Reciprocal Relations Between Meaning in Life, Beneficence, and Psychological Needs for Autonomy, Competence

Meaning in life has been established as a key factor of human well-being and flourishing. Beneficence and the three psychological needs of self-determination theory—auton-omy, competence, and relatedness—have all been individually studied as antecedents of meaningfulness. Yet, no previous research has examined them neither longitudinally nor simultaneously as predictors of meaning over time. In a three-wave longitudinal study in Chile (n: T1 = 1477, T2 = 820, T3 = 487) we examined the reciprocal relations between autonomy, competence, relatedness, beneficence, and meaning, using cross-lagged panel analysis. Taken individually, each of the four factors predicted meaning, and when entered simultaneously into the same model, competence, relatedness, and beneficence predicted meaning over time. Furthermore, we found that meaning predicted all four factors over time. Our results thus advance research on meaning in life by examining key predictors of it and showing that meaning itself predicts the same factors dynamically over time.


Introduction
Meaning in life is often considered a key component of human well-being, flourishing, and a life well lived (Camus, 1955;Frankl, 1963;Martela, 2020;May, 2015).It has also become an important focus of empirical research (Heintzelman & King, 2014;King & Hicks, 2021;Martela & Steger, 2016), with research showing that experiencing meaning and purpose in life is associated in several studies with better health (Czekierda et al., 2017), longevity (Cohen et al., 2016;Hill & Turiano, 2014), adaptive coping strategies (Thompson et al., 2003), lower allostatic load (Zilioli et al., 2015), lower incidence of depression and suicidal ideation (Heisel & Flett, 2004;Mascaro & Rosen, 2005, 2008).Accordingly, understanding what makes life meaningful is an important research topic that can help both individuals and various professionals like therapists, coaches, teachers, and even policymakers to support people's well-being and ability to live a good life.
Self-determination theory (SDT) has proposed that humans have three psychological needs-autonomy, competence, and relatedness-the fulfillment of which is crucial for human well-being, flourishing, and meaningfulness (Ryan & Deci, 2000, 2017).There is strong empirical support for their importance for well-being in various contexts and across cultures (Cerasoli et al., 2016;Chen et al., 2015;Martela & Ryan, 2023;Van den Broeck et al., 2016;Vansteenkiste et al., 2020) and it has been argued that they could also be important for meaning in life (Martela et al., 2018;Weinstein et al., 2012).Autonomy is about a sense of volition, choice, and self-endorsement.It is proposed to enhance meaning in life as people find significance from being able to pursue selfselected goals and from living authentically (Schlegel et al., 2009(Schlegel et al., , 2011).One's life arguably feels more meaningful for oneself, when one is able to live according to selfendorsed values and strivings.Competence is about a sense of mastery, accomplishment, and efficacy.It is proposed to enhance meaning through people finding intrinsic value from a sense of mastery (Martela, 2020;H. Zhang et al., 2018).Exercising one's capacities and being able to accomplish something are valued and sought after experiences that can enhance one's sense of meaningfulness.Relatedness is about a sense of caring mutual relationships.It has been proposed to be a key source of meaning, as people value building strong and deep ties with each other, with research showing that close relationships tend to be key sources of meaning for most people (Lambert et al., 2010(Lambert et al., , 2013;;Stillman et al., 2009).All three needs have thus been independently proposed to be important for meaning in life.However, longitudinal research confirming these relations over time has been lacking.
Accordingly, we propose to study the following longitudinal hypotheses: Hypothesis 1a Autonomy predicts meaning in life over time, even when controlling for baseline levels of meaning in life.

Hypothesis 2a
Competence predicts meaning in life over time, even when controlling for baseline levels of meaning in life.

Hypothesis 3a
Relatedness predicts meaning in life over time, even when controlling for baseline levels of meaning in life.
Furthermore, several studies within SDT have examined beneficence-defined as a sense of prosocial impact-alongside the three needs as a fourth key enhancer of human well-being and flourishing (Martela & Ryan, 2016b, 2020;Titova & Sheldon, 2022;H. Zhang et al., 2021), and it has been shown to be an especially important predictor of meaningfulness (Martela & Ryan, 2016a;Martela et al., 2021).The rationale for beneficence as a source of meaning relies on people tending to find moments where they are able to have a positive impact in the lives of others as highly significant (Klein, 2017;Martela & Ryan, 2016a, 2021;Van Tongeren et al., 2016).Humans come to see their own lives as more valuable when they see that they are able to contribute something of value to other people.Accordingly, we propose the following fourth longitudinal hypothesis:

Hypothesis 4a
Beneficence predicts meaning in life over time, even when controlling for baseline levels of meaning in life.
A rationale thus exists to study each of the four predictors individually.However, instead of studying single sources of meaning in a siloed manner, researchers have called for studies examining multiple sources simultaneously to identify the most robust and direct predictors of meaning (Allan, 2017;Martela et al., 2018;Rosso et al., 2010).Some factors might directly enhance meaningfulness, others might come to empirically predict it mainly due to their confounding relationship with a more direct predictor, making it important to isolate their independent predictive power.Accordingly, it is important to examine which of the four predictors-autonomy, competence, relatedness, and beneficence-predict meaning in life when controlling for the influence of each other.In studies pitting the four proposed variables against each other, the results have been slightly inconsistent.Lambert et al. (2010) found that only competence but not autonomy and relatedness predicted meaning when controlling for each other, while Hicks et al. (2012) that both competence and relatedness predicted meaning, and Martela et al. (2018) found that autonomy, competence, relatedness, and beneficence were all associated with meaning in life simultaneously.However, all of this work has been cross-sectional.
The existing longitudinal studies we are aware of showed that autonomy and beneficence predict future levels of meaningful work (Martela et al., 2021).That study, however, was conducted in a specific domain-the workplace-instead of examining general meaning in life.Additionally, autonomy, competence, and relatedness, when taken as individually or as an aggregated whole, predict future levels of meaning in life (S.Zhang et al., 2022).However, Zhang et al. (2022) did not pit the needs against each other and did not include beneficence.Accordingly, no previous work has examined autonomy, competence, relatedness, and beneficence longitudinally as simultaneous predictors of meaning in life.We believe that each has its own unique pathway to a sense of meaningfulness, as outlined above, and thus they should remain as significant predictors of future levels of meaning, even when the influence of the three others is controlled for:

Hypothesis 1b
Autonomy predicts meaning in life over time, even when controlling for baseline levels of meaning in life, and for the influence of competence, relatedness, and beneficence.

Hypothesis 2b
Competence predicts meaning in life over time, even when controlling for baseline levels of meaning in life, and for the influence of autonomy, relatedness, and beneficence.

Hypothesis 3b
Relatedness predicts meaning in life over time, even when controlling for baseline levels of meaning in life, and for the influence of autonomy, competence, and beneficence.

Hypothesis 4b
Beneficence predicts meaning in life over time, even when controlling for baseline levels of meaning in life, and for the influence of autonomy, competence, and relatedness.
The proposed hypotheses will be explored in a three-wave longitudinal research design using a Cross-Lagged Panel Model (CLPM) that allows to control all constructs by its own baseline level as well as by the other lagged constructs making it possible to identify, which of these four potential predictors are the most robust prospective psychological predictors of meaning in life.This answers the repetitive calls to go beyond mere cross-sectional research in the study of meaning in life (Czekierda et al., 2017;Martela et al., 2018).The study also follows calls to diversify research of meaning in life beyond North American and European samples (Heintzelman et al., 2020;Steger et al., 2008) by studying the topic in Chile.
While we are interested in how the four proposed factors predict meaning in life over time, we should not rule out the other direction of influence: Meaning in life predicting autonomy, competence, relatedness, and beneficence over time.Previous research has shown that while the three needs for autonomy, competence, and relatedness predict positive affect over time, positive affect can also predict need satisfaction over time, making the relationship reciprocal (Unanue et al., 2023).In the case of meaning in life predicting future levels of autonomy, competence, relatedness, and beneficence, we don't have a specific rationale for this direction of influence.But it is important to examine it for comprehensiveness, thus warranting an exploratory analysis without specific hypotheses about which of the four factors meaning in life could predict.Thus, our empirical analyses will examine simultaneously both directions of influence between the factors.

Procedure and Participants
To collect our data, we hired Netquest company, which uses and online panel data (OPD).Netquest meets the ISO 20252 standard, specifically for market, opinion, and social research.OPDs have a series of benefits in comparison with convenience samples and faceto-face surveys.For example, they allow researchers to collect a large number of participants in a relatively short period, with less bias than with convenience samples and at a significantly lower cost than face-to-face surveys, especially in long-term longitudinal studies (for more details, see A. Newman et al., 2021;Peer et al., 2017;Porter et al., 2019).In order to guarantee a correct data collection process, the company was asked to follow the recommendations of Porter et al. (2019) regarding the optimal use of OPDs.The study was approved by the research and ethics committee of a university in Chile, informed consent was acquired from all participants and followed APA and BPS ethical guidelines.
Participants received an online invitation to participate in our three-wave longitudinal survey (6 months between each wave).Data collection for the first wave began in December 2018.Our study sample was representative of the panel of individuals available to Netquest from the capital of Chile, using quotas to ensure representative gender and age distribution.Thus, the sample approximates the adult population living in the capital of Chile (41.18% of the total Chilean population), both in terms of age and gender.The sample size was settled in advance, with the aim of collecting at least 460 participants who answer all three waves.Sample size recommendations for SEM vary widely (see e.g., Catena et al., 2003;Hair et al., 2014).Some suggest at least 200 participants or 10 per parameter (Weston & Gore, 2006), some others indicate at least 5 per parameter (Worthington & Whittaker, 2006), while recent Monte Carlo simulation research found that anywhere from 30 to 460 participants may be needed, depending on the number of factors, indicators, loadings, and correlations.In our study, our sample size exceeded the range suggested by Wolf et al. (2013), matched previous studies like Burkholder and Harlow (2003), and produced no improper estimates (Wolf et al., 2013), indicating it was sufficient.To ensure the robustness and validate the appropriateness of the minimum of 460 cases, we conducted a post-hoc power analysis following Moshagen and Erdfelder's (2016) approach.Using their R package semPower, we assessed the power achieved in our study by calculating the power to reject the model based on the RMSEA relative fit index (RMSEA > 0.05).We conditioned it on the study sample size, focusing on the participants who answered all three waves (a demanding criterion) (n = 487), and the model degrees of freedom of the most complex model (df = 50).We also replicated this procedure for the complete sample size reference (n = 1477).In both scenarios, the achieved power appears adequate for the fitted model (over 0.80).In total, 1477 adults completed the survey at T1, ranging between 19 and 78 years old (41.30% female; mean age = 41.43,SD = 12.81).At T2, 820 participants (41.80% female; mean age = 43.09,SD = 13.20)responded the survey.At T3, 487 individuals (40.70% female; mean age = 44.53,SD = 12.87) responded the questionnaire.Thus, 487 respondents participated of the whole study.

Autonomy, Competence, and Relatedness
We used the Basic Psychological Need Satisfaction Scale developed in three studies by Sheldon and Gunz (2009), building on a previous scale (Sheldon et al., 2001).The scale includes four items for each need (autonomy, competence, and relatedness).Example items are "I was free to do things my own way" (autonomy), "I took on and mastered hard challenges" (competence), and "I felt a strong sense of intimacy with the people I spent time with" (relatedness).The items were rated on a scale from 1 (not at all true) to 7 (very true).The mean of each subscale was computed by averaging their four items.The three subscales showed good internal reliability, with all reliabilities > 0.70 (see Table 1).

Beneficence
We used the beneficence scale developed and validated in four studies by Martela and Ryan (2016b).The scale included four statements (e.g., "In general, my influence in the lives of other people is positive") ranging on a scale from 1 (not true at all) to 7 (very true).The mean of beneficence was computed by averaging their four items.Reliabilities were good, with all reliabilities > 0.85 (see Table 1).

Meaning in Life Scale
We used the recently developed Three-Dimensional Meaning in Life Scale (3DM), validated in four studies by Martela and Steger (2023).The scale aims to provide a more comprehensive assessment of meaning in life by including items capturing the three identified facets of meaning (coherence, purpose, significance; see King & Hicks, 2021;Martela & Steger, 2016) through 4 items for each one (12 items in total).Examples items are "I can easily make sense of my life" (coherence), "I pursue one or more big purposes in my life" (purpose), and "My life is full of value" (significance), rated on a scale from 1 (not at all true) to 7 (very true).As we were interested in the overall sense of meaning in life, and had no specific hypotheses about the sub-dimensions, we computed a scale for general sense of meaning in life by averaging the 12 items.Reliabilities were good, with all reliabilities > 0.90 (see Table 1).

Analysis
We used two kinds of CLPMs to test our hypotheses.For example, Model 1 assessed the prospective relationships between each single need (autonomy: M1a, competence: M1b, relatedness: M1c, and beneficence: M1d) and meaning in life, while controlling for the baseline value of each other.Thus, Model 1 includes two constructs (one need and meaning in life) and three waves.In Model 2, we tested the prospective relationships of each of the four needs and meaning in life, while simultaneously controlling for the baseline level of each predictor.Thus, Model 2 includes 5 constructs (4 needs and meaning in life) and three waves.
For each CLPM (either in model 1 or Model 2), all constructs were represented as possible antecedents and consequences of all the other constructs, controlling for autoregressive effects.For each model, each observed variable was allowed to freely covary with all the other observed variables within the same wave, following the recommendations of Kline (2016).To assess model fit, we use Chi-Square statistic, RMSEA, and CFI, with the following standards for good (or acceptable) fit: RMSEA < 0.06 (< 0.08), and CFA > 0.95 (> 0.90) (Hu & Bentler, 1999).We made simplified assumptions in our models, by constraining both the standardized autoregressive and cross-lagged paths to be time-invariant.That is, for example, that the coefficient from T1 to T2 is settled to be identical to the coefficient from T2 to T3 (Cole et al., 2005).Indeed, because the distance between each wave is the same (6 months), we didn't have any conceptual or theoretical reason to assume that the coefficients would differ across the time points.This procedure allowed us to estimate more parsimonious models and gain statistical power.For robustness, the above-mentioned assumption was statistically checked.The constrained model was compared with the unconstrained model using Chi-square tests (Kline, 2016).We used Mplus 8.0 (Muthén & Muthén, 2017) for all our structural analyses.

Results
The means, standard deviations, Pearson correlations, and reliabilities for all the study variables are shown in Table 1.Below, we analyze the results of all the research models.

Model 2: All four factors and meaning in life
We used a CLPM model to test H1b, H2b, H3b, and H4b, by examining whether previous prospective relationships from each need (autonomy, competence, relatedness, and beneficence) to meaning in life remained significant when controlling for the other three needs and baseline levels of meaning in life (see Fig. 5).In other words, we included in Model 2 the 4 needs and meaning in life plus across the three waves.Further, all constructs were represented as potential antecedents and potential consequences of the other constructs while controlling for stability effects.As in Model 1, we started with a model without any constraint (free model).The fit of this model was good χ 2 (25) = 157.69,p < 0.001; CFI = 0.97; and RMSEA = 0.06 (90% CI: 0.05, 0.07).Then, we constrained all paths (to maximize statistical power) to be equal between waves.The constrained model fit remained acceptable: χ 2 (50) = 177.13,p < 0.001; CFI = 0.97; and RMSEA = 0.04 (90% CI: 0.03, 0.05), and this more parsimonious model showed no significant loss of fit compared to a model where all structural paths were estimated freely: (∆χ 2 (25) = 19.44,p = 0.78).Based on the constrained model, only H1b was not supported.We found that autonomy was not a significant prospective predictor of meaning in life (β = 0.02, p = 0.42) when controlling for the influence of the other three constructs.However, H2b, H3b, and H4b were fully supported.We found that competence (β = 0.06, [95% CI 0.01, 0.11], p = 0.03), relatedness (β = 0.07, [95% CI 0.02, 0.12], p = 0.01), and beneficence (β = 0.07, [95% CI 0.02, 0.12], p = 0.01) were all significant and positive prospective predictors of meaning in life, while controlling for the influence of the other three needs and baseline levels of meaning in life.Also, we found that meaning in life were significant and positive prospective predictors of autonomy (β = 0.

Discussion
The aim of this study was to examine whether beneficence and the needs for autonomy, competence, and relatedness would predict meaning in life over time.The results mainly supported these hypotheses.When taken individually, and when controlling for baseline meaning in life, each of the four factors was a significant positive prospective predictor of future levels of meaning in life.When all four were examined together in one model, competence, relatedness, and beneficence remained as significant predictors of future levels of meaning in life, while autonomy became insignificant.This advances research on key predictors of meaning in life (see King & Hicks, 2021) by demonstrating that competence, relatedness, and beneficence seem to be robust predictors of meaning in life over time.While cross-sectional work has associated psychological needs with meaning in life (Martela et al., 2018), this is the first longitudinal study to provide evidence for directional influence of psychological needs on meaning, thus confirming a key theoretical hypothesis put forward by SDT (Weinstein et al., 2012).The results also partially contrast with previous longitudinal work on meaningful work, that found that autonomy and beneficence were the most robust predictors of work meaningfulness, when controlling for the other factors (Martela et al., 2021).Perhaps autonomy is more challenged in work contexts, making its effects stand out better in that context compared to life in general.However, when examined alone without controlling for the three other predictors, autonomy predicted meaning in life over time, making it important to conduct more work to identify under what conditions autonomy contributes to meaning in life.Interestingly, we found that meaning in life was a positive prospective predictor of autonomy, competence, relatedness, and beneficence, both when taken individually and when controlling for the influence of the other factors.This was part of our exploratory analysis and thus warrants future research.But it is congruent with one previous longitudinal study that found in a Chinese college sample that, when taken individually, meaning in life prospectively predicted autonomy, competence, and relatedness (S.Zhang et al., 2022).High sense of meaning might motivate people to pursue self-congruent goals and mastery, and invest in relationships, thus contributing to future levels of autonomy, competence, relatedness, and beneficence.The results also revealed some interesting longitudinal dynamics between the four predictor variables.Autonomy prospectively predicted future levels of competence, relatedness prospectively predicted future levels of competence and beneficence, and beneficence prospectively predicted future levels of competence.These dynamics are interesting but, as they were not hypothesized, they merit future research before being confirmed.Nevertheless, identifying meaning in life as a potential predictor of psychological need satisfaction is not only relevant for research on meaningfulness but also for research in self-determination theory (Ryan & Deci, 2017;Vansteenkiste et al., 2020) that aims to identify key factors enhancing need satisfaction.Knowing that meaning in life could have potential to enhance need satisfaction can inform future interventions aiming to find ways to improve sense of autonomy, competence, and relatedness.
Besides theoretical contributions, these results can also inform practitioners aiming to enhance meaning in life in their own lives or as a counsellor, therapist, coach, or teacher.Meaning in life is a relatively abstract construct, and thus even individuals feeling a lack of meaning might not always know what to do to improve their sense of meaning in life.The finding that competence, relatedness, and beneficence can predict it over time helps people to focus on activities where they can experience competence, relationships where they can experience relatedness, and prosocial acts where they can experience beneficence as pathways to experiencing more meaning in life.
Certain limitations need to be acknowledged as regards the study.First, all data were collected through self-report, which can introduce common method bias.However, for psychological need satisfaction and sense of meaning, self-report is the most reliable source of data.Second, the sample was collected in one country, Chile, and thus merits to be replicated in other cultures.Third, although the longitudinal design provides evidence of temporal precedence from certain variables to others, this design does not rule out the possibility that a third variable could be involved, making it crucial to replicate the results while controlling for any identified potential confounders.Fourth, our research has focused on the main effects in the whole sample.Future studies with larger sample sizes could engage in various sub-group analyses to see whether, e.g., the initial level of meaning in life of the participants would affect the results.Fifth, while the identified estimates were of relatively small size, it is important to notice that they are lagged estimates that were controlled for by the baseline levels.Indeed, cross-lagged paths are usually smaller than concurrent estimates, but very relevant as they account for systematic variation controlling for individuals' previous states (Geiser, 2021).Further, taking a closer look at a recent meta-analysis analyzing effect sizes through various disciplines in psychology (Orth et al., 2022), the effect sizes we find in the current study are mostly relatively substantial compared to previous studies, as it is quite unrealistic to expect effects sizes above 0.12.Indeed, based on the advice provided by Orth et al. (2022), we interpreted the effect sizes of cross-lagged parameters as follows: 0.03 (small), 0.07 (medium), and 0.12 (large).
All in all, this study answers the call to examine potential predictors of meaning in life longitudinally and simultaneously, and finds that autonomy, competence, relatedness, and beneficence each prospectively predicts future levels of meaning in life when examined alone.When examined together, competence, relatedness, and beneficence remained significant prospective predictors of meaning.Furthermore, meaning in life itself prospectively predicted all four factors.This corroborates previous suggestions (S.Zhang et al., 2022) about a mutually reinforcing relationship between basic psychological needs and meaning in life that points out towards a possibility of a positive upward spiral (see Fredrickson & Joiner, 2002).The close mutual dynamics between the need satisfaction and meaning can help practitioners identify the most potential leverage points through which to support the well-being and meaningfulness of various individuals.