Tasks and impact of school social work in Switzerland as perceived by teachers, principals and school social workers – a multilevel analysis

To investigate whether the perceptions of school-based professionals regarding the tasks and impact of school social work (SSW) converge or diverge, this study collected survey data among 638 teachers, 41 school social workers, 62 principals, and 23 special education teachers distributed over 92 Swiss schools. After constructing several scales measuring the tasks and the impact of SSW via principal factor analyses, ANOVA’s were carried out to compare the mean perceptions of the included professionals. To prove for related perceptions between teachers and school social workers multilevel analyses were performed by including additional exploratory variables such as school context and personal factors. Most results indicate considerable concordance between the professions included in the study regarding the tasks rated as important as well as with respect to the outcomes of school social work. While all professions agreed that social problem solving, i.e. the intervention approach, is the most important approach of school social work, some discordance was found with respect to preventive tasks such as project work. Overall, teachers and principals tended to underestimate the effects of school social work implying some potential to destabilize their collaboration with the school social workers. While the multilevel analyses revealed a high variability between schools due to local factors, the perceptions of the teachers and the school social workers located at the same schools proved to be independent over most investigated dimensions.


Introduction
School social work (SSW) first emerged, selectively but continuously, in major urban US centers from the beginning of the 20th century (Shaffer, 2006). After 1950, it increasingly evolved in various countries as an additional profession working in schools to support students' health and well-being. In Germany, social work in schools emerged in the early 1970s (Abels, 1971(Abels, , 1972Speck, 2014), while in Switzerland, the first professional school social work services began in the 1980s. From this point forward, SSW in Switzerland slowly developed until the arrival of the new millennium, followed by its accelerated expansion over the last 20 years (Baier, 2011a;Ziegele, 2014). The actual Swiss picture is very mixed, with the supply of SSW varying across many urban communities and cantons-from full access, where every pupil has access to SSW, to minimal or even zero access; the remarkable number with zero access is predominantly situated in rural areas mainly due to financial and political reasons (Seiterle, 2014;Ziegele, 2014). In accordance with the federal Swiss education system in general, cantonal and, in part, even local laws and contracts regulate SSW, resulting in the aforementioned mixed picture and in the concentration of SSW among urban and suburban communalities and cantons that are more concerned with social flashpoints than are the rural ones. With regard to education, most Swiss school social workers (SSWers) acquired a bachelor degree in social work and an additional certificate of advanced studies in SSW.
From a theoretical point of view, the implementation of SSW principally refers to at least two different, but not mutually exclusive, perspectives. According to modernization theory, the differentiation of the involved professions at school is in response to insufficient socialization conditions of children and adolescents (i.e., value shifts, changes of living environment, and troubled family lives), aiming to support the students in their efforts to cope with these conditions (Speck, 2014). Accordingly, most intervention approaches nowadays have their roots in resilience theory or resource theory (Early & Vonk, 2001;Leyba, 2009) targeting different multiple intervention agents by promoting home-schoolcommunity linkages (Franklin & Kelly, 2009). In addition, they focus on the strengths of the students and their ecological systems, including parents, neighborhoods, medical care, and local associations with their attendant resources and opportunities (Isaksson & Sjöström, 2016).
In contrast, the educational systems perspective focused on the academic underachievement of socially disadvantaged or underprivileged children and determined that SSW supports these children's academic careers as well as contributes to the schools' development (Speck, 2014). Accordingly, social work values, such as dignity and worth of the person and social justice, align with this perspective, supporting children's rights in order to address their developmental and age-appropriate needs (Baier, 2011b).
To our knowledge, no empirical data exist thus far showing whether school social workers (SSWers), teachers, or school principals adopt one of these perspectives or instead a combination of them. However, it seems rather likely that the different professions working at schools do not adopt the same perspectives on SSW (Speck, 2014) due to their distinct professional backgrounds. Consequently, it is also likely that such divergent perspectives enhance the tendency toward role confusion regarding the tasks of SSW, as observed in some previous studies (e.g., Peckover, Vasquez, Van Housen, Saunders & Allen, 2012). Moreover, as SSWers follow an anti-oppressive approach and see themselves as advocates of the students, "there is some potential for inter-professional conflict" (Isaksson & Sjöström, 2016, p. 10), which in turn can undermine their collaboration with teachers.
Previous research demonstrates that good inter-professional collaboration (Lawson, 2004) between teachers and SSWers proved to be an important precondition of success in SSW (Lynn, McKernan McKay, & Atkins, 2003;D'Agostino, 2013;Baier, 2011c;Speck, 2014). For that reason, it seems quite relevant whether these professions mutually agree about the prioritization of tasks completed by SSWers and whether they expect the same or different outcomes of SSW, as such shared understandings and visions would be important to supporting inter-professional collaboration (Lawson, 2004;D'Amour, Ferrada-Videla, San Martin Rodriguez & Beaulieu, 2005;D'Agostino, 2013). The present study addresses these gaps in research by comparing the understanding held by four types of professionals working at Swiss schools.
According to most scholars, SSW is responsible for (1) the support of students (individually and in groups) and their families, (2) collaboration with and support of teachers and school authorities, (3) collaboration with other services to give children access to community-based resources, (4) primary and secondary prevention, and (5) administrative work (Frey, Alvarez, Anlauf Sabatino et al., 2012;Kelly, Thompson, Frey, Klemp, Alvarez, & Cosner Berzin, 2015;Lynn et al., 2003). Despite this broad approach, SSWers are mostly engaged to reduce absenteeism (Newsome, Anderson-Butcher, Fink, Hall, & Huffer, 2008), aggressive or disruptive behavior, and emotional or health problems (including drug abuse) (Jonson-Reid, Kontak, Citerman, Essma, & Fezzi, 2004;Kelly et al., 2015). Certainly, SSWers address psychosocial problems that are mainly caused by extracurricular factors-e.g., child maltreatment, abuse, or neglect (Jonson-Reid et al., 2004;Kelly et al., 2015), or drug abuse-and that are only rarely caused by school factors. Obviously, the roles and practices of SSWers are very diverse, comprising networking over different contexts to mitigate risks and build up protective factors, counselling single children and groups as well as collaborating with parents, teachers, and school authorities. This diversity informed our scale development process to record the tasks of SSW in Switzerland, as described below.

The present study
In this study, our aims were threefold. First, we assessed the tasks of Swiss SSWers and explored how school-based professionals (i.e., SSWers, teachers, principals, and special education teachers) perceive the importance as well as the impact of these tasks. Thus, the first goal of our study was to develop and use an appropriate instrument by creating survey items, conducting the survey, and establishing underlying factors by exploratory factor analyses. Second, we explored by ANOVA whether and to what extent particular perceptions of the included professions differ from each other. Third, we explored, using multi-level modelling, whether and to what extent the perceptions of the included professions are interrelated, and whether individual and contextual factors such as gender, schools, and cantons interfere with the identified associations.

Sample
Sample recruitment occurred in 2017. First, the School Social Work Association of Switzerland supported the recruitment by sending emails to the SSWers registered in their database, asking them to participate in the study. Second, we asked selected principals, together with some teachers from their schools, either directly or mediated by cantonal and local education departments to participate. A school could participate if at least ten teachers, the principal, and a SSWer were willing to fill in the questionnaires. As soon as we received an expression of interest by a specific school, we asked for the email addresses of the probable participants. Subsequently, we sent the participating teachers, principals, and SSWers a personalized email containing a link to the survey. The only incentive promised to the individual participants was a report of the main results of the study. Participants, who did not complete the survey within two weeks, received a reminder email. Overall, we sent 952 personalized invitations containing the link to the questionnaire to teachers, principals, and SSWers, and we received 764 (80%) valid questionnaires usable for data analysis. The remaining 188 questionnaires were not returned or were excluded because of (too many) missing data (e.g., lacking personal details). Due to the implemented convenience sampling method, we are not able to estimate the true response rate of this study.
The participants (N = 764) were distributed over 92 schools. The sample predominantly comprised teachers (n = 638); remaining participants were school principals (n = 62), SSWers (n = 41), and special education teachers (n = 23) ( Table 1). The sample consisted of an uneven gender distribution, i.e., while teachers were predominantly female, more school principals were male than female (Table 1).  The average age of the participants was 43.5 years. School principals and special education teachers were older than the teachers and the SSWers (Table 1). Accordingly, the professional experience of the participants varied considerablyfrom less than 12 months to 44 years, with teachers averaging more work experience than school principals, and SSWers averaging less work experience than school principals (Table 1). Ninety percent of the teachers were 50 percent or higher part-time employed or in full-time positions.
The abovementioned 92 schools were located in 35 municipalities, the majority of them situated in rural, small-town, or suburban areas distributed over eleven cantons of the German-speaking part of Switzerland.

Instrument
In order to compare the respective expectations and estimations of the participants, we developed an online questionnaire identically applicable among all groups included in the survey (SSWers, teachers, and school principals). The instrument consisted of four sections: 1. The first section presented specific activities or tasks of SSWers at a medium level of abstraction, e.g., by differentiating counselling single persons from working with groups, but not mentioning all of the reasons behind particular interventions. We based these items on various performance specifications and job descriptions for SSW services in the German-speaking part of Switzerland (see, for example, Bildungsdirektion des Kantons Zürich [Department of Education of the Canton Zurich], 2011). The participants were asked to prioritize these activities (across four response levels, ranging from not important to very important).
2. We asked the respondents to estimate the impact of SSW with respect to dimensions targeting students (across four response levels, ranging from small to large). We constructed the items based on the work of Bye, Shepard, Partridge, and Alvarez (2009)  3. The next section aimed to collect assumed effects of SSW on the teachers working in the school (across four response levels, ranging from small to large). Again, we based these ad hoc-constructed items on previous studies (see, for example Gutzwiller-Helfenfinger & Wicki, 2008).
4. Finally, we collected personal data of the participants (age, gender, function at the school, work experience) and information about their respective schools.
After initial drafting, we developed the questionnaire further. SSW experts from the Lucerne University of Applied Sciences (School of Social Work) validated all items (face validity), and subsequently, the instrument was qualitatively tested by means of the think-aloud protocol with one SSWer, one principal, and one teacher. Finally, we piloted the instrument among 8 teachers (7 females and 1 male) and 13 SSWers (10 females and 3 males). Based on the feedback received, we once again supplemented and modified the questionnaire. The final version included 32 items regarding specific activities or tasks of SSW, 22 items measuring the expected impact of SSW on students, and 18 items measuring the effects estimated to address the impact on teachers' work besides some items assessing sociodemographic data.

Data analysis
Data concerning the items related to the importance and impact of SSW were analyzed by means of exploratory factor analyses using SPSS version 24, and scales were created on this basis. Principal axis factoring was used as the extraction method, followed by Varimax and Oblimin rotation. Scale reliability was assessed by Cronbach's alpha. Values above 0.7 indicate good internal consistency. All additional data analyses were done with R version 3.4.4 (R Development Core Team, 2018). To find out whether the assessments differ between the professions, we used Tukey post-hoc tests for pairwise comparisons following ANOVAs. Tukey's method is the most common to control the experiment-wide error rate when comparing all possible group pairings. To examine the relationships between the perceptions of the teachers and those of social workers and school principals, we carried out multilevel regression analyses, as teachers of the same school tend to be more similar than teachers in different schools. To take this into account, mixed effects models were fitted with a random school effect using the R packages lme4 (Bates, 2015) and multilevel (Bliese, 2016). The teachers' scale values served as response variables. The explanatory variables of main interest were the mean values of the same scale of social workers and school principals at the same school. In some schools, far fewer than ten teachers took part, despite prior confirmation. For the multilevel regression analyses, we excluded schools with less than five teachers. Additional explanatory variables were gender, age group, part/full-time employment, work experience, and canton.
When examining the predictors of the regression models, we included non-linear terms and interactions. We carried out all regression analyses by means of R version 3.4.4 (R Development Core Team, 2018) and based the variable selection on the Akaike information criterion (AIC). For all statistical tests a significance level of 0.05 was chosen.

Importance and impact of tasks
With respect to the importance of the tasks, the exploratory factor analysis (Principal axis analysis, Varimax rotation) yielded a measure of sampling adequacy (Kaiser-Meyer-Olkin, KMO) of 0.86. The scree plot pointed to the following seven factors: -Social problems. Example: "Counselling and support of single students affected by social problems." -Project work. Example: "Organizing support offers for students at the school (e.g., training in social competencies or health promotion projects)." -Cooperation with parents. Example: "Counselling and support of parents of a student affected by social problems." -Establishing contacts and providing information. Example: "Establishing contacts with appropriate services." -Exchange. Example: "Exchange of observations and estimations with the principal." -Documentation. Example: "Documentation of the activities (statistics, dossiers, and so on)." -Targeting learning problems. Example: "Counselling and support of single students with learning problems." Total variance explained was 54.3 percent. The Cronbach alpha varied between 0.69 and 0.82, and the corrected item-total correlation ranged from 0.34 to 0.70.
With respect to the impact for students, we eliminated two items because they had low communalities and correlations with the remaining items of this construct. A factor analysis suggested a 4-factor solution (KMO = 0.91; total variance explained, 65.9%). The four subscales were: -Social and cognitive competencies. Example: "To reduce violence among students at the school." -Student-teacher relationship. Example: "To confide in teachers when involved in peer conflicts." -Reduction of addictive behaviors. Example: "To reduce experimenting with drugs." -Use of extracurricular activities. Example: "To encourage students to use other services at the school, e.g., midday meals or tutoring." The Cronbach alpha varied between 0.78 and 0.96, and the corrected item-total correlation ranged from 0.52 to 0.92.
A factor analysis including all items addressing the impact for teachers revealed a single factor solution (KMO = 0.91; total variance explained, 44.4%). Cronbach alpha was 0.93, and the corrected item-total correlation ranged from 0.46 to 0.74. Example: "To improve the teachers' social skills (empathy, self-competencies, and so on)." Table 2 provides an overview of all scales and their psychometric properties.

Importance of tasks as perceived by different professions
In general, all participant groups similarly perceived most SSW tasks as important or even very important, with social problem solving as the most important task, conducting projects and documentation as somewhat less important, and targeting learning problems as the least important ( Figure 1). For the purpose of this study, the differences between the involved professions were of special interest. SSWers estimated projects as more important than did teachers (Tukey p < .05) and principals (Tukey p < .05). Similarly, SSWers found learning issues to be more important than did teachers (Tukey p < .05) and principals (Tukey p < .05). SSWers also found collaboration with parents and documentation to be more important than did teachers (twice Tukey's p <. 05). However, for special education teachers, documentation was more important than for their colleagues teaching typically developing children (Tukey p < .05). Teachers estimated information exchange between the different professions and with parents to be less important compared with the principals (Tukey p <. 05).

Impact perceived by different professions
The general picture reveals that the respondents perceived the impact of SSW on students and teachers to be moderate, i.e., the ratings averaged between "rather small impact" and "rather strong impact." However, it is also obvious that special needs teachers and the SSWers themselves were more convinced of the impact of SSW than were teachers and principals (Figure 2).

Figure 2: Perceived impact by professions
In particular, teachers perceived the impact of SSW on the students' social and cognitive competence less optimistically than did principals (Tukey p < .05) and social workers (Tukey p < .01). In a similar vein, social workers were-compared with teachers-more convinced of their positive impact on the relationship between teacher and students (Tukey p < .05). An even more pronounced difference became evident regarding extracurricular activity attendance and the impact on teachers' quality of work. Both comparisons evidenced more optimistic ratings among social workers on the one hand, and teachers and principals on the other (Tukey p < .05 and p < .01, respectively, for extracurricular activities; and twice Tukey p < .001 for teachers' quality of work). Finally, special needs teachers showed more optimistic impact ratings than teachers and principals with respect to extracurricular activities (twice Tukey p < .01), and they outperformed teachers regarding the impact on their work quality (Tukey p < .05).

Multilevel analyses
In the following, we present the results of the multilevel models developed, as described in the method section. All models relate to one of the seven scales measuring the tasks of SSW or to one of the five impact scales, respectively. Because of our focus on associations between the perceptions of the included professions, we limit the presentation to the results revealing such associations.

1) Importance of the tasks of school social work
Overall, we found many gender differences, indicating that female teachers perceived most of the tasks as more important than did male teachers. As these gender differences are not the focus of this study, we do not specify the respective results further.
Conducting projects. The school factor was significant (ICC = 0.11), indicating that this tasks' importance rating varied over schools. In addition, schools situated in the Canton of Lucerne perceived the task as less important than schools of the Canton of reference (i.e., Aargau; β = -.36, p < .001).
Cooperation with parents. The importance of cooperation with parents also varied across schools (ICC = 0.124). In addition, we found a trend for a positive linear relationship between the teachers' and the social workers' perceptions (β = 1.26, p < .10).

Improvement of contacts and information.
Improvement of contacts and information did not vary among schools; therefore, we used an ordinary multiple linear regression model. We found that high importance ratings among principals related to an increase of importance rated by teachers (β = 1.29, p < .05), while the opposite was true for social workers (β = -2.39, p < .05). The more the social workers perceived contact improvement as very important, the less teachers evaluated the tasks as important.
Exchange. Schools varied regarding their view on the importance of exchange activities (ICC=0.084). Principals' ratings were positively associated with the teachers' ratings (β = .20, p <.05).
Targeting learning problems. Schools varied significantly in this regard (ICC = 0.08). Again, principals and teachers were positively related (β =.22, p = .001), while social workers' ratings were not related to the teachers' ratings. In addition, we found some variation between the Cantons. Compared with the reference Canton (Aargau), teachers working in the Canton of Solothurn found this task more important (β = .53, p < .05), while the opposite was true among the teachers of the Canton of St. Gall (β =-.57, p < .05).
Documentation. Variation among schools was significant (ICC = 0.08). Apart from the gender effect, we did not found any additional predictors.
2) Perceived impact of school social work The analyses revealed that the schools rated the impact of SSW differently over all dimensions (all ICCs between 0.07 and 0.13). In addition, we found several gender differences, indicating that female teachers always perceived more impact than did male teachers.
Student-teacher relationship. We found a significant interaction term involving the gender of teachers (β = .37, p < .01). The ratings of social workers were negatively correlated with the ratings of female teachers, whereas for male teachers, the social workers' ratings had a positive relation with their own ratings. Teachers aged between 40 and 49 rated a lower impact regarding improvements of the student-teacher relationship than did teachers aged below 29 years (β = -.42, p < .001).
Reduction of addictive behaviors. Apart from the gender difference indicating that female teachers rated this impact more optimistically than did male teachers (β = -.21, p < .05), we found only a small negative correlation between the teachers' and the principals' perceptions (β = -.12, p = .05).
Use of extracurricular activities. We found a trend indicating a negative association between teachers' ratings and the respective ratings of the social workers (β = .97, p < .10). In addition, schools of the urban Canton of Zurich showed higher ratings than the Canton of Aargau (β = .81, p < .05), while the opposite was true for the more rural Canton of Lucerne (β = -.66, p < .01).
Impact for teachers. Apart from the variation among schools, teacher age predicted the respective impact ratings. Compared with the teachers aged younger than 29 years, the older age groups perceived less impact for themselves (p < .05 for all comparisons).

Discussion
In general, our results prove a considerable convergence between the professionals working in the Swiss compulsory education system regarding the perceived importance and the expected impact of the tasks of SSW, but some interesting discordances as well. The four professions clearly agreed that social problem solving is the most important task of SSW, which is in line with both a modernization theory point of view and an educational system position (Speck, 2014). However, we found some discordances with respect to other tasks rated more important by the SSWers and the special education teachers compared with teachers and principals. For example, SSWers rated conducting projects as more important than teachers and principals, eventually indicating a greater interest in primary prevention among the former groups. Similarly, the difference regarding cooperation with parents refers to the notable ecological systems approach pursued by SSWers (as also found by Isaksson & Sjöström, 2016), which teachers probably did not completely share with them.
Interestingly, the pattern that special education teachers and the SSWers share many perceptions changed somewhat regarding learning problems. SSWers perceived this task as more important for SSW than did the remaining professions (including the special education teachers), thus indicating some affinity for the educational systems perspective (i.e., they felt more responsible for students' learning problems than their professional peers expected they should feel).
Although the perceptions of principals and teachers mostly converged, they diverged with respect to the importance of the information exchange task by SSW. We interpret this result as at least partly caused by shared goals of principals and SSW regarding information exchange involving teachers, parents, external services, and others.
With respect to the perceived impact of school social work, it is not surprising that SSWers themselves were more optimistic compared with the teachers; however, we did not expect that special education teachers would converge almost completely with the SSW, while principals went along with the more pessimistic ratings of the teachers. Teachers and principals probably systematically underestimate the impact of school social work, implying the potential to compromise a productive inter-professional collaboration (Lawson, 2004;Lynn et al., 2003). However, more research is necessary to support such a conclusion.
Beside the fact that all professions and all schools emphasized social problem solving-i.e., an intervention approach, in comparison with a more preventive approach-we found that female teachers perceived this task (as well as other tasks) as more important than did males. This gender difference is probably due to a general response pattern; however, these differences are beyond the focus of this paper.
Schools varied considerably regarding the majority of their importance ratings (project work, cooperation with parents, exchange, targeting learning problems, documentation) and regarding their impact ratings (over all dimensions). This result indicates that local factors (e.g., cantonal legal requirements and guidelines, status of local school development, composition of students and teachers, ecological and social aspects of neighborhoods) not only have an impact on the way teachers perceive the tasks of SSW, but even more so on the way they perceive its outcomes. We believe that such local factors are still a neglected topic of research.
There are some age-related differences indicating that especially young teachers highly appreciate SSW. This might be because work overload and a lack of experience is more prominent among younger teachers; however, we cannot exclude that at least some older teachers cultivate prejudices against school social work.
Remarkably, the multi-level analyses only rarely revealed that the teachers' ratings were associated with the perceptions of the SSWers. The importance of cooperation with parents and the perceived impact on the students' competencies both related (positively) to the respective ratings of SSWers, while the opposite was true for the importance of contact and information improvement (negatively correlated). While it seems quite difficult to interpret the significant interaction term with gender as it concerns the impact on the student-teacher relationship (male teachers correlated positively, female teachers negatively with SSWers' views), our results, in general, indicate a rather high independence between the perceptions of teachers and SSWers. This is not necessarily problematic, as both follow their own professional goals, which need not to be completely congruent to each other (Isaksson & Sjöström, 2016).

Limitations of this study
There are some limitations of this study. First, a random sampling was not feasible. Therefore, we investigated a convenience sample yielding results that are not representative in a strict sense. Perhaps a random sampling would reveal more critical teacher ratings because of a selection bias that we cannot completely exclude. Teachers with more positive attitudes regarding SSW were probably more motivated to participate in the study. Second, the sample does not comprise schools from the biggest German-speaking Swiss cities (i.e., Zurich, Basel, Berne), probably resulting in an overrepresentation of surrounding agglomeration areas and rural areas. Third, our survey data are correlational in nature, and are therefore unsuitable for producing causal conclusions.

Conclusions and recommendations
Despite the exploratory character of this study, some conclusions are possible. First, there are many perceptions that teachers, principals, and SSWers share in order to support a strong and loyal cooperation. Second, there are, nonetheless, some threats to this cooperation, as evidenced by the fact that at least some teachers tend to disapprove of the secondary preventive tasks of SSWers, such as project work and cooperation with parents, and by the fact that teachers and principals tend to underestimate the impact of SSW for students and teachers as well.
In our view, SSW should further develop in the future as an independent profession not only working in schools or for schools but rather cooperating with schools and all the other professions working there (Gherardi & Whittlesey-Jerome, 2018). That implies a strong and loyal cooperation between different professionals; it also means supporting each other and seeing the limitations and opportunities of others' work, but it does not mean abandoning one's own professional beliefs.
As we found many differences due to different contexts (e.g., schools, cantons), future studies should consider these differences in a more systematic way-for example, by systematically varying contextual factors in order to achieve a better understanding of underlying causal factors.