Nola Simpson (Central Queensland University).

The presence of computers and associated technologies in schools is now commonplace. Indeed for more than twenty years schools have responded to the development of new computerised technologies with significant financial outlay by state education departments and independent schools. To be sure, there has been a massive outlay of funds to equip schools with computers and more recently information and communication technologies (ICT's) to the point that now they are a routine part of daily school life.

Although the physical presence of these technologies is evident, less apparent is what teachers use these technologies for in the classroom and how and why this has occurred. Much research has been undertaken distinguishing a variety of facets of computer use. Whilst too numerous to produce an exhaustive list here, by and large this research ranges from areas such as technology and literacy to social justice issues and pedagogical considerations of using technologies. Indeed the use of technologies is a well-researched area within the broader scope of educational research. But there appears to be little research on the events and strategies used by state education departments to ensure teachers use these technologies in the manner intended, as indicated by system policy documents. Indeed to answer such questions is to examine the political manoeuvrings involved in the introduction of these technologies. Sarason (1971) made the point almost thirty years ago that understandings of change are flawed unless the issue of power is addressed.

It appears that this is still an issue that needs to be addressed, as more recently Hargreaves (1998) indicates that research which can provide a political understanding of the process of implementing change or introducing innovations in schools is needed. Thus while we have ample research to draw on for understanding what technologies are used in schools and how they are used, there is little research addressing the questions of why certain outcomes have resulted from the introduction of technologies in schools. What is warranted then, is research that can attend to the underlying reasons why certain ways of using computers and technologies have occurred. The type of questions that need to be asked involve ascertaining who or what factors were influential in the way in which how computers and technologies are used in schools.

The current literature also indicates there is another issue in researching educational technologies that needs to be addressed: the predominance of technically determined research approaches. Studying technologies in schools poses a methodological challenge due to the juxtaposing of highly technical elements in a highly social setting of schools and classrooms. The result of this has been research that has been either technically or socially determined. Indeed many have argued (Bromely 1997; Lyon, 1991;Mackay, 1991; Young, 1991) within the scope of technologies in education, for too long decisions have been made by technically determined analyses.

While Mackay (1991) and Young (1991) advocate for more accounts to incorporate the social factors, Bromley (1997) reflecting the argument in the broader field of technologies, (Dutton, 1996; Williams & Edge, 1996) argues that a worthwhile research approach does not 'take sides'. Rather, both the social and technical factors need to be incorporated into a worthwhile study of educational technologies. Bromely (1997) attempts to address this issue and offers an analytical model 'with the intent of accommodating both perspective's.' (p. 57).

Unfortunately as Bromely (1997) concedes the model has drawbacks. The boundaries between the social and the technical are still maintained. It is not a case of dissolving them, but rather providing a means of moving from one to the other in a continual and cyclical manner.

The research from which this paper is drawn addressed these two areas of concern evident in current literature. The first issue being the need for an analytical approach that could address political issues concerning the use of technologies in schools. Research questions were concerned with asking why things happened as they did. Secondly the research sought to utilize an analytical approach that did not rely on either a socially determined or a technically determined approach. Actor-network theory provided an analytical approach that addressed both of these identified problems. This paper reports on research that employed an actor-network theory approach in examining the way in which a state education department, Education Queensland, deployed an innovation, the integration of information and communication technologies into the curriculum.

This paper contrasts two approaches to telling stories about innovations: Rogers' (1995) theory of diffusion of innovations and an actor-network theory approach. There has been some interest in comparing diffusion innovation and actor-network theory; McMaster, Vidgen & Wastell (1997) examined the failure of a UK City Council to adopt a structured method for systems development, and Tatnall (2000) examined a curriculum innovation within a university faculty.

Before examining the way these approaches might be used to frame the development of a specific innovation, specifically, the integration of ICT's into the curriculum by teachers in the geographically large state of Queensland, this paper will briefly describe the two approaches to the study of innovations. This paper concludes with a comparison of these two approaches to conceptualising innovation, and identifies the strengths and weaknesses of both approaches.

Diffusion theory.

Rogers' theory of diffusion, beginning with his first publication of Diffusion of Innovations in 1962, is and continues to be widely used as one way of thinking about the way in which innovations are introduced and adopted. He defines diffusion as the process by which an innovation is communicated through certain channels over time among members of a social system. Stemming from this framework for studying innovations, then, are the key concepts of innovation, communication channels, time and social systems.

Innovation refers to an idea, practice or object, which is perceived by an individual as new. It must be new to the individual, called the adopter, although the innovation may not be new per se. The rate at which an innovation is taken up by a group of potential adopters is influenced firstly by characteristics of the innovation itself. These include: relative advantage - the degree to which it appears superior to an existing product or practice; compatibility - the degree to which it matches values and experiences of individuals in the community; complexity -the degree to which it is relatively difficult to understand or use; trialability - the degree to which an innovation may be experimented with on a limited basis; and observability - the degree to which the results of an innovation are visible to others. Secondly, the rate of adoption may be influenced by characteristics of the adopter. These include level of education, social status and 'cosmopolitanism'. Finally, organizational issues such as organizational structure, size, and degree of decentralization or centralization may also influence the rate of diffusion.

The diffusion theory employs a number of 'categories' for individuals. One such group is that of opinion leaders. These individuals influence adopters of innovations either positively or negatively. Their influence varies in accordance with their level of technical competence, social accessibility, level of conformity to social norms and their degree of support for the innovation. A change agent, according to Rogers (1995) is 'an individual who influences clients' innovation-decisions in a direction deemed desirable by a change agency.' (p. 335). Change agents have to ensure that a need for change is developed, establish an information-exchange relationship, and create intent for change in the client. This intent must then be translated into action and the adoption stabilized to prevent discontinuance of the innovation (Rogers, 1995). This occurs due to the attainment of a 'critical mass' (Rogers 1995). When enough individuals have adopted the innovation, the rate of further adoption becomes self-sustaining.

The second element, communication channels, is the means by which information about an innovation is exchanged between individuals or groups of individuals. They may include the mass media as well as face-to-face exchanges.

Time is seen to affect the diffusion process in three ways. Firstly in the decision process, it involves the passage of time through which the adopter passes from first knowledge of the innovation to the decision to adopt or reject. The degree of innovativeness of the individual is the second factor with early or late patterns of adoption equating to the innovativeness of the adopter. Rogers (1995) uses five categories to describe adopters: innovators, early adopters, early majority, late majority and laggards. Thirdly the overall rate of adoption within a community is gauged within a time period. Hence if the number of individuals adopting a new idea is plotted over time, it usually forms what Rogers' (1995) terms, a basic S-shaped curve.

According to Rogers (1995), the social system refers to the 'bounded' community in which the innovation diffuses. The social system thus is made up of the 'units.' These units represent potential innovation adopters and can be individuals, groups of individuals or organizations. For example, in studying the diffusion of the specific innovation of teachers' use of computer generated reporting system, a school would represent the social system, and teachers the units within that social system. Another example could be a state school system as the social system, and schools as the adopting units.

Rogers (1995) uses the term 'homophily' to indicate the degree which individuals, within the social system, share the same or similar interest. He argues that the more similar the individuals the more likely that effective communication will occur. Some 'heterophily' is necessary to ensure that new ideas and practices are able to diffuse from individual to individual. The structures and norms of the social system are also recognized as important components of the social system. For example in bureaucratic organizations, higher-ranking officers generally, would assume that their directives are followed when deploying a directive regarding the uptake of an innovation. Unsuccessful uptake of the innovation then would be framed in terms of attributes of the innovation or attributes of the potential adopter or adopters, who are organizational 'subordinates.'

Actor-network theory.

Actor-network theory takes quite a different perspective in its treatment of innovations, characterised by tracing the complex interplay of influences that impact upon the innovation. Emerging from social constructivist studies of scientific knowledge production in the 1970's (Brey 1997), actor-network theory is underpinned by the assumption that the production of knowledge is not an objective exercise, (Latour 1987), but rather is influenced by social factors (Law 1986). Actor-network theory in adherence to the principles of generalized symmetry and free association (Callon 1986b) offers a mode of analysis that sees binaries like the social and technical as something to be explained rather than as an explanation. It does not make a priori assignments of attributes to humans or non-humans but rather seeks to trace the mutual negotiation of roles. Rather than viewing innovation as a linear process moving along a predicable and visible path (e.g. scientific principles leading to technological innovation, leading to micro-economic considerations, leading to marketing and consumption), 'right from the start, technical, scientific, social, economic or political considerations have been inextricably bound up into an organic whole.' (Callon, 1987, p. 86). Actor-network theory attempts to trace the innovation process as it develops or does not develop. Seminal actor-network theory studies include Callon's (1986a) study of the development of the electric vehicle in France; Latour's (1996) study of the development of an automated transit system for Paris; Nespor's (1996) study of curriculum in university faculties in the USA; Callon's (1986b) study of the population depletion of scallops in St, Brieuc Bay in France; Law's (1986) study of the work of a scientist; and Latour's (1988) examination of the work of Louis Pasteur.

At base, an actor-network theory approach to the study of an innovation involves "following the actors," human and non-human (Latour, 1996), tracing the moves, counter-moves and compromises that go to make up the negotiation of any innovation. Network building or making alliances involves a mutual negotiation of roles for actors, which is typically seen in terms of recruitment, negotiation of roles, policing of roles and then once the alliance is firm, being able to speak on behalf of an alliance and represent the alliance in a space and a time removed from the actors.

The negotiation and stabilisation of roles applies equally to human and non-human actors. Indeed, much effort is required to ensure nonhumans remain true to their negotiated roles. For example, the Sydney harbour bridge requires constant work to ensure it remains true to its role of providing safe transport across Sydney harbour.

Some negotiations with actors are easily achieved, while others may require much persuasion or even coercion within the process of negotiating roles . In essence, network building, and then ensuring actors remain faithful to their role is a story of the relations of power. Here Callon (1986b) indicates the significance of network building,

The stability of networks is never assured. The network builder must continually ensure the actors remain true to their negotiated roles. In some situations, it is possible to delegate roles in such a way that they require minimal policing. The control at a distance, which is a common feature of all innovations, is then made manageable. Traffic lights and speed bumps are examples of non-human actors put in place to provide control at a distance. While this control is not and never can be perfect (you can ignore both at your peril), these non-human actors replace the need to have people monitoring traffic flow and speed.

The innovation.

In this section of the paper an innovation will be read through the lens of both diffusion theory and actor-network theory. Education Queensland's commitment to providing learning technologies has been played out in all Queensland government schools. In broad terms the overall goal was stated as:

Following this publication, Education Queensland initiated several projects to support the development and progress of the innovation. These include the Schooling 2001 project, which provides professional development for teachers, and the ConnectEd project which provides the infrastructure for ICT's in schools. These are umbrella projects comprising many smaller initiatives, each with specific goals. For example within the Schooling 2001 project, the Lighthouse initiative provides funds for schools to explore and deliver professional development programs for clusters of school.

In addition to these projects Education Queensland produced a number of support texts. These were designed to act as a resource from which teachers could draw information. For example the Guidelines for the use of computers in learning (Department of Education 1995) text provides an extensive array of examples of how teachers can use technologies in each of the key learning areas.

A set of minimum standards was formulated by Education Queensland and perceived as an indication of the skills and knowledge that teachers required to integrate learning technologies into the curriculum. If teachers had not already acquired these skills and knowledge from their past experiences, they were obligated to achieve them via professional development. Attainment of minimum standards became incorporated into the enterprise bargaining agreement formulated between Education Queensland and the Queensland Teachers Union in 1997. This accord is highlighted in a 1999 Education Queensland publication:

Additionally both technical and educational advisors were made available by Education Queensland to assist teachers with setting up equipment and protocols in their school communities. This assistance varied and was made available both at a state level, and also within district regions. This assistance was provided not only by Education Queensland but also from companies such as Telstra. For example Telstra, as a provider for Internet access, set up help-desk services and conducted seminars to assist teachers to setting up email accounts and web pages.

This paper reports on a study that was framed methodologically in actor-network theory terms. In other words, an actor-network theory perspective, not a diffusion theory-based perspective underpinned collection of the data. As different research approaches influence the type of data collected, a different data set would have resulted from a study framed by Rogers (1995) diffusion theory. Nevertheless, it is useful to read this innovation in both frames largely because of the dominance of diffusion theory in the educational and technology literature concerned with innovation.

In diffusion theory terms, the provision of professional development can be read as an attempt to reduce the degree of difficulty associated with adopting and using the innovation; in simple terms, to make it easier for users to take up the innovation. Using Rogers (1995) concepts of complexity, compatibility and relative advantage as discussed above, the professional development can also be read as a means of increasing the compatibility of the innovation: the degree to which it matches the values and experiences of the individual adopters. By delivering professional development in which teachers are instructed and given many examples of how ICT's can be used in education programs, the promoters of the innovation aim to make the innovation more compatible with the experiences of teachers. The professional development also serves to reduce the complexity of the innovation. Instruction on how to use ICT's and examples of curriculum applications both serve to make the innovation 'easier' for teachers to use.

The professional development can also be seen in terms of addressing what Rogers (1995) terms relative advantage, the degree to which the innovation appears superior to an existing product or practices Through professional development programs, teachers are made aware of the benefits of using learning technologies as well as allaying fears that the adoption of the innovations is not as difficult as anticipated. Such an approach assumes to know in advance what the major difficulties of teachers are in terms of taking up ICT's in their classrooms, and also what teachers would understand as an improvement to an existing practice.

In diffusion theory terms, teachers as potential adopters are influenced in several ways. Firstly the assumed shared values of teachers as being caring educators, wanting the best for his/her students, is evident in the following words from Frank Peach, (the Director General of Education at that time). 'The ultimate goal for the future development of the use of information technology in our classrooms is to improve students' learning outcomes.' Implicit in this comment is that teachers do want what is best for students, and the use of learning technologies is beneficial for students.

The role of the Queensland Teachers Union in negotiating that teachers will achieve a set of minimum standards as part of the Teachers' Award (1998), can be read in terms of Education Queensland gaining the support of a significant opinion leader. The inclusion of attainment of minimum standards in the Teachers' Award represents a considerable influence on the actions of teachers.

Communication channels are developed through the production of texts and through the appointment of educational advisors and nominated specialist contact personnel associated with the various projects and initiatives. Some of the initiatives (e.g. Lighthouse project) were developed specifically to open up avenues of communication between teachers within schools and school clusters.

Time is an important element in a diffusion theory and is necessary for making statements about the rate of adoption of the innovation. The researcher can nominally set time periods and from these, rates can be calculated. This would provide a quantitative understanding of the adoption of the innovation. However there is scope for disagreement on what is actually classified as adopting. Certainly there would be many ways in which teachers could integrate learning technologies, thus a clear working definition of adoption would be required. It appears that within a diffusion theory approach, the researcher determines this definition.

The missing elements from a diffusion analysis are the underlying reasons for these actions. Diffusion theory is unable to provide a compelling argument as to the reasons behind the actions of individuals beyond those concerned with attributes of the innovation or of the adopters. Working from the assumption that the innovation possesses enough momentum to diffuse through the social system of its own accord, the analyst becomes concerned with answering what factors it was about the innovation or adopters that resulted in successful diffusion. Or conversely, what were the attributes of the innovation or adopters that impeded successful diffusions. Diffusion theory appears to be unable to weave political elements into an explanation of the diffusion of an innovation.

The innovation through an Actor-network theory perspective.

An actor-network analysis would trace Education Queensland's moves in attempting to assemble a large network of actors to support the innovation. Specifically they have negotiated roles with various texts, advisory personnel, computers and the Queensland Teachers Union. Via these actors, Education Queensland has attempted to negotiate certain roles with teachers, roles that support the integration of learning technologies in the curriculum. How successful Education Queensland has been in this regard is not something that can be easily judged but in the school in which I did my research, teachers were clearly supporting students' use of ICT's and incorporating the use of ICT's in their curricula.

Secondly, an actor-network theory framing draws attention to the non-human elements in Education Queensland's network. For instance, the computers need to be persuaded to function in schools in ways directed by teachers. Indeed, the success of the entire innovation hinges on the 'behaviour' of the computers and related technologies. Simply ensuring that teachers behave in a particular way is a necessary but not sufficient condition for the network to be stabilised, as Education Queensland desires. Thus actor-network theory is particularly interested in how actors are recruited; what negotiations take place to arrive at roles that are mutually agreed; and then what mechanisms are put in place to police the various heterogeneous actors as they perform their delegated roles.

Of course Education Queensland, in actor-network theory terms, is itself a network. Thus it recruits and assembles actors into a network, which in turn can work to recruit other actors and extend the network. Importantly, in actor-network theory terms, it is important to pay attention to both human and non-human actors, which are configured into this network. A number of people (such as advisors, technicians), things (the Teachers' Award, computers, cables, routers, hubs and so on) and texts (policy documents, teaching guideline texts and others) are assembled and perform roles in teachers' integration of ICT's into the curriculum.

To recruit computers and ensure they perform their roles, they are provided with electricity, in some instances placed in air-conditioned rooms and monitored by surveillance devices. To integrate ICT's into education programs, then much equipment is provided in schools, hubs, routers, ISL cables and so on. These actors enable (persuade) computers in schools to become part of the network that Education Queensland is attempting to assemble. It may seem odd to say that computers need to be persuaded to accept particular roles, none-the-less they can and do act in ways unintended and undesired. The role the computers have is particularly demanding. They have to operate in such a way that at once, they support how teachers make use of them in their classrooms, support the work of student, and conform the demands of Education Queensland (e.g. not accessing web pages with undesirable content on them). Thus computers have to be persuaded to only perform certain tasks. To ensure undesirable web pages are not accessed, Education Queensland enlists the assistance of filters. These also act to ensure that students adhere to their role, which involves not accessing sites deemed undesirable! Help-desk services, technicians and specialists are all enlisted to ensure that computers perform their delegated role. These personnel also serve to assist teachers to perform their roles.

Thus, an actor-network perspective would have us view Education Queensland as assembling a network to ensure that teachers actively integrate learning technologies into their education programs but in particular ways. Education Queensland does more than simply ask teachers to teach in this new way. They recruit the Queensland Teachers Union and put in place a set of minimum standards. They offer money to schools to put in place plans to develop computing and related technologies in particular ways. They publish and distribute papers, deploy support people, put cables and other equipment into school. Education Queensland assembles a network of actors, human and non-human, to persuade teachers to do a specific task: integrate learning technologies into the curriculum.

Finally we see how computers are persuaded to perform particular roles. The filters replace the need to have a person physically monitoring what web sites are accessed; the enterprise bargaining agreement ensures that teachers will have a minimum set of competencies for using the computers; and system-wide and school-based rules are deployed to determine appropriate and inappropriate use by teachers and students. Each of these delegations and moves by Education Queensland works to provide a measure of control-at-a-distance. If, on occasion, particular actors do not behave in the manner negotiated, Education Queensland or a school can deploy other actors to direct any errant elements in the assemblage back into place. For example, help-desk services can aid teachers encountering difficulties with various hardware or software. Software packages enable teachers to identify undesirable web access by individual students. Each of these actors works to 'encourage' actors to perform certain roles.

An actor-network framing of this innovation draws attention to the processes by which Education Queensland assembles and then tries to stabilise a network to deliver it's educational goals for all of its students. Role negotiation while seemingly democratic was in many instances one-sided, that is, teachers had little say over many of the roles they were required to perform, likewise for the computers. The story that emerges from such a study identifies the high contingency of the innovation: it could have been otherwise, and makes plain the ongoing and additional work that is required of many, many actors to stabilise the network and enable Education Queensland to talk on behalf of its IT initiatives in schools.

In some respects it might be argued that actor-network theory and diffusion theory have some strong similarities in that they both are concerned with the adoption of innovations, they both attend to influences that impinge on the adoption of an innovation and both attend to matters of persuasion and subsequent reinforcement of the adoption. This paper argues that these apparent similarities hide profound differences. Diffusion theory attempts to develop a general model for the adoption of innovations and is framed in terms of the characteristics of innovations, the adopters and the social system in which the adoption is to be adopted. What is less clear is how such assignments are made and, in particular who makes them. For example, who judges the complexity or compatibility of an innovation? Who decides if a person is a change agent, a laggard or an opinion leader? Diffusion theory-based studies assign these roles after the event. So, for instance, if a person is a positive influence on the adoption of a particular innovation then they are labelled a change agent. Who decides what is positive and how influence is attributed is unclear. In providing a framework and a strongly typed lexicon, diffusion theory limits the way in which an innovation is followed and imposes a relatively linear logic, that is from idea, to product, to adoption or rejection of the product.

Actor-network theory on the other hand is based on an assumption that all of the actors involved in the negotiation of an innovation attempt to get their own way, that is operate in their own self interest in any attempts to assemble them into an assemblage that is ordered for a particular purpose. This means that judgements about what a particular actor can or cannot do come from tracing their engagement in the negotiations of the network, not from any a priori judgement that they are good at or predisposed to do something. Actor-network theory sees any innovation as fundamentally contingent, as peculiar, as something that could have been otherwise. What is of interest is how the network or assemblage is made durable over time. Rather than seeking to "explain" why a particular innovation succeeded or failed, an actor-network theory approach traces the complex, negotiated and contingent processes associated with the ordering that any innovation requires. In this way it offers a means of telling a different account about an innovation than diffusion theory does. Actor-network theory stories don't partition out the non-humans, requiring that, in terms of analysis, all actors be treated in the same way. Rather than the cause and effect approach of diffusion theory, actor-network theory traces the manoeuvres, compromises, twists and turns of a negotiation as it changes or is translated during the process of adoption.

The scope for an actor-network theory analysis to yield a broader understanding can be seen in relation to the provision of professional development. A diffusion analysis would posit that the provision of professional development would effect a reduction in the complexity of the innovation. An actor-network theory perspective would argue that the provision of professional development is an attempt to control the way in which the innovation is implemented. That is, by teaching teachers how to use these technologies, there is less scope for them to use the technologies as they wish. Actor-network theory has the potential to reveal what a collection of diffusion 'cause-and-effect' (such as professional development causing a decrease in complexity) actions produce, in this case control over what teachers do. Thus diffusion theory does not appear to be able to go beyond a cause-and-effect treatment of events. As this paper has shown, an actor-network theory approach has revealed attempts at seeking to control at a distance. That is, Eduction Queensland assembles a network to ensure a particular goal is achieved (the innovation is implemented in the manner intended) and that Education Queensland does not have to continually monitor the actions of individual teachers in performing a particular role.

The diffusion theory, as Rogers (1995) warns has a weakness due to the tendency toward a pro-innovation bias: the implication that a given innovation ought to be adopted and therefore will diffuse through a social system. This is not always the case, Rogers (1995) uses the examples of cigarettes, nuclear weapons and crack cocaine. This bias is problematic in that it can sometimes lead researchers to make unnecessary assumptions that the innovation should be adopted. Research underpinned by such an assumption then influences significantly the way in which the research is conducted. Actor-network theory, while obeying the principle of agnosticism, can provide a means of avoiding this pro-innovation bias (to the best of the researcher's ability.) Thus instead of being 'blinkered' in a research approach of ascertaining why things did or did not happen in a particular way, an actor-network theory approach is primarily concerned with tracing the complex and contingent factors involved in the overall innovation process. It is from this empirical evidence then that an understanding of the innovation arises, what factors influence the trajectory of the innovation. And to be sure, regardless of whether these influences are either social or technical in nature, both are treated in the same manner.

Actor-network theory avoids an essentialist notion that innovations possess an 'essence' which is responsible for successful or unsuccessful diffusion. While the diffusion theory examines attributes of the innovation and frame the innovation's success or failure in terms of those attributes, actor-network theory does not make the judgement that the innovation is inherently 'good' or 'bad', but simply reveals the influences that contribute to the fate of an innovation.

As well, the diffusion theory is unable, analytically, to incorporate non-human entities to the extent of actor-network theory. To assume that computers and other non-humans will perform as desired is to dismiss the extensive work involved in ensuring that these actors do perform as required.

This article then has served to support the use of research framed by actor-network theory. It argues that we need to move beyond approaches that offer a limited view on the innovation processes such as that offered by Rogers (1995) diffusion theory.

NOTES