This paper reports an analysis of policy on 'quality' in Australian higher education, from its global context to localised contexts in individual universities. The central argument is that the raison d'etre of such quality policy is to enhance the accountability of universities to external stakeholders, especially government. In the Australian higher education context, it could be argued that specific policy on quality, which appeared in the 1990s, constituted 'unfinished business' from the Dawkins reforms1 of the 1980s. Minister Dawkins clearly set an agenda to achieve tighter Commonwealth Government control across all sectors of education in order to serve 'the national interest' and he focused on higher education first because he was able to use financial 'carrots' and 'sticks' more directly in that sector. His White Paper (Dawkins, 1988) foreshadowed the use of performance indicators as the basis for funding universities, but by 1991, when he moved to become Treasurer, the performance indicator project had stalled. His own Performance Indicators Research Group was cautioning against their use (Linke, 1991), and there was also a growing negative reaction from the sector generally. Quality policy then provided an alternative mechanism to achieve a less direct form of control.

With Higher Education: Quality and Diversity in the 1990s (Baldwin, 1991), the Commonwealth minister established a defined quality policy cycle which extended from 1991 to 1995, and which made a significant impact on increasing the accountability of the higher education sector to the Commonwealth Government. After the initial ministerial policy statement of 1991, the issue of quality was referred to the Higher Education Council (HEC) during 1992 to refine the policy text. The Committee for Quality Assurance in Higher Education (CQAHE), on which academics predominated, then operated between 1993 and 1995 to conduct three rounds of quality reviews of all 36 Australian universities. The incentive for universities to participate was $76 million annual rewards to be distributed differentially according to their quality league table ranking. Beyond 1995, the HEC produced further reports about proposed new directions for quality policy in Australian higher education, until that advisory body itself became defunct in 1998. By the end of the 1990s, quality policy became 'mainstreamed' with other accountability mechanisms and more explicitly aligned with 'hard nosed' quantitative performance indicators. This paper focuses on the period 1991 to 1995, when quality policy enjoyed a unique status, relatively independent of the other accountability requirements of the Commonwealth bureaucracy (Department of Employment, Education and Training2).

For analytic purposes, three levels in the quality policy process were distinguished: the macro level consisting of the minister and Higher Education Council (HEC); the intermediate level consisting of the Committee for Quality Assurance in Higher Education (CQAHE); and the micro level of individual universities. In this analysis 6 university case studies were selected, representing three 'types' of institutions labelled as 'traditional' (old, established and well endowed), 'alternative' (established in the 1960s/70s) and 'former colleges' (transformed from teacher colleges and then colleges of advanced education with the Dawkins' reforms of 1988). The ensuing analysis of the policy process was based on both documentary and interview data, with heavier reliance on documents at the macro level and interviews at the micro level.

In the education policy literature, tensions between analytic frameworks which emphasise state control of policy (eg Dale, 1989) and those which emphasise micro-political agency (eg Ball, 1994) have been highlighted over the last decade (Troyna, 1994; Lingard, 1996; Taylor, 1997). The analysis presented here was designed to bridge the polarity between these positions by employing a theoretical framework of a policy trajectory which incorporates both dimensions. The main components of the policy cycle are the context of influence (where interest groups struggle over construction of policy discourses); the context of policy text production (where texts represent policy, although they may contain inconsistencies and contradictions); and the context of practice (where policy is subject to interpretation and recreation) (Bowe, Ball & Gold, 1992; Ball, 1994). However, three modifications were added to this analytic framework. The first modification was to extend the terrain from within an individual nation-state to the global context, given the increased attention to global policy directions. The second modification was to incorporate a stronger element of state-centred constraint into the policy cycle. The approach is state-centred to emphasise the Government's central role in the quality policy, but not state controlled which implies a top-down implementation model. The assumption was that when a state-centred perpective is combined with a micro-political approach which explores the complexities and 'messiness' of policy processes, a more 'complete picture' is likely to emerge. The third modification to Ball's policy cycle was to highlight explicitly the interlinkages between different levels and contexts, by examining how they continually relate to each other throughout the policy process. For example, possible avenues and mechanisms for feedback from micro level policy practices to both the contexts of influence and policy text production at intermediate and macro levels are explored.

As this paper proceeds, it moves through the contexts of influence, policy text production and practice, with a special focus on 'bigger picture' effects of the Australian higher education quality policy. Then, a trajectory is synthesised for the particular policy under investigation to highlight the messy realities of the policy process, and finally the theoretical framework employed in the analysis is evaluated.

Global economic, political and ideological shifts provided the context for the emergence of quality policies in higher education in many countries3. More specifically, the significant global trends were a transition from elite to mass higher education at a time when public expenditure was being constrained; changing power relations between governments and universities to enhance centralised control of higher education; and an ideological shift towards the New Right which would lead to the growing privatisation and corporatisation of higher education. Markets and customers are central to the New Right ideology. This involves a shift from a producer-dominated culture to a greater legitimisation of the role of external stakeholders, which has been a key feature of quality policies across the globe (Neave, 1991; Green, 1993). Governments, business/industry and students become the customers for the outputs of university teaching, research and community service. Governments are claiming that as they are carrying the biggest burden of the cost of higher education, they can demand value for money. Business and industry claim a stake, too, as employers of an increasing number of graduates, and also as sponsors of research in line with government policies to privatise and commercialise research. Students are also redefined as customers, especially as they are required to contribute a greater share of the cost of their education with fees in various forms.

In Australia, all of the global economic, political and ideological trends in higher education identified above were evident prior to the announcement of the quality policy in 1991. The transition from elite to mass higher education created by the Dawkins' White Paper of 1988 had occurred at a time of growing concerns about a budget deficit, and arguably a crisis of legitimation for the Australian Labor Government (Dudley & Vidovich, 1995). The Government moved to enhance financial accountability right across the public sector and higher education was no exception. In particular, the New Right discourse, or economic rationalism, which became dominant in Australian public policy from the late 1980s (Pusey, 1991; Marginson, 1993; Dudley & Vidovich, 1995), would privilege economic constructions of the notion of quality in higher education, thereby excluding other possible constructions.

Economic factors were cited by macro (minister/HEC), intermediate (CQAHE) and micro (institutional) level respondents in this study as the main initiator of the quality policy under investigation. However, there were different emphases at each level of the trajectory. For example, CQAHE respondents who had been trained by a director of the Higher Education Quality Council of the United Kingdom were acutely aware of global forces for change, whereas institutional respondents were more likely to focus within Australia on increasing demands by Government for financial accountability.

As the global ascent of the quality agenda for higher education mostly pre-dated Australian quality policy, it initially appeared that Australia was simply responding to pressure to take up a global trend. However, it is argued here that the localised context in Australia also created its own unique set of influences which in turn operated to construct an Australian quality policy for universities adapted to suit those local conditions. The Australian approach could be differentiated from the 'global' model emerging from Europe and North America (Vught and Westerheijden, 1994) in that it involved ranking institutions into bands based on performance, and a direct relationship between performance and funding, although the rewards were only marginal, incentive funds. Further, Australia adopted a whole institutional approach, rather than the discipline reviews used elsewhere. These features can be related to the financial power of the Commonwealth Government, the economic constraints facing Australian universities, the backlash against the tight controls foreshadowed in the Dawkins' reforms and the desire to avoid the type of quality policy being subject to heavy criticism in the United Kingdom. Thus, a combination of global and more localised Australian factors provided the context of influence for Australia's higher education quality policy.

The quality policy text was being continually reconstructed throughout the progression along the trajectory from macro to micro levels, as presented in the three separate sections below. In some instances the details were refined and modified but in several others the policy completely changed direction.

Minister Baldwin (1991) set the broad parameters of the quality policy in Higher Education: Quality and Diversity in the 1990s. This document contained multiple constructions of the concept of quality from 'excellent standards' to 'quality assurance' (QA) to 'quality improvement' (QI). However, despite the confusing and often contradictory discourses, it is argued here that all of these 'mutations' of quality were about accountability. In 1991, Minister Baldwin referred the matter of quality to the HEC, which in turn consulted with interest groups both within and outside the higher education sector. Arguably, though, the HEC set the parameters for the contributions from these groups by establishing a framework for quality drawn from an economic model of production and based on the three components of inputs, processes and outcomes. The 1992 HEC report, Achieving Quality, set out the advice to the minister on how the quality policy should be conducted during 1993-1995, but lack of clarity in the discourses, especially between quality and quality assurance was still evident. In the year between the policy documents Higher Education: Quality and Diversity in the 1990s (Baldwin, 1991) and Achieving Quality (Higher Education Council, 1992), there had been an elision from the notion of quality with an emphasis on outcome standards, to the notion of QA with an emphasis on processes.

After Minister Baldwin, two additional ministers were responsible for higher education over the life cycle of this particular quality policy, and they continued to produce policy text. Worthy of specific note was the role of Minister Beazley in adding outcomes to the processes dimension prescribed by the HEC in Achieving Quality (1992). Arguably, this late change which was retained by the subsequent Minister, Crean, contributed to confusion at the institutional level about just what was being assessed by CQAHE - processes or outcomes. Minister Beazley also wanted to reward only half the institutions but Minister Crean later changed this to allow all universities a share of the $76 million annual rewards. The changing policy text in relation to outcomes assessments and the distribution of rewards were major shifts in the policy parameters coming from the ministerial level as the policy was being operationaised.

CQAHE respondents indicated that during the lifetime of this committee (1993-95) it changed the policy text in only a limited way. For example, it changed the league table format for reporting results from six bands to three bands to a matrix; it allocated money on the basis of student numbers rather than operating grants; and it provided institutions with clearer guidelines on the criteria to be used for assessment as well as more detailed reports back to institutions after the assessments. All of these changes were as a result of pressures from the institutional level.

CQAHE enjoyed relatively more autonomy in making the judgements about how to rank institutions on their performance than the procedures to be adopted for the reviews. However, within an economic rationalist discourse designed to serve the interests of the Australian economy in the global marketplace, the particular ranking of institutions within the league table was unimportant. As long as the judgements were made and institutions were seen to be accountable to external stakeholders, the political 'masters' were satisfied that a 'quasi-market' in higher education was being managed. Details of how the judgements were made about institutional performance were largely withheld from the public domain. However it was clear that CQAHE used the input - process - outcome model for the assessment of quality, as the HEC had earlier established, thereby de-emphasising the non-economic dimensions of higher education.

CQAHE produced its policy text for making judgements about institutions with an emphasis on qualitative peer review rather than quantitative performance indicators. Part of the CQAHE brief was to take institutional contexts into account in making its assessments, however CQAHE respondents indicated that consideration of the large range of localised contexts within which the 36 public universities operated had been a difficult undertaking. They argued, though, that it was the qualitative nature of the assessments which could best make allowances for differing contexts, and therefore diversity across the sector. This raises the issue of the implications for diversity of the more recent move towards quantitative performance indicators.

Competing interests struggled for influence on CQAHE, and the policy text it produced was the product of multiple compromises between the various influences from 'above' (especially ministers); 'within' the committee itself (industry, bureaucracy and universities of different types); and 'below' (institutions). Thus, CQAHE was located at the interface of the macro and micro levels of the policy process, and it was required to interpret and re-create (Bowe et al., 1992) the original ministerial policy text to make it useable at the institutional level. Therefore, CQAHE members might be seen as policy gatekeepers or "key mediators of policy in any setting who are relied upon by others to relate policy to context" (Ball, 1994, p. 17).

Individual universities at the micro level of the policy trajectory reconstructed the policy text within their own localised context for each of the 3 years of the quality reviews . They were required to further refine the CQAHE-generated policy text by constructing an institutional quality portfolio each year. Since this portfolio was intended as an exercise in self-assessment, universities were, in theory, given a considerable degree of freedom in the way they constructed their own text. The CQAHE team visits were designed to check the veracity of claims made in the portfolios, and therefore they were also an integral part of the policy text production by individual universities.

Each university had a quality 'story' to tell, and according to respondents at most case study universities, there was an intensive phase of story construction in preparation for both the portfolios and CQAHE team visits. Many policies were generated in the few weeks prior to the submissions and then extensive hours were spent disseminating the story to those who would be interviewed during the visits. Rehearsal for the one day stage production often included hiring private consultants to role play the CQAHE team visits and second guess the questions which might be put to those institutional representatives carefully auditioned for the production. Even most of those academics opposed to the notion of applying quality policies in universities were prepared to learn their lines for the production because their university's reputation (and de facto their own reputation) was at stake.

Interviews revealed that the institutional quality policy text was constructed largely by university managers. The short time frame between the issuing of CQAHE guidelines and the need to submit a quality portfolio would have impeded widespread consultative processes. The quality policy was yet another example of external imperatives demanding rapid responses; a scenario conducive to top-down and streamlined decision making which arguably was one of the key agendas of the quality policy.

As CQAHE guidelines were very general, managers in all six case study universities indicated that they had experienced difficulty in knowing how to go about constructing their own quality policy text. Given the rewards of money, and importantly status, universities wanted more explicit guidelines. However, CQAHE wanted to avoid the notion of a single 'gold standard' which would encourage universities to adopt a common approach to quality. Thus, CQAHE handed responsibility back to individual universities to construct their own policy text. Considerable tension was generated between CQAHE and institutional managers over this issue, with unclear guidelines being the biggest problem identified by institutional respondents in constructing their quality policy texts at the micro level.

There were many common responses to the quality policy in terms of the nature of the text produced across the six university case studies. For example, compulsory student evaluations of units and awards for excellence in teaching were both identified by respondents as emerging from the quality policy across all case studies. However, different 'types' of universities ('traditional', 'alternative' and 'former colleges'), and individual institutions within these types, produced some different forms of quality policy texts. For example, some universities were responding by abolishing and downsizing committees, whereas others were creating new (often centralised) committees. Some were eliminating non-core activities and others were forcing research selectivity. These differential responses suggest that the self assessment component of the program may have created spaces to allow for localised context, although the degree of manoeuvre for micro text production was strictly limited within the guidelines set by the ministers, HEC and CQAHE higher up the policy trajectory.

Respondent views of the effects of the quality policy differed considerably depending on whether they were at the intermediate (CQAHE) or micro (institutional) level of the policy trajectory.

Decision making styles within universities had changed towards more streamlined structures and processes as a result of the quality program, according to nearly all CQAHE respondents, but only half of the institutional respondents. The term 'corporate managerialism' was used to describe these changes by around half of the respondents, but there was no clear pattern of responses for different types of universities. Outcomes of core university activities had improved as a result of the quality program, according to just over half of CQAHE respondents and a large majority of institutional respondents. Teaching was identified by both groups as the area of greatest improvement, with those at 'traditional' universities even more likely to cite teaching improvements than others. Competition between universities had increased according to a large majority of both CQAHE and institutional respondents, with little difference evident between types of universities. Closely associated with competition was the redifferentiation of the sector into a hierarchy of different status institutions. This effect was reported by just over half of the CQAHE respondents, but by an overwhelming majority at the institutional level (the greatest consensus of any question put to institutional respondents). Although there was little difference between university types in observing redifferentiation, it was respondents at 'traditional' universities who emphasised the need to create some world class Australian universities which could compete in the global marketplace for student and research dollars. Not one CQAHE respondent believed that institutional autonomy was threatened by the quality policy (the only category to have zero responses), and only a minority believed that diversity was threatened. By contrast, institutional respondents were evenly divided about effects on institutional autonomy, and a majority believed that diversity had been threatened.

Thus, not only were perceptions of the quality policy effects different at different levels of the trajectory (intermediate and micro), but they were also different across the case study universities (at the micro level), pointing to the messy realities of the policy process. In some cases, contexts were sufficiently similar within types of universities to create similar policy effects but in others, each individual university revealed its unique responses.

It is argued here that the overarching effect of Australia's higher education quality policy was to increase Government control of higher education, although the mechanism employed of 'steering at a distance' was more subtle than direct control. It is also argued here that the quality policy under investigation exacerbated inequalities both between and within universities, and that the resultant differentiation further served the interests of the Government. These two 'bigger picture' policy effects are examined separately below, before briefly highlighting several effects which were identified by respondents as the more positive outcomes of the quality policy.

Australian higher education's quality policy of 1991-1995 was an example par excellence of the mechanism of 'steering at a distance' (Kickert, 1991). According to Kickert, instead of government steering by means of legal prescriptions, prohibitions and regulations, in the new paradigm autonomy and self-responsibility of institutions come first. This type of outcomes-directed steerage has been described by Lyotard (1984) as 'performativity'. Both steering at a distance and performativity were evidenced in Australia's higher education quality policy. 'Performativity' is suggested by the use of incentive funds of $76 million to be distributed annually for each of three years on a differential basis, with the biggest rewards (both financial and, more importantly, status) going to those universities demonstrating the highest quality. 'Steering at a distance' from the Government is suggested by the QA program being voluntary, as well as CQAHE being independent of the Government bureaucracy and dominated by academics making meta-level peer review type qualitative judgements following institutional self assessment.

'Steering at a distance' has also been referred to as 'self-regulation' (Neave & Van Vught, 1991) or 'tight-loose coupling' (Lawton, 1992) or 'steering, not rowing' (Osborne & Gaebler, 1992) or 'separating the steering and rowing functions' (Cuttance, 1994) or 'remote control' (Goedegebuure, Kaiser, Maassen and de Wert, 1994). However, despite the implications of the terms 'distance', 'loose', 'self' and 'remote', it is important to emphasise that steering at a distance does not mean the lessening of controls by governments, but merely a different form of steerage which is more covert. Kickert (1991) explained that its adoption as a guiding paradigm of governance in Dutch higher education was designed to increase the effectiveness of steering and therefore increase efficiency, and it was definitely not a form of government withdrawal. Similarly, Marginson (1997) has identified steering at a distance as a mechanism which facilitates an advance of government control.

'Steering at a distance' involves the simultaneous but contradictory changes which characterise the New Right ideology towards de-regulation, decentralisation and devolution on the one hand, yet greater centralisation and government intervention on the other (Goedegebuure, Kaiser, Maassen and de Wert, 1994). These apparently competing tensions can be separated out as a de-regulation of the 'means' (processes), but a centralisation of the 'ends' (goals). Marginson (1993) has explained the contradictions in terms of local sites gaining process control but losing product control. Using definitions of autonomy by Berdahl and McConnell (1994), the policy mechanism of steering at a distance involves institutions gaining procedural autonomy (over processes) but losing substantive autonomy (over goals). With the quality policy in Australian higher education from 1991 to 1995, not only did universities lose substantive autonomy, but there are serious doubts about whether they gained (or even maintained) procedural autonomy. The quality judgements were about processes as well as outcomes and therefore it might be contended that assessments about processes could threaten procedural autonomy, in addition to the impact of assessments about outcomes on substantive autonomy.

Enhanced control by Government points to the relevance of state-centred models of policy making. However, the mechanism of 'steering at a distance' suggests the need to 'decentre' this approach, and examine how the state operates through localised sites. Further, if steering at a distance is characterised by centralised goals and decentralised means, then a state-centred approach (Dale, 1989) may be most pertinent to understanding goal setting, and a policy cycle approach (Bowe et al., 1992; Ball, 1994) may be most suitable for understanding the processes which occur at local sites to achieve, transform or resist those centrally mandated goals.

Ball (1994) added two more contexts to his original framework of a policy trajectory: a context of outcomes (relating to justice, equality and individual freedom) and a context of political strategy (to tackle inequalities). He argued that the market ideology effectively reproduces disadvantage because it "provides a mechanism for the reinvention and legitimation of hierarchy and differentiation via the ideology of diversity, competition and choice" (Ball, 1994, p. 123), and this effect can be readily seen in the example of Australia's higher education quality policy.

It is argued here that the quality policy increased inequalities between universities. Enhanced competition between universities (a key element of the market ideology) was identified as one of the most significant effects of the quality policy by a large majority of both institutional and CQAHE respondents. Furthermore, although diversity had featured as a prominent goal of higher education, a large majority of institutional respondents perceived that diversity had declined, and instead redifferentiation of the sector into a status hierarchy of institutions had prevailed. This redifferentiation of the sector had the effect of exacerbating inequalities because the greatest rewards (money and status) went to those institutions which performed best in the quality 'stakes', and so the rich just got richer.

Although CQAHE respondents acknowledged the divisive impact of the quality league tables and the negative effect on competition for overseas students for those universities in the lowest bands, on the whole, they did not believe that the program had exacerbated disadvantage for those universities outside the 'Group of 8' ('traditional' universities). Nevertheless, concerns about re-establishing a status hierarchy amongst Australia's universities have persisted in the media (Powell, 1994; Armitage, 1996; Illing, 1997; Richardson, 1998). Redifferentiation could be seen as a way out of the Dawkins' reforms because simply returning to the pre-1988 binary divide of universities and colleges would have been politically unacceptable given the extreme 'pain' that the sector had been put through to bring about those expensive reforms, especially in terms of establishing a research culture in 'former colleges'. Quality provided an apparently 'rational' and 'neutral' legitimating mechanism for selecting those universities which were most capable of competing in the international arena, and supporting them at the expense of the other universities.

Now turning the focus inside institutions, it is also argued here that the quality policy increased inequalities within universities. Not only did respondents report a culture change in universities as a result of the QA program but they also reported a growing culture gap between managers and grassroots academics, with a deepening of the status divisions between the two groups. Concerns about such a gap have been increasingly evident in the literature (Bessant, 1995; Moses, 1995; Illing, 1998). Thus, the traditional (albeit rose-tinted) conception of a university as a community of scholars lead by the 'first among equals' became fragmented with the introduction of corporate managerial models, which many respondents perceived had been an effect of the quality policy. This fragmentation involved a clearer differentiation of Chief Executives and line managers from those involved directly in the core activities of research, teaching and community service.

It appears then, that the Australian quality policy enhanced the power differentials and inequalities both across the sector and within universities. The way in which quality policies impact on existing structures of inequalities needs to be further investigated. For example, on the issue of gender, as women are already overrepresented at the bottom of the hierarchy in universities, they may be disproportionately affected by the changes, even as far as being more likely to lose their jobs with the restructuring associated with quality policies because they are more vulnerable in contract and casual employment. However, when Luke (1997) examined the effects of the 1991-95 quality policy in one Australian university (not one of the case studies used here), she came to the conclusion that in that instance, with the transformation from an informal model of management to one with open systems of accountability, equity target groups such as women became a more visible focus. Thus, the quality policy had brought some new opportunities for women in that particular case. Although more research is required on equality, Luke's work provides an example of potentially positive effects of quality policies. This issue is taken up further in the following section.

Ball and colleagues who have emphasised that policy effects should not always be viewed negatively as new possibilities may arise with the intersection of national and local level policies (Bowe et al., 1992). Similarly, Foucault has argued that power is positive and productive as well as negative and repressive (Mc Houl & Grace, 1993). Respondents in this study certainly found some positive policy effects to counterbalance what most perceived as the negatives of enhanced Government control and increasing inequalities.

Both CQAHE and institutional level respondents clearly identified a major positive as the opportunity for self reflection by institutions, and some even described a positive 'culture change' towards acceptance of the value of such reflection. Of course, this culture shift was not perceived as positive by all respondents, but instead some saw it as further evidence of a managerial takeover of collegial culture. This culture shift also included a frequently lauded renewed emphasis on the quality of teaching. Improved teaching as a result of the quality policy was cited often by both CQAHE and institutional respondents, although it was the 'traditional' universities which saw this effect in the most positive light. An OECD study into Australian higher education (Organisation for Economic Cooperation and Development, 1997) also acknowledged the 'dramatic impact' of the quality policy on teaching.

Finally, most CQAHE respondents emphasised, as a positive factor, that within the global context, the quality policy for Australian universities had achieved enhanced accountability of universities in a way which was more respectful of university culture and autonomy than in most other countries, especially the United Kingdom. However, institutional respondents did not appear to be aware of, or care about, comparisons with other OECD countries.

Therefore, although a number of positive effects of the quality policy were identified, there was never a consensus that any positive effect described was beneficial across the sector, and for different types of institutions and academics. Overall, institutional level respondents showed a greater tendency than CQAHE respondents to be concerned about the adverse effects of increasing steerage by Government and increasing inequalities both between and within universities.

Here the components of Australia's higher education quality policy, which were separated for analytic purposes, are brought back together as a continuous policy trajectory. Identifying the multiple arenas of action and highlighting the continuous two-way interlinkages between the different levels (macro, intermediate and micro) and contexts (influence, policy text production and practice) has enabled the non-linear dimensions of the policy process to be teased out, as shown in Figure 1 below.

Influences frame the policy process (shaded box) and they can be subdivided into those at macro, intermediate and micro levels. These influences fed into the policy text production at each level, as represented by the horizontal arrows moving from left to right within the shaded box. Policy text production at the three different levels is interconnected by using two-way arrows, and the size of the arrows indicates the relative strength of top-down and bottom-up policy text production. The localised context of individual institutions (micro level) influences the nature of effects at that site (as represented by the oblique arrow). The policy effects emerging from the bottom of the box are produced by the complex interactions of the influences and text production. Although only two arrows are shown, the effects are multiple, with some cycling back to become influence factors again at the three different levels of the trajectory. Other policy effects do not cycle back as influences. For example, the effect of decreased institutional autonomy reported by institutional respondents did not appear to be influential in the production of subsequent macro quality policy text, and it largely disappeared from the macro discourse by 1995. By contrast, the effect of decreased diversity across the sector continued to influence the macro policy text discourse, and in 1995 the HEC even acknowlegded that quality policy had reduced diversity (HEC, 1995).

Overall, Figure 1 suggests that there was a plurality of contexts and multiple trajectories occurring simultaneously throughout the quality policy process between 1993 and 1995. It is important to indicate that the arrows shown on Figure 1 are not the only ones possible but they represent the major flows in the policy process in this particular case. The intention of Figure 1 is to begin to depict the messiness of the policy process, but not to be so overwhelmed by the messiness that the policy process is rendered beyond systematic analysis. Therefore, although Figure 1 is a gross oversimplification, as models usually tend to be, it is presented to facilitate a visualisation of the 'bigger picture'.

The theoretical framework employed in the study revolved around the concept of a policy trajectory with the main components being the contexts of influence, policy text production and practice. The principal modifications were to add a global context; incorporate both state-centred constraint and micro-political agency; and highlight the continual interlinkages between the different levels and contexts. The contributions made by each of these modifications are briefly evaluated below, before concluding with a general reflection on the usefulness of the conceptual tools employed here for globalising, yet simultaneously localising, Australian higher education quality policy.

For the Australian higher education quality policy, the trajectory was extended 'upwards' from the level of the nation-state to a global context. This global context was essential in this particular example. It was cited by the minister, all CQAHE respondents and a majority of institutional respondents as a key initiating factor for the quality policy. A global perspective provided a basis for 'locating' the Australian policy by enabling comparisons and contrasts between the Australian approach and that taken elsewhere.

One theoretical conception of globalisation which sees it as simultaneously contributing to homogenisation on the one hand and localised differentiation on the other (Hall, Held & McGrew, 1992; Sharp, 1994/95) was borne out in this particular policy example. Whilst the Australian quality policy reflected a number of the elements of Van Vught and Westerheijden's (1994) 'global' model, the Australian approach could also be differentiated from the wider global trends. Further, it was not a fixed entity announced by the minister and then fully enacted, but it evolved through multiple levels and contexts during its life cycle, creating localised variations along the way. Thus, the localised plurality incorporated into some conceptualisations of globalisation was evident in this quality policy example, not only at the level of the Australian nation-state but all the way down to individual institutions at the micro level of Australian higher education.

The approach in this study was to bring together the two often juxtaposed perspectives of state-centred constraint (eg Dale, 1989) and micro-political agency (eg Ball, 1994) . For the Australian higher education quality policy, both were applicable, but there appeared to be a graduated effect from constraint to agency with movement from macro to micro levels of the trajectory. The closer the policy process came to the institutional level, the more academics appeared to be given "legitimate voices" (Ball, 1994, p. 15). Arguably this was designed to convey a sense of ownership, thereby increasing the chances that the quality policy would be successfully implemented. However, the ownership was more to do with institutions being involved in deciding which processes they would use to achieve the pre-determined policy goals.

Despite the formal inclusion of university representatives in the ongoing policy text (re)construction, it must be emphasised that the degree of manoeuvre for both CQAHE and institutions was strictly limited within the parameters set by the ministers involved. Thus, it is important to highlight that top-down policy text production predominated, suggesting a state-centred but not state control model because some reverse flow influences were clearly in evidence. However, the state-centred explanation was unsatisfactory alone as there were still sufficient spaces to allow for variations in policy practices and effects between and within universities. Therefore, more pluralist, policy cycle perspectives which emphasise micro-political agency were also required.

The balance between macro constraint and micro agency would be expected to vary with different policies, but in general terms there is a need to establish more formal bridges between state-centred and micro-political approaches to policy analysis. Policy networks may be one possible way to address the gap between different top-down and bottom-up theoretical approaches. A meso level concept of 'policy networks' which Raab has defined as "a generic label for different types of relationships between the state and interest groups in the policy process" (Raab, 1994, p. 13), and which highlights the interconnections between the state and other actors may represent a nascent form of such a bridge. Raab believes that this analytical approach requires further development both theoretically and empirically.

Capano (1996) has taken a similar approach in examining relations among all actors involved in the whole policy process. Locating the components of relational networks, he argued, is particularly important for understanding the impact of external changes on public policy and for comparative studies between countries. As with Raab (above), he calls for further empirical research on policy networks. Certainly the combined effect of the multiple trajectories for quality policy in Australian universities in the early 1990s (as depicted in Figure 1) looks like a network of interrelationships in which neither the state nor academics had absolute power. Thus, the findings of this study suggest that policy networks may be worthy of further investigation in terms of their usefulness as a tool for policy analysis.

This study indicates the importance of highlighting interlinkages between levels and contexts to reveal the complex and contested nature of education policy as a process rather than an end product (Ball, 1994), and especially to reveal the non-linear nature of the policy process. The synthesis of the multiple trajectories for the Australian higher education policy process of 1993-95 (Figure 1) was made possible by highlighting these interlinkages. The continuous two-way interlinkages between levels and contexts meant that the policy text was being actively reconstructed during the ongoing policy process. Some policy text reconstruction was due to effects cycling back as influences between earlier and later rounds of the quality reviews. Some other policy text reconstruction was due to direct negotiations between the minister and CQAHE even before the policy was operationalised. Still other policy text reconstruction was due to new influences which emerged during the three year period.

In general terms, most of the quality policy text reconstruction during 1991-95 worked to ameliorate early negative reactions from universities and helped to keep all 36 institutions in the voluntary program, and therefore under the control of the Government. If some universities had withdrawn from the program, as several had threatened, the power of the Government would have been considerably weakened. By highlighting interlinkages between levels and contexts, the mechanisms by which the Government increased its chances of achieving its desired goals came into focus. Pathways which connect micro level processes and the 'bigger picture' effect of increasing Government control, could be traced.

Overall, the theoretical framework of a policy trajectory through the contexts of influence, to policy text production, to practice (effects) along with all of the modifications employed in this study has provided a valuable tool for the analysis of Australia's higher education quality policy of the early 1990s from global to local levels. It has allowed a degree of separation of the policy process into discrete phases for closer examination, but then invites a synthesis of the multiple pathways involved in a continuous policy process. As the policy trajectory framework is not based on a simplistic assumption of policy formulation followed by implementation, it can readily accommodate both the negotiations over policy beyond the macro level and the creation of variations in policy effects at different sites. Its emphasis on context is crucial. As the three contexts of influences, text production and practice will occur in multiple arenas, the policy process at any particular site can be constructed as a unique configuration of these dimensions.

Although the specific example here was quality policy in Australian higher education of the early 1990s, I believe that the notion of a policy trajectory could be fruitfully applied in other policy analysis studies, in education and other domains. However, to 'coin' an expression from the quality literature, 'continuous improvement' is always possible, and the theoretical framework employed here could certainly stand further refinement before application in other policy contexts. The process continues.

1. The Dawkins' reforms have been described as creating a 'revolution' in Australian higher education. The main thrust was to create a 'Unified National System' which combined universities and colleges of advanced education. Key features were growth in the system on the human capital assumption of increasing the skills base of the economy; amalgamation of institutions on the assumption of economies of scale, enhanced accountability through the mechanism of an institutional profile which would be negotiated with the Government according to national priorities, management reforms, and changes to funding arrangements such as fees, privatisation and commercialisation.

2. The Department of Employment, Education and Training (DEET) became the Department of Employment, Education, Training and Youth Affairs (DEETYA) and later the Department of Education, Training and Youth Affairs (DETYA) over the 1990s as the Commonwealth bureaucracy underwent several restructurings.

3. Certain supranational organisations such as the Organisation for Economic Cooperation and Development (OECD) and the World Bank (both of which have principally economic roles) have been actively promulgating globalisation agendas, including the reform of higher education generally, and more specifically the need for quality policies.