Self-determination theory and relationship quality

To explain how relationship quality influences work outcomes, the present study draws on Self-Determination Theory (SDT) [13]. The significance of SDT in partner relationships lies in the explanatory role of basic psychological need satisfaction in the experienced quality of interpersonal relationships [14]. SDT posits that all human beings share three universal basic psychological needs: autonomy, relatedness, and competence. The fulfillment of these needs is predictive of human well-being, regardless of individual differences, while the frustration of these needs is detrimental to well-being [13]. The need for autonomy involves a sense of volition, choice, and personal causation for the behaviors in which individuals engage [15]. Relatedness pertains to the need to feel cared for, care for others, feel at home in relationships with significant others, and belong [16]. The need for competence refers to feeling effective in one’s behaviors in playful exploration [17].

When the partner relationship is perceived as supportive of one’s sense of autonomy, it leads to fewer negative emotions and more positive behaviors [18]. Also, in the case of a relational conflict, it was found that autonomous reasons for being in the relationship and higher levels of need fulfillment of all three basic needs were predictive of satisfaction [19]. Moreover, a diary study revealed that autonomous reasons and need fulfillment in the relationship also contributed to the quality of the post-disagreement relationship [20]. The more one’s basic psychological needs are met, the easier it becomes to securely attach oneself and experience a close connection with one’s partner [21,22]. Overall, satisfying basic psychological needs is crucial for experiencing high-quality close relationships.

Opposite to the satisfaction of basic psychological needs, SDT posits the frustration of those needs. Need frustration is considered a separate concept from needs satisfaction and entails more than just not finding one’s needs for psychological need satisfaction met [23]. It involves a more active undermining of the partner’s needs, such as intentional need thwarting [24]. For instance, when partners feel pressured to behave in specific ways, the need for autonomy may be frustrated, which may concur with anger and irritation [21]. When a partner feels actively rejected, the need for relatedness is frustrated, which may coincide with emotions such as hurt and sadness [25]. In summary, it is expected that relationship quality positively relates to need satisfaction and negatively to need frustration. If relationship quality increases, it should reflect an increase in the satisfaction and a decrease in the frustration of one’s basic psychological needs. When the relationship quality drops, it should reflect a decline in satisfaction and an increase in the frustration of basic needs.

Studies in the domain of SDT also showed that the satisfaction and frustration of basic psychological needs often mediate the relationship between the quality of interpersonal relationships and health- and performance-related outcomes. Leadership studies found that an autonomy-supportive leadership style strengthens the employees’ sense of autonomy, competence, and relatedness, stimulating positive work outcomes and engagement [26] and that a leadership style characterized by controlling motives increased employees’ need frustration and was subsequently associated with poor work outcomes [27]. In sports coaching, athletes perform better and report higher psychological well-being when their coaches employ an autonomy-supportive coaching style that fulfills the athlete’s basic psychological needs [28]. When teachers in higher education adopt an autonomy-supportive approach that satisfies students’ basic needs, their motivation, well-being, and academic achievements benefit [29,30]. The same counts for parental autonomy support and children’s well-being and parental psychological control and children’s ill-being [31].

Through its mediating role, basic psychological needs theory adds explanatory value to the relationships under study [23,32]. SDT posits that basic psychological needs should be considered an underlying mechanism that explains the integration of extrinsic social elements and their impact on a person’s well-being and psychological growth [24,33]. Building on this theoretical and empirical work, we argue that if a partner relationship nurtures a person’s needs for autonomy, competence, and relatedness, it will facilitate positive integration and foster psychological well-being and growth, which may reflect on work-related states. If, in contrast, a private relationship frustrates or even thwarts a person’s basic psychological needs, it will lead to negative integration and ill-being.

Relationship quality, work engagement, and burnout

In the present study, we conceptualize relationship quality as an individual’s perception of the extent to which their romantic partnership meets their emotional and practical needs, emphasizing mutual support, commitment, and positive relational behaviors [34]. This functional definition avoids conflating relationship quality with related but distinct constructs such as satisfaction, stability, or agreement on major life domains. Instead, it focuses on how the relationship contributes to overall psychosocial well-being. More practically, it represents an approach that emphasizes functionality in the relationship, rather than focusing on happiness or relationship problems, and allows for diverging preferences in committed partner relationships [35].

Work engagement, as developed by Schaufeli and colleagues, is defined as a positive, fulfilling work-related state characterized by vigor (high energy and mental resilience), dedication (a sense of significance, enthusiasm, and inspiration), and absorption (full concentration and immersion in one’s work) [36]. Yet engagement is not unconditionally beneficial. When taken to extremes, high levels of engagement can lead to work–family conflict and relational strain, particularly when employees become overly involved in their jobs [37]. Similarly, burnout—defined by exhaustion, cynicism, and reduced efficacy—not only impairs work functioning but also negatively affects partner relationships and family dynamics [38,39]. One longitudinal study found that reducing work-family conflict reduced burnout [40]. Another study showed that employees high on burnout score lower on positive affect and work engagement in reaction to negative external stressors [41]. Theoretically, high-quality romantic partner relationships may bolster work engagement by providing emotional stability, enhancing resilience, and reinforcing intrinsic motivation [9]. Simultaneously, they may act as protective factors against burnout by alleviating stress and reducing the accumulation of work-related strain.

By investigating how the satisfaction and frustration of basic needs within partner relationships may mediate the association between the quality of the partner relationship and work-related states, we contribute to a more nuanced understanding of how relationship quality can shape work-related outcomes.

Hypotheses

H1. Relationship quality is positively associated with work engagement.

H2. Relationship quality is negatively associated with burnout.

H3. The satisfaction of basic psychological needs mediates the association between relationship quality and work outcomes, such that it (a) positively mediates the link with work engagement and (b) negatively mediates the link with burnout.

H4. The frustration of basic psychological needs mediates the association between relationship quality and work outcomes, such that it (a) negatively mediates the link with work engagement and (b) positively mediates the link with burnout.

We tested the hypotheses in a parallel mediation design, as depicted below in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study design.

https://doi.org/10.1371/journal.pone.0335404.g001

Sample

A self-report survey questionnaire was used to gather data from participants over 18 years old who were in committed relationships, worked at least eight hours per week, lived in the Netherlands, and were proficient in Dutch. Data were collected between September 20th, 2022, and December 20th, 2022. All participants provided written consent, confirming that they had understood the study’s information and voluntarily agreed to participate. Additionally, they consented to the study being used for scientific purposes and acknowledged that their information would be handled confidentially. Participants could stop completing the survey at any time and did not receive financial compensation. The online questionnaire comprised 89 items. We used a combination of convenience and snowball sampling. Convenience sampling was conducted via LinkedIn and Facebook, selecting participants who met the criteria, were available, and willing to participate. The additional snowball sampling helped us access participants through referrals in trusted social networks, which was important given the personal nature of the topic, resulting in N = 434 responses. After excluding incomplete responses, we retained N = 317 complete surveys.

The average age of the participants was 48 years (SD = 12), and 67% were female. On average, the respondents reported working 30 hours per week (SD = 10). Of the participants, 65% were in permanent employment, 67% were married, 7% had a registered partnership, 17% lived together without a registered partnership, and 9% reported not living together. Of all respondents, 95% reported being in monogamous relationships. The average duration of their relationships with their partner was 20 years (SD = 13). All participants had completed elementary education – the mandatory primary eight-year education in the Dutch schooling system between the ages of 4 and 12 – of which 10.5% did not pursue further studies, and 6% followed secondary vocational education. 38.5% had obtained higher vocational education, and 45% had pursued university studies.

Before the study, ethical clearance was obtained from the Ethical Review Board of Eindhoven University of Technology (ETC) granted on July 26, 2022, under reference number ERB2022IEIS21.

Instruments

The following measures were applied. The quality of partner relationships was assessed using the 9-item survey developed and validated by Chonody et al. [34]. The instrument was selected for its strong psychometric properties, including high internal consistency and demonstrated convergent and known-groups validity. The scale was developed using a large, diverse community sample (N = 8,132) from the Enduring Love Project (a mixed methods study on long-term adult couple relationships), which spanned the United Kingdom, the United States, and Australia. The sample included a broad range of ages, sexual orientations, relationship types (e.g., married, cohabiting, dating), and ethnicities, making the scale especially suitable for diverse populations. It provides a strengths-based evaluation of relationship quality, emphasizing positive relationship aspects such as commitment, mutual enjoyment, and shared values, and employs inclusive language, making it appropriate for various relationship types. The Dutch version of the original English survey was constructed using Brislin’s [42] back-translation procedure. Items were rated on a 5-point Likert scale ranging from 1 (Completely untrue) to 5 (Completely true). An example item for the quality of a partner relationship is: ‘I am totally committed to making this relationship work.’ Reliability for the relationship quality scale was α = .91.

Basic psychological need satisfaction and frustration in romantic relationships were measured using the 24-item scale developed and validated by Vanhee et al. [43]. The scale assesses basic psychological need satisfaction and frustration through the subscales of autonomy, relatedness, and competence. An item for autonomy satisfaction is: ‘In my relationship with my partner, I feel that my decisions reflect what I really want.’ An item for autonomy frustration is: ‘In my relationship with my partner, most of the things I do feel like I have to do.’ Items were rated on a 5-point Likert scale ranging from 1 (completely untrue) to 5 (completely true). The reliability for satisfaction and frustration levels of autonomy was α = .85 and α = .87, respectively. An item for competence satisfaction is ‘I feel capable at what I do’ (α = .82). For competence frustration ‘I feel insecure about my capabilities’ (α = .89). An item for relatedness is ‘I feel connected with him/her’ (α = .93). For relatedness frustration ‘I feel that our relationship is just superficial’ (α = .88). The reliability for the satisfaction of basic psychological needs scale as a whole was α = .923 and α = .924 for needs frustration.

To measure work engagement, we used the shortened 9-item Utrecht Work Engagement Scale [44] with the subscales of vigor, dedication, and absorption (α = .94). The number of occurrences rated item questions over time: (0) Never (1) Once a year or less; (2) At least once a year; (3) At least once a month; (4) At least a couple of times a month; (5) At least once a week; (6) A couple of times per week or daily. An example item for vigor (α = .88) is: ‘At my job, I feel bursting with energy.’ An example item for dedication (α = .90) is: “I am proud of the work that I do.” For absorption (α = .85): “I get carried away when I am working.”

Burnout was measured with the shortened 12-item Burnout Assessment Tool (BAT) [45], measuring exhaustion, mental distance, cognitive impairment, and emotional impairment, showing an overall reliability score of (α = .87). Items were rated on a 5-point Likert scale ranging from 1 (never) to 5 (always). An example item for the exhaustion subscale (α = .81) is: “At work, I feel mentally exhausted.” An example item for mental distancing (α = .80) is “At work, I do not think much about what I am doing, and I function on autopilot.” An example of cognitive impairment (α = .79) is “At work, I struggle to think clearly.” An example of emotional impairment (α = .78) is “At work, I feel unable to control my emotions.”

Lastly, we added a one-item question about relationship problems on a 5-point scale from never to continuously: “Have relationship issues been affecting your personal situation last month?” (M = 1.40, SD = .80), because we suspected relationship issues may influence the results. Other variables we controlled for were age, average number of hours worked per week, and form of employment (permanent, temporary, or self-employed/entrepreneur).

Preliminary analyses

We prepared the data for structural equation modeling (SEM) in JASP (Version 0.19.02). First, we estimated the correlation matrix (Table 1). The correlation matrix shows that the quality of the partner relationship strongly correlates with the satisfaction and frustration of basic psychological needs. The correlations between quality of the partner relationship and the work engagement subscales only indicated a significant correlation with vigor (p < .05). The correlation between quality of the partner relationship and burnout is more robust, showing two significant correlations for exhaustion (p < .01) and cognitive impairment (p < .05). The other correlations from relation quality to the subscales of work engagement and burnout were nonsignificant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Pearson’s correlation matrix.

https://doi.org/10.1371/journal.pone.0335404.t001

Measurement model

We then tested how the measurement model fitted the data by conducting a confirmatory factor analysis in the structural model in JASP and entered the syntax in lavaan. Our initial model (Fig 1) measured all scales as one factor: Relationship Quality (RQ), Needs Satisfaction (NDS), Needs Frustration (NDF), Work Engagement (WE), and Burnout (BO). We set the model estimation to WLSMV (weighted least square mean and variance-adjusted), which is a robust estimation method that does not require a normal distribution of data [46] and operates on the assumption that continuous latent responses follow a multivariate normal distribution [47]. Additionally, we selected the option to obtain additional fit measures. Model 1 fitted the data to an acceptable level, considering the general rules of thumb for assessing model fit. The chi-squared divided by the degrees of freedom was χ²/df < 3, but the CFI (Comparative Fit Index,.89) and TLI (Tucker-Lewis Index,.88) were < .90 [48]. Also, the RMSEA (Root Mean Square Error of Approximation (.064) was > .06. And, although the width of the confidence interval was small (.060 −.068) it returned significant where non-significance is preferred when estimating PCLOSE, the p-value of close fit [49,50]. Hence, we continued testing a second model for a better model fit. We estimated the dependent variables for work engagement and burnout through their subscales of vigor, dedication, and absorption for work engagement. For burnout, these were the subscales of exhaustion, mental distancing, and cognitive and emotional impairment. Model 2 had 10 factors against five for model 1 and showed a good fit to the data: CFI (.92) and TLI (.91) were > .90, the RMSEA showed a closer fit (.051), and the PCLOSE returned nonsignificant (p = .285).

Hypothesis testing

To test the hypotheses, we estimated model 2 in a parallel mediation model in JASP SEM with standardized estimates, a 95% confidence interval, and 5.000 bootstrap replications set to the bias-corrected percentile. Additional parameters were the total indirect effects and the standardized path coefficients (please see the supporting information S1 Table for all path coefficients). First, we examined the direct effects of the quality of the partner relationship on the satisfaction and frustration of basic psychological needs and on work engagement and burnout (Fig 2). We found a significant and positive path from the quality of the partner relationship to the satisfaction of basic needs and a negative path to the frustration of these needs. In contrast, and contrary to our assumptions about the positive path from quality of the partner relationship to work engagement, we found a negative path to vigor and dedication, and no association with absorption. We had to reject H1. The path from the quality of the partner relationship to burnout was positive and significant for exhaustion, mental distance, and emotional impairment, leading to rejecting H2 as well, as we hypothesized a negative path.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Model 2, direct and indirect effects.

https://doi.org/10.1371/journal.pone.0335404.g002

We then studied the indirect effects via the satisfaction of basic psychological needs and the frustration of those needs to test hypotheses 3 and 4. The results showed that the signs of the direct and indirect paths for the quality of the partner relationship to work engagement and burnout via need satisfaction were opposite. Usually, a mediation model’s direct and indirect effects share the same sign. There may be a suppression effect in the case of opposite signs [51]. We followed the procedure described by MacKinnon et al. [52] to establish whether there was a suppression effect in this mediation. In short, the procedure consists of finding whether the population’s third variable effect is zero, positive, or negative (⍺β = τ – τ’) and then calculating the difference between the total effect τ and the direct effect τ’ (Table 2). The analysis indicated that our model was a mediation model, and there was no suppression. Next, we conducted the Sobel test for the standard errors and calculated the confidence intervals (Table 3). We found that need satisfaction mediates the relationship between the quality of the partner relationship and work engagement for vigor and dedication (H3a) and with burnout for cognitive and emotional impairment (H3b). This confirmed the indirect effects we found in the mediation analysis (Fig 2). For need frustration, we found confirmation of the mediation of the association between the quality of the partner relationship and work engagement (H4a, vigor) and burnout (H4b, exhaustion, mental distance, and emotional impairment). However, contrary to our hypothesis, we found that need frustration associated positively with vigor and negatively with three of the four levels of burnout. Hence, we had to reject both hypotheses H4a and b.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Mediation or suppression for needs satisfaction & frustration.

https://doi.org/10.1371/journal.pone.0335404.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Confidence interval estimation of the mediation effect using the Sobel test.

https://doi.org/10.1371/journal.pone.0335404.t003

Strengths, limitations, and further research

The present study has notable strengths. As far as we were able to ascertain, this is the first study to explore the relationship between partner relationship quality and work engagement through the lens of basic psychological needs theory. The study’s design and specifically the examination of the mediating role of both basic needs satisfaction and needs frustration helps to gain insights into the dynamics between the quality of the partner relationship and work outcomes. Despite these strengths, several limitations should be considered. First, the study’s cross-sectional design precludes any conclusions about causality. While the findings suggest a relationship between quality of the partner relationship and work engagement through basic needs frustration and satisfaction, the directionality of these effects remains uncertain. Future longitudinal and experimental studies are needed to clarify whether relational need frustration leads to increased work investment or, for example, whether pre-existing tendencies toward high work engagement influence perceptions of relationship quality. Also, the role of need satisfaction in the partner relationship and its relation with work engagement and burnout would benefit from a longitudinal design. It would be of interest to know how this association may fluctuate over longer time periods.

Second, we relied on self-report measures, which can introduce the risk of common method bias. This bias might inflate or distort the relationships we observed [64]. To mitigate this potential issue ex ante, we utilized well-researched and validated psychometric instruments with clear and neutrally worded items. Ex post, we conducted Harman’s single-factor test, which revealed that the first factor accounted for 39% of the total variance, a figure well below the commonly accepted threshold of 50% [65]. We also performed a common latent factor analysis (CLF) using confirmatory factor analysis (CFA), comparing the applied measurement model with a modified model that incorporated a common latent factor allowed to load on all observed variables. The inclusion of the CLF did not significantly improve model fit (ΔCFI = .012, ΔRMSEA = .002), and the variance extracted by the common latent factor was AVE = 0, indicating that common method variance is unlikely to affect the validity of our findings [64].

Third, the generalizability of the findings is limited. Although we controlled for background variables our sample characteristics highlight potential constraints on generalizability: 45% of participants held a university degree, with an average age of 48 years (SD = 12) and an average relationship length of 20 years (SD = 13). This raises the question of whether the same patterns would emerge in younger generational cohorts, particularly individuals under 35 in relationships shorter than five or six years. Furthermore, the study sample was limited to Dutch-speaking citizens, which may restrict the applicability of our findings to populations with different cultural backgrounds or work environments. Variations in relationship norms, occupational structures, occupational categories, and socio-cultural influences on work engagement may yield different outcomes in other research contexts. Future research should prioritize replication in diverse samples and cross-cultural studies to assess the robustness, ecological validity, and generalizability of these effects. A last aspect that may limit the generalizability of our findings was the use of a convenience sample and snowball sampling, which brings the risk of selection bias, as participants often refer to similar others, which may lead to sample homogeneity.

Ultimately, the study confirms the prevailing notion that a strong partner relationship fosters work engagement and mitigates burnout. However, it challenges the assumption that a bad partner relationship that thwarts one’s basic needs decreases engagement and increases burnout. Nonetheless, the present study does not account for individual differences in coping strategies. Some individuals may seek refuge in work to compensate for relational distress. In contrast, others may disengage from work and personal commitments altogether. There are myriad ways of dealing with adverse circumstances in the relationship with one’s partner at home. In sum, this study underscores the complexity of the relationship between partner relationship quality and work engagement through the lens of basic psychological needs theory. Addressing the outlined limitations through longitudinal research, diary studies, and multi-method approaches will be essential to disentangle the mechanisms at play further.

Practical implications

The recent rise in work-related stress, increased employee absenteeism due to psychosocial factors, the growing number of burnouts, and the decline in work engagement make it essential for employers to safeguard the psychological well-being of their employees [11,66]. A recent survey of 26.000 employees across 35 countries showed that workers value a healthy work-life balance over pay or career advancement [12]. Also, employee retention rates and turnover intentions were found to correlate with work-life balance and the measures organizations take to support their staff. A recent study published in Nature revealed that hybrid work arrangements increased employee retention without negatively affecting performance and opportunities for promotion [67].

However, the blurring of boundaries between work and home—especially since the pandemic—has made employee well-being more complex and interdependent across life domains. While promoting work-life balance through flexible arrangements and the right to disconnect is essential [68], organizations should recognize that these measures alone may not be sufficient. Training managers to develop their capacity to recognize early warning signals—such as exhaustion, disengagement, or sudden shifts in motivation—and to engage in respectful, autonomy-supportive conversations with employees may be critical skills that help address issues before they escalate.

Expanding the role of managers in supporting employee well-being inevitably raises questions about the ethical boundaries of such involvement. Some may argue that such involvement intrudes on privacy and goes beyond the legitimate scope of managerial responsibility. Others may contend that when signs of distress—such as exhaustion, disengagement, unexpected spikes in energy, or a sudden aspiration for career advancement—become apparent at work, a sensitive and autonomy-supportive conversation that gently explores underlying factors, including relational well-being is both ethical and necessary [69].