1 Introduction

The prevailing high levels of ill-being in teachers and their detrimental consequences for the educational system have been widely reviewed and studied internationally (e.g. De Heus and Diekstra 1999). If we (as a society) want outstanding education for future generations, we must think about how we can create the best possible conditions for those who are supposed to provide it. While the development of health-promoting conditions at the political level is difficult and can only progress slowly by nature, opportunities for change at individual schools are much easier to discern and can be implemented on a low-threshold level. One starting point for fostering teachers’ well-being could be perceived autonomy support by the principal as it has often been identified as a relevant social-contextual factor for teachers’ well-being (e.g. Klassen et al. 2012), particularly in a current “era of strict accountability” for teachers (Webb 2002, p. 47). However, it seems worthwhile to first take a closer look at the underlying psychological mechanisms that mediate the relationship between perceived autonomy support and well-being or burnout in teachers. Within the framework of self-determination theory (Deci and Ryan 2000; Ryan and Deci 2000), satisfaction of the basic psychological needs has been shown to mediate this relationship (Klassen et al. 2012). There is also some initial evidence that frustration of the basic psychological needs is important for teachers as well (Bartholomew et al. 2014). However, in the teaching context need frustration and need satisfaction have not yet been examined simultaneously. The purpose of the present study is to examine how need satisfaction and need frustration differentially impact the relationship between autonomy support and positive as well as negative aspects of teachers’ well-being, and thus extending our understanding of the conditions of teachers’ health.

1.1 Teacher well-being and autonomy

Internationally, the teaching profession has long been associated with high levels of stress, emotional exhaustion, burnout, and health impairments (De Heus and Diekstra 1999) as well as high work attrition (Macdonald 1999). Moreover, teachers’ impaired well-being has been shown to be associated with impairments in instructional performance (Klusmann et al. 2008) and may lead to harm in the classroom and the school. However, while the negative aspects of teachers’ well-being have dominated in the literature, the teaching profession has also been shown to offer the potential to experience personal fulfilment (Roth et al. 2007), occupational well-being (such as job satisfaction; Klusmann et al. 2008) as well as general well-being (such as flourishing; Collie et al. 2016).

Drawing from a Positive Psychology perspective, focusing on subjective well-being might be particularly helpful for understanding this versatile potential of the teaching profession. Pavot and Diener’s (2013) concept of subjective well-being is defined as high life satisfaction, frequent positive, and infrequent negative affect. High subjective well-being has generally desirable personal consequences, such as better health, better health behavior, and longevity (Lyubomirsky and Lepper 1999). For teachers in particular, positive (i.e. enjoyment, happiness, enthusiasm, and satisfaction) and negative affect (i.e. anger, frustration, and anxiety) have been shown to influence educational outcomes in the classroom (Sutton and Wheatley 2003). First, teachers’ affect may influence instructional behavior and students’ motivation and learning. For example, teachers’ positive affect corresponds with self-reported frequency of using effective teaching strategies, such as asking students to write down key words on the topic described and creating links between different topics and subjects (Moè et al. 2010). Teachers’ positive and negative affect also correspond with students’ perceptions of teacher behavior, such as comprehensibility, autonomy support, and support after failure (Frenzel et al. 2009b). Moreover, Kunter (2013) showed in a one-year study with repeated measurements that mathematics teachers’ earlier enthusiasm for teaching, conceptualized as habitual positive affective experience in the exercise of the profession, explained changes in students’ mathematics achievement and joy in mathematics. Second, a number of studies have shown direct effects of emotional transmission in the classroom. For example, there are associations between teacher and student habitual enjoyment of mathematics classes, even when controlling for mathematics enjoyment in the previous year (Frenzel et al. 2009a). Teachers’ and students’ joy as well as anger and anxiety are also associated when using experience sampling methodology, even when controlling for students’ previous affect (Becker et al. 2014).

While teachers’ capacity to draw well-being from their work varies with personal factors, there is extensive literature on the effects of job-specific social-contextual demands (potential stressors), such as discipline problems, low student motivation and of job-specific resources, such as a supportive social climate and autonomy (Skaalvik and Skaalvik 2017). In particular, social-contextual factors related to (perceived) autonomy could be crucial for teachers’ well-being, considering that the teaching profession requires professional discretion on the one hand, and strict accountability on the other (Webb 2002). A valuable approach to further understand how these socio-contextual factors might affect teachers’ well-being lies within self-determination theory.

1.2 Need satisfaction and autonomy support

An integral component of self-determination theory (Deci and Ryan 2000; Ryan and Deci 2000) is the concept of three basic psychological needs for autonomy, competence, and relatedness (basic psychological need theory). The need for autonomy entails the desire to perceive oneself as the source of one’s own behavior. The need for competence is fulfilled when one experiences oneself as self-confident and able to act effectively. The need for relatedness refers to feeling connected with others, being important to others, and being accepted by them. The theory assumes that motivation, social development, and well-being are largely a function of the extent that the social-contextual conditions, in particular perceived autonomy support, affect the satisfaction of the basic psychological needs.

When the basic psychological needs are satisfied, intrinsic motivation, social development, and well-being are facilitated. It should be noted that while overall need satisfaction has shown to be important (e.g. Deci et al. 2001), each need generally also predicts unique variance (Van den Broeck et al. 2016). The relation of need satisfaction and well-being has been extensively shown in general as well as in specific life domains and in pervasive social contexts (see Ryan and Deci 2017 for an overview). In the educational field, the importance of need satisfaction has been amply shown. As an example for students, Ratelle and Duchesne (2014) conducted a longitudinal study over six years and found that high levels of need satisfaction were associated with higher levels of academic achievement and social as well as personal-emotional adjustment. For teachers, need satisfaction (at work) is associated with positive and negative aspects of well-being and outcomes in the classroom, e.g. job commitment and burnout (Abós Catalán et al. 2017), work enthusiasm (Aldrup et al. 2016), and observable teaching behavior (Korthagen and Evelein 2016).

Another assumption of self-determination theory is that perceived autonomy support is a crucial social-contextual antecedent for the satisfaction of the basic psychological needs and consequently for motivation and well-being (Ryan and Deci 2000). In work organizations, autonomy support “requires supervisor’s understanding and acknowledging their subordinates’ perspectives, providing meaningful information in an informational manner, offering opportunities for choice, and encouraging self-initiation” (Deci et al. 2001, p. 931). Accordingly, it could be shown that perceived autonomy support not only predicts (overall) need satisfaction, which in turn predicts motivation and well-being, but also that (overall) need satisfaction plays a mediating role in this relationship (Deci et al. 2001). It should be noted that self-determination theory shows that autonomy-supportive leaders who try to understand their employees’ perspective are more likely to facilitate competence and relatedness satisfaction in addition to autonomy satisfaction (Baard et al. 2004; Deci et al. 2001). Correspondingly, recent meta-analytic reviews of leader autonomy support in the workplace show that leader autonomy support and employee satisfaction of each of the three needs were strongly and positively correlated, though the strongest association was found with autonomy satisfaction (Slemp et al. 2018; Van den Broeck et al. 2016). Perceived autonomy support affects students (e.g. Froiland et al. 2016; Kaplan 2018) and principals (Chang et al. 2015) as well as teachers: For example, Klassen et al. (2012) showed that teachers’ satisfaction of each of the basic psychological needs at work, but mostly autonomy and relatedness with colleagues, mediated the relationship between perceived autonomy support by the principal and work engagement, emotions of anger, anxiety and joy as well as emotional exhaustion. Likewise, Collie et al. (2016) found that perceived autonomy support had a positive effect on teachers’ satisfaction of the basic psychological needs at work, in particular on satisfaction of the needs for autonomy and relatedness with colleagues, and, to a lesser extent, on competence and relatedness with students. In turn, need satisfaction of each need, but mostly competence satisfaction, affected general well-being (flourishing), and satisfaction of the needs for autonomy and relatedness with colleagues affected occupational well-being.

1.3 Differentiation of need satisfaction and need frustration

Although Deci and Ryan (2000) propose the additional psychological mechanism of need frustration (or need thwarting) in order to explain degradation and ill-being, research in self-determination theory has traditionally focused on need satisfaction, optimal development, and well-being. It has thus been argued that self-determination theory is more suitable to explain positive (e.g. positive affect) than negative aspects (e.g. negative affect) of human nature (Pyszczynski et al. 2000; Van den Broeck et al. 2016). However, in more recent years the distinction between need satisfaction and need frustration has been taken up in order to better capture the “darker side of human existence” (Bartholomew et al. 2014, p. 105).

In this line, it is assumed that need frustration is more than just the mere absence of need satisfaction: The need for autonomy is frustrated by feeling controlled by external or self-imposed pressure. The need for relatedness is frustrated by the experience of exclusion and loneliness. The need for competence is frustrated by feelings of failure and doubts about one’s own effectiveness (Chen et al. 2015). As an example for the teaching context, low competence satisfaction may be experienced when a teacher feels that he or she does not have sufficient skills to perform well. By contrast, the need for competence might be frustrated when the school environment and/or colleagues are overly critical or demeaning (Cuevas et al. 2015, p. 2).

In accordance with these theoretical assumptions, need satisfaction and need frustration could be identified in factor analyses as separate dimensions (Chen et al. 2015; Cordeiro et al. 2016) that predict different criterion variables (Bartholomew et al. 2011a, b; Chen et al. 2015; Vansteenkiste and Ryan 2013). For example, it has empirically been shown that whereas need satisfaction mainly affects positive aspects of well-being (e.g. vitality) and not negative aspects of well-being, need frustration mainly affects negative aspects of well-being (e.g. depressive symptoms, somatization) and not positive aspects of well-being (Chen et al. 2015; Cordeiro et al. 2016). For the sports context, Bartholomew et al. (2011a) examined need satisfaction and need frustration simultaneously within a more complex model. Amongst other variables, they related athletes’ sports-related need satisfaction with perceived trainer autonomy support and positive as well as negative aspects of well-being. On the one hand, trainer autonomy support had a positive effect on need satisfaction. In turn, need satisfaction positively influenced positive aspects of well-being and, to some extent, also negatively influenced negative aspects of well-being. On the other hand, autonomy support negatively influenced need frustration, meaning that low autonomy support was associated with high need frustration, and vice versa. In turn, need frustration only positively affected negative aspects of well-being but was not associated with positive aspects of well-being.

The frustration of basic psychological needs has already been proven relevant for teachers. Need frustration at work, in particular competence frustration, was shown to predict burnout in physical education teachers (Cuevas et al. 2015). Furthermore, Bartholomew et al. (2014) found that the frustration of each of the basic needs at work, but mostly autonomy frustration, mediated the relationship between work pressure (e.g. from school authorities) and burnout of physical education teachers, whereas only competence frustration mediated the effect of work pressure on somatic complaints.

1.4 The present study

Although there is ample evidence that both satisfaction and frustration of the basic psychological needs explain the relationship between the work context and teachers’ well-being, the two dimensions have not yet been examined together in the teaching context. However, since the teaching profession is characterized by both the risk of high ill-being and the potential for high well-being, the joint examination of need satisfaction and need frustration appears especially promising. The aim of this study is therefore to explain the underlying psychological mechanisms of positive and negative aspects of well-being in teachers more precisely by examining need satisfaction and need frustration as potential differential mediators.

As Pavot and Diener’s (2013) concept of subjective well-being might be particularly helpful for understanding the versatile potential of the teaching profession, we focus on life satisfaction and the frequency of positive affect as positive aspects of well-being and on the frequency of negative affect as a negative aspect of well-being. In addition, emotional exhaustion (as the leading symptom of burnout; Maslach and Jackson 1986) is included as another negative aspect of well-being as it seems to be particularly affecting teachers (De Heus and Diekstra 1999).

Our central assumption is that need satisfaction is a mediator for the positive aspects of well-being (life satisfaction, positive affect) and not the negative aspects of well-being, whereas need frustration is a mediator for the negative aspects of well-being (negative affect, emotional exhaustion) and not the positive aspects. Since it can be assumed that autonomy support affects all three basic psychological needs (Baard et al. 2004), the hypotheses are formulated for the overall scores of need satisfaction and need frustration which are aggregated over the three basic psychological needs. The following hypotheses are also graphically illustrated in Fig. 1.

Fig. 1
figure 1

Hypothetical relationships between autonomy support, need satisfaction, need frustration, and positive and negative aspects of well-being. The dashed lines represent the relationships that are assumed to be insignificant. H: hypothesis

Full size image

Hypothesis 1 (H1)

The effect of autonomy support on life satisfaction is mediated through need satisfaction, but not through need frustration.

Hypothesis 2 (H2)

The effect of autonomy support on the frequency of positive emotions is mediated through need satisfaction, but not through need frustration.

Hypothesis 3 (H3)

The effect of autonomy support on the frequency of negative emotions is mediated through need frustration, but not through need satisfaction.

Hypothesis 4 (H4)

The effect of autonomy support on emotional exhaustion is mediated through need frustration, but not through need satisfaction.

These hypotheses are additionally tested for each of the individual needs, meaning that the pairings of autonomy-satisfaction and -frustration, competence-satisfaction and -frustration, and relatedness-satisfaction and -frustration are each assumed to be differential mediators. This serves the purpose to determine to which of the individual needs the effects of overall need satisfaction and need frustration can primarily be attributed.

2 Method

2.1 Sample and procedure

The study was approved by the State Supervisory School Authority. Initially, this study involved 56 secondary school teachers from six schools in a major German city. Due to missing values in the indispensable predictor variables for each hypothesis, autonomy support, need satisfaction as well as need frustration, the final sample size was N = 49. A large majority of participants (n = 46) were higher secondary school teachers (“Gymnasium”) and n = 3 were lower secondary school teachers (“Realschule”). Among the participating teachers, there were 31 women (63%) and 18 men. This gender distribution corresponds to the German average of 62% female teachers and the OECD average of 64% female teachers in secondary education (Organization for Economic Co-operation and Development [OECD] 2017). On average, the participants were 41.80 years old (SD = 9.65) with an age distribution of 4% under 30 years, 75% 30–49 years, and 21% 50 years or older. This age distribution is below the average age distribution of secondary school teachers in Germany (7% under 30 years, 48% 30–49 years, and 45% 50 years or older) and slightly below the OECD average age distribution (10% under 30 years, 54% 30–49 years, and 35% 50 years or older; OECD 2017). Of the participating teachers, 30 worked in a full-time position, 18 in a part-time position and one person made no statement. A majority of 40 participants lived in a permanent relationship, while seven participants did not live in a permanent relationship, and two persons did not give any information.

The teachers were recruited through a project that examined well-being in students and teachers by means of an enquiry to all secondary school administrations of the respective German city. Participation was voluntary and informed consent was obtained from each teacher. For data collection, a questionnaire was completed by the teachers, which took about 15 min. Some of the teachers (n = 21) completed the questionnaire within one allocated class hour, while their pupils completed another questionnaire. The other questionnaires were distributed to and returned by the teachers in the schools during the same period of time.

2.2 Measures

2.2.1 Perceived autonomy support by the school principal

Perceived autonomy support by the school principal was assessed with our own translation of the short form of the Work Climate Questionnaire (WCQ; Baard et al. 2004). For this purpose, two fluently English-speaking psychologists first translated the questionnaire independently from English into German. These two translations were discussed and a solution was worked out for each item. Afterwards, another fluently English-speaking psychologist translated the text back into English. The resulting ambiguities and open questions were discussed and clarified until a consensus was reached. On a five-point scale ranging from 1 (strongly disagree) to 5 (strongly agree), six items were used to assess the extent to which teachers perceived their school principals as autonomy-supportive (e.g. “My school principal tries to understand how I see things before he or she suggests a new way to do things”).

2.2.2 Satisfaction and frustration of basic psychological needs

Satisfaction and frustration of basic psychological needs was measured by the German translation of the Basic Psychological Need Satisfaction and Frustration Scale (BPNSFS; Original: Chen et al. 2015; German version: Heissel et al. 2018). On a five-point scale ranging from 1 (completely disagree) to 5 (completely agree), the experience of need satisfaction and need frustration was assessed with 24 items. The BPNSFS is multi-dimensionally structured with three subscales for need satisfaction and three subscales for need frustration with four items for each subscale. The three subscales for need satisfaction contain the satisfaction of the need for autonomy (e.g. “I feel that my decisions reflect what I really want”), competence (e.g. “I feel competent to achieve my goals”) and relatedness (e.g. “I experience a warm feeling with the people I spend time with”). The three subscales for need frustration contain the frustration of the need for autonomy (e.g. “I feel forced to do many things I wouldn’t choose to do”), competence (e.g. “I feel disappointed with many of my performances”), and relatedness (e.g. “I feel the relationships I have are just superficial”). The three subscales for need satisfaction and the three subscales for need frustration were also combined into overall scores of need satisfaction and need frustration, respectively.

2.2.3 Life satisfaction

As the cognitive component of subjective well-being, general life satisfaction was assessed with the eponymous subscale of the Habituelle Subjektive Wohlbefindensskala (HSWBS; Dalbert 1992). On a six-point scale ranging from 1 (not true at all) to 6 (exactly true), teachers rated seven statements on general life satisfaction (e.g. “I am satisfied with my life”).

2.2.4 Frequency of experiencing positive and negative affect

As the affective component of subjective well-being, frequency of experiencing positive and negative affect in the last four weeks was measured with the Scale of Positive and Negative Experience (SPANE; Original: Diener et al. 2010; German version: Rahm et al. 2017). On a five-point scale ranging from 1 (very rarely or never) to 5 (very often or always), the frequency of positive affect (e.g. “pleasant”) and the frequency of negative affect (e.g. “unpleasant”) were assessed with six items each.

2.2.5 Emotional exhaustion

Emotional exhaustion as the leading symptom of burnout was recorded with the eponymous subscale from the Maslach Burnout Inventory (MBI; Original: Maslach and Jackson 1981, 1986; German version: Büssing and Perrar 1992). On a six-point scale ranging from 1 (never) to 6 (very often), the frequency of feeling emotionally exhausted in relation to work was assessed with five items (e.g. “I feel emotionally drained from my work”).

2.2.6 Socio-demographic data

As socio-demographic data, terms of employment (full-time/part-time) was assessed as well as gender, age and relationship status (living in a permanent relationship/not living in a permanent relationship). Terms of employment was assessed because part-time work has been shown to have mixed effects on well-being, in particular in interaction with gender (e.g. Montero and Rau 2015). Gender and age were assessed because there are mixed findings on gender and age differences in well-being (see Lucas and Gohm 2000). Relationship status was assessed because marital status is a well-established, important predictor of well-being with married persons reporting higher well-being than unmarried persons (e.g. Myers 2000).

2.3 Data analysis

The statistical analysis was performed with IBM SPSS Statistics 24 and the SPSS add-on module PROCESS 2.16.3 (Hayes 2013). Parallel multiple mediation analyses were calculated according to Preacher and Hayes (2008) and Hayes (2013). The estimation of indirect effects in a parallel multiple mediator model with need satisfaction and need frustration as potential mediators allows the simultaneous testing of both psychological mechanisms, taking into account the relationship between them. Thus, each of the analyses included autonomy support as a predictor, need satisfaction and need frustration as parallel mediators, and, depending on the hypothesis, an indicator of well-being or emotional exhaustion as a criterion. Due to pairwise deletion of the scattered missing values in the respective criteria (depending on the hypothesis), the sample size in question varied from N = 46 to N = 48.

Figure 2 depicts the analyzed statistical model according to Hayes (2013, p. 127). This model shows that the criterion must be predicted by three regression equations in order to determine the indirect effects: In the first and second step, two regression analyses are carried out with autonomy support predicting the two mediators respectively (effects a1 for need satisfaction and a2 for need frustration). In the third step, a third regression analysis is carried out with autonomy support and the two mediators as predictors of the criterion. The following effects are estimated simultaneously in this regression analysis: (1) the direct effect c′ of autonomy support on the criterion taking into account the two mediators, (2) the effect b1 of need satisfaction on the criterion taking into account autonomy support and need frustration, and (3) the effect b2 of need frustration on the criterion taking into account autonomy support and need satisfaction. For these three regression analyses, the model summaries and standardized regression coefficients are presented in the results section.

Fig. 2
figure 2

Statistical model of the parallel multiple mediation analyses for hypotheses 1–4, depending on the criterion (Hayes 2013, p. 127). Path coefficients a1 and a2 are estimated through simple regression analyses with autonomy support as a predictor, path coefficients b1, b2, and c′ are estimated in one multiple regression analysis with autonomy support, need satisfaction, and need frustration as predictors

Full size image

Subsequently, the specific indirect effects of autonomy support on the respective criterion through the two parallel mediators, need satisfaction and need frustration, are estimated and tested for significance. Each specific indirect effect is calculated as the product of the two corresponding path coefficients: a1b1 for the specific indirect effect of need satisfaction and a2b2 for the specific indirect effect of need frustration. In Table 2 in the results section, these specific indirect effects are reported as completely standardized indirect effects which are a measure of effect size in mediation analysis (Hayes 2013, pp. 184–188). Significance testing of the indirect effects is performed using 95% bias-corrected confidence intervals (BC 95% CI) from 10,000 bootstrap samples (Hayes 2013, pp. 405–406). According to this approach, the indirect effects are considered to be significant if the 95% confidence interval does not include zero.

In addition to the first set of tests with the aggregated overall scores of need satisfaction and need frustration, the hypotheses were also tested with a second set of tests for each of the individual basic psychological needs. For each hypothesis, three additional parallel multiple mediation analyses were carried out according to the procedure described above, with the satisfaction and frustration of a single basic psychological need as parallel mediators (AUT = need for autonomy, COMP = need for competence, REL = need for relatedness). Thus, analogous to the first set of tests, the second set of tests also includes two mediators, need satisfaction and need frustration, in one domain. For example, in hypothesis 1 in order to predict life satisfaction, there was a first additional analysis with autonomy-satisfaction and -frustration as potential mediators (AUT), a second additional analysis with competence-satisfaction and -frustration as potential mediators (COMP), and a third additional analysis with relatedness-satisfaction and -frustration (REL). For these analyses, only the completely standardized indirect effects and their bias-corrected 95% confidence intervals are reported in Table 2 in the results section.

3 Results

Table 1 shows the descriptive statistics, reliabilities (Cronbach’s alpha) and Pearson correlations for the variables in the study. The reliabilities of the main scales can be regarded as satisfactory to good with .78 ≤ α ≤ .91, even though the reliabilities of some of the subscales of need satisfaction and frustration are slightly lower. For all variables, significant correlations were found in the expected direction, i.e. the predictor (autonomy support) correlated with the mediators (need satisfaction and need frustration), and the mediators correlated with the criteria (life satisfaction, positive affect, negative affect, emotional exhaustion). The sociodemographic characteristics age and terms of employment collected in addition to the variables presented in Table 1 do not correlate with any of the criterion variables. Gender correlated significantly with the criterion variable positive affect (see Table 1), with women showing more favorable scores than men. Also, relationship status correlated significantly with the criterion variable emotional exhaustion (see Table 1), with respondents not living in a permanent relationship showing less favorable scores than respondents living in a permanent relationship, which is in accordance with the literature, e.g. Myers (2000). In order to control for these effects, for each hypothesis two additional analyses were carried out, one with gender and another one with relationship status as a covariate.

Table 1 Descriptive statistics, reliabilities (Cronbach’s alpha) and Pearson correlations for variables
Full size table

In the following section, for each hypothesis the results of the respective parallel mediation model are reported. The results of the first and second step of each analysis slightly differ because of varying sample sizes due to the pairwise deletion of missing data.

Hypothesis 1

The parallel multiple mediation analysis for predicting life satisfaction included N = 48 teachers. The first step showed that autonomy support significantly predicted need satisfaction with a1 = .44, p < .01; R2 = .19, F (1, 46) = 11.09, p < .01. The second step showed that autonomy support significantly predicted need frustration with a2 = − .57, p < .001; R2 = .33, F (1, 46) = 22.12, p < .001. In the third step, it was shown that life satisfaction was significantly predicted by autonomy support, need satisfaction, and need frustration taken together, R2 = .50, F (3, 44) = 14.46, p < .001. However, autonomy support had no direct effect on life satisfaction with c′ = .10, p = .45, whereas need satisfaction with b1 = .40, p < .01 and need frustration with b2 = − .33, p = .03 had significant effects on life satisfaction. The results of the fourth step are shown in Table 2. The specific indirect effect through need satisfaction of a1b1 = .18 was significant according to the BC 95% CI. The specific indirect effect through need frustration of a2b2 = .19 was also significant. Moreover, it turned out that under control of relationship status, need frustration did not remain a significant mediator (even though relationship status itself was not a significant predictor). Additional testing at the level of the individual basic psychological needs showed that these findings were accounted for only by the need for autonomy, since only autonomy satisfaction (a1b1AUT = .21) and autonomy frustration (a2b2AUT = .24) were significant mediators (see Table 2). The results therefore indicate that the effect of autonomy support on life satisfaction is not only mediated through need satisfaction, as expected, but also through need frustration, although this effect is not robust against socio-demographic control variables. Thus hypothesis 1 is only partially confirmed.

Table 2 Indirect effects of autonomy support on well-being through overall scores of need satisfaction and need frustration, and each of the basic psychological needs
Full size table

Hypothesis 2

The parallel multiple mediation analysis to predict positive affect included N = 47 teachers. The first step showed that autonomy support significantly predicted need satisfaction with a1 = .44, p < .01; R2 = .19, F (1, 45) = 10.61, p < .01. The second step showed that autonomy support significantly predicted need frustration with a2 = − .57, p < .001; R2 = .32, F (1, 45) = 21.07, p < .001. In the third step, it was shown that positive affect was significantly predicted by autonomy support, need satisfaction, and need frustration taken together, R2 = .41, F (3, 43) = 9.94, p < .001. However, autonomy support had no direct effect on positive affect, c′ = .18, p = .21, need satisfaction had a significant effect with b1 = .41, p < .01 and need frustration had no effect with b2 = − .17, p = .28. The results of the fourth step are shown in Table 2. It was found that the specific indirect effect through need satisfaction of a1b1 = .18 was significant according to the BC 95% CI, whereas the specific indirect effect through need frustration proved to be insignificant. Moreover, need satisfaction did not remain a mediator when controlling for gender (even though gender itself was not a significant predictor). Additional testing at the level of the individual basic psychological needs showed that this finding was exclusively accounted for by autonomy satisfaction (a1b1AUT = .20, see Table 2). The results therefore indicate, as expected, that the effect of autonomy support on positive affect is only mediated through need satisfaction and not through need frustration. However, the mediator effect of need satisfaction is not robust against the sociodemographic control variables. Therefore, hypothesis 2 is only partially confirmed.

Hypothesis 3

The parallel multiple mediation analysis to predict negative affect included N = 46 teachers. The first step showed that autonomy support significantly predicted need satisfaction with a1 = .43, p < .01; R2 = .19, F (1, 44) = 10.23, p < .01. The second step showed that autonomy support also significantly predicted need frustration with a2 = − .58, p < .001; R2 = .33, F (1, 44) = 21.85, p < .001. In the third step, it was shown that negative affect was predicted significantly by autonomy support, need satisfaction, and need frustration taken together, R2 = .45, F (3, 42) = 11.56, p < .001. However, autonomy support had no direct effect on negative affect with c′ = .10, p = .48, need satisfaction also had no effect with b1 = − .11, p = .46, whereas need frustration had a significant effect with b2 = .66, p < .001. The results of the fourth step are shown in Table 2. It was found that the specific indirect effect through need satisfaction was not significant, whereas the specific indirect effect through need frustration of a2b2 = − .38 was significant according to the BC 95% CI. Even when controlling for gender and/or relationship status, need frustration remained a significant mediator. Additional testing at the level of individual basic psychological needs showed that these findings were significantly accounted for by the frustration of all three basic psychological needs: Frustration of autonomy (a2b2AUT = − .36), competence (a2b2COMP = − .14), and relatedness (a2b2REL = − .16) represented significant mediators (see Table 2). As expected, the results show that the effect of autonomy support on negative affect is only mediated through need frustration and not through need satisfaction and that this effect remains robust against sociodemographic control variables, thus confirming hypothesis 3.

Hypothesis 4

The parallel multiple mediation analysis to predict emotional exhaustion included N = 47 teachers. The first step showed that autonomy support significantly predicted need satisfaction with a1 = .44, p < .01; R2 = .19, F (1, 45) = 10.80, p < .01. The second step showed that autonomy support significantly predicted need frustration with a2 = − .59, p < .001; R2 = .34, F (1, 45) = 24.15, p < .001. In the third step, it was shown that emotional exhaustion was significantly predicted by autonomy support, need satisfaction, and need frustration taken together, R2 = .58, F (3, 43) = 19.57, p < .001. Autonomy support had no direct effect on emotional exhaustion with c′ = − .14, p = .28, need satisfaction had no effect with b1 = .07, p = .55, whereas need frustration had a significant effect with b2 = .71, p < .001. The results of the fourth step are shown in Table 2. It was found that the specific indirect effect through need satisfaction was not significant, whereas the specific indirect effect of need frustration of a2b2 = − .42 reached significance according to the BC 95% CI. When controlling for gender and/or relationship status, need frustration remained a significant mediator. Additional testing at the level of the individual basic psychological needs showed that these findings were significantly accounted for by the frustration of all basic psychological needs: Frustration of autonomy (a2b2AUT = − .42), as well as competence (a2b2COMP = − .21) and relatedness (a2b2REL = − .10) were significant mediators (see Table 2). The results indicate, as expected, that the effect of autonomy support on emotional exhaustion is only mediated through need frustration and not through need satisfaction and that this effect remains robust against sociodemographic control variables, thus confirming hypothesis 4.

4 Discussion

Overall, the findings of this study suggest that need satisfaction and need frustration are differential mediators of the relationship between perceived autonomy support by the school principal and positive and negative aspects of well-being in teachers. Contrary to our expectation, for the relationship between autonomy support and life satisfaction, both need satisfaction and need frustration were significant mediators. However, according to our expectations, only need satisfaction emerged as a significant mediator of the relationship between perceived autonomy support and positive affect. This means that teachers experience positive affect when their basic psychological needs (especially the need for autonomy) are satisfied by an autonomy-supportive principal. Because effect sizes appear similar in size and do not statistically differ, as additional analyses show, it needs to be considered that the non-significant mediating effect of need frustration on positive affect might be due to low sample size. Therefore, the interpretation of differential mediations through need satisfaction and need frustration on positive affect can only be provisionally and cautiously supported.

Again, according to our expectations, only need frustration played a significant role in mediating the relationship between perceived autonomy support and negative affect as well as emotional exhaustion. This means that teachers experience negative affect and emotional exhaustion when their basic psychological needs (needs for autonomy, competence, and relatedness) are frustrated by a lack of autonomy support. The mediation effects through need frustration on negative aspects of well-being were robust against sociodemographic control variables. Even more, the effect sizes for need frustration are significantly larger than for need satisfaction, as additional analyses show. These results further strengthen the interpretation of need satisfaction and frustration as differential mediators for negative aspects of well-being.

Altogether, it can be concluded that teachers’ positive and negative aspects of well-being at one point in time can be explained by different psychological mechanisms and considering them separately accounts more adequately for the versatile potential of the teaching profession. These findings correspond to current theoretical considerations (e.g. Vansteenkiste and Ryan 2013) and extend the empirical findings on the distinction between need satisfaction and need frustration (Bartholomew et al. 2011a, b; Chen et al. 2015; Cordeiro et al. 2016; Van den Broeck et al. 2016) and on teachers’ basic psychological needs (Bartholomew et al. 2014; Klassen et al. 2012).

Moreover, the findings at the level of the three basic psychological needs also indicated that the amount of autonomy support by the principal is more important for frustrating basic psychological needs than it is for satisfying them: Autonomy support by the school principal only had an effect on the positive aspects of well-being through the need for autonomy and not through the needs for competence and relatedness. By contrast, autonomy support by the school principal had a more extensive effect on the negative aspects through the frustration of all three basic psychological needs. These findings further emphasize the need to consider need frustration in explaining the high levels of ill-being in teachers (Bartholomew et al. 2014; Pyszczynski et al. 2000). Additionally, these findings show that the effects of each of the three needs should also be considered separately (Van den Broeck et al. 2016).

Contrary to our expectation, the effect of autonomy support on life satisfaction was mediated both through need satisfaction and need frustration, even though need frustration did not remain a significant mediator when controlling for relationship status. The literature is inconsistent on this issue: Cordeiro et al. (2016) showed for Portuguese adolescents and university students that only need satisfaction and not need frustration predicted life satisfaction longitudinally. Conversely, Chen et al. (2015) found in a longitudinal cross-cultural study with adolescents and university students that life satisfaction was predicted by need satisfaction and to a lesser extent by need frustration. Nevertheless, it seems reasonable to assume that life satisfaction as a global cognitive evaluation of one’s own life (Diener et al. 1985) is influenced by positive and negative experiences of need satisfaction and need frustration — possibly all the more in adult teachers than in adolescents and university students because of a more balanced spectrum of life experiences with age.

4.1 Practical implications

Consistent with the literature (e.g. Klassen et al. 2012), the findings of the present study stress the importance of autonomy support by the school principal for teachers’ well-being. Furthermore, we were able to show that the amount of autonomy support by the principal in particular influences the ill-being of teachers mediated through the frustration of basic psychological needs. The present findings can thus be used to derive strategies to reduce teachers’ ill-being and to increase their well-being. First, by means of consulting, supervision, and coaching, principals could be trained in providing autonomy support and in particularly be sensitized on how their actions could lead to the experience of need frustration. There is some initial evidence that leaders’ training consistent with self-determination theory, specifically including how to support the autonomy of employees, can be beneficial for leaders as well as for their employees (Deci et al. 1989; Hardré and Reeve 2009). Thus, sensitizing for autonomy support and in particular need frustration might be a promising approach that could contribute to the reduction of teachers’ need frustration. Second, interventions and measures to promote teachers’ well-being (or to reduce their ill-being) should take into account experiences of need frustration and foster coping skills, e.g. through the practice of mental processing strategies or systematic problem solving. Since mentoring interventions on basic psychological need satisfaction have already been shown to foster need satisfaction at work in the short-term (Lewis et al. 2016), this might also be a feasible approach.

4.2 Limitations and future perspectives

The main limitation of the present study is the rather small sample size, so that some of the non-significant results might be due to low power. However, analyses with control variables and considering the effect sizes point to the fact that the interpretation of differential mediation effects appears conclusive at least for the negative outcomes. Furthermore, since in our sample the age distribution slightly differs from that of the German and OECD secondary teacher population, mainly upper secondary schools took part, and teachers participated on the basis of self-selection, a replication should include a more representative sample of teachers. Due to these limitations in the sample, the findings can only be considered as first indications of differential mediation effects that should be replicated with larger samples. Additionally, with a larger number of participating schools and teachers per school, multi-level analyses could be applied to systematically separate effects at the school level from effects at the individual level. In order to examine whether the same psychological mechanisms are responsible for the medium and long-term development of teachers’ well-being and ill-being, longitudinal studies should be carried out in addition to cross-sectional studies. In the present study, general need satisfaction and need frustration were assessed in order to explain subjective well-being as a global construct. However, while general and domain-specific need satisfaction are related, it has been shown that they also uniquely contribute to well-being (Milyavskaya et al. 2013). Therefore, need satisfaction and need frustration at work (Schultz et al. 2015) should also be utilized in order to differentially explain effects on teachers’ occupational well-being and motivation.

The results of this study emphasize the role of principal autonomy support. However, it must also be considered that principals themselves face autonomy conflicts between meeting the specific needs of their schools and complying with top-down mandates and policies, and are affected by their superintendents’ autonomy support (Chang et al. 2015). Thus, the ways in which principals might actually be able to provide autonomy support for their teachers within a relatively narrow framework and how their behavior might be perceived by the teachers need to be further examined.

Considering the importance of self-determination, especially need satisfaction, for students (Ratelle and Duchesne 2014), teachers (Klassen et al. 2012), and principals (Chang et al. 2015), another future perspective might lie in investigating the interplay of perceived autonomy support and need satisfaction as well as need frustration in teachers, students, and principals. Clarification of these relationships might be worthwhile in order to develop whole-school level approaches promoting self-determination.

5 Conclusions

Despite the limitations mentioned, the present study provides differentiated insights into the underlying psychological mechanisms of how perceived autonomy support by the principal as a relevant social-contextual factor affects teachers’ positive and negative aspects of well-being. Our results extend the use of self-determination theory in the educational field by showing that considering need satisfaction and need frustration together accounts more adequately for the versatile positive and negative potential of the teaching profession than considering need satisfaction alone. Moreover, need frustration was shown to be particularly informative for the explanation of negative aspects of well-being in teachers and might be useful in better understanding the development of burnout and somatic symptoms. These findings offer guidance for school administrators and for teachers themselves to reduce the high level of ill-being experienced by teachers and to increase their sense of well-being.