Comparing sub-national policy workers in Canada and the Czech Republic: Who are they, what they do, and why it matters?^*

* The work of Arnoš Veselý was supported by the Czech Science Foundation Grant (P404/12/0725). The work of Adam Wellstead was in part supported by the Social Sciences and Humanities Research Council of Canada. The authors would like to thank two anonymous reviewers for their valuable comments and suggestions.

Abstract

This article compares profiles and policy-related activities of policy workers (PWs) in thirteen Canadian provinces and territories with PWs in the Czech Republic regions. Canadian data come from 13 separate surveys conducted in provinces and territories in late 2008 and early 2009 (N = 1357). The Czech data are from analogical large-scale survey carried out at the end of 2012 (N = 783). First, the paper compares basic characteristics of Canadian and Czech PWs. In the two countries the proportion of men and women is similar and PWs are equally highly educated. Examining other characteristics, however, reveals substantial differences. When compared with the Czech PWs, Canadian PWs tend to be older, more often having social science educational backgrounds, more frequently recruited from academia, stay in a single organization for a shorter period of time and anticipate staying in their current position for only a short time. Second, a comparison of policy-related work activities discerns three basic clusters of policy tasks: policy analysis work, evidence-based work, and consulting/briefing. Canadian PWs are much more involved in evidence-based work, especially in evaluation and policy research. They also deal more with policy analysis activities such as identification of policy issues and options. In contrast, Czech PWs are more engaged in consulting with the public and briefing managers and decision-makers. The article concludes with implications for further research and theory building.

1 Introduction

Despite long-standing scholarly interest in the characteristics and work of “policy workers”, actual empirical evidence was, until recently, very rare.¹ This changed during the mid-2000s with qualitative case studies in the United Kingdom and the Netherlands (Colebatch, 2006; Page & Jenkins, 2005). A decade later a series of large-scale quantitative studies of Canadian federal and provincial policy analysts were undertaken (Howlett & Newman, 2010; Howlett & Wellstead, 2011; Wellstead & Stedman, 2010; Wellstead, Stedman & Lindquist, 2009).

Despite the interesting findings, the Canadian work was conducted without an international comparative component. This substantially limits the potential of the findings in terms of theory building. A number of articles on policy work in Canada have been published during the last decade (see Kohoutek, Nekola & Novotný, 2013 for review). Given the absence of data from other jurisdictions, the findings from Canada have often been the only source for theory development on the nature of policy work worldwide, and have been implicitly and unreflectively generalized. However, it is not clear to what extent the Canadian results are idiosyncratic, and to what extent they reflect broader global trends (such as a move to new public governance, described in the introduction to this special issue), or structures in which they are embedded. For instance, Howlett (2009a) found that Canadian provincial policy analysts tend to be relatively young, and primarily engaged in fire-fighting activities on a day-to-day basis. The question is how to interpret such findings. What portion of fire-fighting activities at the subnational level is “normal” or “necessary”? Even characteristics such as age or years of experience alone do not tell us much because there is no absolute baseline to which they can be related. For instance, what does it mean to say “policy analysts are young”? While for some, it might be twenty years old, for others thirty and still for others even forty. The same holds for any other characteristics of policy work — there is no clear norm as to how much time should be spent on consulting with the public, etc.

The results for any country can only be meaningfully compared if there is a baseline establishing what is most typical, or what the average is. Comparison is also of crucial importance for theory building. We need to know whether the patterns of policy work found in Canada are also observed in others. In other words, does the current policy work have some inherent features that are generalizable across contexts? For instance, on the basis of Canadian data it has been suggested that older technical forms of analysis are being replaced by newer more participatory ones (Wellstead, Stedman, & Howlett, 2011, p. 355). Is this a general pattern to be found in all developed countries, or is it specific feature of Canadian policy work?

The only way to overcome the problems mentioned above is to compare the Canadian findings with results from across jurisdictions, and to see the (dis)similarities. So far the Canadian findings have been practically unchallenged by different sources of evidence. The present article contributes to filling this gap in the literature as we, for the first time, compare results from Canadian provinces and territories — the dominant source of evidence in current literature on policy work — with findings from a similar large-scale empirical survey in the Czech Republic's regions. While these two countries are geographically rather distant, such comparison is theoretically justified for several reasons.

Although not a federal country, the thirteen Czech regions (kraje) resemble the ten Canadian provinces and three territories in many respects.² Although geographically smaller and more densely populated, the Czech regions are autonomously governed and have a wide range of competencies and responsibilities, similar to those of Canadian provinces and territories, especially in social and health services, education, infrastructure development, environment protection, etc. Their administration is concentrated in regional authorities (offices) which are analogical to provincial and territorial ministries or departments. They are governed by a Board of Councilors (rada, analogical to government) and a Regional Council (zastupitelstvo, analogical to Legislative Assembly), the latter elected in fully democratic regional elections. In contrast to Canadian provinces and territories, Czech regions were only established in 2001 as a result of extensive decentralization (Baun & Marek, 2006). Regional authorities are thus rather new institutions. In general, however, Czech regional authorities play in the Czech Republic governance system very much the role provinces and territories play in Canada. It thus can be reasonably assumed that the nature of subnational work in regional officials and provincial and territories should be similar. On the other hand, as explained below, Canada and the Czech Republic differ substantially in other respects, most importantly in the fact that while policy analysis in Canada has long been institutionalized for a long time, it has started to be established in the Czech Republic only very recently. When comparing these two countries, it is thus possible to discern which macro factor seems to matter more: position in a governance system or institutionalization of policy analysis (we consider some other factors, too).

There are two basic aims of the paper: (1) comparison of similarities and dissimilarities in the profile of Canadian and Czech provincial/regional PWs, (2) analysis of policy work activities in the two countries with a view to identifying differences and similarities. Because the comparison leads to serious theoretical and methodological questions, a supplementary and secondary aim of the paper is a discussion of possibilities and limits of such cross-national comparisons. The paper is structured as follows. First, we briefly review the current state of the art and formulate theoretical background and research questions. Second, we describe the data used for the comparison and assess the methodological strengths and limits of our approach. Third, we compare the basic profile of policy workers in the two countries. Fourth, we analyze and compare policy tasks undertaken by policy workers, and describe patterns for these tasks. We conclude with preliminary findings and suggestions for future research.

2 Theoretical background and research questions

In this paper we ask two basic research questions: (1) To what extent the profile of Canadian subnational policy workers corresponds with the profile of their Czech counterparts? (2) How the proportion of particular policy work tasks found in Canada differ or resemble the proportion in the Czech Republic? These two questions relate to one more general and theoretical question: is current subnational policy work in different jurisdictions more or less similar, or are there substantial differences? As noted above, given the lack of other evidence, the Canadian data are sometimes implicitly taken as a source for general claims on policy work. A competing hypothesis, however, is that subnational policy work in different countries differ profoundly depending upon institutional structures, state-society relationships, civil service codes, etc. In other words, other authors (e.g. Howlett & Newman, 2010) assume that in different countries and jurisdictions quite different policy advice systems exist.

In this respect, comparison of Canada and the Czech Republic can be quite interesting. As noted above, while provinces and territories play similar role in governance structure as regional authorities in the Czech Republic, the history, culture and many other factors in these two countries are quite different. The Czech Republic public administration, as other Central European countries, has gone through a turbulent era after 1989. The result of unceasing, unsystematic reforms combining parts of different traditions is that ‘a patchy agency landscape has gradually evolved in most post-transition EU-10 countries, characterized by a broad and poorly regulated plethora of various, often idiosyncratic, organizational forms and a variety of different (and frequently changing) modes of autonomy and control’ (Randma-Liiv, Nakrošis, & Gyorgy, 2011, p. 164). As a result, public administration in the Czech Republic is assumed to have relatively low capacity (van Thiel, 2011) and be less stable than in the most developed countries. Moreover, traces of the communist past are still recognizable, especially in everyday public administration practices, especially the hierarchical style of communist governance (Kabele & Hájek, 2008). On the other hand, the Czech Republic shares with Canada the introduction of new public management ideas in 1990s (Veselý, 2013d). Despite the extensive scholarship on public administration in CEE countries, empirical evidence is still very much limited, and just a few large scale empirical surveys have been conducted in the region.³ Even these exceptions, however, provide very little information on what people in public administration actually do and what are their basic characteristics. The aim of this paper is to fill this gap in knowledge by presenting hitherto unpublished data from two large-N empirical surveys that were recently carried out in the Czech Republic.

There are two general and competing hypotheses on policy work: one claiming that policy work in analogical entities is likely to be similar, while the other suggesting that macro-structural societal factors strongly influence the nature of the policy work, and because they are different, the policy work is also likely to differ across jurisdictions. In empirical research, however, we need to specify the general ideas outlined above into more concrete hypotheses. As for the first question, there is no clear theoretical background that would help us to formulate hypotheses concerning an expected profile of policy workers in Canada and the Czech Republic. As for gender, there is no clear factor that would lead us to assume that the proportion of male and female PWs in the two countries should be different. As for age, we could assume that average age might be influenced by the history of the organizations the PWs work in. Because regional offices in the Czech Republic were established relatively recently (followed by the creation of new job positions), we assume that the Czech PWs are on average younger. We expect that in both countries PWs are highly educated. Because policy work is knowledge-intensive we assume that a high proportion of PWs come from academia in both countries. As for educational background, we assume that such approximates areas for which the offices are mostly responsible (health, environment, education, etc.).

As for the second research question, two contradictory hypotheses can be raised. First, because the structural organization in the governance systems of Canadian and Czech offices is similar, the policy work of subnational PWs should be similar. It can also be argued that both countries have been influenced by the same macro-structural changes during the last two decades, such as new public management and new public governance (see introductory chapter of the issue). Consequently, government PWs are now expected to “engage in greater consultation, consensus building and public dialogue” as a core part of their professional role (Wellstead et al., 2009, p. 37). Accordingly, it can be assumed that consultation and negotiation has evolved as a separate activity in both countries, and that it is also prevalent to a similar degree.

The opposite hypothesis states that the nature and structure of policy work in the two countries will differ substantially because of the completely different contexts and traditions. While in Canada policy analysis is now a traditional and clearly recognized profession, in the Czech Republic it has never been fully institutionalized. Indeed there is no equivalent for ‘policy analysis’ in the Czech language, and it is taught in just a very few places in the country. As a consequence, few people have training in many activities studied under the rubric of “policy analysis” (such as identification of policy problems, evaluation, etc.). There are also other contextual factors that might influence the nature of policy work, such as organizational practices, culture or labor force structure.

The question of the allocated proportion of various policy-related tasks is closely related to typologies of policy workers, which are widely discussed in the literature (Dluhy, 1981; Durning & Osuna, 1994; Hoppe & Jeliazkova, 2006; Mayer, Bots, & van Daalen, 2004; Radin, 2000; Thissen & Twaalfhoven, 2001). Despite some obvious overlap with these typologies, empirically created clusters do not fully correspond with any of them. Wellstead et al. (2009) discovered two major policy related roles. They labeled the first one as “policy work” (including collecting policy-related data, identification of policy issues and options) and the second “networking” (including negotiation and consultation). Howlett (2011, p. 619) found four distinct types: policy appraisal (including collecting data and conducting policy related research), implementation (including delivering policies and programs, consulting and negotiation), strategic brokerage (including reports preparation and consultation with decision makers) and evaluation (evaluation of policy processes and results). So far all empirically based typologies of PWs were always restricted to one particular country. Our data, for the first time, enable us to analyze cross-national policy works patterns.

3 Methodology and data

We compare two sets of data. The Canadian data was derived from 13 separate provincial and territorial surveys conducted in late 2008 and early 2009 (N = 1357; response rate = 43.3%). For more details on the Canadian methodology see Howlett (2009b). The Czech data came from a survey of public servants in regional authorities that was conducted in late 2012 (N = 783; response rate = 32.5%).⁴ The Czech survey drew on the Canadian experience, and in cooperation with Canadian researchers. In both countries, the questionnaire was piloted to ensure that questions are clear to the respondents. In both countries a self-administered online questionnaire was used. Howlett and Wellstead administered the Canadian survey and the Czech survey was administered by renowned public opinion research agency Ppm factum research.

In both countries a strong effort was made to comply with standards of research in public administration (Lee, Benoit-Bryan, & Johnson, 2012), and as a result we claim to obtain highly reliable and robust data. It is fair, however, to explicitly mention several reasons for caution regarding the methodology with regard to comparison. First, in both countries, data were gathered through online surveys, which may lead to a nonresponse bias (Lee et al., 2012). Nevertheless, in both countries the surveys were realized as censuses (all members of the target group were invited to take part). The overall response rate was slightly higher in Canada. Nevertheless, there is no reason to believe that nonresponse distribution should systematically differ for these two countries (e.g. that in any either country females are more likely not to take a part in a survey than in the other country).

The second reason for caution concerns sampling procedure. While in Canada positive criteria for inclusion were developed (i.e. people in positions, such as policy analyst, researcher, etc. were searched for and then included), in the Czech Republic negative criteria for inclusion were formulated (i.e. the list of all public officials working for the regional offices was created and then from this full list those fulfilling criteria for non-inclusion were excluded).⁵ The reason for different population identification procedures is obvious and was necessary. In the Czech Republic no such position as “policy analyst” (or similar) exist and it was completely unclear who a priori counts as a “policy worker”. While it might be argued that the sample frame in the two countries differed, in practice it does not have to lead to severe distortions in the final populations. As reflected also in response rates (which was lower in the Czech Republic), in both countries there were a relatively high proportion of those respondents who opened the online questionnaire and after several questions dropped off. We might assume that those were persons who realized that the survey questions were not addressed pertinent to their actual roles (e.g. do not carry out any of the suggested tasks or policy methods). The pros and cons of the methodological procedures and their comparability are discussed in length elsewhere (Veselý, 2013b).

The third reason for caution concerns the research instrument. The Czech questionnaire was developed on the basis of a 64-item questionnaire originally developed and used by Adam Wellstead, Richard Stedman, and Evert Lindquist in 2007 survey of federal government PWs (see Wellstead et al., 2009) and applied at the sub national level by Howlett (2009a) However, the original Canadian questionnaire was not primarily designed to be used in an international context. Differences in tradition, discourse and overall context led the Czech researchers to modify the original Canadian questionnaire in some respects. For instance, classification of education in Canada was based upon national customs and not on international standards such as ISCED. There were also severe translation problems. While terms such as “policy” and “policy work” are common in Anglo-Saxon practice, there is no direct equivalent of these terms in the Czech language. Thus sometimes it was necessary to change the wording of questions. For instance, while one of the Canadian questions is on involvement in “collecting policy relevant data or information”, the Czech questionnaire simply asked about “collection of data and information.” Explaining the meaning of “policy relevance” during the course of the survey would be time consuming and may not have necessarily guaranteed that all the respondents understood its meaning. Similarly, what is and what is not policy related research is blurred in the Czech cultural context. Indeed, an important methodological lesson learned was the difficulty in developing, preparing and implementing a survey in the Czech Republic based on an instrument originating from a different jurisdiction. Despite this challenge, we were able to develop a survey reflecting the Czech policy work and undertake comparative analysis with the results from the Canadian study.

In spite of these methodological limitations, however, the data analyzed in the paper are quite robust, and constitute the best available evidence we can have at the moment. For the comparison in this article, we have included only questions that were identical in both countries in terms of scaling, and that were either identical in terms of wording, or only slightly modified.

4 Empirical findings

4.1 Comparison of policy workers’ profile

We will first compare the basic profile of PWs in the two countries. A 2008–2009 survey of Canadian PWs found that provincial policy analysts tend to be relatively young, well-educated, from a social science background, mobile, relatively inexperienced and untrained in formal policy analysis or analytical techniques (Howlett, 2009a, 2009b, 2009c). Table 1 provides comparison of the PWs’ basic characteristics. It reveals that the only feature in which both countries are more or less similar is the gender proportion. Women prevail in both countries, and a bit more so in the Czech Republic (however, the difference is not statistically significant). The Czech PWs are in general younger. As formulated in our hypothesis stated above that may be a result of the fact that the regional offices in the Czech Republic were established relatively recently, which provided openings to new positions.⁶ However, though the overall difference in age is statistically significant, it is not extreme. The difference is caused mainly by the youngest generation (which is overrepresented in the Czech Republic), and more senior age categories (which is more prevalent in Canada).

Table 1 Basic profile of subnational policy workers in Canada and Czech Republic.

		Canada	Czech Republic
Gender	Female	58.4	61.7
	Male	41.6	38.3
Age	30 or younger	15.9	20.9
	31–40	27.1	30.5
	41–50	27.6	23.2
	51–60	24.1	22.7
	Over 60	5.3	2.6
Employed in present organization	Less than 1 year	14.4	7.3
	1–5 years	43.7	38.1
	6–10 years/6–9 years	15.4	31.5
	10–14 years/More than 10 years	8.0	23.1
	15–20 years	8.5	n.a.
	More than 20 years	10.1	n. a.
Educational Attainment	ISCED 2 (Basic education)	n. a.	.1
	High School/ISCED 3A (Upper Secondary Education)	2.2	11.5
	College or technical/ISCED 5B (Tertiary vocational schools)	5	1.9
	University/ISCED 5A — Bachelor	32.2	10.1
	Graduate or Professional/ISCED 5A — Master	60.7	70.2
	ISCED 6 (PhD.)	n.a.	6.1
Previous work experience ^a	Academia	22.0	4.7
	Federal Government/Ministry	12.9	4.3
	Other organization of central state administration than Ministry	n.a.	7.4
	Municipal government department or agency/Regional (or municipal) authority	11.0	22.3
	Other department or agency in your current province/Other organization of regional administration	44.2	6.3
	Non-state non-profit sector	26.1	5.7
	Private sector	16.7	22.9
	Department or agency in another country/Foreign (or supra-national) organization of public administration	5.4	0.3
	No previous experience	n.a.	12.5
Anticipation of being in present department/in current organization	Less than 1 year	12.1	2.4
	1–5 years/1–2 years	54.9	12.7
	3–5 years	n.a.	27.2
	6–9 years/6–10 years	14.5	19.0
	10–14 years/More than 10 years	10.7	38.7
	15–20 years	4.5	n.a.
	More than 20 years	3.3	n.a.

a The wording in the Czech questionnaire was more restrictive because it focused on formal working groups and committees.

Though, as explained above, a strictly exact comparison of educational attainment is not possible, PWs in both countries are highly educated. This suggests that working for subnational authorities is a rather attractive career option for highly educated people in both countries. However, the Czech Republic's PWs seem to vary more in this respect. There are more policy workers with “only” secondary education as well those with the highest achievable education.

There are striking differences in terms of previous work experience. Canadian PWs are recruited mainly from academia, other provincial departments and agencies, and the non-profit sector. In contrast, Czech PWs come from municipalities, the private sector or directly after graduation. Despite their high educational attainment, just a few of them come from academia. This reveals quite different career paths in both countries.

There are also substantial differences in future expectations. While in Canada, only 33% of PWs anticipate staying in their current department for more than six years, it is 58% in the CR. This also concerns young people. In Canada, 18% of policy workers under 30 anticipate they will be in their present department for less than one year, contrasted to just three percent in the CR. Though these two questions are not completely comparable and the results should be verified by more detailed questions, policy work in the CR seems to be more stabilized and there seems to be lower turnover but also less mobility, compared to Canada.⁷

Educational background is another important feature of PWs. Because the classification of subject areas in tertiary education slightly differs between Canada and the Czech Republic, for the sake of comparison we grouped the original Czech data in line with the Canadian questionnaire.⁸ Table 2 reveals substantial differences between the two countries. While social sciences dominate in Canada, most Czech officials have a degree in engineering, education or agriculture. The only exception is economics where the proportion is higher in the Czech Republic.

Table 2 Education and degree subject area.

	Canada%	Czech Republic%
Political Science	16.0	.9
Business Management	14.2	7.7
Economics	11.7	15.5
- /Other social sciences	10.3	n.a.
Public Administration	9.9	3.8
Sociology	7.8	1.2
Natural Sciences	7.0	6.5
Public and Social Policy/Public Policy	6.6	1.7
Environmental Studies	6.5	8.5
Law	6.5	5.5
Geography	6.4	5.8
History	5.8	3.0
Education	4.8	11.1
Fine Arts/Humanities or fine arts	4.7	.6
Planning	4.6	4.6
- /English	4.6	n.a.
- /Natural Resource Management	3.4	n.a.
- /Other health sciences	3.0	n.a.
Journalism and Library Studies/Communications or journalism	2.4	.3
- /Other arts or humanities	2.2	n.a.
Engineering	2.1	12.6
Languages or linguistics	2.1	.7
Computing Science	1.6	2.3
Medicine	.4	1.3
Agriculture	n.a.	10.0
Social Policy and Social Work	n.a.	5.2
Psychology	n.a.	2.9
Mathematics	n.a.	2.7
Philosophy, theology	n.a.	2.7
Transport	n.a.	2.0
Military	n.a.	1.7
Architecture	n.a.	1.6
Interdisciplinary	n.a.	1.0
International Relations	n.a.	1.0
Physical Education and Sport	n.a.	1.0

The difference in education attainment can be partly explained by the structure of tertiary education degrees in the general population. For instance, in the Czech Republic the number of graduates in humanities and social sciences is below the OECD average, while the number of graduates in sciences and agriculture is above the OECD average (OECD, 2012).⁹ Particularly noteworthy is the example of degrees in agriculture. Although regional offices are not primarily responsible for agricultural policy and agriculture plays a minor part in the Czech economy, 10% of officials at regional offices have a degree in agriculture. This finding can be at least partially attributed to the over-representation of agriculture graduates in the Czech Republic as compared to other countries.¹⁰ This reminds us that every policy advisory system is embedded in a broader context which influences its structure.

Although comparative data in terms of degree subjects are not available for Canada, we can assume that the background of Canadian policy workers also at least partly reflects the general structure of tertiary education, e.g. the proportion of graduates in social sciences. One of the most striking differences between the two countries is the proportion of Canadian policy workers with political science, public administration, public policy and sociology degrees (Table 2). While these subjects are among the most common in Canada, they are insignificant (or virtually nonexistent in the case of political science) in the profile of Czech PWs. This is not because these fields do not exist in the Czech Republic. In fact, in a comparative perspective, the proportion of graduates with social science degrees in the Czech Republic is above the OECD average. Yet, especially in the case of political science and sociology, these degrees are not usually understood as a path for work in public administration. They rather lead to a career in academia, politics or advisory organizations (in the political science case) or in survey agencies (in the case of sociology).

4.2 Comparison of policy tasks

Moving to the second research question, Table 3 shows the mean scores of time spent on different policy-related work activities.¹¹ Out of 16 activities, there is only one in which there is no statistical difference between Canadian and Czech PWs: implementing and delivering policies and programs. In 12 activities Canadian respondents report more involvement, while the Czech policy workers are more active just in three policy related tasks.

Table 3 Comparison of policy-related activities.

Tasks	Canada			Czech Republic
	N	SD	Mean	N	SD	M
Appraise policy options ^**	1365	1.13	3.57	783	1.56	3.38
Collect policy-related data or information ^***	1392	1.00	3.99	783	1.46	3.22
Conduct policy-related research ^***	1401	1.07	3.81	783	.87	1.51
Identify policy issues ^***	1410	1.00	3.93	783	1.56	3.33
Identify policy options ^***	1405	1.00	3.72	783	1.56	3.41
Implement or deliver policies or programs	1335	1.14	2.90	783	1.56	2.91
Negotiate with stakeholders on policy matters ^***	1347	1.19	2.84	783	1.31	2.54
Negotiate with central agencies on policy matters ^***	1298	1.24	2.62	783	.98	2.21
Consult with the public on policy matters ^***	1292	.975	1.97	783	1.61	2.53
Consult with stakeholders on policy matters ^***	1382	1.00	2.76	783	1.31	2.54
Prepare reports, briefs or presentations for decision-makers on policy matters ^***	1391	.960	3.75	783	1.36	3.09
Consult with decision-makers on policy matters ^***	1388	1.09	3.44	783	1.26	2.02
Brief lower or mid-level policy managers ^***	1350	1.13	3.47	783	1.51	4.19
Brief high level decision-makers ^**	1372	1.14	2.85	783	1.44	3.02
Evaluate policy results and outcomes ^***	1363	1.15	2.93	783	1.02	1.97
Evaluate policy processes and procedures ^***	1361	1.17	2.89	783	.92	1.78

*** Significant at the .001 level.

** Significant at the .01 level.

The most striking difference between Canadian and Czech PWs is in time spent on conducting policy-related research. First and foremost, in contrast to Canada, there are almost no policy researchers in Czech regional administration. While nearly 90% of Canadian PWs are engaged in research quarterly, the same applies to just about 10% of Czech officials. This finding may be partly caused by the sampling specifics explained above, but it is supported by other evidence (in-depth qualitative interviews): policy research as such is virtually non-existent in the Czech Republic's regional authorities and all research-intensive analyses are outsourced. In contrast, Canadian provinces and territories are far more engaged in policy research on a rather regular basis.

As for “traditional” policy analysis (identification of policy issues, identification of policy options, policy options appraisal), the comparison is less clear. Despite the differences being statistically significant in favor of Canadian PWs, they are less pronounced than in policy research. The Czech PWs also display higher variation in these activities (as is reflected in standard deviations), i.e. there is a higher proportion of those reporting either daily involvement in the activities (on the one side of continuum) as well as those reporting “never” on the other side. For instance, though on average Canadian PWs display more engagement in activities such as identification of policy issues and identification of policy options, a substantial proportion (around 10%) of Czech PWs, in contrast to Canadian PWs, report daily involvement in the first steps of policy analysis, namely gathering of information, issue identification, and identification of possible solutions. Canadian respondents report involvement in these activities on weekly or monthly basis.

Interpretation of these findings, however, is not completely clear. It might be that, for instance, “appraisal of possible options” has a different meaning for respondents in the two countries. In Canada, where policy analysis is institutionalized and most PWs are probably familiar with it, “appraisal of possible options” can be associated with particular methods, described in policy analysis textbooks. In the Czech Republic, where only a very minor proportion of PWs are at least a bit familiar with Anglo-Saxon policy analysis, appraisal of possible options is likely to be associated with particular work practices.

In any event, however, the data suggests despite that policy analysis is not formally institutionalized in the Czech Republic, it is, at least partially, practiced. As is clear from the factor analysis (see below) where policy analysis forms one particular cluster, there are commonalities between Canadian and Czech PWs in terms of associations between different policy analysis steps. An important conclusion is that policy analysis activities such as problem identification, identification and comparison of possible solutions are: (1) practiced regardless of whether they are trained or not; (2) interconnected in practice, even if their intertwining is not formalized and “imposed” by common training.

In general, however, many more differences than similarities are found among the work of Canadian and Czech PWs. One of the most important is evaluation. It is clear there is a sharp difference in involvement in evaluation of processes, procedures and results. In contrast to Canadian PWs, only a very small proportion of Czech PWs are involved in these activities. This conforms to the finding that overall evaluation capacity in Central Europe is generally quite weak (Potluka, 2010).

Canadian PWs also spent substantially more time on negotiation/consultation with stakeholders on policy matters, negotiation with central agencies on policy matters, consulting with decision-makers on policy matters and also preparing reports, briefs or presentations for decision-makers on policy matters. Interestingly, however, Canadian PWs report less time spent on consulting with the public on policy matters. While just about seven percent of Canadian PWs engage in consultations with the public on a monthly basis, 27% of Czech PWs report doing so.

Czech PWs claim to spend significantly more time briefing mid-level regional managers (heads of departments) — a weekly activity for 53% of them (contrasted with 17% of Canadian officials). They are also more engaged in briefing high level decision-makers. This may emerge from the finding (not shown in the table) that only 34% of respondents in Canada, compared to 68% in the Czech Republic, belong to any policy-related interdepartmental or intergovernmental committee within their province/region (formal or informal).¹²

In summary, the nature of the prevailing work of Canadian and Czech policy workers differs in almost all respects. While Canadian PWs certainly spent much more time on research and analysis, Czech PWs spend more time on internal, and partially also on external, networking. Though these differences are profound, their interpretation is not easy. It is likely that because of the non-institutionalization of policy analysis, and partially because of the limited focus on policy research, in the Czech Republic, at least some policy analysis activities are in general less prevalent than in Canada. They also may be more likely to be outsourced and “externalized” (Veselý, 2013c). The administrative culture is also very different. For instance, in the Czech Republic, it is normal to hold departmental meetings on a weekly basis, and spend considerable time on internal coordination activities. That might explain more involvement in briefing (which does not, however, necessarily lead to better alignment of activities of different bodies of public administration).

A factor analysis in Table 4 compares the structure of policy work, more specifically policy-related tasks from Table 3, of the Canadian and Czech respondents, enabling us to measure similarities and differences in broader cross-national policy work patterns. In a similar way, it is interesting to analyze general patterns of policy work across countries. So far, all attempts at policy work classification (by factor analysis and similar statistical techniques) have been conducted on national data only. It is quite interesting to see how the policy work activities cluster when data from different jurisdictions are combined.

Table 4 Factor analysis of policy work tasks (Canadian and Czech PWs combined).

	Component
	Policy analysis work α = .871	Evidence based work α = .830	Consulting and briefing α = .651
Appraise policy options	.832	.189	.292
Collect policy-related data or information	.722	.344	−.026
Conduct policy-related research	.355	.707	−.304
Identify policy issues	.851	.287	.196
Identify policy options	.866	.239	.263
Implement or deliver policies or programs	.484	.118	.387
Negotiate with central agencies on policy matters	.180	.456	.376
Consult with the public on policy matters	.087	−.015	.698
Consult with stakeholders on policy matters	.197	.372	.554
Prepare reports, briefs or presentations for decision-makers on policy matters	.356	.532	.327
Consult with decision-makers on policy matters	.167	.768	.173
Brief lower or mid-level policy managers	.255	−.050	.696
Brief high level decision-makers	.100	.335	.600
Evaluate policy results and outcomes	.234	.776	.129
Evaluate policy processes and procedures	.172	.816	.107

Extraction method: principal component analysis. Rotation method: Varimax with Kaiser normalization. Rotation converged in 6 iterations. 61.5% of the variance explained.

A factor analysis of these 15 items was conducted (with 61.5% of the variance explained) and it produced three distinct items consistent with public management literature and confirmed in previous surveys in Canada and the Czech Republic: “policy work”, “evidence based work,” and “consulting and briefing.”

The new summed variable, “policy analysis work” consisted of appraising policy options, collecting policy-related data or information, identification of policy issues, identification policy options and implementing policies or programs. From the second factor, labeled “evidence based work”, items included conducting policy-related research, evaluation of policy results and outcomes and evaluation of policy processes and procedures. Also belonging to this category is negotiation with central agencies on policy matters, consulting with decision-makers on policy matters and preparing reports, briefs or presentations for decision-makers on policy matters. The third factor, designated as “consulting and briefing” included consulting with the public on policy matters, consulting with stakeholders on policy matters, briefing lower or mid-level policy managers and briefing high level decision-makers.

Table 5 illustrates differences between Canadian and Czech PWs in terms of three composite factors derived from Table 4. Canadian PWs are much more involved in evidence-based work and in policy analysis work than their Czech counterparts. In contrast, the Czech PWs dedicate more time in consulting and briefing activities.

Table 5 Differences between Canadian and Czech PWs in three clusters of policy-related tasks.

Tasks	Canada			Czech Republic
	N	SD	Mean	N	SD	M
Policy analysis work ^***	1275	.81	3.63	783	1.36	3.25
Evidence based work ^***	1216	.77	3.25	783	.72	2.10
Consulting and briefing ^***	1221	.72	2.75	783	1.06	3.07

*** Significant at .001.

When we return to theoretical questions on policy worker types, we can discern three broad clusters of policy work: “routine” policy analysis (including implementation), policy research (including evaluation) and networking. It would be, however, misleading to reify these findings, i.e. to take these abstract categories as a real thing with clear boundaries. In practice, all the tasks are closely interrelated. It also should be stressed that though the three clusters are statistically independent, the variables loadings are often shared. It would be fallacious to conclude that each particular PW could be conclusively classified into any of the clusters. As was already observed, in practice quite unexpected combination of tasks might be found, e.g. “those who did policy work were also likely to do street level activities, meaning that these positions and tasks were far from mutually exclusive.” (Wellstead et al., 2011, p. 363).

5 Conclusions and implications for further research

This article tries to answer the general question of whether there are more similarities or dissimilarities in policy work in Canada and the Czech Republic. We have found that characteristics of Canadian and Czech PWs differ in almost all possible respects. The differences are less visible in the nature of policy work, but still substantial differences between Canada and the Czech Republic have been observed.

Admittedly, some of these differences might be methodological artifacts. However, most of these differences are so deep and substantial, that it is extremely unlikely that they are not real and reflective of specific historical institutional factors and the peculiarities of governance arrangements within each state. This leads to a clear conclusion that any attempts of general context-free theory of policy work are not likely to be successful. It is less clear how the observed differences might be explained. The present data does not allow us to test particular theories about how macro-structural factors influence nature of policy work. At this moment, any explanation necessarily remains highly speculative.

However, the results suggest that many more macro factors than usually assumed influence the nature of policy work. Though, our findings are in general agreement with new governance theory (i.e. networking as an important policy work activity), such theory alone only very partially explains the cross-national differences. More “mundane” factors such as labor market structure, structure and organization of higher education and institutionalization of policy analysis, organization practices are likely to influence how the policy work is done. For instance, the structure of tertiary education, number of graduates from various fields as well as contents of curriculum in fields such as public policy or public administration is very likely to have impact upon the profile of PWs. It is why comparative analysis of policy work is so useful. It points to factors that are usually taken for granted and remain unreflected when only one jurisdiction in the focus.

Cross-national comparison led to some particular new findings, too. Our analysis suggests new dimensions that seem to differentiate between policy work tasks, and that have not been discussed in the literature so far. The first dimension is knowledge-intensity. As follows from policy work clusters, the policy tasks differ profoundly in terms of how much — and how difficult — information processing they require. On the one side of continuum, there are routine tasks, not requiring any special intellectual input. On the other side, there are research related tasks demanding considerable individual creativity and intensive information processing. The second dimension, so far not discussed in the literature, is the type of network that PWs belong to. Though, only one general factor concerning networking was discerned (consulting and briefing), closer look reveals that in fact networking is rather heterogeneous category, spreading across the factors. The question here is not whether one particular PW is connected or disconnected, but with whom he/she is connected (public, decision-makers, medium or high-level managers), what is the nature of this connection (hierarchy, parity) and contents of these connections (reporting, briefing, consultation).

More research is needed to test propositions stated above. It would be useful if more countries with different cultural and historical conditions were surveyed. International comparative analysis of policy work is very useful, but very hard to do. Our experience with comparing Canada and the Czech Republic demonstrated that difficulties start at the very definition of comparable target populations. Job positions as well as job tasks are labeled and structured differently. In the future it will be necessary to create a more standardized methodology that would enable exact international comparison (including clear operationalization of basic concepts). Similarly, it would be useful if our knowledge is enhanced by in-depth qualitative analysis, giving more consideration to the contextual nuances.

Notes

* The work of Arnoš Veselý was supported by the Czech Science Foundation Grant (P404/12/0725). The work of Adam Wellstead was in part supported by the Social Sciences and Humanities Research Council of Canada. The authors would like to thank two anonymous reviewers for their valuable comments and suggestions.

1 In this paper we use term “policy workers” (or “PWs”) interchangeably with “policy bureaucrats”, “policy analysts” or “bureaucratic policy analysts”. For discussion see introductory paper to this special issue and Veselý (2013b).

2 The capital city of Prague is governed on the basis of a special statute and was not included in the survey.

3 The exception to this include web-based survey of civil servants in the ministerial bureaucracies developed and managed by OECD in co-operation with RAND Europe (Meyer-Sahling, 2009) and executive survey on Public sector reform in Europe under project Coordinating for Cohesion in the Public Sector of the Future (COCOPS), which was realized in Hungary, Lithuania, Serbia and Croatia. In other countries such as the Czech Republic and Slovakia, there have also been several narrower and more specialized studies (Drulák, Česal, & Hampl, 2003; Scherpereel, 2004).

4 During June and July 2012, complete lists of workers at regional authorities were created based on information from official regional websites. A total of 2615 public servants were included in the final selection based on information on their function or job position. An invitation to fill out a self-administered online questionnaire (Computer-Aided Web Interview — CAWI) was emailed to these selected public servants. We got 783 fully completed questionnaires, 200 contacts bounced as invalid (for more information see Veselý, 2013a).

5 Excluded were those who focused on inner functioning of the office only, i.e. those persons working at secretariats and departments responsible for management of the regional office: financial/economic, investment and IT departments, business licensing authorities, technical maintenance and drivers.

6 Though it should be noted that newly established regions replaced abolished districts (okresy), and that many people from districts moved to regions (often people in senior positions).

7 This is supported also by the length of employment in current organization. While 14 percent of Canadian PBs report is employed in current organization for less than one year, only 7 percent of the Czech PWs report so. And this so, even though Czech policy workers are on average younger and regional authorities were only recently established.

8 In the Czech questionnaire: “In what area(s) did you work before taking up the position in the Regional Authority?”. In the Canadian questionnaire: “Do you have professional policy-related work experience in any of the following environments?”

9 The classification of degree subject areas in the Czech questionnaire was more detailed than in the Canadian one (it included 57 educational fields). Respondents in both Canada and the Czech Republic were free to check as many subject areas as applied to them.

10 Comparative data for Canada are not available.

11 In countries for which comparative data are available, only Turkey and Indonesia have higher proportions of agriculture graduates (OECD, 2012). This is caused by several agriculture universities with a long tradition that are still able to attract a lot of students. This in turn can be explained by the availability of a number of job positions in public administration and by the existence of a well-established agricultural policy advisory system in the Czech Republic (Veselý, 2012).

12 The question was as follows: “How often are you involved in the following types of policy-related work?” The scale was: 1 = never, 2 = yearly, 3 = quarterly, 4 = monthly, 5 = weekly, 6 = daily.