific activities.However, novelty seeking was borderline significant for change (negative slope) in overall engagement in lifelong learning.More knowledge is therefore needed to understand the of role of novelty seeking in relation to both level and change in lifelong learning.Because novelty seeking seems to be an important factor, it might be a great challenge for any society, company or organisation to understand how individuals with low levels of novelty seeking can be triggered to engage in learning activities when needed.Correspondingly, it may also be important to understand how to develop strategies for both fostering and maintaining the feeling of novelty for individuals with high levels of this personality trait, in order to maintain their curiosity and interest and reduce their feelings of boredom.
Finally, in terms of novelty-seeking, it is important to note that the results of this study may be in alignment with those identified by Offerhaus (2013) who found that individuals with higher levels of openness, a personality factor derived from the five-factor model of personality, were more likely to participate in employment-related further education and training.Openness, as previously mentioned, is associated with curiosity and a tendency for engaging in new experiences (Costa & McCrae, 1992), which to a certain degree conceptually overlaps with TCI's novelty-seeking.
Previous research has also found a significant correlation between openness and novelty-seeking (Aluja & Blanch, 2011;Capanna et al., 2012;MacDonald & Holland, 2002).Furthermore, this current study found a relationship between self-transcendence, defined by Cloninger et al. (1993) as related to spirituality and an individual's perception of themselves as part of the universe, and overall engagement in lifelong learning.This factor has also been shown to significantly overlap with openness (Aluja & Blanch, 2011;MacDonald & Holland, 2002), probably because openness as well reflects an individual's tendency to explore their inner life (Costa & McCrae, 1992), suggesting a conceptual overlap between these factors.Therefore, the overall findings of this study are somewhat consistent with those reported by Offerhaus (2013).
However, it may still be somewhat challenging to interpret the relation between lifelong learning and self-transcendence.Speculatively, however, it may be that such individuals are less anxious when it comes to trying new learning activities.Research has shown that self-transcendence is related to more positive emotions, optimism, and higher self-esteem, and to lower levels of depression and neuroticism (Reischer et al., 2020).Such aspects may have a positive impact on both the probability of engaging in new learning activities and the expectations of the outcome of such engagement.Future studies will need to confirm whether self-transcendence is a personality trait that may be related to lifelong learning.If so, investigating how and why this trait might influence lifelong learning could be fruitful, and could potentially inform methods to promote lifelong learning.
Finally, the latent growth curve analyses did not show a relation between harm avoidance and overall engagement in learning activities.However, additional analyses revealed that this trait was negatively related to the probability of changing occupation.Individuals with high levels of worry, pessimism, and doubt thus seem to avoid such a change.Changing occupation may, depending on the circumstances, be important both for individual development and for adaptation to societal change, but it may also be more challenging on an individual level in comparison to the other learning activities included in this study.The finding that harm avoidance may play a role in occupational change is important knowledge; for instance, when it comes to supporting a career change in cases where this may be necessary and/or fruitful.It is possible that individuals with high levels of harm avoidance need more encouragement and support to give them the courage and/or motivation to be able to change occupation.
None of the other personality factors in this study were related to level of engagement in lifelong learning.However, it is still possible that these traits might be related to learning activities other than those included in this study, and/or to engagement in activities not actively initiated by the individual.As an example, individuals highly driven by reward or relief of punishment, who react strongly to reinforcement and approval from others, and who are dependent on social connection (i.e.reward-dependence), as well as individuals with high levels of persistence in the presence of fatigue or frustration (i.e.persistence), may engage to a greater extent in learning activities that are not primarily driven by intrinsic motivation and curiosity; for instance, in situations when learning activities have been initiated by an employer and/or when there is a social expectation of the individual.Similarly, it may be that individuals who are responsible, dutiful, and self-accepting (i.e.self-directness) and/or who identify with and accept others (i.e.cooperativeness) have other motives for engagement in learning activities.Future studies should therefore investigate the impact of these personality traits on other features of lifelong learning, and on engagement in activities not primarily undertaken due to the individual's own interest or motivation.
Overall, the results from this study show that novelty seeking is an important factor to consider in relation to lifelong learning.One factor that may be important in the context of novelty seeking and the curiosity to learn is that previous positive learning experiences influence learning selfefficacy, motivation, and the likelihood of engaging in new learning activities (Sanders et al., 2015).Conversely, previous negative experiences of learning can decrease an individual's likelihood of engaging in new learning activities (Illeris, 2006;Sanders et al., 2015.New experiences of teachers, pedagogues, and working methods may therefore be important to (re-)create curiosity and desire for learning, and perhaps even more so for individuals who have negative experiences of learning.Considering also the predictive power of number of years in school for engagement in future learning, as confirmed in this study, and the fact that positive learning experiences influence the willingness to learn new things, it is likely that the school environment at young ages, where some of the desire and curiosity to learn is developed, not only has a great impact on the number of years in education, but may also increase the likelihood of engagement in lifelong learning.

Interpreting additional findings of this study
Among the confounders included in the analyses, a greater number of years of education was significantly and positively related to lifelong learning, and older individuals were found to be less likely to engage in learning activities.Both these findings are in agreement with previous research (see e.g.Macleod & Lambe, 2007;White, 2012).That individuals tend to be less inclined to engage in new learning activities as they age is, of course, part of the challenge when it comes to motivating individuals to lifelong learning.Age was however not related to slope (i.e.change).Thus, individuals who are already older and engaged in learning activities may not necessarily become less engaged over time.Again, the great challenge seems to lie in creating a curiosity for learning early in life, as this then seems to follow the individual through their lifetime.
The results also revealed that in comparison to men, women showed higher engagement in taking courses over their working lives, and decreased probability of changing occupation.One possible explanation for these findings could be that women are overrepresented in certain occupations, such as the caring professions, which may give less opportunity to take courses.However, the data available for this study did not allow us to confirm or disprove this.Similarly, the data at hand do not show whether working in certain professions gives less opportunity to change occupation.
Higher levels of occupational complexity were related to higher levels of attending courses; this could be because occupations with high complexity are usually correlated with higher education, so the individual may already be interested in competence development and learning new skills.Finally, the results also showed that having more rooms in one's household (i.e.indicative of higher socioeconomic status) was related to less engagement in new educations.This result presents a challenge in terms of interpretation.One potential explanation could be that individuals with more rooms in their household have already attained a socioeconomic standing that reduces their necessity for pursuing further education; this reasoning, however, is highly speculative.

Strengths and limitations
This study has several strengths.The participants were selected through stratified random sampling, and it has previously been shown that the Betula sample has good population validity based on aspects such as education, income, gender, marital status, and number of persons in the household (Nilsson et al., 1997).Other strengths are the broad follow-up period and the use of a relatively large sample size.Thus, the results from this study can be considered as robust and reliable.However, some limitations should be acknowledged.One is the lack of information about motives that may influence both engagement in learning activities and the relationship between personality and lifelong learning.Previous research has shown that factors such as better employment prospects, higher income, social benefits, and health issues may play a role (see e.g.Maclean et al., 2013).Other factors may also cause individuals to make changes in their lives; for instance, individuals may change occupation due to layoffs and/or closures.Unfortunately, such information is not available in the Betula database in relation to learning activities.Another limitation of this study is the number of learning activities included.We have used the information available in the database, but we are aware that this information is not fully comprehensive.
Another factor that can limit and affect the opportunity to engage in learning activities is the number of children in a household.The number of children is a potential confounder for which, unfortunately, there is no specific information available in the Betula database.We used 'number of people in household' as a covariate in the analyses but are aware that this does not fully capture this aspect.However, it should be noted that in Sweden the majority of households do not practise generational living (Joelsson & Ekman Ladru, 2022), and it is often the children who increase the number of people in the household.Future research should, however, consider the family situation by using more precise measures regarding the number of children in the household.
As previously noted, the total sample size significantly decreased over time.Even though the sample characteristics remained relatively stable over time, and attrition could to some extent be explained by the study design and exclusion criteria, there is always some risk of bias in the results due to changes in the sample size over time.This should be carefully considered when interpreting our findings.
Finally, the last data collection in this longitudinal study was executed between 2008 and 2010.Since then, society has undergone substantial changes, which have brought both new professional challenges and contributed to new forms of learning.For instance, the rapid digital transition that occurred during the Covid pandemic still influences society and learning activities today.We cannot rule out that the impact of personality on engagement in new learning activities may demonstrate a pattern different from those found in this study.However, it is also possible that personality traits such as novelty seeking and harm avoidance may have an even greater influence as society's demands for new forms of learning increase.For instance, even though it may not be immediately linked to learning activity, it has been shown that novelty seeking does have a positive effect on the use of modern technology, which has seen significant growth since this study had its last data collection, for example in terms of the use of Mobile Instant Messaging (Cruz-Cárdenas et al., 2021).Future studies should, however, further investigate the role of the personality traits included in this study in engagement in both other and new forms of learning activities, by following individuals over time.

Conclusions and future directions
The present study demonstrates the importance of considering personality in relation to engagement in lifelong learning.Novelty seeking seems to be a factor that plays a key role, but selftranscendence and harm avoidance also appear to be influential.However, more studies are needed to confirm these findings, and there is a need to increase the understanding of how and why such traits may influence engagement in learning activities.Such knowledge could potentially increase the likelihood of finding methods to promote lifelong learning.
In addition, there is a need to investigate the impact of personality on learning activities other than those included in this study, and other personality dimensions should be considered too.Future studies should also investigate personality in relation to the motives and/or the cause for engagement in learning activities.Finally, none of the variables included in this study were predictive of change in learning.Future studies should therefore aim to identify factors that may predict change, in order to increase the understanding of what makes individuals more (or less) likely to engage in lifelong learning over time.

Table 2 .
Percentage distribution for engagement in learning activities over the 15-year follow-up period including four test waves (T1-T4), each separated by five years.

Table 3 .
Summary of the results from the conditional latent growth model including predictors of level and change in lifelong learning.

Table 4 .
Results from linear regression analyses with different learning activities as dependent variables.Involvement in each learning activity is calculated and weighted based on the amount considering the entire follow-up period (up to 15 years).