Introduction

With modern electronic mapping devices, the relationship between a place's specific location in a district and its place in a country, or indeed the world, can be explored through a zooming process between the micro and the macro. In the study of the policy process relating micro relationship to the wider structure is not so easy, and there is often an impression of different kinds of studies talking past each other unable to connect. Schlager has defined the state of public policy studies as having ‘mountain islands of theoretical structure … occasionally attached together by foothills of shared methods and concepts, and empirical work … surrounded by oceans of descriptive work’ (Reference SchlagerSchlager, 1997, p. 14). But the issue about the relationships between studies is made even more complicated by the extent to which such theoretical development as has occurred tends to be very location specific. This is what makes comparative studies important, testing the scientific precept that a theoretical generalisation needs replication in a variety of contexts. So there are here two related issues: about the relationship between theories and about the universality of the assumptions made by those theories.

These remarks have been made about a variety of approaches to policy analysis, including notably Sabatier's advocacy coalition theory and Kingdon's multiple streams approach (Reference Kingdon1995). In each of these cases proposition developed in studies in the United States have been taken up elsewhere in the world, with, in each case, the identification of reservations about generalisation (see Sabatier's own defence on this in Sabatier & Weible, Reference Sabatier, Weible and Sabatier2007 and the exploration of the generalisation of Kingdon's work in Cairney & Jones, Reference Cairney and Jones2016). But this concern particularly applies to the subject of this article: street-level bureaucracy theory. On this many examples can be found of writings that simply indicate that street-level discretion is pervasive. A typical example of this is the following:

This prompts the questions, ‘but how’, and under what conditions ‘may’ that happen? There is a need to demonstrate how the mechanisms involved work, and to explore how they differ from context to context, hence explaining empirical diversity in policy practice. Comparison is needed to answer these questions and to go beyond approaches to the subject that simply focuses on very general assumptions about the personal characteristics and attitudes of street-level bureaucrats. The national context of street-level bureaucracy is of particular importance to explain how empirical diversity matter to the way street-level tasks are carried out in the encounter with citizens, because street-level organisations are, in effect a microcosm of the state. This is true in a formal sense, where policies regulate practice, but also, in an informal sense, where ideas about authority, civic identity and social roles are derived from a country's macro structure (Reference DouglasDouglas, 1986, pp. 58–59).

In the next section, the issues about developing cross-national comparative approaches to street-level bureaucracy will be explored and the limited amount of already published work that has contributed to this will be outlined. In the following section, it will be argued that good comparative work must rest upon precise specification of the contexts to be compared and it will be suggested that what is particularly important here is to avoid comparing tasks that seem similar, but do in fact serve different functions in different contexts. This will lead into a section in which one particular task – pre-school child care – is selected in these terms. The next section will then consider ways in the approach, developed through the discussion of the specific example of child care, may be used as a model for similar comparative work, and will be followed by a brief conclusion highlighting some practical issues to be considered.

Developing comparative approaches to street-level bureaucracy

Michael Lipsky, who may appropriately be described as the ‘founding father’ of research on street-level bureaucracy, identifies the subject with the observation that ‘ … the decisions of street-level bureaucrats, the routines they establish, and the devices they invent to cope with uncertainties and work pressures, effectively become the public policies [our emphasis] they carry out (Reference LipskyLipsky, 1980, xii)’. There are then two particular problems about applying that formulation. One is the simple comparative question: does that mean, always and everywhere? The other is about how far to take a statement that stresses ‘agency’ as opposed to structure? On both of these questions Lipsky is ambiguous, with the inevitable result that amongst those who have been influenced by his work there are varied emphases about how to develop street-level bureaucracy theory (see Reference Hill and HupeHill, 2003; Reference HupeHupe, 2019). As far as the first of those questions are concerned, in the introduction to his revised edition (Reference Hill, Long, Marzanati, Bönker, Wollmann and Marcou2010), Lipsky describes his approach as an ‘essentially comparative’ but clarifies this in a very generic way, writing of the need to identify ‘which features of people processing are common, and which are unique to the different occupation milieux in which they arise’ (Reference Hill, Hupe and Buffat2016). This implies many different dimensions for comparison. This article is concerned with one of these, cross-national comparisons of the impact of street-level bureaucrats on public policy and particularly the mediating effect of ‘the milieux’ of the street-level including the place and policy space claimed by government (Reference PollittPollitt, 2013, p. 50).

This particular theme is important for the comparative study of implementation, helping to put it in the framework of studies of the public policy process more generally. Sætren's reviews of implementation studies (Reference Sætren2005, Reference Sætren2014) show a growth of comparative studies so that now about half of all studies are comparative. But only about 20% of these studies are cross- national. He comments:

Lipsky picks up his other issue about the influence of structure in the tenth chapter of his original book with the observation that there is a reciprocity ‘between the larger society and the structure of bureaucratic institutions’ and these institutions are embedded in a larger system that creates and fortifies working conditions’ (Reference LipskyLipsky, 1980, p. 180). Hence, Lipsky also sees the mechanisms that shape frontline work and rule appliance towards the citizenry as the result of a contextual relationship between ‘the larger society’ and ‘bureaucratic institutions’, but he is less explicit about why this should be discussed in terms of variable conditions across street-level settings. Exploration of this so far given pays too little attention to what extent national culture matters in general to street-level work? Also, scholars seem to agree that structural factors such as the style of bureaucracy and the political economy of the welfare state can affect street-level bureaucrats’ ‘room of discretion’.

However, there is still a lack of more structured analyses of the interaction between macro- and micro-level sources of influences at the street-level to confirm or reject this in order to expand insights about national aspects of frontline mechanisms. The impact of structure on the street-level has never been tested systematically across countries. Since people live in places, not in policies, expectations about structural similarities may very well be proven wrong (Reference DuboisDubois, 2010, p. 37; Reference PollittPollitt, 2013, vii; Schneider & Ingram, Reference Schneider and Ingram1997, p. 2). Governments are actors privileged to claim, structure and institutionalise policy space on behalf of society. They govern through persons in places where street-level bureaucracy is of particular interest because it is there where citizens are particularly involved with the state. Street-level bureaucrats materialise policy as they make judgments about who should get what, when and how (Reference Schneider and IngramSchneider & Ingram, 1997, p. 2). They categorise people, problems and cases to comply with law of rule, policy intentions and citizen conditions and through these interactions mediate the social order of the state through public policy procedures and the everyday life of ordinary citizens. Street-level bureaucrats serve in a double role as state agents and as citizen agents. In a way, they have two ‘bodies’. A state body shaped of neutrality and bureaucracy and a personal body shaped by normativity and lived experience (Reference DuboisDubois, 2010, p. 73). However, citizens also influence the state by engaging with the street-level bureaucrats in various ways. They build identity and moral careers as they learn lessons about their worth as seen in the eyes of the state (Reference DuboisDubois, 2010, p. 31; Goffman Reference Goffman1961). These encounters happen in local, situational contexts. They represent the micro-dynamics of society's macro-order and serve as a window to study, not only how civic identity is shaped by the state but also how the state teaches citizens about their deserved place in society. Hence, dynamics of categorisation and identity construction are enacted in site-specific moments or in contexts’ of space and place. So far, these relations between governments’ design of the state, the world of street-level bureaucrats’ and the lived life of the citizenry have been studied as separate dynamics in society. Zacka argues that ‘[W]e must look at both the structural pressures to which bureaucrats are exposed, and at the everyday practices they deploy to modulate the effect their environment has on them’ (2017, p. 244).

This takes us to the issue of the separation of street-level bureaucracy theory from macro theory about the policy process. There is by now a well-established consensus within the welfare state and public administration literature that countries differ significantly in terms of welfare state regimes (Reference Esping-AndersenEsping-Andersen, 1990), market orientation (Reference Hall and SoskiceHall & Soskice, 2001) and national administrative cultures (Reference Pollitt and BouckaertPollitt & Bouckaert, 2004). Hence comparative theory is being developed that contributes towards an understanding of the larger political context of policy delivery work but it is still unclear whether and how such differences matter to the street-level. There is some work that tackles this issue. Stone's (Reference Stone1984) comparative street-level study of medical insurance policies in France, the U.S. and England identifies the core task of social insurance provided by a street-level bureaucrat and studies how political and economic characteristics of three countries shape both local regulation and professional reflection about social insurance. Knight and Trowler's (Reference Knight and Trowler2000) country-comparative study of department-level cultures and the improvement of learning and teaching uses in-depth interviews with academics in England and Canada to compare social mechanisms that facilitate change processes at the street-level of teaching. [Author]'s in-depth study of caseworker reasoning in Sweden and Denmark shows that frontline responses to managerial pressure are filtered through welfare state regime types (Møller, Reference Møller and Stensöta2017). Finally, Jewell's (Reference Jewell2007) study of social assistance in Sweden, Germany and the U.S.A. explores the impact of welfare state types on how the state handles social assistance and services for unemployed people at the street level.

However, other perspectives on public services (Reference PollittPollitt, 2013), encounters at the street level (Reference DuboisDubois, 2010) and policy design (Reference Schneider and IngramSchneider & Ingram, 1997) argue that, in general, the encounter between state and citizen is highly site specific. Therefore, research needs to be creative to find methods and analytical strategies that can illuminate both how state/citizen encounters are nested within local contexts and to what extent systematic patterns across national contexts can be identified.

A few scholars studying frontline organisation have paid substantial attention to the impact of physical and social settings in public offices as an independent driver framing the interaction style with citizens (see, e.g. Reference DuboisDubois, 2010). Social interaction theory points at the physical and social setting of an organisation as a ‘scene’, which scripts how people meet and interact (see Reference DuboisDubois, 2010, p. 42). This perspective claims that frontline organisations are certainly politically and bureaucratic scripted places bound to regulative rules, but they are also bound to normative rules that follows their professional standards as organisational and occupational experts of knowledge, as well as they are bound to cognitive rules shaped by more general social rules of interactions (Reference DouglasDouglas, 1986; Reference GoffmanGoffman, 1990; Reference ScottScott, 1995; Reference ZackaZacka, 2017). Hence, the frames of action in frontline organisations are seen as constituted by political visions and bureaucratic statecraft, and also by physical and social settings. Street-level staff are agents scripted by rules and at the same time made capable of action by them (see Reference Hill, Long, Marzanati, Bönker, Wollmann and MarcouHill, 2010). They are expected to select, reflect and act based on regulative, normative and cognitive rules echoing bureaucratic scripts, professionalism, as well as the available conditions for social interaction given by the place or the locality of the encounter.

The underlying question in such studies is to what extent aspects of national culture matter for street-level work? We suggest there will be differences at the level of the individual street-level bureaucrat, which will include their role conceptions, identities, beliefs, values and policy preferences. These will be particularly shaped by national culture as well as by universal laws about cross-pressured organisations as suggested by Brodkin (Reference Brodkin2007, p. 12, footnote 11). This implies a need to relate these to other conditions, at the system and organisation level and therefore potentially to differences between nation states.

The choice of key conditions is obviously also a choice of the key argument to be tested in a comparative study. Within the literature on street-level bureaucracy there is sometimes an implication that despite differences in settings and in the personal characteristics bureaucrats will (all other things being equal) display similar types of behaviour such as interaction styles towards their clients. This is challengeable in two respects: One is that personal characteristic differences can be identified in some systematic ways as sources of behaviour variation. The other is that different settings will have the measurable effects. Differences in locality/settings will include differences in tasks (which need to be held constant through the examination of a task that is very similar across contexts) and differences in the organisational arrangements established for public policy delivery. Furthermore, there is a particular challenge to explore whether it may be possible to relate the study of this topic to cross-national comparative theory, by exploring the relationship between setting and national policy delivery systems.

However, these two views are not simply alternative explanations of differences in behaviour at the street level. Both may apply, the two may interact inasmuch as people assume roles in specific contexts and conversely contexts may be influenced by culture. In that sense, cultural differences are more than simply system differences.

A national culture can be grasped as the reproduction of a sense of ‘peoplehood’ and as the normative macro-order of society. This sense works as a prerequisite for political community and for a legitimate relation between the state and its citizenry. Stories of peoplehood are continuously in the making both outside and within the political system (Reference EliasElias, 1994). Leaders construct stories of peoplehood (and through them group identity), as they try to mobilise followers around a shared identity (Reference JenkinsJenkins, 2008). As in smaller groups, group identity is also an important part of national culture as it produces social coherence and social trust between individuals and its communities. National culture takes its shape or meaning from social constructed boundaries that are always also at play in the interaction between the state and the citizenry either as explicit administrative guidelines or as social categories informing the street-level worker about how to interpret different groups of citizens (Møller, Reference Møller2016). Thus because groups tend to define themselves in terms of ‘the other’ who they are not, policy design, and particularly the identification of target populations (see Reference Schneider and IngramSchneider & Ingram, 1997), delivers not only access to benefits but gives also normative messages on worthiness from the state to the citizen in civic life and to the client seeking assistance through the street-level bureaucracy. Just as rules, administrative routines and social dimensions differ between countries, we should expect that so do ‘senses of peoplehood’.

Precise specification of context and task

At the core of Lipsky's formulation of the issues about street-level bureaucracy is the notion that there is a large group of public sector workers with much in common with each other. Where, as noted above, he concedes ‘occupational milieux’ may differ this implies a case for cross-occupational comparative studies. This, as noted above, is not the concern here. Rather for cross-national comparative studies it is important to try to explore the performance, in different national contexts, of a similar task. How do street-level bureaucrats, with different individual characteristics, working in different contexts, deal with a similar task? Do different national contexts lead to differences in outputs or outcomes related to that task?

The chosen task must meet the requirement that it is general enough to feature in all compared systems. To explore, for example, overall health policy or education policy overall is difficult, since such policies encompass mixtures of complex tasks performed by a variety of street-level bureaucrats. Even looking at a generic task like nursing or teaching is difficult, and even within those occupations specifically identified professional groups may be used differently in different societies. However, there are some risks with respect to the choice of task. It is important to avoid tasks that may be regarded as trivial or merely minor aspects of some more important task. There is a need to avoid the danger of what Bourdieu refers to as ‘confusing nominal similarities with functional similarities from one social space to another social space:

In other words, there is a need for care when interpreting presumed similarities and differences. In the discussion to follow, questions are raised about whether ‘parenting’ and ‘child care education’ in two social spaces (such as in two countries) could serve different functions in society and hence only present a nominal, but not a functionally similar task. The next section thus explores a task which seems – from an inspection of comparative literature – seems to be susceptible to the kind of analysis proposed here. The aim is to outline an approach and explore the difficulties in adopting it. But it must be read in terms of the contribution it makes to a grounded account of what a comparative study of street-level bureaucracy is likely to involve.

Finally it should be observed that – inasmuch as the purpose of comparison is to advance the elaboration of a theoretical perspective – everything that is said here can be seen as relevant to within country comparisons.

It is to try to contribute to meeting the implicit challenge in that last section that the focus here is upon cross-national comparison, but this may also be justified in terms of the goal of linking with the kinds of macro-theories discussed above. However, we follow Pollitt (Reference Pollitt2013) and Dubois (Reference Dubois2010) advice of thinking about public policy provision as the materialisation of policy space and not implementation of policy content, as this approach emphasises the simple fact that people and street-level bureaucrats live and work in places not in polices.

Developing the approach using a specific example

This discussion provides an example of how to select a ‘task’ suitable to be compared across countries. The case is pre-school care and education (defined in terms of organised public provisions for children in the period between the age of four and school entry) since it is provision for this group that has become apparently very similar across western European nations. Bonoli has given the topic particular attention in his The Origins of Active Social Policy: Labour Market and Childcare Policies in a Comparative Perspective (Reference Bonoli2013). While he acknowledges the education component in this activity his emphasis is on child care provision as a facet of employment policy, linking government efforts to stimulate labour market participation (particularly by women). His analysis gives attention to data on the quantity of care provision rather than its quality. But issues about education and child development have also been emphasised in the official discourse about child care policy. In a sense then it may be that it is left to street-level actors – as either providers or regulators – to achieve the accommodation between education and care.

In order to interpret whether ‘pre-school care and education’ is indeed similar across countries, we undertook an examination of published – largely official (Eurofound, 2013; European Commission, 2014; OECD, 2017) evidence to explore similarities and differences across a limited number of western European nations all of which have developed extensive pre-school public care and education policies for children aged 4 or above (Belgium, Denmark, Germany, France, Italy, the Netherlands, Sweden and the United Kingdom). All the countries come close to 100% early childhood education (ECEC) participation by the age of 4, which we argue is a first indication that ‘pre-school care and education’ is not only nominal but also a functional equivalent across these countries, as children that age are placed in the realm of the state during a significant amount of their day. There are much greater variations in provision before that age, hence this discussion (as far as possible) leaves that out of consideration.

So this seems to provide a basis for a study which can explore the activities of the street-level workers, assuming that they are performing broadly the same task. The two slightly deviant countries are Italy and the U.K. As far as Italy is concerned, Bonoli (Ibid) draws a contrast between a tradition of extended family care on the one hand and a pioneering role in pre-school education, so that once reaching the years immediately before primary school the situations for Italian children are not so different to that in other countries. In the U.K., case change has been comparatively recent and involved a variety of subsidies to private providers. In that sense development may have been particularly driven by labour market participation concerns.

There is considerable variation in the settings within which child care provision occurs across the selected countries. Complexities arise, for example, in Italy where about three quarters is state provided (Reference PennPenn, 2014, p. 437), in Germany where there are variations in settings according to Lander and in the U.K. where there is considerable (and growing) public support but provision is in a mix of agencies many of which are voluntary or profit-making private organisations.

Provision is free in Belgium, France, Italy and Sweden, except for charges for meals and for any extra hours. Fees are charged in Denmark and they vary from place to place. In the Netherlands, there is no parental fee for pre-primary programmes in schools available on a voluntary basis but local authorities may charge for other provisions. There are provisions for funding support for children from disadvantaged backgrounds. In the U.K., recent legislation provides for free entitlement of up to 30 h per week, but only for children with working parents. In Germany, there are fees in most Lander, but special exceptions in some (particularly for children in the last year before primary school).

The European Commission report (2014) distinguishes staff in terms of three categories: ‘education staff’, ‘care staff’ and ‘auxiliaries’. Often, within the same country, a number of different types of professionals are involved with the everyday care and education of children. In addition, staff performing similar roles may also have different types of job titles. A particular complication is the importance of ‘pedagogues’ in the Danish system. In a recent published article, the Danish pre-school childcare is shown to be highly professionalised, but based on developing a particular balanced social competence, which is different from education in its normal meaning. Furthermore, parents are expected to take part in their children's education as competent and balanced performers within ‘the social’ context.

This concept of ‘pedagogy’ may indicate that these professionals have a different function in Denmark compared to Germany or England. However, if the focus is on the interactions between professionals and children in a context of public child care, the ‘task’ may very well be similar to the interactions in Germany and England. Exploration of this theme indicates that what is involved here are professionals with particular skills to assist with the socialisation and development of children rather than the more didactic expertise of educationalists.

With the exception of Germany, bachelor level education is required of teachers and pedagogues. The countries are more split on the issue of expectation of training for care staff, with just Denmark, Sweden and the U.K. having explicit educational expectations. With respect to both of these items evidence on ‘expectations’ leaves questions both about educational content and, particularly, about the extent to which those expectations are realised in practice. The data on ‘heads’ seems to imply some curious contradictions (with explicit requirements only in Belgium, France and Italy) these are perhaps reflections on the extent to which (particularly when the providers are often private or voluntary organisations as in the U.K. and Germany) such requirements are unenforceable.

According to the Eurostat report rules about maximum number of children per staff member were absent in most countries, exceptions to this being England where 13 and Germany where 14 were specified as the maximum for 4 year olds. However, OECD (2017, p. 111) reports child to teacher ratios, ranging from 22 for France, through 16 for the U.K. and the Netherlands, 15 for Belgium, 14 for Italy, 10 for Germany, 9 for Denmark and 5 for Sweden. While bearing in mind, the difficulties of definition observed above these are quite striking differences.

In all the countries considered, there are steering documents coming from the national or regional level but the European Commission's report stresses (2014, pp. 122–3 emphasis in the original).

Most countries have no centrally recommended support materials and allow institutions to select or create their own materials to meet the needs of children and the activities planned. The wording there indicates that there may be differences in the extent to which more formal teaching is expected to occur. Structured timetables are recommended only in Francophone Belgium. The only specific support material recommended in the countries discussed here are materials designed for language assessment tests recommended in Denmark. Health and safety standards are specified everywhere.

Bonoli's analysis (which provided the starting point for this discussion) shows the eight nations have a great deal in common. All are now offering a considerable amount of provision, free or extensively subsidised, for children close to the age at which they progress to primary education. But, as already highlighted in relation to the Danish case, there is evidence to expect that the care/education balance is struck rather differently in different systems. The regulations about staffing suggest a strong emphasis on education across the board. Where, however, the Eurostat data is lacking is in the provision of evidence about the extent to which official goals are realised in practice. This calls for deeper within-case reviewing of, e.g. ethnographic research such as the analysis by Dannesboe et al., Reference Dannesboe, Bach, Kjaer and Palludan2018 referred to above, that goes beyond statistical reports in order to interpret such similarities and differences across national contexts.

Developing a logic of critical explanation and functional comparison

Since the aim here is not to make a case for a specific study of pre-school care and education but rather to establish an effective approach for research, it is not appropriate to go into the more detailed questions about how that topic might be more fully developed or the difficulties it would encounter except inasmuch as they help to explicate some of the key questions about using such an approach.

Success with such a study depends very much upon being able to delineate a task that is broadly similar across research sites (countries) but where there are differences of perspective. One difficult issue here is about the extent to which apparently similar systems can be seen to be on very different pathways. This makes the contextual exploration of the kind explored here in relation to child care important. For example, variations, in publicly funded and regulated health-care systems in respect of private sector roles and the use of insurance (see Böhm, Schmid, Götze, Landwehr, & Rothgang, Reference Böhm, Schmid, Götze, Landwehr and Rothgang2013) may have such strong effects upon the behaviour as to render behavioural choices by street-level actors difficult to compare. A related issue is the existence of different system differences in the same country, not by any means an argument against comparative study but a challenge for cross-national ones. Any study will need to tackle a sampling issue – is a number of separate sites needed and if so how should they be chosen? In the pre-school care and education case additional considerations would be how to deal with variations between Lander in Germany or the variation in the mix of private and public provision (particularly in the U.K.) would pose problems.

These considerations, often in practice subject to pragmatic solutions based upon resources availability (for example, Jewell's (Reference Jewell2007) selection of single cities in the hope that they are a typical of national systems). But that leads us to another question, inasmuch as hypotheses about the cultural and institutional influences upon street-level behaviour are relevant can they be used to give shape to comparative studies? The discussion of the case of pre-school care and education indicates a lack of applicability of regime theory. The particular focus of street-level theories within institutional arrangements where central–local relationships are likely to be important points to have regard to administrative typologies (Reference Kuhlmann and WollmannKuhlmann & Wollmann, 2014; Reference Painter and PetersPainter & Peters, 2010; Reference Pollitt and BouckaertPollitt & Bouckaert, 2004). But the problem with using these is that it is important to see the structuring of street-level bureaucrat activity in terms of ‘action scales’ (Reference HupeHupe, 2013) or ‘layers’ (Reference Hill and HupeHill, 2003) where managerial influences that may be quite locally determined may be of key importance. While context specification or comparison along the lines outlined above may be helpful there has in the end to be a certain arbitrary character to the structuring sources that are considered and those that are left out. The child care and education example suggests a dilemma about nations (from the list discussed) that might be left out of a study. Prima facie there seems a case for arguing that the peculiarities of the systems in the U.K., Germany and Italy might point to leaving them out. Against that of course, it may be argued that those very peculiarities are important to highlight, for example, the impact of privatisation or federalism. Here again we see the importance of background study and of clear problem specification. Is it more instructive to explore whether a study is confined to what are expected to be very similar systems, or is it good to use expected difference to provide the focus for analysis?

But this last question brings us to a very fundamental issue. This article is based upon a view that comparative work is needed to develop the study of street-level bureaucracy and embed it better in comparative studies. But where can advances be most easily made? It seems to us that any temptation to assemble multi-nation data should be resisted. Instead we suggest to do a ‘background study’ such as the sample of nations explored here in order to select more carefully, what cases/countries to ‘zoom in on’ in a systematic qualitative research design. Here, we suggest confining attention to very detailed analyses of a sample of national systems where intense attention can be given to the spatial context and to institutional and historical analysis. We are not talking here of abandoning the kind of quantitative analysis used above, but of ensuring that it is backed up by qualitative contextual material.

Such work, even if actual decisions cannot be explored or measured, may involve interaction styles, consisting of both a ‘mental’ aspect and a ‘physical’ aspect. The mental aspect includes the way the street-level bureaucrat conceptualises, understands and perceives the case/problem at hand, and the importance of this for his/her inclination to act. The physical aspect refers to the physical setting of the interaction and from this to what the street-level bureaucrat actually does or refrains from doing (Reference GoffmanGoffman, 1990, p. 32). We argue that since these signs and meanings which frame the place of social interactions, encourage certain ways of performing as well as they suppress others, they are important factors to include in street-level theory on the provision of public policy service delivery. However, we suggest to reserve these micro-relational observations to in-depth analyses of contexts selected based on a background study as the one we put forward here, as they require solid analytical attention and material.

There are important issues here, only to be solved by careful examination of contexts, in which prior consultation of ‘experts’ within the countries to be studied will be as important as the available statistical data in the background study. In the discussion of child care and education, it was suggested that questions to street-level bureaucrats might be structured so that place-specific orientations in respect of care, socialisation and education might be explored. Similar distinctions are likely to be possible with other street-level tasks – the extent to which opportunities to use discretion are taken up rather than simple rule conformity, or the extent to which punitive or helping roles are assumed. Beyond this lies conceptions of what it means to act as a liaison between the state and the public. As suggested above, such differences may be identified through observing, interviewing and surveying in particular selected sites where the material should be classified in systematic (qualitative or quantitative) ways, e.g. by using network analyses and methods of organising, coding and scaling important components such as ‘interaction styles’. This will enable the researcher to identity empirically significant patterns within and across places, which again is the fundamental prerequisite for critical explanation and for elaborating on the theoretical understanding of street-level bureaucracy as a particular site for state–citizen encounters. But there are substantial difficulties about studying these, not least limitations in the scope for observation and measurement. In relation to the study of child care arrangements both ethical and practical concerns are relevant here. In practice, much research on street-level bureaucrats focusses on roles and role conceptions, accepting the validity of the ways people explain their activities (see, e.g. Møller, Reference Møller and Stensöta2017; Reference Møller and Stone2012), whereas few studies include citizen perspectives or administrative data as a way to validate such insights on public policy delivery (see Reference Soss, Fording and SchramSoss, Fording, & Schram, 2011 for an exception).

Conclusion

The aim of this article has been to link practical and theoretical concerns to advocate a way towards additions to the so far very small number of actual studies. The case for such studies rest primarily on the aspiration to consolidate the body of knowledge about street-level bureaucracy and locate it better within the broader corpus of public policy studies. But it has a certain practical validity inasmuch as policy transfer is widespread (Reference Dolowicz and MarshDolowicz & Marsh, 1996; Reference Holzinger and KnillHolzinger & Knill, 2005). Allegedly successful policies are copied, with much consultative work advocating ideas from other countries (often adjacent ones or speaking the same language). Transfer failure is often a consequence of insufficiently detailed attention to exactly how a policy works at the street level, something that comparative study can highlight.

Bearing in mind how simplistic policy innovations are carried from country to country the particular emphasis in this article has been upon on (a) the need for careful efforts to delineate comparable tasks, (b) the importance of the recognition that assembling contexts for comparison is difficult and (c) the case for very careful specification of the characteristics of those contexts emphasising the dimensions of policy space and local place. More specifically, within the article the following have been identified as issues needing attention:

• The chosen task must meet the requirement that it is general enough to feature in all compared systems.

• But it is important to avoid tasks that may be regarded as trivial or merely minor aspects of some more important task.

• There is a need to consider the extent to which ‘surface similarities’ may confuse, and while this need not deter comparison its adds complexities needing attention.

• Preliminary work must involve an examination of existing data, and the consultation of experts in the substantive field.

• Macro classification models need to be used with caution inasmuch as they may point towards over-simplifications

• The number of countries including in any comparison will need to be carefully selected, and even within those countries there may be sampling problems because of within country differences. It may be added here that of course the issues identified may apply as much to within country comparisons as to those between countries explored here.

These practical considerations need to be reinforced by recognition of the considerable difficulties about establishing and funding a network of scholars to carry out a study that requires very precise examination of contexts and behaviour. It is not surprising that little work, of the kind discussed here, has been carried out. And inasmuch as the comments about selection of tasks and about methods of comparison may be seen to be a little lacking in sophistication the purpose here has been to explore how to open up a neglected area of study, largely using inductive methods, rather than to specify a very specific programme of hypothesis testing.

The complexity of the subject together with the methodological and practical problems of assembling micro-level comparative research of macro-systemic influences on street-level work can be overcome by technology and advanced systematic qualitative methodology. That may prevent otherwise leaving the field of development of street-level bureaucracy theory to those who focus on de-contextualised local studies, giving attention to psychological differences amongst street-level bureaucrats. Not only the advancement of micro-studies in comparative administration but also the development of the full potential of street-level bureaucracy theory needs scholars to try to ‘grasp the nettles’ identified in this article and set up research enterprises along the lines we have described.