Personal Narratives as a Predictor of Trait Change and State Fluctuations in Self-Esteem and Life Satisfaction during the Transition from Education to Work

ABSTRACT

Narrative identities are not only a product of lived experiences, but also predict new experiences and developmental outcomes over time. In this study, we examined whether the characteristics of students’ narratives about a previous turning point in their lives predicted self-esteem and life satisfaction during their education-to-work transition. We examined predictions of trait change and of stability in state self-esteem and life satisfaction in daily life. Students (aged 24 years on average) in diverse master’s programs in the Netherlands who expected to graduate before the end of the study took part in an 8-month longitudinal (n = 216) and 14-day diary (n = 192) part of the study. Our findings show that more agentic and redemptive turning point narratives did not predict significant change in trait self-esteem and life satisfaction during the next 8 months of the transition. Narrative agency and redemption were no consistent significant predictors of higher day-to-day stability in state self-esteem and life satisfaction. Yet, higher levels in narrative agency tended to be associated with higher stability in daily self-esteem. Our findings show promising possibilities to further examine how and when the characteristics of people’s narratives relate to the development and stability of psychological adjustment.

KEYWORDS:

• Self-esteem
• life satisfaction
• daily diary
• narrative identity
• work transition

People construct narratives about their past as part of the creation of an internalized life story, called narrative identity. By linking the past to the present and imagined future, a coherent narrative identity arises that provides a feeling of personal continuity and purpose (McAdams, 2001). Yet, one’s narrative identity is not only a product of lived experiences but has also been shown to predict new experiences and developmental outcomes over time, such as general mental health (e.g., Adler, 2012; Adler et al., 2015), adaptive formation of commitments (Van Doeselaar et al., 2019), sobriety (Dunlop & Tracy, 2013), and life satisfaction (Dunlop et al., 2022). In the present study, we extended the examination of personal narratives’ outcomes by predicting self-esteem and life satisfaction, key indicators of psychological adjustment, in two important ways. First, in addition to examining whether narratives predict changes in trait self-esteem and life satisfaction, we are, to our knowledge, the first to examine whether narratives predict stability in state self-esteem and life satisfaction in daily life. In doing so, we provide novel evidence relevant for understanding the mechanisms behind narratives’ effect on psychological adjustment, since instability at the daily level could make people’s psychological adjustment vulnerable (Sowislo et al., 2014). Second, we examined these questions during the sensitive developmental phase of the education-to-work transition. This transition is an ideal context to study narrative identity’s link with psychological adjustment: it entails many changes in people’s lives that can evoke changes in psychological adjustment. It is especially during such challenging times that narratives are thought to be influential (Adler et al., 2015; Lilgendahl & McLean, 2020).

The transition from education to work

The transition from education to work is a major developmental milestone, marking the end of youth and beginning of adulthood (Arnett, 2000). This transition is challenging. Young adults leave the familiar educational environment and need to find a job that, ideally, not only matches their education, but also their personal values (Sortheix et al., 2013). When having found a job, they need to find out what their new role as a member of the workforce is and how to perform well, while they also need to adjust to a new social group of colleagues (Bauer et al., 2007). People differ in how they experience and adjust to this transition, with vast differences in how their self-esteem and life satisfaction changes (Filosa et al., 2022; Haase et al., 2012; Reitz et al., 2020). In the present study, we examined whether a salient personal narrative helps explain these individual differences in self-esteem and life satisfaction.

Changeability and stability of self-esteem and life satisfaction traits and states

Self-esteem and life satisfaction are two key indicators of psychological adjustment, both predictive of better outcomes in a wide range of domains, such as mental health, physical health, relationships, and work (Diener et al., 2018; Fergusson et al., 2015; Orth & Robins, 2014, 2022). Self-esteem refers to individuals’ subjective judgments of their worth as a person (Donnellan et al., 2011). Life satisfaction refers to individuals’ subjective judgments of the quality of their life (Diener et al., 1985). Self-esteem and life satisfaction are both relatively stable traits (Lucas & Donnellan, 2007; Orth & Robins, 2014), but also changeable in response to major life events (Luhmann et al., 2012; Tetzner et al., 2016). During work transitions, change trajectories of self-esteem and life satisfaction show multidirectional change patterns (Filosa et al., 2022; Haase et al., 2012; Reitz et al., 2022, 2020). For example, although starting a fulltime job after graduation is on average characterized by a small increase in self-esteem across approximately 2 years, there are significant individual differences in young adults’ self-esteem change (Reitz et al., 2020). Similarly, even though life satisfaction seems to remain on average stable in the year after graduation, individuals differ in these trajectories with some experiencing increases and others decreases (Haase et al., 2012).

In addition to trait change across periods of months and years, self-esteem and life satisfaction change can be conceptualized as changes in states, which have been shown to fluctuate across days (Alessandri et al., 2016; Kernis et al., 1989; Willroth et al., 2020). The degree to which states of self-esteem and life satisfaction are unstable (i.e., fluctuate) from one day to the next differs between individuals. Self-esteem and life satisfaction are thought to fluctuate across days as individuals’ evaluation of their self-esteem or life satisfaction is affected by their current emotions (Clasen et al., 2015; Sowislo et al., 2014; Willroth et al., 2020). Individuals whose self-esteem or life satisfaction is more reactive and fluctuates more are believed to be more vulnerable to develop mental health issues (Clasen et al., 2015; Kernis et al., 1989; Willroth et al., 2020). In line with this notion, instability in daily self-esteem has been shown to relate to mental health outcomes, such as paranoia (Thewissen et al., 2008), borderline personality features (Zeigler–Hill & Abraham, 2006), and depressive symptoms (Sowislo et al., 2014; Van Tuijl et al., 2018). Similarly, instability in daily life satisfaction has been found to be negatively associated with psychological health (Willroth et al., 2020). These associations exist regardless of individuals’ trait levels of self-esteem and life satisfaction, signifying the importance of not only studying traits but also the daily stability of self-esteem and life satisfaction.

Given that both, individual differences in traits and in stability of self-esteem and life satisfaction states have important consequences for mental health, it is important to understand their antecedents. Yet, not much is known about why people differ in the trait development and state fluctuations of these two key psychological adjustment indicators.

Narrative agency and redemption

In the present study, we examined whether individual differences in self-esteem and life satisfaction are predicted by how people narrate an experienced turning point. McAdams considers turning point narratives as a part of people’s narrative identity, the layer of personality that reveals most about human individuality as it conveys who one is to oneself and others (McAdams, 2001; McAdams & Pals, 2006). During times of transition, people’s interpretations of and lessons from previous turning points might be reactivated and be predictive of how they cope with this new challenge. Although each turning point narrative is unique, narrative characteristics have been identified that are associated with psychological adjustment (McLean et al., 2019). Two such characteristics are agency and redemption.

Agency refers to strivings of the individual to be autonomous, to protect and assert themselves, and to be in control (Bakan, 1966). Agentic individuals tend to act purposely and serve their own personal interests, thereby increasing opportunities for self-actualization. People value themselves more when they see themselves as agentic (Wojciszke et al., 2011). Narratives reflect a high degree of agency when protagonists are able to affect the course of their own lives and achieve some control over the course of their experiences, instead of being completely at mercy of circumstances (Adler, 2012). Previous work has shown that narrating in a more agentic manner was predictive of increases in psychological adjustment across time (Adler, 2012; Adler et al., 2015; Van Doeselaar et al., 2019).

Redemption refers to the presence of a sequence in a story in which something good results from something bad (McAdams et al., 2001). The storyteller states that a negative event (e.g., feeling lonely when studying abroad) transformed into something positive (e.g., one is pleased to have learned from this to become more open to others; McAdams et al., 2001). By finding good in bad, individuals likely focus on meaningfulness in randomness and might see their strengths instead of blaming themselves (Janoff-Bulman, 1998). Presence of redemption in adults’ narratives has been shown to be associated with increases in psychological adjustment across time (Adler et al., 2015; Dunlop & Tracy, 2013).

Narrative agency and redemption predicting self-esteem and life satisfaction

Previous work points to several mechanisms through which narrative agency and redemption might influence self-esteem and life satisfaction during the transition to work. First, being reminded of an agentic or redemptive narrative, such as by new events during the transition to work, can boost people’s self-esteem and life satisfaction (Adler et al., 2016). When repeatedly being reminded of a personal narrative characterized by agency and redemption, this might result in higher trait self-esteem and life satisfaction, whereas being reminded of having no agency or stories that ended negatively might result in lower trait self-esteem and life satisfaction. Second, a memory of a past event and how it unfolded might guide people’s attitudes and behavior during new events (Philippe et al., 2011). Agentic and redemptive narratives might be an inspiration and encouragement of overcoming new setbacks, which lead to an increase in self-esteem and life satisfaction. Third, individuals might want to maintain and hold on to their view about themselves (Dunlop, Wilkinson et al., 2020; Merton, 1948; Swann, 1983). Individuals who view themselves as agentic and who experienced redemption during personally salient events might want these self-views to be confirmed. Fourth, the presence of agency and redemption in highly personal narratives might be indicative of people’s explicit and implicit motivational and emotional preferences (Adler et al., 2016; McClelland et al., 1989). These preferences might predict their future behavior and attitudes. That is, people who narrate in a more agentic manner likely tend to behave in a more agentic manner as well. Similarly, sharing redemptive narratives might be an indicator that an individual has a redemptive mind-set (Dunlop, Wilkinson et al., 2020).

Most studies on the link between narrative agency and redemption and self-esteem and life satisfaction examined cross-sectional associations at the trait level. This research has shown that narrative agency correlated positively with self-esteem (Holm et al., 2020; Steiner et al., 2019), but not significantly with life satisfaction (rs = .20 and .16, respectively; Bauer & McAdams, 2004; Grossbaum & Bates, 2002). However, these studies on life satisfaction had limited power (ns= 76 and 49, respectively). For redemption, research has shown a small to moderate positive association with life satisfaction (Dunlop, Harake et al., 2020; McAdams et al., 2001) and generally pointed to a moderate association with self-esteem (McAdams et al., 2001; McLean & Breen, 2009; Steiner et al., 2019). One longitudinal study recently found that a redemptive self-defining academic narrative was associated with increases in students’ life satisfaction but not with changes in self-esteem (Dunlop et al., 2022). To draw conclusions about the predictive effects of narrative agency and redemption on trait self-esteem and trait life satisfaction, more longitudinal research is needed.

Previous research has so far ignored narratives’ associations with daily fluctuations in psychological adjustment. Yet, narrative characteristics might not only affect psychological adjustment across months and years, but also across days. Agentic and redemptive narratives might be a source of stability in psychological adjustment in daily life, in similar ways as how they are thought to affect psychological adjustment across longer periods. If rooted in a salient agentic and redemptive narrative, self-esteem and life satisfaction might fluctuate less. Personally salient agentic and redemptive narratives might guide people’s attitudes and behavior, motivate them to maintain this view of themselves, and be exemplary of their explicit and implicit motivational and emotional preferences. All in all, agentic and redemptive narratives might predict consistency in people’s everyday behavior and attitudes and help them when experiencing daily obstacles or setbacks. For instance, having to write yet another job application or receiving a rejection letter might be less impactful for people with an agentic or redemptive personal narrative. As a result, their daily self-esteem and life satisfaction might be less susceptible, reflected in higher day-to-day stability.

To date, no study has, to our knowledge, examined the predictive effects of personal narratives on day-to-day stability in state self-esteem and state life satisfaction. However, such research is important for the following reasons. First, the daily stability of self-esteem and life satisfaction are associated with psychological health and developmental outcomes (e.g., Diener et al., 2018; Orth & Robins, 2014; Sowislo et al., 2014; Thewissen et al., 2008; Zeigler–Hill & Abraham, 2006). Second, the nature of the association between narrative characteristics and daily stability of self-esteem and life satisfaction provides insights into the mechanisms underlying the influence of narratives on psychological adjustment. Instability at the daily level could make people vulnerable for more maladaptive changes in their psychological adjustment (Sowislo et al., 2014), whereas stability at the daily level might be indicative of resilience and lead to psychological adjustment over time.

The present study

The present study examines the association between narrative characteristics and psychological adjustment during the transition from education to work. The first aim was to examine whether higher levels of agency and the presence of redemption in students’ turning point narratives before the transition to work were associated with a more positive change in self-esteem and life satisfaction across the next 8 months, compared to students with less agentic or non-redemptive turning point narratives. The second aim was to examine whether students with more agentic and redemptive turning point narratives experienced less fluctuation in their daily self-esteem and life satisfaction than those with less agentic or non-redemptive turning point narratives.

We conducted an intensive 8-month longitudinal study, which combined coded narrative, longitudinal, and daily-diary data. Students in diverse master’s programs in the Netherlands who expected to graduate before the end of the study participated. Since changes in individuals’ traits have been found across a few months during major life transitions (Luhmann et al., 2014; Van Scheppingen et al., 2018), we expected changes in trait self-esteem and life satisfaction during the 8 months of the transition to work. At the start of the study, students wrote about a previous turning point. In addition to three assessments on trait self-esteem and trait life satisfaction that took place every 4 months, participants reported on their daily self-esteem and daily life satisfaction for 14 days at the start of the study. Our hypotheses, methods, and strategy of analysis were preregistered at https://osf.io/zd3n9.¹

Method

Participants

Data were drawn from project GradLife, a multi-wave longitudinal study on the development of self and identity that captures the daily life experiences during the transition from university to work. Participants were Dutch-speaking students enrolled in diverse master’s programs at universities in the Netherlands who expected to graduate prior to the end of the third measurement wave. The present study used data from an online questionnaire at the start (T1) and 4 (T2) and 8 (T3) months later, and daily assessments collected at T1. At T1, 309 students started the online questionnaire which included a turning point narrative prompt. Of these students, 281 wrote a narrative (18 dropped out before and 10 did not write a narrative). Since we aimed to examine the transition to work, two participants were excluded who had worked already for more than 5 years without studying. In addition to these inclusion criteria, we used additional criteria to select two subsamples: a longitudinal subsample to examine trait change across time and a daily-diary subsample to examine fluctuations in daily life. The two subsamples were overlapping, but not completely similar since some participants only completed longitudinal data and others only daily diary.

To examine research questions on predicting change in trait self-esteem and life satisfaction, we used a longitudinal subsample of 216 students (M_age = 24.25 years, SD = 1.62, 75.9% female), who met the inclusion criterium of a minimum of half of the online questionnaires on self-esteem and life satisfaction traits completed across the three waves. In this subsample, only 8.56% of the self-esteem and life satisfaction variables were missing across waves. From those participating at T3 (n = 182), 88% were no longer studying at that time and 79% were working. Participants of this subsample did not differ from other participants in narrative agency, t(279) = 1.56, p = .119, d = .22, redemption, χ2 (1) = 1.53, p = .216, φ = −.07, or life satisfaction at T1, t(287) = 0.63, p = .530, d = .09. Yet, participants in this subsample reported slightly lower self-esteem at T1, t(297) = 2.28, p = .024, d = .29.

To examine research questions on predicting stability in daily state self-esteem and life satisfaction, we used a daily-diary subsample of 192 students (M_age = 24.43 years, SD = 2.02, 75.5% female), who met the criterium of at least 5 days of daily-diary data (M = 11.59 days, SD = 2.71). This subsample did not differ from unselected participants in narrative agency, t(279) = 1.21, p = .229, d = .15, or redemption, χ2 (1) = 0.06, p = .814, φ = .01. Since there exists, to our knowledge, no consensus on the minimum number of completed days needed to calculate fluctuation in daily self-esteem and life satisfaction reliably and validly, we checked whether findings altered when setting the cutoff not at 5, but at 6 (n = 186), 7 (n = 173), or 8 (n = 167) completed days (see Footnote 2).

Procedure

Project GradLife was approved by the Ethics Review Board (ERB) of Tilburg University (RP158). For a complete description of the project and its procedure see https://osf.io/fwak9/. Dutch-speaking master students were invited to participate via multiple channels of Tilburg University (e.g., e-mail and the website), university papers, and social media. Students were informed about the project via the projects’ website and asked to complete the consent form.

At each wave, participants were invited to complete an online questionnaire as well as four brief surveys per day via a mobile app (www.ethicadata.com) for 14 days using the experience sampling method (ESM). In the present study, we only used data from the online questionnaires and from the 14 ESM surveys that were prompted at the end of each day, which contained questions about reflections on that day (i.e., available from 21.00 until 4.00), henceforth referred to as daily-diary data. Participants received monetary rewards (e.g., €7 to €8 for the online questionnaires and up to €7 for the ESM part), personalized reports, and newsletters. Moreover, participation in the mobile app was encouraged via text messages.

Measures

Turning point narrative

In the online questionnaire at T1, participants were asked to describe a turning point in their self-understanding in a prompt formulated by McAdams (2007) and adapted by McLean et al. (2010). Participants were asked to identify a particular episode in their life story in which they experienced an important change about how they understand themselves. They were asked to describe what happened, when it happened, who was involved, what they were thinking and feeling, why the experience was significant, and what it could say about them and their personality (for the full prompt see the codebook of project GradLife; https://osf.io/fwak9/). In addition to their expected salience for students going through a transition, turning points were of interest as they elicit redemption (McAdams et al., 2001) and vary in agency (Van Doeselaar et al., 2019). Participants could use as many words as needed. Narrative length ranged from 16 to 595 words (longitudinal subsample: M = 174, SD = 97; daily-diary subsample: M = 176, SD = 101).

Agency was coded using the coding manual by Adler et al. (2008) on a 5-point scale: 0 (protagonist lacked agency, completely at mercy of circumstances), 1 (somewhat at mercy of circumstances), 2 (both non-agentic and agentic elements, or lacked information on agency), 3 (somewhat agentic), and 4 (completely agentic, able to affect own life; see, Table 1). Additionally, it was specified that a lack of agency (0 or 1) was only coded when the protagonist was negatively affected by circumstances; not having a role in the course of a positive event was not deemed to signify a lack of agency (Van Doeselaar et al., 2019). Moreover, if the narrative described a change in agency, the final state of agency was coded. The final code was the average of the codes from three coders.

Table 1. Illustration of turning point narratives and coding systems.

Example turning point narrative	Agency score	Redemption score
“The choice to go on exchange to England. I had a difficult childhood, because of a father with an alcohol problem. My father sometimes told me that nothing would come of me, etc. During my exchange, I proved to myself that I can achieve anything I want. Since this period, I have the confidence to find my way. I am independent and motivated to make the best of everything. Reflecting my own life in relation to my father gives motivation.”	4	1
“Two weeks ago, I cycled to the station with my internship supervisor to go home. I told him that I found it difficult that my other supervisor was often too busy for me. We concluded that I should discuss this with him. At first, I found this difficult to do, because I want to be kind. Eventually I did it, and since then he made more time for me. This says about my personality that I sometimes find it difficult to stand up for myself.”	2	0
“Last year, after I had submitted my thesis, my supervisor told me that I would have to redo a large part of it. This surprised me. It initially provided renewed motivation to improve the thesis, but after a few weeks I lost all motivation and confidence. I had expected to graduate soon and find a new job, but it took a very long time. This event evoked my feelings of hopelessness to ever find a job. I lost my determination.”	0	0

Note. The examples presented here are composed of parts of narratives by various participants. Additionally, slight changes were made to guarantee anonymity.

Redemption was coded using the description of redemption imagery in the manual by McAdams et al. (2001). Redemption was present in a narrative if there was a sequence from negative to positive (see, Table 1). We considered the narrative to contain negative valence when there was significant negative affect or pain (e.g., fear, sickness) or when the event described was so negative that a person experiencing it would generally experience negative affect or pain (e.g., death of a friend). As suggested in the manual by McAdams et al. (2001), a mild negative state (e.g., feeling nervous) or a minor setback (e.g., meeting an unfriendly person) was not coded as negative. The narrative was coded to contain redemption if the negative state was followed by a positive state (also if it was less intense than the negative state). The final code was the code for which three coders reached consensus (i.e., either the code on which coders immediately agreed or on which they agreed after discussing discrepancies).

Two separate teams of three coders, blind to the research questions and other data, coded every narrative on either agency or redemption. First, coders were trained. Since we used the average agency code, turning point narratives of a different dataset (Van Doeselaar et al., 2019) were used to train agency coding. Redemption coding was trained using 40 turning point narratives of project GradLife. Second, all narratives were coded independently and discrepancies were discussed each time after coding about 40 narratives to reach consensus and to prevent coder drift (Syed & Nelson, 2015). The one-way random intra-class correlation coefficients of .90 and .89 indicated that agency was coded reliably in the longitudinal and daily-diary subsample, respectively. The Fleiss’ кs of .75 in the longitudinal subsample and .76 in the daily-diary subsample, based on narratives that were not part of the training phase, indicated that the agency coding was also reliable.

Trait self-esteem

At each wave, trait self-esteem was assessed with the Dutch translation (Franck & De Raedt, 2007) of the Rosenberg Self-Esteem Scale (Rosenberg, 1965). Participants rated their agreement to 10 statements (e.g., “I take a positive attitude toward myself”) on a scale from 1 (strongly disagree) to 5 (strongly agree). Cronbach’s alpha ranged from .87 to .89 across the three waves in the longitudinal subsample.

Fluctuation in daily self-esteem

Fluctuation in daily self-esteem was assessed at the end of each of the 14 days using four items of the Rosenberg Self-Esteem Scale, previously used by Nezlek and Plesko (2001). To assess daily self-esteem, items were preceded by information that they referred to today and slightly rephrased (e.g., “I took a positive attitude toward myself”). Based on these scores, we calculated within-person standard deviations. Within-person standard deviations are the most widely used measure of fluctuation in daily self-esteem and relate to psychosocial health (e.g., Kernis et al., 1989; Sowislo et al., 2014). Also, they are the most parsimonious indicator of daily dynamics, as it was argued that more complex indicators have no incremental validity (Dejonckheere et al., 2019; Wendt et al., 2020). Cronbach’s alpha of daily self-esteem was on average .77 and ranged from .68 to .82 across the 14 days in the daily-diary subsample.

Trait life satisfaction

Trait life satisfaction was assessed at every wave with the Satisfaction With Life Scale (Dutch translation by Arrindell et al., 1991; Diener et al., 1985). Participants were asked to rate five statements (e.g., “In most ways my life is close to my ideal”) on a scale from 1 (strongly disagree) to 7 (strongly agree). Cronbach’s alpha ranged from .80 to .85 across the three waves in the longitudinal subsample.

Fluctuation in daily life satisfaction

Fluctuation in daily life satisfaction was assessed at T1 using one item on daily life satisfaction (“All things considered, how satisfied are you today with your life as a whole?”), based on single-item measures that have shown to assess trait life satisfaction reliably and validly (Cheung & Lucas, 2014; Lucas & Donnellan, 2012). Based on these scores, within-person standard deviations were calculated. Previous research has shown that fluctuations in daily life satisfaction assessed with one general life satisfaction item were similar to those assessed with three items (Willroth et al., 2020). Moreover, this previous work showed that collecting daily assessments across 14 days is enough to obtain reliable estimates of fluctuation in daily life satisfaction.

Strategy of analysis

We performed two analytical steps. First, we examined whether agency and redemption predicted changes in trait self-esteem and life satisfaction (aim 1) with latent growth curve (LGC) modeling in Mplus 7 (Muthén & Muthén, 1998–2015). LGC modeling estimates a mean-level (intercept) and the mean change rate (slope). To take individual differences in the time between waves into account, latent growth curves were estimated based on the exact time of each of the waves using the TSCORES option (Coulombe et al., 2016). Thus, intercepts reflect the estimated trait at T1, and slopes reflect changes across 8 months. Latent growth curves for self-esteem and life satisfaction were estimated in one model to appropriately handle missing data using full information maximum likelihood (FIML). After describing the average latent growth curves, we tested whether agency and redemption predicted individual differences in the growth curves, by alternately including them as predictors of the intercepts and slopes. In predicting the slopes, participants’ starting point on a specific trait was added as a control variable (i.e., each slope was regressed on the corresponding intercept). Next, we checked whether adding narrative length as predictor of the intercepts and slopes and reanalyzing the model without 38 participants with a narrative that lacked information on agency (n = 178), altered the findings.

Second, we tested associations of narrative agency and redemption with students’ degree of fluctuation in daily self-esteem and life satisfaction using four multiple regression analyses (one for each narrative coding and each outcome variable; aim 2). As fluctuation across days often relates to its mean (Wendt et al., 2020), we accounted for the mean of daily self-esteem and life satisfaction across the 14 days. Moreover, we checked whether adding narrative length as a control variable and reanalyzing the model without 36 participants with a narrative that lacked information on agency (n = 156), altered the findings.

Results

Descriptive statistics are shown in Table 2. In the longitudinal subsample the distribution of the given agency codes (for examples see, Table 1) of 0, 1, 2, 3, and 4 (for no agency, almost no agency, both elements or lacking information, some agency, completely agentic) was 4%, 12%, 29%, 38%, and 17%, respectively. In the daily-diary subsample this was 3%, 12%, 29%, 38%, and 17%. Correlations between the variables are shown in Table 3.

Table 2. Descriptive statistics.

	Longitudinal subsample (n = 216)				Daily-diary subsample (n = 192)
	Min	Max	M	SD	Min	Max	M	SD
Agency	0.00	4.00	2.53	0.94	0.00	4.00	2.53	0.92
Redemption	0.00	1.00	0.36	0.48	0.00	1.00	0.39	0.49
Self-esteem T1	1.60	5.00	3.65	0.65
Self-esteem T2	2.00	5.00	3.76	0.62
Self-esteem T3	2.00	5.00	3.76	0.58
Life satisfaction T1	2.20	7.00	5.20	1.06
Life satisfaction T2	1.80	7.00	5.11	1.11
Life satisfaction T3	1.20	7.00	5.04	1.16
Daily self-esteem M					2.09	4.98	3.75	0.48
Daily self-esteem SD					0.07	1.38	0.49	0.21
Daily life satisfaction M					1.73	5.00	3.63	0.66
Daily life satisfaction SD					0.00	1.40	0.57	0.25

Table 3. Correlations.

	Longitudinal subsample
	Redemption	Narrative length		Self-esteem					Life satisfaction
				T1	T2		T3		T1	T2		T3
Agency	.07	.00		.28***	.20**		.30***		.25***	.20**		.19*
Redemption		.09		−.08	−.02		.02		−.17*	−.06		−.03
Narrative length				.06	.07		.07		.04	.13		−.06
Self-esteem T1					.77***		.72***		.58***	.49***		.34***
Self-esteem T2							.79***		.57***	.59***		.45***
Self-esteem T3									.54***	.54***		.55***
Life satisfaction T1										.71***		.53***
Life satisfaction T2												.59***
	Daily-diary subsample
	Redemption	Narrative length		Daily self-esteem					Daily life satisfaction
				M		SD			M		SD
Agency	.12	.05		.25***		−.19**			.24***		−.11
Redemption		.18*		−.06		.07			.00		.09
Narrative length				−.01		.09			−.06		.11
Daily self-esteem M						−.25***			.69***		−.16*
Daily self-esteem SD									−.10		.56***
Daily life satisfaction M											−.11

* p < .05; ** p < .01; *** p < .001.

AIM 1: predicting change in trait self-esteem and trait life satisfaction

Latent growth curve modeling showed an average increase in self-esteem during the transition, intercept = 3.67, p < .001, slope = 0.18, p = .001, and average stability in life satisfaction, intercept = 5.19, p < .001, slope = −0.19, p = .114. Students differed in their intercept of self-esteem, σ2 = 0.31, p < .001, and life satisfaction, σ2 = 0.86, p < .001, but not in their slopes of self-esteem, σ2 = 0.14, p = .189, and life satisfaction, σ2 = 0.92, p = .062.

Results of the latent growth curve models in which agency and redemption were added as predictors are displayed in Figure 1. Agency was significantly positively associated with the intercepts of students’ self-esteem and life satisfaction but was not significantly associated with their slopes. Presence of redemption was associated with a significantly lower intercept of life satisfaction but was not significantly associated with the intercept of self-esteem or with the slopes of self-esteem and life satisfaction. Adding narrative length as a control variable or excluding cases on which consensus was reached that they lacked information on agency did not change the results.

Figure 1. Agency and redemption as predictors of the latent growth curves of self-esteem and life satisfaction in the longitudinal subsample.

Note. Results from two latent growth curve models with either agency or redemption included as predictor of the intercepts and slopes of self-esteem and life satisfaction. Correlations between the intercepts of self-esteem and life satisfaction and between the slopes of self-esteem and life satisfaction are not shown but were included in the tested models. Coefficients before the slash indicate predictions by agency. Coefficients after the slash represent predictions by redemption. Numbers are standardized regression coefficients representing the difference in y in y standard deviation units for a standard deviation change in x (i.e., StdYX). For the binary redemption variable, the standardized estimates representing the difference in y in y standard deviation units between no or presence of redemption (i.e., StdY) are −.17 and −.36 for the intercepts and .43, and .33 for the slopes of self-esteem and life satisfaction, respectively. Unstandardized estimates, standard errors, and p are displayed Supplemental Table S1 in the supplemental material.* p < .05; ** p < .01; *** p < .001.

AIM 2: predicting fluctuation in daily self-esteem and life satisfaction

Regarding the associations between narrative agency and redemption and fluctuation in daily self-esteem and life satisfaction, we first looked at the correlations (see, Table 3). Here, higher levels in agency were related to significantly less fluctuation in daily self-esteem. Agency was not significantly correlated with fluctuations in daily life satisfaction. Moreover, redemption was neither correlated with fluctuations in daily self-esteem nor with fluctuations in daily life satisfaction.

In four multiple regression analyses we tested these associations again and controlled for the mean of the daily assessments in predicting the corresponding fluctuations (see, Table 4).² This association between mean and fluctuations was significant for daily self-esteem but not for daily life satisfaction. When controlling for average daily self-esteem, agency was no longer significantly associated with fluctuations in daily self-esteem. The p was .054, approaching significance. Moreover, agency was again not associated with fluctuations in daily life satisfaction, and redemption was neither associated with fluctuations in daily self-esteem nor with fluctuations in daily life satisfaction.

Table 4. Results of the regression analyses on agency and redemption predicting fluctuation in daily self-esteem and life satisfaction in the daily-diary subsample.

	SD daily self-esteem
	b	p	SE	β	F	R2	p
M daily self-esteem	−.10	.003	.03	−.22	8.55	.08	< .001
Agency	−.03	.054	.02	−.14
M daily self-esteem	−.11	< .001	.03	−.25	6.81	.07	.001
Redemption	.02	.481	.03	.05
	SD daily life satisfaction
	b	p	SE	β	F	R2	p
M daily life satisfaction	−.03	.242	.03	−.09	1.84	.02	.161
Agency	−.02	.237	.02	−.09
M daily life satisfaction	−.04	.134	.03	−.11	1.87	.02	.156
Redemption	.05	.227	.04	.09

When including narrative length as a predictor of fluctuations in daily self-esteem, β = .10, p = .162, in addition to average daily self-esteem, β = −.22, p = .003, the positive association between agency and fluctuations in daily self-esteem did become significant, β = −.15, p = .044. The other associations of agency and redemption with fluctuations in daily self-esteem and life satisfaction remained not significant when adding narrative length.

When excluding students with turning point narratives that lacked information on agency, agency was again significantly correlated with fluctuations in daily self-esteem, r = −.21, p = .010, but not with fluctuations in daily life satisfaction, r = −.12, p = .153. We then ran the regression analyses, in which we accounted for average daily self-esteem and life satisfaction, again in this group. In these analyses, agency was no significant predictor of fluctuations in daily self-esteem, β = −.14, p = .083; the effect of average daily self-esteem was β = −.20, p = .013. Also, agency did not predict fluctuations in daily life satisfaction, β = −.11, p = .208; the effect of average daily life satisfaction was β = −.03, p = .756.

Discussion

In the present study, we examined whether two characteristics of a personally salient narrative from students in the midst of their transition to work predicted their self-esteem and life satisfaction. We studied the effect of narrative characteristics on trait changes across 8 months and on daily fluctuations in self-esteem and life satisfaction states. The findings showed that students who, before their transition to work, narrated a previously experienced turning point in a more agentic manner and redemptively did not increase significantly more in self-esteem and life satisfaction during the next 8 months of their transition, compared to those with less agentic or non-redemptive turning point narratives (aim 1). Moreover, narrative agency and redemption were not clear significant predictors of higher stability in students’ daily self-esteem and life satisfaction (aim 2). However, narrating a personal turning point in a more agentic manner was associated with higher stability in daily self-esteem, although inconsistently. A finding asking for further examination.

Predicting change in trait self-esteem and life satisfaction

That presence of agency and redemption did not predict changes in self-esteem and life satisfaction is not in line with previous work in which they predicted increases in psychological adjustment (Adler, 2012; Adler et al., 2015; Dunlop et al., 2022; Dunlop & Tracy, 2013; Van Doeselaar et al., 2019). A likely explanation for our non-significant findings is the study duration of 8 months. Previous work mostly focused on longer time frames (Adler et al., 2015; Dunlop et al., 2022; Van Doeselaar et al., 2019). Moreover, those who did find predictive effects across short time frames of 14 weeks (Adler, 2012) and 4 months (Dunlop & Tracy, 2013) focused on clinical subgroups which are possibly more susceptible to change than students transitioning to work. Although we expected substantial variations between students in their changes in self-esteem and life satisfaction during the transition to work based on previous one- (Haase et al., 2012) and two-year (Reitz et al., 2020) longitudinal studies on the transition to work and (re)employment (Reitz et al., 2022), our findings showed no significant individual differences in change (i.e., slopes) during 8 months of the transition to work. Possibly, individual differences in the changes of trait self-esteem and life satisfaction take longer than 8 months to become noticeable during the transition to work.

It remains possible that the construction of turning point narratives predicts changes in trait self-esteem and life satisfaction during the transition to work over a time frame longer than the examined 8 months. Our findings indicate this seems to be especially the case for redemption, since the non-significant predictive effects of redemption on self-esteem and life satisfaction were both in the expected direction. Predictive effects of agency on trait changes in self-esteem and life satisfaction are less likely since the findings indicated non-significant effects very close to zero. Additionally, it is possible that during 8 months of the transition to work narrative agency and redemption are predictive of individual differences in changes of other aspects of psychological adjustment, with less stability in individual differences in it across time, such as depressive symptoms (Wouters et al., 2013).

Predicting fluctuation in daily self-esteem and life satisfaction

Overall, we found no evidence that the presence of agency and redemption in a salient narrative is predictive of individual differences in the stability of daily self-esteem and life satisfaction, with one exception. Narrative agency’s prediction of stability in daily self-esteem almost reached significant and was significant when controlling for narrative length.

Lengthy narratives might indicate a greater engagement with sharing and possibly constructing the personal turning point narrative (e.g., reflected in positive associations with autobiographical reasoning; Van Doeselaar et al., 2019). Therefore, narrative length is a common control variable when examining associations with characteristics of personal narratives (e.g., Adler, 2012; Dunlop & Tracy, 2013; Van Doeselaar et al., 2019). The current study shows that longer narratives are not significantly more agentic, which is in line with previous findings (Adler, 2012; Van Doeselaar et al., 2019), nor are they associated with significantly more stability in daily self-esteem. Yet, we accounted for the small variations in the stability of daily self-esteem that might simply be explained by more engagement in sharing and constructing narratives, reflected in the length of narratives. When omitting this source of noise, narrative agency did have a small significant association with stability in daily self-esteem.

Previous work found that people who view themselves as agentic value themselves more (Wojciszke et al., 2011). Our findings provide first evidence that when this agentic view of oneself is rooted in relevant personal narratives it could make one’s self-esteem more robust and less susceptible to negative influences during the day. If confirmed in future studies, this link between narrative agency and stability in daily self-esteem could explain one way in which narrative agency eventually predicts the development of psychological adjustment (Sowislo et al., 2014), a predictive effect found in previous work (Adler, 2012; Adler et al., 2015; Van Doeselaar et al., 2019). Although the mixed findings ask for a careful interpretation, they thus also call for further examination of whether giving oneself a more agentic role in one’s narratives fosters higher day-to-day stability in one’s self-esteem.

The fact that narrative agency did not relate to individual differences in the stability of daily life satisfaction and narrative redemption did neither relate to individual differences in the stability of daily life satisfaction or in daily self-esteem might indicate that the effect that personal narratives have on experiences at the daily level are limited. Possibly, the effect of a personal narrative is stronger in a daily situation that is more directly related to the narrative (e.g., a narrative about how one previously managed to learn new skills during an internship might mitigate the effect of difficulties with executing a current job). Nevertheless, before drawing conclusions about a limited effect of personal narratives at the daily level, replication studies are needed that overcome some of the limitations of the present research.

Strengths and limitations

Strengths of the present study were the combined use of coded narrative, longitudinal, and daily-diary data. Combining different methods decreases the risk that associations are due to shared measurement variance. Another strength was that the hypotheses, methods, and strategy of analysis were preregistered. Moreover, our sample sizes of 216 (longitudinal subsample) and 192 (daily-diary subsample) were sufficient to detect associations of small to moderate size (r = .23 and r = .22, respectively; Faul et al., 2009). There were, however, also limitations that need to be acknowledged.

First, our study focused on a time frame of 8 months during the transition from education to work. Our findings not only indicate that during these 8 months narrative agency and redemption are not predictive of individual differences in trait change in self-esteem and life satisfaction, but also that this period is too brief to find individual differences in trait change in self-esteem and life satisfaction. Likely, it takes longer for self-esteem and life satisfaction to be affected by the transition to work. Previous studies demonstrate interindividual differences in changes in life satisfaction and self-esteem during the transition to work across periods of one and two years, respectively (Haase et al., 2012; Reitz et al., 2020). Future studies should examine whether agency and redemption are significant predictors of changes in life satisfaction and self-esteem during these longer intervals.

Moreover, although at baseline all participating master’s students had expected to graduate during the 8 months of the present study, at the final measurement wave a small percentage (i.e., 12%) did not graduate and not all participants (21%) started working yet. The present study included all participants, since they all anticipated to transition to work, which could already impact their self-esteem (Filosa et al., 2022; Reitz et al., 2022). Nevertheless, previous work showed that especially students who transition to fulltime work, compared to those who have parttime jobs or are unemployed, experience less rank-order stability in their trait self-esteem (Reitz et al., 2020). Possibly, agentic or redemptive personal narratives might help people during this specific phase of the transition, as these narratives might affect whether and how one experiences successes and failures at work, experiences that are likely related to changes in trait self-esteem (Reitz et al., 2020). Future studies with a larger sample should examine the predictive effect of personal narratives on psychological adjustment across time as well as at the daily level and compare young adults who have and those who have not yet made the transition to work.

Furthermore, we focused on one narrative. Yet, the narratives that form one person’s narrative identity can have different characteristics (McLean et al., 2016). For instance, some narratives are redemptive and others not. Thus, it is key to focus on those narratives that are salient during a specific time and for a specific outcome. Turning point narratives were chosen as they were expected to be particularly salient for those facing a new turning point, the education-to-work transition. Nevertheless, during the transition to work, individuals might not only be reminded of this turning point narrative. For example, when having trouble deciding on what job might be a good fit, one might also be reminded of a self-defining academic narrative (Dunlop et al., 2022). Possibly, taking multiple narratives that are likely to be evoked by the transition to work into account would make it possible to detect effects on the development and daily stability of self-esteem and life satisfaction in future studies.

Our study examines the transition to work of university master’s students. Results can therefore not be generalized to transitions to work from other educational levels. For instance, university graduates have generally less chance of unemployment, compared to graduates from vocational education (Centraal Bureau voor de Statistiek, n.d.). The transition to work could thus be more challenging for students from vocational education. This challenge could increase the impact of students’ narrative characteristics on psychological adjustment, since this impact is expected to be stronger during challenging times (Adler et al., 2015; Lilgendahl & McLean, 2020). Future studies should examine whether narrative characteristics are predictive of changes in and daily stability of psychological adjustment during the transition to work from other educational levels than university.

Lastly, we have assessed daily life satisfaction with a single item, based on a frequently used item to assess trait life satisfaction. Although this single item has shown to assess trait life satisfaction reliably and validly (Cheung & Lucas, 2014; Lucas & Donnellan, 2012), less is known about its psychometric properties when used at the daily level. Future studies should assess whether this single item suffices to assess daily life satisfaction.

Conclusion

Personal narratives have previously been shown to predict psychological adjustment. Adaptive personal narratives, showing agency and redemption, could help people adjust during challenging times. We examined the effect of agentic and redemptive narratives on self-esteem and life satisfaction across months across months (i.e., trait change) and across days (i.e., stability in states). Our findings indicate that narrative agency and redemption do not predict individual differences in changes in trait self-esteem and life satisfaction across 8 months of the transition to work. Also, agency did not relate to higher stability in daily life satisfaction, and redemption not to higher stability in daily self-esteem and life satisfaction. However, we found first evidence to suggest that an agentic view of oneself rooted in salient personal narratives could make job beginners’ self-esteem more robust and less susceptible to negative influences in their daily lives. This finding warrants further examination since it was inconsistent. Yet, it could be one explanation for why narrative agency has previously been shown to predict a more adaptive development of psychological adjustment. Examining the link between narrative agency and fluctuations in daily self-esteem further will help understanding the roots of individual differences in psychological adjustment and promoting a better psychological adjustment during the challenging, important transition to working life. Together, our study suggests that personal narratives might be important to consider when examining psychological adjustment to major life transitions, also at the daily level.

Acknowledgments

This project (GradLife) has received funding from the European Union’s Horizon 2020 research and innovation program under the MarieSkłodowska-Curie grant agreement No 846839, awarded to Dr. Anne Reitz. We thank the following coders: Janou Boermans, Tijn van der Burg, Eline Delmée, Demi de Groot, Yentl van Schijndel, and Marthe Verhoeve.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

The data that support the findings of this study are openly available at the Open Science Framework (https://osf.io/ak5vu/).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15283488.2022.2106229

Additional information

Funding

This work has received funding from the European Union’s Horizon 2020 research and innovation program under the MarieSkłodowska-Curie grant agreement No 846839, awarded to Dr. Anne Reitz.

Notes

1. In our strategy of analysis, we later added that we would check whether controlling for narrative length altered the findings. Narrative length is a common control variable when examining associations with narrative characteristics (e.g., Adler, 2012; Dunlop & Tracy, 2013; Van Doeselaar et al., 2019).

2. To our knowledge, there is no consensus on the minimum number of completed days needed to calculate fluctuation in daily self-esteem and life satisfaction reliably and validly, and previous studies vary in the used cutoff (e.g., Meier et al., 2011; Zeigler-Hill et al., 2015). Therefore, we checked whether findings altered when setting the cutoff for the number of completed days of daily diary data not at 5, but at 6 (n = 186), 7 (n = 173), or 8 (n = 167). Reanalyzing the four regression analyses in these subsamples showed that only in the subsample in which participants completed at least 6 days higher levels of agency predicted significantly less fluctuation in daily self-esteem, β = −.15, p = .047. In all other regression analyses, the same associations were (non-)significant as in the main analyses.