Principals’ identity – developing and validating an analysis instrument in the Swedish context

ABSTRACT

The aim of this article is to develop and validate an analysis instrument of principal’ identity in the Swedish context. A point of departure is taken in the work of Crow, Day, and Møller (2017) and three dimensions of the concept: professional, personal, and situated or socially located. The article builds on a survey on Swedish principals covering questions of the three identity dimensions. The survey reached approximately 3,000 out of the 3,600 Swedish principals, generating 1,286 replies (n = 1,286). The analyses identified questions with strong and weak correlation with the three dimensions. Thus, the findings provide components for a more refined analysis instrument to analyze how principals’ identity is formed and negotiated in different contexts without overlooking the fact that construction of identity is situated in time, space, and place. These findings also contribute to a more comprehensive image of principal’s professional identity. Additional studies, however, are requested for a more in-depth understanding of the concept of identity in various leadership contexts.

Around the globe, the concept of principals’ identity has gained increased attention in school leadership research (e.g. Crawford, 2007; G. M. Crow & Scribner, 2014; Møller, 2005; Murakami et al., 2016). However, and in support for further studies, G. Crow and Møller (2017) argued that an identity perspective becomes an important counterweight to a technocratic view on principals’ work, bringing ‘real life’ back into the discussion on school leadership. Therefore, former works have emphasized that principals’ identities are situated in time, space, and place, and that this should also constitute a point of departure for understanding the concept (Crow et al., 2017). Hence, principals’ identities are understood as individual and collective constructions linked to negotiations in different times, rooms, and places (cf. Wenger, 1998). Comparably, Whiteman et al. (2015, p. 579) concluded that ‘professional identities have several elements; they are dynamic, multiple, socially negotiated, and developmental.’ From such a point of departure, specific roles and leadership assignments could be designed (e.g. in educational policy), but one cannot design the identities that will be constructed through these assignments and roles. Against this backdrop, developing a professional identity becomes essential, not the least for newly appointed principals in their first years in the profession (cf. Ritacco & Bolivar, 2019; Saarukka, 2014).

Previous research have also generated a considerable number of frameworks and models for understanding the (re)construction processes of both teachers’ and principals’ identities (e.g. Day & Lee, 2011; Hargreaves, 1998, 2005; Leithwood & Beatty, 2008; Lumby & English, 2009; Thomson, 2011). Without overlooking these former works, this article argues that we still need to explore analysis instruments to conduct analysis on principals’ identity from a nationwide level. In this regard, this article offers a significant contribution to the current body of research by developing and validating an analysis instrument for analyzing school principals’ identity. In addition, the Swedish case as such offers a novel perspective on principals’ identity due to a former tradition of strong state governance and control – a system that has been modified into a system characterized by market adaptation and decentralization in the latter decades (e.g. Lindensjö & Lundgren, 2000; Lundahl, 2002a, 2002b).

A cornerstone in the current body of research is the profound work conducted by Crow et al. (2017) providing a synthesis of existing works and a tentative framework for further research (see also G. Crow & Møller, 2017). Regarding principals’ professional identity, Crow et al. (2017) outline a dividing line between three dimensions, namely: the ‘professional, “the “personal,” and the “situated or socially located” dimension. Principals’ identity is to be understood as an interplay between these three dimensions (also detailed further below). Worth highlighting, there are also adjacent concepts visible in the current body of research. For example, the work of Saarukka (2014) offers a model built up by the “personal,” “professional,” and “positional” levels of principals” identity. In this model, principals’ identity construction is described as a relational process formed by activities and communication in which teachers adapt to the principal as a person, in the professional role, and in the leadership position.

That declared, one point of departure in this article is that the three dimensions in the work of Crow et al. (2017) could reinforce a profound analysis instrument for exploring how principals’ identity is formed and negotiated in different contexts. A key argument for such a quest is that a strong link between theory and empirical data is an essential component for quantitative data analysis in social sciences (e.g. Djurfeldt et al., 2010). Thus, this article further examines the three dimensions detailed in the work of Crow et al. (2017) in order to build a more robust analysis instrument.

Thus, by using data from a nationwide survey on Swedish principals (subsequently detailed), the aim of this article is to explore the three dimensions of principals’ identity and to develop and validate an analysis instrument of principal’ professional identity in the Sweden context.

The article is structured as follows: First, the Swedish case is introduced. Next, the theoretical perspectives are detailed. Then, the empirical studies and the questionnaire are presented. In the results section, descriptive statistics are first presented, followed by a factor analysis. The article ends with a discussion and implications for future research.

Theoretical background and literature review

The Swedish case – a contextualization

As noted above, principals’ identities are situated in time, space, and place (cf. Crow et al., 2017; Wenger, 1998). Accordingly, an analysis instrument needs to consider characteristics of the specific context. Regarding Sweden and the Swedish school system, there are several factors to consider with relevance for principals’ identity. In the postwar period, Sweden was often described as a ‘Social Democratic welfare state,’ having one of the most centralized school systems in the western world (see, e.g. Jarl et al., 2007; Lindensjö & Lundgren, 2000). The National Board of Education and the County Administrative Board was the cornerstone of this system, providing detailed directives to the local municipal level, for instance, regarding employment of principals and teachers, salary setting, distribution of financial resources, the content of the syllabuses, and so forth. In this system, a professional principal was, a bit simplified, a talented administrator well familiar with current regulations and directives.

Over time, ideas were initiated on how to involve principals and teachers in local school improvement and develop professional expertise based on the local conditions. A watershed in this development was the investigation on School’s Inner Work (SIA) (Government Bill 1974: 53, 1974, p. 53), where several recommendations were provided regarding principals’ work and school management. The investigation, for example, suggested that principals should develop the school’s organization and quality work to promote good educational results and job satisfaction. These ideas were also in line with the school research during that time, where the importance of understanding the organization and culture of the local school is emphasized (see, e.g. Lortie, 1975; Sarason, 1971). These ideas, at least theoretically, opened up for a professional leadership more rooted in the local context, integrating principals’ personal driving forces and ideas.

However, the challenges of educational reform are well-known today (see, e.g., Cuban, 1990, 2013), and it was not until the 1990s that the decentralization ideas were truly implemented in Sweden (Lundahl, 2002a, 2002b). The school system was also opened up for private or ‘independent’ school organizers, which together with public schools, from this time on, forms a school market in Sweden. From a pedagogical point of view, there was still confidence in the local professionals and that they could plan, implement, and develop their activities linked to the local context. The new school authority, the Swedish National Agency for Education, therefore, had the directives to ‘stop at the municipal border,’ safeguarding the local freedom. In retrospect, however, it can be noted that many municipalities, principals, and teachers found it difficult to understand and act in the transformed educational landscape (Swedish National Audit Office, 2004, 2011). Among other things, it has been revealed that the reforms were carried out at a time when neither municipalities, principals, nor teachers had the preparedness, resources, and/or capacity to handle the new responsibilities and assignments. In addition, it has also been argued that the Swedish Government and its agencies did not provide sufficient support to local education authorities (LEAs), principals, and teachers as implementers of the reforms (Bill 2014, p. 5).

However, in the past 10 years, a course direction has occurred in Swedish, schools, which has affected principal’s work and profession. This course direction has been described in terms of a re-centralization movement where the state is considerably more present at the local municipal level compared to the 1990s and early 2000s. The establishment of a new agency in 2008, the Schools Inspectorate, is often seen as an important step in this development (Rönnberg, 2011, 2012). The re-centralization movement has also been described in terms of a ‘juridification’ process of the Swedish school system (Novak, 2018; Runesdotter, 2016). A main assignment for the School Inspectorate has been (and still is) to inspect principals’ responsibilities and obligations, for example, regarding the schools’ systematic quality work, equality, grading and assessment, and so forth. In addition to external inspection and an increased number of state directives, a number of large-scale school improvement initiatives are also launched by the National Agency for Education in later years, for instance, ‘Collaboration for the Best School Possible,’ in which LEAs and single schools with poor school results can receive support in collaboration with the National Agency for Education and participating universities.

It is worth emphasizing that LEAs, principals, and teachers do not necessarily have a negative view on external inspection and increased state presence at the local level. The work of Ek (2012), for example, indicates that Swedish municipalities not only exhibit a susceptibility to increased state control but have also internalized an ‘inspection logic,’ where inspection is perceived as an important and accepted part of their work. Comparable results are also shown in recent research, where newly-appointed principals in Sweden express opinions in favor of ‘the strong state’ and a continued re-centralization of the school system (Nordholm & Andersson, 2019).

Against this background, there are several characteristics of the current forms for school leadership and management in Sweden framing principals’ work. To start with, the division of responsibility for education is roughly speaking divided between three levels: the government and its agencies, the LEAs or its independent counterpart, and principals in the local schools. In the (still) rather decentralized governance model, the government level is responsible for the overall framework and structure, for evaluation of school activities, and for providing support to LEAs/its independent counterpart and principals in all school forms. LEAs/its independent counterpart, in turn, are responsible for providing education of good quality in preschools and schools. Regarding the municipal school organization, a majority of the public schools is directed by a school board and one or several superintendents. Somewhat specific for Sweden, the Swedish school boards consist of appointed politicians representing the political parties of the municipality council. Also, worth emphasizing that Sweden has 290 local municipalities; therefore, there is also a variety regarding local school governance (Johansson et al., 2016). Concerning the local school level and principals’ work, there is, here as well, a diversity regarding how education is organized and operationalized. Both public and independent school organizers and single principals are inspected and evaluated regularly by the Swedish Schools Inspectorate. All head principals are also obligated to study at the National Principal Training Program within 1 year from employment. The program is organized into three courses: ‘Educational law,’ ‘Governance, organization and quality,’ and ‘School Leadership,’ each of which generates 10 higher education credits.

Summing up this section, there are several characteristics of the Swedish case with relevance for principals’ professional identity in general and for the three dimensions elaborated on in the work of Crow et al. (2017) in particular. The three dimensions and the theoretical lenses of the article are described further in the following section.

Theoretical lenses

Against the above contextualization and the current forms for leadership and school governance in Sweden, the choice of theoretical starting points becomes important to consider. On an overall level, Burke and Stets (2009, p. 2) pointed out that ‘an identity is the set of meanings that define who one is when one is an occupant of a particular role in a society, a member of a particular group, or claims particular characteristics that identify him or her as a unique person.’ In that sense, the relationship between identities and context is essential. Burke and Stets also emphasized the distinction between person and agent, stressing that an identity is an agent, and therefore, one person has many identities. The importance of building a strong and stable identity, for example, has been demonstrated in the work of Ritacco and Bolivar (2019), showing that successful leadership and school improvement largely depend on principals’ professional identities. Likewise, Saarukka (2014) showed that a strong identity becomes a supportive structure for principals in navigating complex situations of school leadership.

As noted above, the work of Crow et al. (2017) constitutes the theoretical starting point for the analysis of this article. They outlined that principals’ identity is made up of three dimensions (professional, personal, and situated or socially located) and principals’ identities are to be understood as an interplay between these dimensions. Based on the idea that principals’ identities contain several elements (i.e. dynamic, socially negotiated, developmental), identities are not seen as static as they change as the context changes (cf. Whiteman et al., 2015). Crow et al. (2017) described more in detail that the professional dimension involves social and policy expectations on what a good school principal is and leadership ideals. This dimension also reflects influence by social trends and from long-term policy. The professional dimension can also contain competing and conflicting elements linked to the national, regional, and local context. Certain examples include local and national policy, professional roles and development, research, leadership expectations, and so forth. Linked to the systemic-structural aspects highlighted above and the close relationship between identity formation and society stressed by Burke and Stets (2009), it becomes relevant to consider the expectations on the ‘strong state,’ which appear to stand strong at the local municipal level in Sweden. Put differently, despite the decentralization programs in the 1990s and early 2000s, the government and its agencies have a strong position in the current model for school governance, especially in light of the re-centralization programs in recent years. Regarding normative expectations, the syllabuses of the National Principal Training Program are determined and specified by the Swedish National Agency for Education. Perhaps most obviously, the courses ‘Educational law’ and ‘Governance, organization and quality’ express such normative expectations impacting principals’ professional identity.

Regarding the second dimension, i.e. the personal dimension, Crow et al. (2017) sketch this dimensions in terms of personal values and beliefs, yet in relationship to the role as a principal (cf. Burke & Stets, 2009). That stated, the dimension is more linked to the life ‘outside’ of schools, focusing on characteristics closely related to family and social roles. Crow et al. (2017) depicted that this dimension therefore could involve competing elements such as family and relations with friends. Response and feedback from friends and family could also become sources of tension as they might collide with the other two identity dimensions. The importance of this dimension is also confirmed in corresponding leadership research, where the importance of principals’ and other leaders’ personal commitment, values, authenticity, and curiosity has proven to be important for successful leadership (see e.g. Angelle, 2017; Bezzina & Bufalino, 2019). Linked to the Swedish case and the systemic-structural aspects described above, there are relevant characteristics related to this second identity dimension as well. From a historical perspective, Swedish principals were more requested to become skilled administrators executing and fulfilling state directives rather than addressing leadership qualities linked to the personal identity dimension. Even if the decentralization reforms of the 1990s and early 2000s, at least hypothetically, opened up for principals to explore this dimension further, the above review indicates that former expectations and beliefs remain – a development that perhaps has increased in recent years in light of the re-centralization programs and increased state inspection. However, even if goals in the syllabuses in the National Principal Training Program, arguably, have an emphasis on qualities within the professional identity dimension, there is also content, particularly in the third course ‘School leadership’ in which the principals are encouraged to explore more personal leadership qualities. Nevertheless, there is still a need to examine this identity dimension more in detail to study how this dimension is formed in the Swedish context.

The third identity dimension, the situated or socially located dimension, is more linked to local conditions and a specific context, i.e. a school district, a school department, a specific school, or even a certain classroom. According to Crow et al. (2017), this identity dimension, therefore, holds demographics, school culture, socio-economic conditions, principals’ support, and feedback. The efforts to involve principals and teachers in local school improvement activities to reinforce a professional expertise based on the local conditions since the 1970s becomes noteworthy from this perspective, not the least because the above review indicates that such ideas have been difficult to realize. There are, as noted, also questions regarding the government’s support and resources during the 1990s and early 2000s to empower such capacity and understanding at the local municipal level. Against this background, it becomes essential to explore what Swedish principals put into the situated or socially located identity dimension – not the least because capacity to understand and adapt leadership in the local context is verified as a key factor for school leadership and successful school improvement (cf. Hallinger & Heck, 1996; Seashore, 2015).

Research design

A pilot study was conducted with a strategic sample of principals, with regard to gender, number of years in the profession, responsibility for different school forms, school context, and so forth. The evaluation of this pilot study, in total ten (n = 10) principals, led to some adjustments and clarifications for the main survey.

Regarding the main survey, e-mail addresses to principals were attained from the Swedish National Agency for Education. According to the official statistics (Swedish National Agency for Education, 2019), there are 3,600 principals in Sweden, in full-time employment. In the current study, a narrower sample was used, based on the aim and efforts of the article. To start with, deputy principals were excluded because these principals have different responsibilities compared to head principals, for instance, when it comes to legal matters and school budget issues, i.e. responsibilities linked to the professional identity dimension. Deputy principals were also excluded because they generally have less time in the profession; therefore, they also have limited possibilities to answer how identity is shaped over a longer period of time (cf. Ritacco & Bolivar, 2019). They are not obligated to participate in the National Principal Training Program, which is an important argument in our opinion because the program declares prevailing expectations on Swedish principals, particularly from a governmental perspective. Comparably, temporary appointed or hired-in principals were also excluded because they also have reduced possibilities to answer detailed questions on identity, particularly on the situated or socially located identity dimension.

The sample included head principals in both public and independent schools. The key motive for integrating and merging principals from the different forms of school ownership is that further studies (Author) revealed that there were no significant differences regarding ownership. Regarding additional systemic-structural aspects, head principals, regardless of ownership, have to complete the National Principal Training Program, arguably forming the Swedish principals’ identity. An analogous consideration was to include principals with responsibility for both preschools and schools in the final sample because preschool principals have to attend the National School Leadership Training Program. One further argument holds that these principals, especially in rural areas, are responsible for both preschool and compulsory school.

Based on these considerations, the survey targeted approximately 3,000 Swedish principals. The e-mail addresses attained from Swedish National Agency for Education, however, were not entirely updated; therefore, new e-mails were sent to these principals. In addition, two reminders were sent out to all principals. Despite these challenges, we argue that the data collected meet the purpose and objectives of this article. The sample of principals (n = 1,286) represents 249 out of the 290 local municipalities. There is also a nice spread between the four types of municipalities defined by the Swedish Board of Agriculture (large city, city, rural areas, and sparsely populated rural areas). In addition, there was a good spread in terms of age, number of years in the profession, size of the school unit, gender, and a satisfying spread in terms of school ownership.

Regarding homogeneity and heterogeneity of the final sample, we believe that there is a satisfying spread in terms of age, gender, time in the profession, principals’ local work context, and so forth. At the same time, when it comes to homogeneity, we argue that the sample is based on criteria targeting important characteristics of Swedish principals’ identity. In other words, the final sample makes it possible to frame the Swedish case and to address principals’ professional identity.

A non-response analysis highlighted some possible factors explaining missing replies. One explanation holds that Swedish principals tend to change jobs more often than their European colleagues (OECD, 2016); hence, positions are vacant and principals are replaced by hired-in principals. In addition, Swedish principals participate in a great number of studies, which might lead to ‘survey fatigue.’ As noted, we also (a bit surprisingly) had difficulty in obtaining updated e-mail addresses from the National Agency for Education leading to further investigations in order to attain these addresses. Lastly, some e-mails got stuck in the municipal spam filters, resulting in all reminders being sent directly to the principals in the local schools.

The questionnaire

This section details the analysis instrument used in this article to explore the three identity dimensions. The questionnaire consisted of 42 questions. However, only questions regarding the principals’ backgrounds and professional identity were used in this analysis.

Accordingly, some of the questions in the questionnaire were constructed according to the different dimensions of identity, sketched by Crow et al. (2017). More precisely, the professional dimension included questions on the influence of steering documents and the government agencies, current research, normative assumptions in the system, and so forth. Similarly, the personal dimension included questions about values and beliefs, on personal engagement in the school, and in the local society. Lastly, the situated or socially located dimension comprised questions on the influence of local factors such as expectations from other principals and teachers, school culture, and so forth.

The principals were asked to state the extent to which they agreed with each statement, by indicating on a scale ranging from 1 to 6. The selected questions and how they were formulated are presented below.

To what extent do these aspects influence you as a principal (choose between 1 and 6, where: 1 = disagree and 6 = agree completely)

The professional dimension:

Steering documents and policies

The Agency for Education

The Swedish School Inspectorates

Own education or further education

Current research

The personal dimension:

My own personal values

My personal engagement and interest in the school

My interest in the local society

My leisure interest

The situated or socially located dimension:

The school’s culture

Expectations from teachers

Expectations from the local society

Expectations from other principals

Data analysis

The overall aim of this study was to initiate the development of an analysis instrument to measure school leaders’ professional identity and compare empirical data with theoretical suggestions made by Crow et al. (2017). One way to compare theoretical constructs with empirical data is using factor analysis techniques (Gorsuch, 1997). Before performing a factor analysis, it is suggested that the data should be carefully scrutinized (Costello & Osborne, 2005). One important suggestion is to test the variability and skewness of the data. Different methods are suggested if data are normal or violating normal distribution. It is further suggested that when doing a factor analysis, the number of factors must be taken into consideration. One common method is to use eigenvalues greater than 1 as criterion for extraction of factors. However, this method is criticized by Velicer and Jackson (1990) for being among the least accurate methods. An alternative is to use the scree test or scree plot (Catell, 1966) as criterion for extracting factors. This method is also criticized for being too subjective. Costello and Osborne (2005) recommend performing a number of factor analysis with different numbers of factors and using the solution with the ‘cleanest’ solution.

In addition to addressing these remarks, the problem of common method variance must be addressed (see, e.g. Podsakoff et al., 2003). In the current study, all data were collected with one questionnaire on one occasion; therefore, there is a potential problem with common variance. Podsakoff et al. (2003) highlight a number of sources of common variance such as common rater, common measurement context, common item context, and characteristics of the item used. In the analytical work, we have used different forms of questions, both open-ended and closed questions in order to reduce the problem with common variance. The respondents were also protected by anonymity, thus, reducing informants’ evaluation apprehension and minimizing socially desirable answers. Finally, we have used a statistical test suggested by Podsakoff et al. (2003), namely, the Harman single-factor test by including all 13 questions listed above (Chang et al., 2010b). The factor analysis was performed using the principal axis methods and specifying one factor. The results showed that one factor accounted for 19.1% of the variance. This indicates a moderate measure of common variance. However, the results of the study are interpreted and presented with an understanding of the challenges stressed by Podsakoff et al. (2003).

Results

Descriptive analysis

In the first step of the SPSS-analysis, descriptive statistics were computed for all items (see Table 1). Descriptive statistics for the 13 items are shown in Table 1.

Table 1. Descriptive statics for all items in the questionnaire.

	M	SD	Min	Max
To what extent do these aspects influence you as a principal
Steering documents	5,45	,73	1	6
Agency for Education	5,11	,95	1	6
School inspection	4,81	1,14	1	6
Own education or further education	4,94	1,02	1	6
Research	4,78	1,14	1	6
Personal values	5,41	,85	1	6
Personal engagement for school	4,71	1,16	1	6
Interest in local society	4,16	1,30	1	6
My leisure interest	2,30	1,31	1	6
The school culture	4,59	1,12	1	6
Expectations from teachers	4,62	,97	1	6
Expectations from the local society	4,56	1,19	1	6
Expectations from other school leaders	3,10	1,32	1	6
N	1286

The table shows differences between the 13 items. The principals rated the influence of both The National Agency for Education and Swedish Schools Inspectorate as high. Expectations from other principals and influence from own leisure interest were valued comparably low.

Factor analysis

With the aim to validate the factors described by Crow et al. (2017), a number of explanatory factor analyses were conducted using a principal component method with varimax rotation. Costello and Osborne (2005) suggest performing a number of different factor analyses with a varying number of factors extracted. We performed three factor analysis, with two, three, and four factors extracted. In this process, we omitted factors loading on more than one factor with factor loading higher than.30. The two-factor model accounted for 39% of the variance and the four-factor model accounted for 60 percent of the variance. However, the four factor model did not show a clean output, with four items loading higher than .30 on more than one factor. It was suggested that the three-factor model be compared to the empirical data one with the ‘cleanest’ solution (Costello & Osborne, 2005). This model includes 10 items.

Descriptive statistics for the remaining 10 items are shown in Table 1. Measures of sampling adequacy and sphericity were performed to screen the data. The KMO measure was found to be .70, which suggests that the sample is adequate for carrying out a factor analysis. The sphericity was found to be significant, $x^2=1856.54$ p < 0.00, which suggests that the correlation between the items was strong.

The factor analysis resulted in a three-factor model for the 10 items. The rotated factor loading matrix for each of the 10 items is shown in Table 2. All factor loadings below .30 were omitted (cf. Djurfeldt & Barmark, 2009). The three factors accounted for 58.42% of the total variance, with 25.28% for the first factor, 17.35% for the second factor, and 15.77% for the third factor.

Table 2. Factor loadings with principal components extraction method.

		Factor loadings
	Professional	Personal	Local
To what extend do these aspects influence you as a principal
Steering documents	.80
Agency for Education	.84
School inspection	.75
Research	.64
Own education or further education	.53
Personal engagement for school		.82
Personal values		.79
Expectations from teachers			.80
Expectations from the local society			.73
Expectations from other school leaders			.64
N	1286	1,57	1,36
Eigenvalues	12,91
% of variance	25,28	17,35	15,77
KMO	0,69
Bartlett´s test	1856,55
Interfactor correlations
Professional	1.00
Personal	.25	1.00
Local	.21	.33	1.00

The items were categorized into the three dimensions: professional, personal, and the situated or socially located dimension of identity presented by Crow et al. (2017). Thus, the factor analysis identified survey questions with a strong correlation to the three identity dimensions. The professional dimension included four items, with statements about the influence of steering documents, the National Agency for Education, the Swedish Schools Inspectorate, and current research. The personal dimension included three items, with statements about personal engagement in school, personal values, and interest in community. The third dimension included three items, with statements about expectations from teachers, the local school culture, and expectations from other principals.

The internal consistency among all items was tested by computing Cronbach alpha coefficients. The Cronbach alpha was .69 and ranged from .66 to .70.

Discussion

The aim of this article has been to explore the three dimensions of school principals’ identity (professional, personal, and situated or socially located) identified in the study by Crow et al. (2017), and to develop and validate an analysis instrument of principals’ professional identity in the Sweden context. A nationwide survey given to Swedish principals contained questions connecting to the three dimensions to enable such an analysis. Quantitative data analyses, including a factor analysis, were completed to detect areas and questions that correlate (and do not correlate) with the framework provided in the study by Crow et al. (2017).

The SPSS-analyses showed that some survey questions had a strong correlation with the professional dimension, more specifically those that targeted the steering documents; obligations, directives, and recommendations from the National Agency for Education; external inspection by the Swedish Schools Inspectorate; and current research. With regard to the second dimension, i.e. the personal dimension, the analyses identified issues correlating to this dimension, more precisely those addressing personal engagement in the school; personal values; and interest in the community. Regarding the situated or socially located dimension, the analyses revealed questions that correlated strongly with this dimension, namely, expectations from teachers; school culture; and expectations from other principals in the municipality. These results, together, provide a more detailed picture of each identity dimension and the relationship between them; consequently, the results also strengthen the relationship between theory and practice. Based on this map, the results of this study can also set the direction for further studies in other national, regional, and local contexts.

Nevertheless, it becomes important to reflect upon the results and the analysis instrument from a wider perspective, given the historical background and the systemic-structural aspects of the current school system in Sweden. A point of departure for the article was that principals’ identities are understood as individual and collective constructions linked to negotiations in different times, rooms, and places (cf. Crow et al., 2017; G. Crow & Møller, 2017; Wenger, 1998). We also assume that the link between identity formation and the surrounding society (i.e. structure) is essential (Burke & Stets, 2009). In other words, it becomes important to stress that identities are not static as they change as the context changes (Whiteman et al., 2015). That declared, there are several characteristics of the Swedish case with relevance for the three identity dimensions and the analysis instrument developed. For instance, the strong influence of the government and its agencies within the professional dimension becomes somewhat logical given the historical background in Sweden, but also in relationship to the content of the National Principal Training Program’s courses Educational law and Governance, organization and quality. These results also become important in upcoming analyses, for instance, focusing in the power relationship between the three dimensions.

Regarding the personal dimension, it is relevant to highlight that the principals’ personal engagement in the school and their personal values appear to constitute important content for this dimension, which, importantly, has been established alongside the decentralization – re-centralization reforms (cf. Angelle, 2017; Bezzina & Bufalino, 2019). When it comes to the third and final dimension, the situated or socially located dimension, the analysis has provided a more detailed image of what Swedish principals actually place in this dimension. Also, it is relevant to note that principals rated factors ‘within’ their local organizations higher compared to the ‘outside’ factors. Thus, the factor expectations in the surrounding community was rated lower compared to expectations from teachers and other principals, and the culture of the school. One potential way to explain these results is that the decentralization programs managed to develop and restrain some structures at the local municipal level, but not others. These results are also important for upcoming analyses, given that understanding and adapting leadership in the local context is verified as a key factor for school leadership and effective school improvement (cf. Hallinger, 2011; Seashore, 2015).

By obtaining a clearer image of what the principal’s identity includes and what it does not include, the results of the article also become important in a broader sense, for example, for school politicians, policy makers, and other stakeholders as well as the universities involved in training principals. At best, the results can contribute to improvements and re-prioritizations to strengthen the forming of a professional identity throughout the entire career (cf. Ritacco & Bolivar, 2019).

Conclusion

Crow et al. (2017) and G. Crow and Møller (2017) encouraged further studies on principals’ professional identity, and we believe that our contribution and the Swedish case have expanded the current understanding of the field. More precisely, the above analysis has contributed to the creation of an analysis instrument for exploring the three dimensions, the relationship between them, and how principals’ professional identity is formed and negotiated in different contexts. For instance, what are the potential differences regarding identity between principals in larger cities and sparsely populated areas, or between those working in socio-economically vulnerable areas and those working in socio-economically strong areas? Moreover, are there differences between principals with longer experience in the profession and those who are new to the profession? Questions of this nature can give us important pieces for obtaining a more multifaceted understanding of school leadership in different contexts.

That declared, some limitations of the current study must be highlighted. To start with, the factor analysis and the empirical results are to be understood as one of the first attempts to compare theoretical constructs on school leaders’ identity with empirical data. As noted above, one of the limitations of the current study is that the current design did not provide opportunities to test the factor structure on a new sample. Accordingly, further studies should address that difficulty to validate our results with another sample, for instance, from another country to generate more reliable conclusions (cf. Podsakoff et al., 2003). In addition, based on the assumption that identities are not static and that identities change as the context changes, there is a need to modify the analysis instrument and its questions, having a sensitivity to different national, regional, and local school contexts. Lastly, but equally important, it is also evident that the findings attained by this analytical instrument needs to be complemented by case studies, for instance, with mixed methods designs, i.e. by biographical interviews, observations, narrative analysis, or focus groups. Only then can we examine the relational aspects of principals’ professional identity which, for example, the work of Saarukka (2014) has revealed as being of importance.

Acknowledgments

The authors would like to thank Mikael Magnusson, Project Administrator at the Center for Educational Management, Uppsala University, for his assistance with the design and administration of the survey, upon which the article is based.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Daniel Nordholm

Daniel Nordholm is a Senior lecturer at the Faculty of Education, Uppsala University, Sweden. His former research has been published in influential leadership journals.

Anders Arnqvist

Anders Arnqvist is a Professor at the Faculty of Education, Uppsala University, Sweden. He has long experience of educational research, both with quantitative and qualitative methods.

Elisabet Nihlfors

Elisabet Nihlfors is a Senior Professor at the Faculty of Education, Uppsala University, Sweden. She has a prominent publication list of scientific articles, books, and book chapters.