1 Introduction
Within psychology, the view of personality as stable throughout life is rapidly changing to one where traits react fluidly to life circumstances (Caspi
1998; Caspi and Bem
1990; Roberts et al.
2008). Despite a mass of evidence that suggests an individual’s personality changes across the complete lifespan (Lucas and Donnellan
2011; Roberts et al.
2006; Specht et al.
2011), core personality traits are still generally considered ‘relatively enduring’, particularly in disciplines outside psychology. As a result of this, the use of personality change measures in well-being research has been limited, with most studies utilising personality measures at one time point to predict well-being outcomes (Boyce and Wood
2011; deBeurs et al.
2005; DeNeve and Cooper
1998; Friedman et al.
2010; Steel et al.
2008) and only a few studies exploring the effect of personality change on well-being.
Studies that have investigated personality change have found an association with subjective well-being measures such as life satisfaction (Boyce et al.
2013; Heller et al.
2007; Specht et al.
2013; vanAken et al.
2006), self-rated health (Berg and Johansson
2014; Magee et al.
2013; Turiano et al.
2012), self-efficacy (Hutteman et al.
2014), psychological turning points (Allemand et al.
2010; Sutin et al.
2010) as well as physical and mental health outcomes (Human et al.
2013; Mroczek and Spiro
2007). Evidence for an association between well-being and personality has been taken to support the social investment perspective on personality development (Roberts and Wood
2006; Roberts et al.
2005) which suggests that committing and successfully adapting to social roles such as marriage and work drives personality development. Whilst the social investment theory considers the effect of societal-determined expectations and goals on personality change, it does not address the importance of striving for authentic, self-concordant goals for personality change. Such a relation forms the basis of an alternative explanation for personality change, proposed by existential and humanistic theories which have not previously been introduced into the contemporary empirical literature on the malleability of personality.
Taken together, the existential and humanistic theories propose that each individual has the freedom and responsibility to transcend the meaninglessness of their existence. Personality change is thought to occur when the individual confronts meaningless in life and has to decide for themselves how to shape their life.
1 If the individual chooses to strive towards fulfilment, personality is likely to develop in potentially positive ways (i.e., perhaps becoming more open to opportunities or more extraverted) because the individual recognises their capacity to choose their own future and is able to take full advantage of opportunities to find meaning to their existence. Alternatively, if the individual is consumed with feelings of despair and fails to engage with themselves and the world around them to achieve their full potential, this may result in changes in the opposite direction (i.e., becoming less open and more introverted). Associating personality change with changes in such ways of functioning would be part of a theoretical movement from seeing personality change as a biological maturation or social investment process towards seeing such change as part of a holistic development of the person in ways that are right for the individual (existential well-being) (Deci and Ryan
1985,
2000; Joseph and Linley
2005).
The existential-humanistic theory of personality change can be tested using a measure of psychological well-being (PWB). Waterman (
1984,
1993) defines PWB as concerned with the feelings associated with an individual’s strive to grow and fully develop oneself amid life challenges. PWB encompasses an individual’s perception of engagement with the self, environment and others (Keyes et al.
2002; Ryan and Deci
2001), thus capturing existential well-being. In terms of measurement, Ryff (
1989), Ryff and Keyes (
1995) operationalize PWB as comprising
autonomy (the extent to which one is self-determining and independent),
environmental mastery (competence in managing the environment and presented opportunities),
personal growth (possessing feelings of continued developments),
positive relations (having strong social ties),
purpose in life (having goals in life or a sense of directedness) and
self-
acceptance (possessing a positive attitude toward the self) (Ryff and Keyes
1995).
In this paper we report on a study that seeks to better understand the relationship between personality change and well-being change through linking changes in personality to an individual’s existential engagement with the world, as represented by changes in PWB. We additionally aim to assess the use of personality change measures as well-being indicators and targets for intervention, through (a) quantifying the size of personality change relative to socioeconomic metrics commonly used in well-being research and (b) comparing the predictive value of changes in personality and socioeconomic factors on changes in PWB. In order to quantify an effect size as large or small, direct comparisons with effect sizes of other variables of interest must be made (Cohen
1992; Glass et al.
1981). Recently, Boyce et al. (
2013) have been the first to compare the magnitude of personality change with that of socioeconomic indicators that are widely considered changeable (e.g. income, marital employment status). They find that personality changes at least as much as socioeconomic factors across a wide age range. Here, we specifically examine whether personality changes more than socioeconomic factors during midlife. Furthermore, we examine how personality change relates to changes in other well-being measures such as depression, hostility, and life satisfaction in order to assess the importance of personality change for PWB over other well-being measures.
3 Results
Stability of personality scores across time were as follows: 0.68 for neuroticism, 0.74 for extraversion, 0.71 for openness, 0.61 for agreeableness, 0.62 for conscientiousness. These coefficients were comparable to those found in similar study by Roberts and delVecchio (
2000). Our stability coefficients represent the correlation between the mean personality score at Time 1 and Time 2 and therefore indicate that personality measures across the sample are generally stable over time. However, the high stability coefficients do not preclude the possibility of personality changes within an individual (Ozer
1986), which is the focus of our study. The less than perfect (i.e.
r < 1.0) stability across the two time points further suggest there may be are individual-level changes in personality. In order to explore this, we examined the number of individuals who showed reliable change (i.e. true change not due to measurement error) in personality measures from Time 1 to Time 2 and the magnitude of this change. Table
4 illustrates the percentage of individuals who experienced true change in personality measures and the lowest and highest magnitude of change (in standard deviation) for these individuals. Our results in Table
4 shows that a proportion of the sample experience change in personality of considerable magnitude. Through the ‘between-to-within variation ratio’ in Table
1, we also show that even at midlife, personality changes as much as other indicators that have traditionally been used to predict human outcomes. Our between-to-within stability ratios were lower for personality traits than for the different categories of educational achievement and marital status in our sample, indicating that an individual’s personality is more likely to change from Time 1 to Time 2 than their educational achievement or marital status.
Table 4
Individual differences in personality traits
Neuroticism | 16.7 | 7.3 | 76.0 | 1.0 | 3.8 |
Extraversion | 8.4 | 6.3 | 85.3 | 1.0 | 3.2 |
Openness | 4.6 | 3.0 | 92.4 | 1.0 | 4.0 |
Agreeableness | 10.0 | 11.1 | 78.9 | 0.8 | 4.7 |
Conscientiousness | 11.5 | 6.9 | 81.6 | 0.8 | 4.7 |
Table
2 demonstrates that personality change relates to significant changes in well-being. Models 2a–2f in Table
2 show that personality alone explained 3–7 times more variation (indicated by R-squared values) in the PWB subscales in our sample than socioeconomic and health indicators together (Models 1a–1f). For example, in Model 2a in Table
2, personality change explained 3 % of the variation in an individual’s level of autonomy, while changes in socioeconomic variables (Model 1a in Table
3) explained only 1 % of the within-person variation. Similarly, personality change explained 7 % of the within-person variation in environmental mastery (Model 2b), 5 % of the within-person variation in personal growth (Model 2c), positive relations (Model 2d) and purpose in life (Model 2e), and 6 % of the within-person variation in self-acceptance (Model 2f) over the 10 years period. Socioeconomic variables together explained only 1 % of the within-person variation in each of the PWB subscales (Models 1a, 1c–1f). Our R squared values are similar to those found in similar models which estimate variation within individuals (Boyce et al.
2013; Ferrer-i-Carbonell and Frijters
2004) and are used here to highlight the stronger relationship between well-being change and change in personality measures compared to change in socioeconomic factors. Furthermore, Models 1a–1f in Table
2 indicates that change in log-transformed income and unemployment status were not significant predictors of change in any of the PWB subscales while Models 2a–2f in Table
2 shows that personality change was significantly associated with change in each of the PWB subscales. For example, in the case of purpose in life (Model 2e), a one standard deviation increase in extroversion, openness, agreeableness and conscientiousness was significantly associated with a 0.12, 0.08, 0.11 and 0.11 standard deviation increase in purpose in life respectively, and a one unit increase in neuroticism was significantly associated with a 0.08 standard deviation decrease in purpose in life. In Models 3a–3f, we further show that personality change remained significantly associated with changes in all PWB scales (effect sizes ranging from 0.04 to 0.15 standard deviations) even after controlling for changes in socioeconomic variables, whereas changes in socioeconomic indicators were only significantly associated with changes in positive relations (becoming partly retired being associated with a 0.08 standard deviation increase in positive relations compared to not being retired).
The importance of personality change for PWB is highlighted by examining how personality change relates to other well-being measures (life satisfaction, depression, hostility) compared to PWB. Models 2g–2i in Table
3 shows that personality change explained only as much within-person variation in life satisfaction, twice as much variation in depression and three times as much variation in change in hostility than change in socioeconomic variables (Models 1g–1f).
4 Discussion
This study extends earlier research suggesting that personality change is in fact meaningful. We use a midlife population, an age group for which research on personality development is limited. Midlife presents an important period for personality development (Lachman
2004; Neugarten
1968) as it is associated with many biological, physical, work, social and psychological changes, amongst others which in turn may result in changes in personality (see Roberts and delVecchio
2000 and Allemand et al.
2007 for a discussion on mechanisms of trait consistency and change in midlife) as well as a period where individuals seek a sense of identity as they reflect on their lives. Obtaining clarity of self and striving towards a fulfilled life is associated with favourable changes in personality, and may be protective of any negative changes associated with midlife. Therefore, studying whether and how personality changes across midlife can give insight into how individuals are coping with midlife challenges.
Through our stability ratios in Table
1, we illustrate that personality changes to a similar extent as socioeconomic variables during midlife. In our sample, personality variables changed more than marital and education status and almost as much as income across 10 years. We further show that personality change is not only an indicator of change in life satisfaction as previously shown, but associated with changes in a wider range of measures over time, specifically PWB, even after adjusting for socioeconomic variables. In our sample, personality change explained up to seven times as much within-person variation in PWB than socio-economic variables, 2–3 times as much within-person variation in depression and hostility and as much within-person variation in life satisfaction. The overall findings highlight the importance of personality change for PWB as well as the need to distinguish between the different well-being constructs (Kahneman and Deaton
2010). The results provide empirical support for existential-humanistic theories of personality change, indicating that personality change is essential to an individual’s strive towards a fulfilled life; an increase in neuroticism indicating poor existential engagement with the world and increases in the remaining traits suggesting positive existential well-being. For all well-being measures, personality change was a better indicator of well-being change than change in socioeconomic variables. Taken together, the results show that an individual’s personality changes over time and that these changes are strongly related to changes in the individual’s existential well-being. These findings emphasise the importance of personality in psychological functioning during midlife, reiterating the need to integrate personality measures into well-being research.
4.1 Limitations
Our study examines the association between changes in personality and well-being using two time points. Therefore we do not model how these variables change continuously but rather across a 10 years period. However, this could be seen as an advantage as we study long term changes in personality rather than temporary changes due to life events. Secondly, the use of a single item measure is a limitation of the dataset. Single item measures of life satisfaction are considered less stable and correlate less strongly with socioeconomic variables such as income and education (Pinquart and Sorensen
2000) than multi-item measures. However, this dataset was chosen as it is a large sample which provides data on life satisfaction, PWB and personality at two time points, allowing us to examine the association between these measures across time. A third limitation is that personality measures may be influenced by mental health status (Fergusson et al.
1989; Hirschfeld et al.
1983)—a depressed individual may report higher scores for neuroticism than they would in their pre-morbid state, since the individual’s mental state may result in more neurotic perceptions of themselves than usual. This would mean that an apparent change in self-reports of personality traits could be due to the effect of a mental disorder rather than associated changes in environmental circumstances or existential struggles. However, Fergusson et al. (
1989) show that even after correcting for the effect of current mental state on neuroticism, neuroticism still remained a significant predictor of depression. Fourth, we excluded a large proportion of the sample (54 %) from our analyses due to missing data. To explore if individuals included in our analyses were different to those excluded from the analyses, we regressed an inclusion variable (which indicates whether subjects are included in the analyses) on each of our well-being outcomes and all control variables. These regression models indicated that education, employment status and physical health were predictors of inclusion into the analyses. For each outcome variable, a weight was then generated from the inverse of the predicted probability of the model predicting inclusion into the analyses. These weights were then included in our difference score models to account for missing data. The ‘weighted’ models produced similar results to our complete case analyses (Tables
2,
3), except for the regression predicting change in depression, which indicated that becoming unemployed was associated with a 0.20 standard deviation decrease in depression status (
p < 0.01), compared to a 0.06 standard deviation decrease in depression in the complete case analysis. However, despite the substantive difference in the regression coefficient for unemployment, the weighted model personality change explained the same amount of variation in depression status as in the complete case analysis. Finally, our analysis can not make any causal inferences or direction about the personality-well-being relationship; whereas we have discussed the life choices people make as antecedents of personality change, it may be that certain personality traits facilitate the growth and development process. More research would be needed to confirm the causal pathways between personality change and well-being. Furthermore, we note that associating PWB change and personality change is consistent with other theories of personality development. We merely present the existential-humanistic theory as an alternative explanation for personality development.