The exchange of social support is an essential aspect of social interactions (Knoll and Kienle, 2007) which involves the qualitative aspect of helping behavior between two persons. Multiple types of such supportive interactions can be distinguished. Perceived or anticipated social support entails the support a person expects to be available in their social network when help is needed (Knoll and Kienle, 2007). Studies indicated that perceived social support is rather a stable than a modifiable characteristic (e.g., Sarason et al., 1987). Thus, perceived social support is rather independent from the actual behavior a specific network member displays and therefore not an optimal indicator to measure supportive interactions (Knoll and Kienle, 2007). In contrast, received social support is a retrospective report that reflects actual support transactions from specific network members (Uchino, 2009). The present study focusses on received social support.

The concept of job crafting (Wrzesniewski and Dutton, 2001; Tims and Bakker, 2010) provides a bottom-up perspective from which we can examine what teachers individually can contribute to the social support they receive and therefore to their satisfaction of the need for relatedness. Job crafting involves a process in which employees individually and proactively adjust job demands and job resources to make their job better fit their needs, preferences, and abilities (Wrzesniewski and Dutton, 2001; Tims and Bakker, 2010). This is in contrast to a top-down perspective that focuses on changes in the work environment initiated by supervisors. Searching for social support is a central element of job crafting that aims at actively increasing the social support one receives (Tims et al., 2012). First studies from occupational health psychology that examined samples outside the context of schools suggest that searching for social support can increase the receipt of social support (Tims et al., 2013; Demerouti et al., 2015). However, whether teachers’ individual and proactive search for social support may result in an actual increase in the receipt of support is unclear.

Furthermore, teachers have two sources at school to which they might turn to for receiving social support: the school principal and colleagues. The nature of the relationship between the one giving and the one receiving support may influence the effects of social support (Sarason et al., 1994). Therefore, we explore possible differences in the effect of teachers’ search for social support on the receipt of social support from the school principal and from colleagues and possible differences in their effect on the satisfaction of the need for relatedness.

The basic psychological need for relatedness involves feelings of respect, understanding, and connectedness which result from strong interpersonal relationships (Deci and Ryan, 2000). The need to form and maintain such relationships is considered to be universal and central to human psychological functioning (Baumeister and Leary, 1995). The social interaction that occurs in the process of searching for social support and receiving it, strengthens interpersonal relationships and contributes to feelings of respect, understanding, and connectedness. By searching for social support from the school principal and from colleagues, teachers build meaningful relationships and feel connected to others at work. Empirical studies demonstrated beneficial effects of searching for social support at the workplace, such as more satisfaction of the need for relatedness (Bakker and Oerlemans, 2019), higher work engagement (Bakker et al., 2016), higher job satisfaction (Tims et al., 2013), and higher self-rated performance (Petrou et al., 2015). The underlying mechanism that can explain these beneficial effects might be found in the aim of searching for social support to increase the amount of social support one receives. This implies that effects on the satisfaction of the need for relatedness are indirect effects through actual changes in the receipt of social support. This mediating process was previously demonstrated in the effects of searching for social support on work engagement, job satisfaction, and burnout (Tims et al., 2013).

To improve educational quality it is important to investigate the individual and proactive role that teachers can play in strengthening the social support they receive at work. So far, it is unclear whether searching for social support results in an actual increase in the receipt of social support among teachers. Moreover, there is limited evidence for processes that explain the benefit that can arise from searching for social support. Therefore, the aim of this study is to examine whether searching for social support results in satisfaction of the need for relatedness and whether this is mediated by an increase in the social support that teachers receive from the school principal and from colleagues, hence an indirect effect. In contrast, a direct effect would imply that the effect of searching for social support on the satisfaction of the need for relatedness is independent of changes in the social support that teachers receive. Moreover, these effects of the search for social support and its receipt may depend on the source of the social support (Sarason et al., 1994). Therefore, we distinguish between the receipt of social support from the school principal and colleagues and examine the different mediating roles. Thus, we advance the following hypothesis:

Hypothesis. Searching for social support results in an increase in the satisfaction of the need for relatedness through an increase in the receipt of social support from the school principal and from colleagues.

Figure 1 shows the hypothesized relationships between searching for social support, receipt of social support, and the satisfaction of the need for relatedness.

Participants were teachers at primary (pupils aged 5--12 years) and lower-secondary compulsory school level (pupils aged 13--15 years) in the German-speaking part of Switzerland. They were recruited through teacher organizations. The participants registered individually for participation by giving written informed consent. Eligible for participation were teachers that: teach at primary or lower secondary school level, have a minimum workload of 10 lessons per week, and work at a school with a formal school principal¹.

In total, N = 1365 teachers gave written informed consent. Of these, n = 1082 (79.3%) completed all three questionnaires. Over the course of the three measurement points n = 283 participants dropped out (20.7%). Dropout was unrelated to the control variables and to the model variables. From the total sample of N = 1,365 participants, data of n = 294 participants were excluded: n = 110 participants did not meet the conditions for participation, n = 154 participants were special education teacher and worked with very small groups of students in contrast to the other study participants working with groups of around 20 students, n = 28 participants did not have the same school principal during the entire period of data collection (T1--T3), and n = 2 participants gave implausible answers. The final sample consisted of N = 1071 teachers (79.5% females, 18.3% males, 2.1% persons did not report gender). Age ranged between 22 and 65 years (M = 42.8 years, SD = 11.3). School level was distributed as follows: 74.7% teachers at primary school level, 23.1% teachers at lower secondary school level, and 2.2% teachers at both primary and secondary school level. There were no significant differences between primary and lower secondary school level with regard to the study variables². The mean teaching experience was M = 17.3 years (SD = 10.9), and the mean workload was M = 80.8% of a full-time equivalent (SD = 20.2). Although the study did not aim to obtain representative data, these sample characteristics corresponded largely to the population of teachers in the German-speaking part of Switzerland in the school year 2016/17. Population characteristics are: 75% females, 25% males, 42% younger than 40 years and 58% older than 40 years, and from the total teacher population in Switzerland 59% teach at primary school level and 29% at lower secondary school level (Federal Statistical Office, 2018).

This section describes the measurement scales which were used for the study variable. We used well-established instruments and adapted them to the school context.

Searching for social support was assessed at T2 because we aimed to measure changes during one school year. In this process of change we assume that “searching for social support” becomes most important for teachers later in the school year instead of at the beginning. Moreover, by this design we aimed to build on and partly replicate the study conducted by Tims et al. (2013). This study applied a comparable design by examining effects of “crafting social resources” at T2 on changes in social job resources and wellbeing between T1 and T3.

“Searching for social support” was measured with an adapted version of the scales emotional support seeking and instrumental support seeking from the Pro-active Coping Inventory (Greenglass et al., 1999). Due to a strong correlation between searching for emotional and instrumental support (r = 0.76, p < 0.000) we combined the two scales into one general scale “searching for social support.” From the subscale “searching for emotional support” we excluded the item “I know who can be counted on when the chips are down.” This item related more to the cognitive level of knowing people that can help. Yet, we aimed to measure the active search for support on the behavioral level. This distinction was also empirically confirmed: the factor loading of this item was considerably lower (0.37) than the factor loadings of the other items (ranging between 0.64, and 0.83). For the same reason we excluded the item “I can usually identify people who can help me develop my own solutions to problems” from the subscale “instrumental support seeking.” This item did not relate to an active search for support but rather to a cognitive level of knowing people that can help. Empirically, the factor loading supported this decision because it was considerably lower (0.48) than the factor loadings of the other items (ranging between 0.68 and 0.78). Sample items of the remaining scale were for example “I asked for support from others in my school” and “I asked others in my school what they would do in my situation” with a 4-point response format (anchors is not true and absolutely true). Cronbach’s alpha was 0.91 (M = 2.73, SD = 0.67).

Received social support from the school principal and received social support from colleagues were both assessed at T1 and T3 using an adapted version of the Berlin Social Support Scales (Schulz and Schwarzer, 2003). The original scale was reformulated to fit the school setting and adjusted to the social support provider: school principal or colleagues, respectively. Both scales consisted of the subscales emotional and instrumental social support which were, due to a strong correlation (r = 0.85 and 0.86 at T1 and r = 0.86 and 0.87 at T3), merged to create one scale. Moreover, from both scales (school principal and colleagues), three negatively worded items were excluded because they did not load on the intended factor but constituted a separate factor. A phenomenon that appears to be rather common for negatively worded items (Barnette, 2000). This resulted in two scales both consisting of ten items. A sample item was ‘‘My principal/colleagues assured me that I can rely completely on him/her/them’’ with a 6-point response format (anchors is not true and absolutely true). The scale’s internal consistency from the scale ‘‘social support from the school principal’’ was Cronbach’s alpha 0.94 at T1 (M = 4.11, SD = 1.26) and 0.95 at T3 (M = 3.93, SD = 1.31), and from ‘‘social support from colleagues’’ was Cronbach’s alpha 0.93 at T1 (M = 4.72, SD = 1.06) and 0.94 at T3 (M = 4.61, SD = 1.07)³.

Satisfaction of the need for relatedness: was assessed at T1 and T3 with the German version of the Work-related Basic Need Satisfaction scale (van den Broeck et al., 2010). The scale consisted of 6 items with a 5-point response format (anchors not true at all and absolutely true). A sample item was “At work I feel as part of a group.” Cronbach’s alpha was 0.81 at T1 (M = 3.98, SD = 0.76) and 0.84 at T3 (M = 3.96, SD = 0.78).

The hypothesized model was tested with structural equation modeling using R version 3.4.3 (R Core Team, 2018) and the Lavaan package version 0.6-3 (Rosseel, 2012).

To estimate the hypothesized model, we used robust maximum likelihood estimation (MLR), which is robust against violations of normality assumptions (Lai, 2018). Full information maximum likelihood estimation (FIML) was used to handle missing data (Graham and Coffman, 2012).

Model fit was assessed using the chi-square/df ratio (χ²/df), the Root Mean Square of Approximation (RMSEA), Standardized Root Mean Residual (SRMR), and the Comparative Fit Index (CFI). Values of χ²/df < 5 indicate good model fit (West et al., 2012). RMSEA values of ≤0.06 indicate a good fit, SRMR values of ≤0.08 are considered as a good fit, and the incremental fit indices CFI >0.95 reflects a good fit (Hu and Bentler, 1999). To test the proposed mediated effects we used bootstrapping instead of the MLR estimator. Bootstrapping is a statistical re-sampling method recommended for testing indirect effects which creates an empirical sampling distribution through resampling the original sample (MacKinnon et al., 2004; Preacher and Hayes, 2008). Estimations of confidence intervals of indirect effects by bootstrapping are more accurate, because it does not assume a normal sampling distribution, which is the case for indirect effects (Shrout and Bolger, 2002). We computed bias-corrected bootstrap 95% confidence intervals from 1,000 resamples (MacKinnon et al., 2004; Preacher and Hayes, 2008). A significant mediation is indicated when the confidence intervals exclude zero.

To measure change over the course of one school year in the receipt of social support and in teachers’ satisfaction of the need for relatedness, we used residualized change scores (cf., Tims et al., 2013). Following the recommendations of Smith and Beaton (2008), standardized residualized change scores were computed by regressing T3 scores on the corresponding T1 scores. The differences between the predicted and the observed scores at T3 were included as standardized residual scores in the hypothesized mediation model. Compared to simple difference scores, residualized change scores have higher reliability because they do not correlate with the initial T1 score (Williams et al., 1987). Residualized change scores indicate which individuals have changed more, or less, than can be expected based on their T1 score (Smith and Beaton, 2008).

Descriptive statistics and zero-order correlations of the variables at T1 and T3 and the model variables are displayed in Table 1. Zero-order correlations between T1 and T3 measurement waves were strong ranging between r = 0.62 and r = 0.73 (p < 0.001). Furthermore, zero-order correlations between all model variables were significant (p < 0.01).

We tested the model fit of the measurement model applying a confirmatory factor analysis. The measurement model yields a sufficient model fit: χ2/df = 3.25, CFI = 0.960, TLI = 0.951, RMSEA = 0.052, SRMR = 0.029 (see Table 2).

To test our hypothesis, we compared the hypothesized full mediation model with a partial mediation model. The partial mediation model included a direct effect from searching for social support on the change in satisfaction of the need for relatedness. Scaled Satorra-Bentler χ² difference tests showed that including this direct effect significantly improved the model fit: Δχ² (1) = 14.16, p < 0.001. Thus, besides the indirect effect of searching for social support on the change in satisfaction of the need for relatedness, there was also a direct effect between those two variables (see Figure 2). The indirect effects consisted of a positive effect of searching for social support on the change in receipt of social support from the school principal (β = 0.14, p < 0.001) as well as from colleagues (β = 0.26, p < 0.001), and a positive effect of the change in receipt of social support on the change in satisfaction of the need for relatedness. Yet, the latter positive effect could only be determined for the change in receipt of social support from colleagues (β = 0.29, p < 0.001).

Bootstrap analysis showed a difference between the indirect effects via the receipt of social support from the school principal and via the colleagues (see Table 3). The change in received social support from colleagues mediated the effect of searching for social support on the change in satisfaction of the need for relatedness (β = 0.08, p < 0.001). Yet, the change in received social support from the school principal did not function as mediator of this effect (β = 0.01, p = 0.337). The latter finding is primarily due to the insignificant direct effect of the change in received social support from the school principal on the change in satisfaction of the need for relatedness (β = 0.04, p = 0.374). This gives further clarity to our hypothesis which can only be confirmed for social support from colleagues.

In line with the concept of job crafting (Wrzesniewski and Dutton, 2001; Tims and Bakker, 2010), our results support the assumption that employees can actively adjust their work environment to make it better meet their needs, preferences, and abilities. The finding that teachers can individually and actively contribute to the social support they receive at the workplace and thereby improve the satisfaction of their need for relatedness is an important insight for collaboration in schools. Actual developments in schools such as team-teaching and an increase of part-time working teachers (Appius and Nägeli, 2017) might influence the extent to which teachers search for social support within the school team. Team-teaching involves collaboration between teachers which could intensify the search for social support making it a normal routine of day-to-day practice. An increasing number of teachers in Switzerland that work part-time presumably also affects the exchange of social support (Brägger and Schwendimann, 2022). However, it is rather unclear how this influence might look like. Either working part-time results in less presence within the school team, and consequently in less opportunities to search for social support, or part-time work compels teachers to intensify collaboration which also promotes searching for social support. We argue that it is important to consider such school developments also in the light of their influence on the exchange of social support within school teams.

Searching for social support had a direct as well as an indirect positive effect on the satisfaction of the need for relatedness. This result partly supports previous studies that showed indirect effects of searching for social support on employee outcomes (Tims et al., 2013; Demerouti et al., 2015). The direct effect found in the present study might be explained by other sources of social support. Because we examined social support explicitly from either the school principal or teaching colleagues, we consequently excluded social support from other sources at schools such as school psychologists and school social workers.

To the best of our knowledge, only one previous empirical study investigated the effects of job crafting on teachers’ basic need satisfaction. The main finding of this study is that job crafting contributed to more basic need satisfaction among teachers (van Wingerden et al., 2017). A nuanced look at the different aspects of job crafting shows that the effects of job crafting on teachers’ basic need satisfaction differ between these aspects. Results suggest that only the aspect “increasing challenging job demands” showed a beneficial relation with basic need satisfaction. The aspect “searching for social support” did not seem to contribute to teachers’ basic need satisfaction. This contradicts the results of the present study and gives reason to further investigate the effects of the different aspects of job crafting.

The present study revealed differences between the social support received from the school principal and from colleagues. There might be several reasons for these differences: Distributed and shared forms of leadership have the possibility to arise in school organizations (Harris, 2008; Fuchs and Wyss, 2016). Informal leaders, such as experienced teachers, might occupy a central position within the school team from which they assist and advice their colleagues. Accordingly, previous studies suggested that social relationships within school teams do not reflect the formal organizational structure of the school (Moolenaar, 2012), and that formal advisory roles in the school organization are not always adopted by the formally designated person (Penuel et al., 2010; Spillane and Healey, 2010). Thus, because colleagues are easily accessible and offer competent support, there might not be a need to request support from the school principal. Additionally, there might be other reasons to rather turn to colleagues for social support than to the school principal. Previous research shows that school principals are less frequently involved in day-to-day social interactions between teachers, such as discussing work related issues (Moolenaar et al., 2014). Moreover, a possible explanation could be that school principals might have less available time for such day-to-day social interactions. Thus, teachers might have experienced that their school principal is less available which strengthens them in searching for social support from their colleagues rather than from their school principal. Further, school principals’ responsibility of controlling and evaluating teachers’ performance might play a role in their more remote position toward their teachers (Rothland, 2007). Teachers might want to avoid showing the school principal a need for social support as this could be perceived as a limitation of teacher’s abilities.

The finding that received social support from colleagues was positively related to the satisfaction of teachers’ need for relatedness whereas received social support from the school principal was not, could be explained as follows: The need for relatedness refers to feelings of social connectedness, being a group member, and feeling close with others (Deci and Ryan, 2000). The important role of social support from colleagues in satisfying this need can be expected. However, as the social interaction between school principals and teachers is less close compared to colleagues (Moolenaar et al., 2014), school principals might have less influence on teachers’ satisfaction of the basic need for relatedness. Although this seems to contradict the importance of leadership behavior that aims at the promotion of employees’ basic psychological needs for employees’ functioning (Hetland et al., 2011; Kovjanic et al., 2012), this does not rule out the influence of school principals on the satisfaction of the other basic needs of autonomy and competence. Especially the satisfaction of the need for autonomy seems to benefit from the receipt of social support from the school principal (Maas et al., 2021). Moreover, instead of directly affecting their teachers, school principals might rather exert the influence of their leadership behavior indirectly through adjustments in job demands and job resources which creates opportunities to satisfy teachers’ needs (Vincent-Höper and Stein, 2019). Besides this, our results suggest that school principals display an important aspect of supportive behavior by being responsive to teachers’ requests for social support and providing social support.

The present study has several strengths. First, to the best of our knowledge, this is the first study that examined effects of teachers’ search for social support differentiated by source. By distinguishing social support from the school principal and from colleagues, we could show important differences in relation to searching for social support and the satisfaction of the need for relatedness. These differences shed light on the exchange of social support in school teams and how it can contribute to the satisfaction of teachers’ basic needs.

Second, by using a longitudinal design and residualized change scores we were able to show relationships over time and actual changes in the receipt of social support and the satisfaction of the need for relatedness. Finding these changes even for a long time-interval of approximately 9 months strengthens our findings. Comparable studies used much shorter time-intervals (Tims et al., 2013; Demerouti et al., 2015). Moreover, by covering almost a whole school year the present study provides valuable insight in the occupational reality of teachers across the school year. Although the longitudinal design also implied that study participants had to fill out three online surveys across one school year which was time consuming, the dropout rate was very low. This led to a large sample size of over thousand teachers: to the best of our knowledge one of the biggest teacher samples in the German speaking part of Switzerland.

Yet, there are also some limitations that are worth mentioning. First, despite the longitudinal design, we cannot draw conclusions about causality. The influence of third variables that explains our results cannot be ruled out (Mackinnon and Pirlott, 2015). Also, short term fluctuations in the receipt of social support and the satisfaction of the need for relatedness could have influenced the changes we demonstrated between the measurement points. Studies that use an experimental design are required to neutralize influences of third variables and short-term fluctuations, and to unfold causal relationships.

Second, by using only self-reports possibly common method variance might have affected our results (CMV; Podsakoff et al., 2003). However, we intentionally used self-reports because we were interested in teachers’ personal experience. Using reports by others for measuring teachers’ search for social support, receipt of social support, and satisfaction of the need for relatedness would be complicated and above that not accurate enough. Moreover, “objective” indicators also have shortcomings, for example, observers’ bias, halo, and stereotype effects (Kerlinger and Lee, 2000).