Approaches to empirical research on inclusive education

Empirical and critical studies of inclusion and inclusive practices in school represent different understandings of inclusion and different approaches to studying the phenomenon. They give a contrasting and diffuse picture of what inclusion is and about how inclusion is practiced. This article discusses two different ways of researching inclusive education and defining inclusion to be found in the Norwegian research literature. It is argued that by adopting either just a macro-orientation, studying the school structures, or only a microorientation, studying what selected teachers do in class, obscures other important issues that determine the quality of what goes on in school and that have to be addressed for inclusion to be operative.


Introduction
Very often, when inclusion in education is referred to and discussed, it is as ideology and policy, and not as experience. Internationally, little empirical research has been done on inclusive education (Emanuelsson, Haug, and Persson 2005). Available empirical and critical studies of inclusion and inclusive practices in school represent different understandings of inclusion as a concept and different approaches to studying the phenomenon, and give a contrasting and diffuse picture of what inclusion is and about how inclusion is practiced (Mitchell 2005;Rix et al. 2005). This is partly due to the fact that there is no 'single form' of inclusive practice (Dyson 1999), but also because of different approaches to researching inclusion. This article will look more closely at two different ways of researching inclusive education and defining inclusion adopted by Norwegian researchers with reference to Norwegian schools.

Definition
Broadly defined, inclusion aims to change school by increasing all pupils' participation within the culture and curricula of school, and minimizing their exclusion from school culture and curricula (Booth 1996). Inclusion may be understood with reference to both vertical and horizontal dimensions. The vertical dimension represents different levels in the education system, ranging from ideology, policy and structures, via teaching-and learning-processes to results. The Index for Inclusion, for instance, identifies three vertical dimensions of school life: policy, culture and practice (Booth et al. 2000). The horizontal dimension consists of all the elements or challenges that could or should be met on different single vertical levels. To exemplify, the frame factor theory identifies three different horizontal systems that constrain, govern and regulate the teaching process: curriculum (goal system), administrative apparatus (frame system) and judicial apparatus (rule system) (Lundgren 1981).
Inclusion has been mostly associated with special education. This has reference to developments in the US, where the term was used to describe the restructuring of special education to permit all or most pupils to be integrated into the mainstream classrooms through reorganization and innovation (Ware 1995). The drive towards inclusion in general can also be traced back to the Salamanca statement about special education (United Nations Educational, Scientific and Cultural Organization 1994), which has been very important for developments in Norway. However, it is noteworthy to realize that the declaration goes beyond special education, and sees inclusion as a characteristic and consequence for the benefit for all pupils with a special concern for the heterogeneity between pupils. Inclusion goes beyond integration and the idea of assimilating children with disability into the main and existing school system.

Different approaches to research about inclusion
In the Norwegian research literature two different approaches to inclusion are particularly prominent, one with reference to different levels on the vertical axis, and the other with a focus on different horizontal aspects. The first is inclusion in a macro-perspective, analyzing the structural characteristics of the whole school system. Studies that have been following this line of thought often use formal aspects of the school system as indicators of inclusion, such as, for instance, Vislie (2003) and even Flem and Keller (2000). A second approach is inclusion in a microperspective, with detailed studies of teaching and learning processes in single classrooms, as in for instance Arnesen (2002), Flem, Moen and Gudmundsdottir (2004), Moen (2004); and even discussing the opportunities for individual pupils' learning, as in Flem, Moen and Gudmundsdottir (2004). Here researchers go into classrooms to look at the organization of teaching, the teaching and learning processes, pupils' activities, how the pupils experience the teaching etc., and from there discuss and draw conclusions about inclusion. This article discusses several issues concerning inclusion with reference to these two approaches to research.

Macro-perspectives on inclusion
The macro-study discussed here is Lise Vislie's analysis of changes when it comes to inclusion in school in 14 western European countries during the time period 1990Á96 (Vislie 2003). Most important for her analysis of inclusion is the percentage of children receiving special education, how many of them are placed in segregated provisions and how many are found in mainstream classes. The main idea is that an increasing number of pupils in ordinary schools (mainstreaming) and a decreasing number of pupils receiving special education within and/or outside ordinary classes could indicate a move in an inclusive direction. From this she concludes that inclusion has not gained much ground in the western European region during this period. The system characteristics used as indicators for the different nations during these years have not changed much, and have also been known at least partly from earlier studies (Meijer, Pijl, and Hegarty 1994).

Research validity
The macro-study identifies some structural and formal aspects of education as of most importance for inclusion. They are easily surveyed, and make comparison between countries possible. However, the validity of the variables used as indicators of inclusion must be discussed. It is questionable how and whether this kind of structural data can in any way provide information about inclusion. The argument here is that there might well be more or less inclusive and exclusive structures, but that they alone cannot reveal the degree of inclusion in practice. The formal organization of school only indicates what opportunities for inclusion exist. Inclusion is also affected by what goes on (processes) and what results are achieved within structures. Theoretically this is a consequence of seeing school as an open system, where the different parts and structures of the system affect each other (Scott 1981). Open systems, however, are not linear or rational in the sense that it is possible a priori to anticipate what would be going on within and between the different parts. In that respect organizations are irrational (Brunsson 2000(Brunsson /1985. First, formal structures are not always what they seem to be. To be together in the same school can take several non-inclusive directions both when it comes to special education, and for all pupils (Fulcher 1999). Most Norwegian children, for instance, go to their neighbourhood school. Teaching, however, can be organized in several segregated ways within school such as tracking, special classes and part-time special education given outside classes in small groups. The national data to which Vislie refers indicate that there are hardly any pupils in special schools in Norway, as is the official policy. In reality, during the 1990s and later, an increasing number of municipalities have established segregated provisions for groups of pupils, organized within the ordinary school (Jahnsen, Nergaard, and Flaatten 2006). Statistically they appear as mainstream provisions; in reality they are not. Data from Statistics Norway show that in 2008, 75% of the pupils receiving special education were taught part-time outside ordinary classes, individually or in smaller groups. A segregated special education paradigm and practice seems to be reproduced in ordinary school, because teaching for the majority of these pupils is given systematically outside class (Grøgaard, Hatlevik, and Markussen 2004).
Second, processes within structures do not necessarily reflect inclusive intentions. To be in a class does not in itself guarantee that the teaching and learning going on there corresponds with inclusive ideas. Recent Norwegian research shows that there is a high level of activity in many classrooms. Much happens: the pupils are working and moving, but the purpose of all this activity is often unclear (Klette 2004(Klette , 2007. To be working and active seems to be of more importance than what kind of learning is taking place. The activities often change both in content and form, the pupils can move from one form of tasks to another at a high speed, reducing the possibilities to concentrate and dwell on a task or an issue, making content and substance systematically less clear, and reducing meaning and understanding. Often the pupils are given work plans for a week or more at a time. Then it is up to them when and how to fulfil them. In some classes pupils are occupied with individual tasks for most lessons during a week. They sit by themselves working individually. Their fellow pupils do the same, but the chances are small that they are working with the same subject and the same tasks at the same time. For this reason it becomes difficult for them to help each other. For the same reason it is not possible for the teachers to give support collectively. They must teach each pupil individually, and very often time is not adequate for that (Haug 2006;Klette 2007). The teacher offers the pupils little direct support in the learning process. Teachers' motivational initiatives, control, repetition, questioning, summing up, sanctioning etc. are sparse (Haug 2006). Norwegian pupils in compulsory school are among those who are the least followed up and controlled by teachers in the OECD-area, according to TIMSS (Grønmo et al. 2004). Initiatives for learning, choices of subjects and tasks, deciding what is best for them are very much up to the pupils themselves (Haug 2006;Klette 2007). With self-directed learning and plans for whole weeks, each pupil does more or less as he or she wants, and independently of the teacher and fellow pupils, as long as the point of reference is the work plan for the week. To conclude, a widespread way of teaching and learning in Norwegian schools gives the pupils freedom to choose both the material and formal aspects they want to look into, and even with whom. Many of the activities are self-directed and self-regulated; the pupils are free to withdraw or respond whenever they want to.
Third, the results achieved within what seems to be inclusive structures need not reflect inclusive intentions. Some groups of pupils benefit more from this way of working in Norwegian schools than others. As a group, boys are graded lower than girls in almost all subjects. Twice as many boys as girls receive special education and the benefits of special education have been questioned. Pupils with parents with low cultural capital (little formal education) systematically achieve less than pupils with parents with a high level of education. Pupils with another first language than Norwegian systematically achieve less than pupils with a Norwegian background. The differences between the extremes are considerable and this is documented in recent research (Haug 2003a;Nordahl 2008), in national statistics (Bonesrønning and Iversen 2008) and in international studies (Grønmo et al. 2004;Kjaernsli et al. 2004;Lie et al. 2001). Especially those pupils with less potential for learning are not benefiting sufficiently from being in school (Opheim 2004). The problem is that these differences in Norwegian education have become greater than before. An interesting issue is that for more than 60 years Norwegian education policy has had a declared intention to reduce these differences, and yet we can conclude that it seems more than likely that they have simply been increasing. Heterogeneity seems to be the problem, not only special education. This is one of the main challenges in school, indicating that even within inclusive structures, processes and results do not necessarily promote inclusion.

Micro-studies of inclusion
In the second approach to empirical studies of inclusion discussed here, researchers go into classrooms to examine teaching organization, teaching and learning processes, teacher-pupil communication etc., and from there conclude about the extent of inclusion (Flem, Moen, and Gudmundsdottir 2004;Moen 2004). Compared to the approach discussed above, the most obvious characteristic of this research is the closeness to practice in school. The analysis of inclusion is thorough, detailed, complicated and complex. The research describes what teachers and pupils have been doing, and why these activities and measures are seen as inclusive for the pupils in the classroom, and in particular for those in need of special help.
The choice of teachers has not been left to chance. Both teacher Ruth (Flem, Moen, and Gudmundsdottir 2004) and teacher Ann (Moen 2004) have been recommended to the researchers. Ruth has been chosen because it was a known fact that she had succeeded in achieving inclusive practice for children with special needs, Ann was chosen because she was competent and communicative.
As an example of this approach Moen (2004) adopted a narrative perspective. She wanted to find out how teacher Ann dealt with inclusive educational practice and how she has developed into the teacher she is by paying attention to narratives from Ann's present and past.
What is it then that inclusive teachers do? Flem, Moen and Gudmundsdottir (2004) emphasized that Ruth (their observed teacher) created a positive atmosphere in the classroom; that she had a good academic insight, a positive cooperation with parents, with the school administrators, the other teachers in the team and with an external specialist, and that she continually reinforced positive qualities in all the children. Moen summarized what Ann does with the words: 'membership, mastery, togetherness, involvement and learning ' (2004, 178).

Validity measures
The focus in the micro-oriented research discussed here is the teacher and the pupils, and especially the teacher's way to meet the challenges of learning in class. An important idea behind this research is to give examples of good inclusive teaching as an inspiration to others, and in this way also contribute to defining inclusion. Beyond doubt, this research informs about certain conclusive aspects of teaching and learning, but it is a moot point whether and what this research could say about inclusion.
The challenge for the researchers is that inclusion cannot be observed directly in classroom activity. It is not possible to register discrete and separate incidents and from them conclude directly about the conditions. Instead it must be summarized from thick observational descriptions and thorough analysis that can be broken down into smaller units. Different aspects of what goes on in class must be looked upon as a whole, and in connection with each other. Moen (2004) concludes that it is due to an intertwinement between all the stories and incidents in her data-material that she as a researcher was able to describe what she has seen as inclusive activities. And she continues: ' . . . inclusive activities can only be understood in terms of a continuous developmental process . . .' (Moen 2004, 181).
Therefore it is crucial how inclusion is understood in these studies. The following sentence provides a clue: 'Our aim is therefore to focus on a learning context that, in our view has succeeded in including children with special needs in an ordinary classroom context' (Flem, Moen, and Gudmundsdottir 2004, 86). It is more than a manner of speaking when the authors use the formulation: ' . . . has succeeded in including children . . . in an ordinary classroom context ' (2004, 86) . That a child is to be or has been included could be understood as the opposite of what inclusion is about, because of a focus on the individual and not on the totality of the situation. As has been pointed out many times, when there is a need for an individual to be included it is a sign of stigmatization and even exclusion (Emanuelsson 1995). In that respect no single individual should be included in a class. That formulation gives preference to the class, and defines those who are invited in as being special, as in integration. What is studied in the micro-research discussed here is most certainly integration, and not inclusion.
On the other hand, if we accepted the use of the term inclusion as has been done in the sentence quoted above, what then are the criteria for deciding that what we observe is inclusion? The articles I refer most to do not present such criteria. The research refers to Vygotsky's sociocultural theories and Bakhtin's perspectives on the relationship between language and development. I do not challenge these theories as fundaments at all, but I ask: what is the relationship between these theories and inclusion? To me it seems that these theories should enable the researchers to understand how the teachers taught, and if and how they approached each pupil to embrace the diversity and difference of their needs (Flem, Moen, and Gudmundsdottir 2004). It is difficult to see that this is what they are doing.
A crucial question then is on what grounds can a teacher decide what a child's needs are. The most obvious solution must be to ask the pupil or perhaps the parents. And as is often the case, the most obvious problem must be that the pupil does not express any need; or worse, when the pupil's needs deviate too much from what is accepted in school or as being an issue for school. This is very often the case, illustrating a crucial point in inclusion and a documented problem in Norwegian schools: inclusion is not synonymous with giving the pupil what the pupil wants or for that matter what the pupil needs as he or she sees it. In the micro-studies some disagreements appear. One is, for instance, that some pupils want to be taught out of class, in small groups with one teacher, even when the teacher wants to teach in class (Aas 2003). This can of course be met by arguing that they do not know what is in their own best interests, even though the children's own perceptions, wishes and choices are important within the inclusive context.
In these cases it is not clear, but I presume that both Ruth and Ann give tests, look into the pupils' writing-books and tasks, talk with pupils and parents, observe pupils during lessons and breaks etc. to get to know the pupils' needs. From this they would know, or at least have some idea about each pupil's aspirations, abilities, interests, status etc. And it must be from this knowledge that they decide what questions to ask, what tasks to give and when and where to help.
In the micro-orientation, inclusion is understood as the way teaching and learning happen. The research is strongly oriented around process quality. The problem, however, is that inclusion is reduced to individual mastering by teacher and pupils. Inclusion becomes personalized and individualized. Other issues on the vertical and horizontal axis that could have had some influence on inclusion are more or less excluded from the analysis. For instance, Moen (2004) admits that Ann does not operate with her class in isolation and independent of her present surroundings. There are many voices influencing how she acts, Moen herself saw the need for research to look into that aspect as well, but she herself did not give it preference.
According to the researchers, what was striking was that what the teachers do was not especially unique. On the contrary, these teachers seem to follow up what some research state is important for teachers to do to make sure that pupils will benefit (Nordenbo et al. 2008). The conclusion is that when teaching is done according to well-known and accepted principles of quality in teaching and learning, the result becomes good in the sense of inclusive.
The fact that the quality of teaching is relevant to inclusion is of course beyond doubt. The question is whether high quality teaching is sufficient to ensure inclusion.
First, there are different opinions about what constitutes good teaching (Stigler and Hiebert 1999). Second, the research primarily studies the teachers' and pupils' actions within the structural regulations imposed upon school. It does not look into the regulations or the conditions within them. Neither do the researchers analyze pupils' benefits or result quality. Good classroom practice might well exist within excluding structures and systems. The argument here is that in order to draw conclusions about inclusion, it is not sufficient only to study teaching and learning quality. It is not relevant or adequate only to refer to inclusion on the basis of teaching and learning practices in school.

Discussion
The research on inclusion discussed here has different purposes and different content. Both approaches study different but central aspects of inclusion. The question is how inclusion is understood, and furthermore, what these approaches can tell us when it comes to inclusion? The approaches refer to different vertical positions, structural system characteristics on the one hand, and practical teaching and learning activities on the other. The former case characterizes inclusion according to national school systems. The latter cases study individual teachers' work in class. Inclusion is defined as dealing with general ideals with reference to special education for a total school system, as involving certain characteristics of teaching and learning in school and even as one individual pupil's activities. The studies take into consideration different horizontal aspects as well. In the first case the orientation is mostly around judicial and administrative regulation of school, and mostly related to special education. In the second case curriculum aspects dominate. In the first case the criteria for inclusion are quantitative, clear and simple, objective and formal. In the second case they are qualitative, hidden, subjective and a matter of individual appraisal and analysis.
It of course is possible to use the term inclusion in all these cases, but is it informative? The concept is very complex, which is also in accordance with what has been emphasized by researchers and supporters of inclusion (Clark et al. 1997). As was mentioned in the introduction to this article, inclusion could be defined by three vertical aspects: structure quality, process quality and result quality. The first case studies mostly structural aspects, the other teaching and learning processes. None of the cases focus on learning results. In addition, inclusion is defined by horizontal variables: one way of structuring this is through curricula, administration and legislation. The Organization for Economic Co-operation and Development (OECD), for instance, characterizes inclusive school systems in a macro-perspective in a much broader way than simply by referring to the amount of segregation and integration in special education. OECD looks into the whole education policy when searching for traces of inclusion on a macro-level, where also elements of teaching quality are seen as crucial (Organization for Economic Co-operation and Development 1999). Horizontally they cover elements like: provisions for children with special needs, assessment, funding, staffing, accountability, teacher education, school support services, parental involvement, school organization, curriculum development, classroom organization and classroom teaching. The range of relevant horizontal variables is even greater; for instance, characteristics of the pupils, their parents, their living area, school culture, school leadership, the municipality, curriculum, colleagues, etc. Different groups of pupils, other than those connected to special education, seen as being in danger of exclusion in school could be identified based on variables like age, social class, ethnicity, gender, sexual tendency, residence, language, etc.
It is obvious that the operationalization of the very complex concept of inclusion is limited both vertically and horizontally in both cases. A simplification is often highly necessary in research, to make an issue researchable. At the same time reductionism is problematic because of the third variable phenomenon; the chance that issues could be explained and understood by variables not included in the research, making the research naïve and uninteresting (Gjessing et al. 1982).
There is also another form of reductionism in this research, namely that we have very little information about the relations between different levels on the vertical dimension in the education system. The question is what are the consequences for another level of ideas, formulation and practices on one level? This is an interesting question to study, and this is of importance in education because schools locally have great freedom to act within limits stipulated by state regulations. In general there seem to be weak connections between steering signals about school, teaching and learning and what goes on in school and the results that are achieved. There do not necessarily seem to be any connections between the qualities of the macro-and the micro-systems, with the exception of the fact that the possibilities for inclusion in a pro-inclusive macro-system must be higher than in a non-inclusive macro-system. This is, for instance, one main assumption behind the Aarauer Curriculum Model (Bachmann 2005). A central, and huge, challenge is that school is supposed to live up to many ideals at the same time, often ideals that are not compatible. A central part of understanding inclusion is all the dilemmas associated with meeting all the demands and requirements. Extensive analyses of Norwegian education policy conclude that the expectations of school are contradictory, representing lots of dilemmas and conflicts with consequences also for inclusion on a macro-level (Engelsen 2008), resulting in different and less optimistic conclusions about inclusion than those reached in Vislie (2003).
In Arnesen's (2002) research no single pattern of teaching or strategy of learning in itself guaranteed pupil inclusion. It is relatively easy to notice that, even in Norway, inclusive ideas are more often talked about now than was the case earlier.
A majority of Norwegian parents, pupils and teachers seem to support inclusion on a general level as a main principle for compulsory school (Nes, Strømstad, and Skogen 2004). What is strange is that the concept of inclusion is little used in school, and when it is used, it primarily refers to special education and is in many instances only a synonym for special education. There seems to be no distinct common definition or understanding of the term. The analyses of use and understanding of the concept of inclusion conclude that it does not provide a common basis for the daily work of school (Nes, Strømstad, and Skogen 2004). Inclusion is not deeply rooted within the teaching profession or within school. It is still primarily a political concept. The transformation from ideal into practice on the vertical axis, therefore, appears to be a key challenge in Norway, as in many other countries.
From this perspective, reducing inclusion to being a matter of mainstreaming or not is too simple, since it reduces inclusion to what selected teachers do in class. This gives a few aspects of the concept more weight than others, and conceals other important issues that determine the quality of what goes on in school and that have to be addressed for inclusion to be operative. This does not mean that classroom orientation and the focus on teacher behaviour are unimportant in the debate about inclusion. Inclusive ideas have often been too focused on the macro questions, on the ideals, on the systems, on the organization, and have often neglected what goes on in teaching and learning. The general challenge seems to be that the closer we come to educational practice and results, the more demanding it is to define what inclusion is actually about. There is no doubt: there is an urgent need for research that studies inclusion in a broader perspective, both vertically and horizontally.
This situation is not unique for this concept of inclusion. What we see here is very common when institutions are subject to politically decided top-down reforms imparted by simple, popular and even opportunistic concepts. Such concepts often get different content on different levels in the institutional system. The word, concept and principle of inclusion therefore are too small to incorporate what it involves, and to cover all its meanings. There is a need for differentiation, concretization and nuances. As with many other concepts in education, inclusion is relational, meaning that the phenomenon must be understood in the context and process in which it appears.
One way of meeting the challenges in researching inclusion could be to define different aspects of inclusion. In another context, four different key characteristics of inclusion have been isolated as criteria for inclusive education (Haug 2003b(Haug , 2004. All children should be a member of a school class and be a natural part of the social, cultural and professional life at school together with everybody else (fellowship). Pupils should be allowed to contribute to the good of the fellowship according to their qualifications and to be given opportunities to benefit from the same fellowship (participation). All pupils shall have the opportunity to comment upon and to influence matters concerning their own education (democratization). All pupils should be given an education to their advantage both socially and substantially (benefit). To be able to conclude about the extent of the practice, studies of inclusion within this definition and following this operationalization must collect data, both vertically and horizontally, on all pupils with respect to the amount and quality of fellowship, participation, democratization and benefit.