Policy Sciences 21:263-277 (1988)

Kluwer Academic Publishers, Dordrecht Printed in the Netherlands

Advocacy coalitions and the practice of policy analysis

H. THEODORE HEINTZ, Jr.' AND HANK C. JENKINS-SMITH^

' Office of Policy Analysis, Department of the Interior, Washington, DC; ^ University of New
Mexico, Albuquerque, New Mexico

Sabatier and Jenkins-Smith (above) have presented a framework for de-

scribing the processes by which public policies change over time, and Heintz
and Weyant (above) have provided an array of case studies illustrating and
elaborating those processes. This article will focus explicitly on the manner in
which the advocacy coalition frjimework bears on the practice of policy anal-
ysis, both from the perspective of the manager of analytical resources, and
from the perspective of the practicing policy analyst.

1. Issues in the practice of policy analysis

For more than a decade there has been an extensive debate over both the
desirability and efficacy of public policy analysis. (See, e.g.. Rein and White
1977a.) On the one hand are scholars who claim that analysis has had very
little effect because it has been used primarily for partisan political purposes.
For example, Horowitz (1970) argues that analysis has become mere a politi-
cal tool used to justify preconceived policy decisions, while Rein and White
(1977-B) claim that analysis has proved to be impotent as a device with
which to correct or improve public policy. Other critics contend that policy
analysis serves merely to bury political debate under spurious complexity, or
to mask political interest under the guise of objective neutrality (Banfield
1980). On the other hand, defenders have rejoined that policy analysis can
and does provide an essential means to assess and balance competing claims
for increasingly scarce public resources, (Moore 1980) and that analysis does
lead to a gradual 'enlightenment' effect over time (Weiss 1977).' What is
interesting about this debate is that all sides are able to point to cases that
bear out their claims. In some instances, then, each of the positions seems to
have validity.
These conflicting views are a mainfestiation of the lack of consensus about
the role of policy analysis, a lack that has serious implications for the social
legitimacy of policy analysis, for expectations about its influence in policy
making, and for its practical conduct. Such conflicts reflect beliefs in funda-
264

mentally different models of the policy making process, models which, as

Nelson has shown, have deep historical roots (Nelson 1987).
Viewed in its broadest sense, the advocacy coalition framework seems to
provide a conceptual synthesis to these conflicting models. Such a synthesis
can shed light on some of the critiques of the social and political legitimacy of
policy analysis. It can also provide practical insights useful in its conduct.
This article will briefly describe the fundamental models that imderly con-
flicting views of policy analysis. It will then suggest ways in which the policy
change framework can synthesize these models. Finally, it will discuss the
implications of such a synthesis regarding the socially legitimate roles of pol-
icy analysis, its management and the styles and tactics best suited to these
roles.

2. Policy analysis in differing models ofthe political process

Nelson's (1987) recent characterization of three views of the roles of econo-

mists in policy making provides a convenient summary of the conflicting
models of the political process that are at the root of conflicts over the role of
policy analysis. Nelson sees the three models as (1) the progressive model,
(2) the interest group competition model, and (3) the ideological conflict
model. Each of these models can be used to characterize a set of beliefs
about the role of policy analysis.

The progressive model

The Progressive model is based on the fundamental concept of separating the
subjective, value-oriented political policy making function from the objec-
tive, neutral scientific function (Wilson 1887.) Furthermore, the progressives
hoped that the rationality of scientific management would gradually displace
much of the irrationality they perceived in the process of govemment. While
the progressive movement did not forsee the development of policy analysis,
the development of analysis was a logical outgrowth of the progressive move-
ment's effort to introduce science and rationality into govemment.
In this model, the policy analyst is the neutral expert, objectively evalu-
ating the impacts of various options so that the most efficient can be identi-
fied. Typically the application of this model would call for the policy analyst
to identify a set of options for achieving politically determined goals, to study
their economic and other impacts, and to report the results to the decision
maker in his organization. The decision maker would review the analysis,
consider other information on the political values involved, and render his
decision. The policy analyst would then assist the administrator to identify
the most efficient way to implement the decision. The separation of political
 265

conflict from scientific analysis and management is achieved by regarding the

decision maker as the primary user of the policy analysis and by assuming
that the decision, once made, would be implemented without further political
involvement. In this model, policy analysts are taught to operate primarily in
a scientific research and report writing mode.

The interest group model

By the mid-point of the twentieth century, events had substantially eroded
the intellectual basis for the progressive expectation that science and ration-
ality could supplant much of the perceived irrationality of the governmental
process. Political scientists began to focus on the ways in which competing
interest groups influenced policy to serve their material ends. In this model,
policy developed through a continual, incremental process in which organ-
ized interests negotiated for favorable treatment through legislative, executive
and judicial processes and through implementation as well as policy making.
Policy makers in executive agencies were expected to represent the interests
and views of the primary constituents served by the agency. Policy analysis
was seen in this model as providing the information which interest groups
and policy makers used to advocate their positions, a role often criticized by
believers in the progressive model.
Efforts to expand the role of policy analysis in the 1960s led to efforts to
synthesize the progressive and interest group models. Notable in these efforts
was Charles Schultze who presided over the Bureau of the Budget during
efforts to install policy analysis as a major component of the Plarming, Pro-
gramming and Budgeting Systems (PPBS) in the civilian agencies of the fed-
eral govemment (Schultze 1968). Schultze rejected the progressive separa-
tion of politics and expertise, but retained the rationalist concept of policy
analysis as a means of clarifying objectives and making evident the trade-offs
inherent in edlocating scarce resources. While the policy process was seen as
largely incremental and political, the role of the analyst was to provide infor-
mation on the efficiency of the various choices confronting decision makers
at each decision point. The policy analyst would thus become an 'efficiency
partisan' (Schultze 1968: 95), thus retaining the efficiency orientation of the
progressive ideal while merging it with the acceptance of advocacy inherent
in the interest group model. Furthermore, the analytical tools to be used were
based on the broader concepts of economic efficiency drawn from welfare
economics rather than the scientific management concepts of the progressive
era that were more narrowly focused on the efficient delivery of govemment
services. This model does not provide for the analyst to advocate interest
other than efficiency, however, thus retaining the concept that analysts were
to stand apart from the partisan or special interest politics.
266

The ideological conflict model

As Nelson points out, modem political scientists have now come to recog-
nize not only that the progressive hope of restricting subjective political con-
flict to a narrow set of decisions had failed to be realized, but also that many
observations of the behavior of political systems are difficult to explain whol-
ly on the basis of interest group behavior. Attention has therefore focused on
the role of ideology in political processes. Public policy decisions, it ap-
peared, were often influenced by deeply held beliefs about the way the world
should be made to work. Decisions seemed often to be regarded as symbols
of the ascendency of an ideology rather than merely as instmments for
achieving material consequences.
Nelson suggests that since in ideology is "a basic way of thinking about
social issues [that] rests on fundamental assumptions and values that involve
some elements of faith," economists (and by extension policy analysts) have
an ideology. In this model the policy analyst is not merely an advocate for
efficiency among competing special interests, but a proponent of an ideology,
one that emphasizes rationality and efficiency and a decentralized system of
social organization (the competitive market) that relies upon the workings of
material incentives and self interest. Other ideologies emphasize such values
as faimess in the distribution of income and wealth, the primacy of the
natural environment over man-made goods and services, or systems of cen-
tralized social organization that rely on planning and regulation. Thus, ideo-
logical issues are inescapable in the application of efficiency oriented analysis
of public policy issues. The policy analyst, wittingly or not, often becomes an
advocate for efficiency.

Synthesis
One of the purposes of the conceptual framework of policy change and
leaming presented in the Sabatier and Jenkins-Smith articles (above) has
been to characterize long term policy change in a manner that synthesizes
these three models. It does so through its focus on the relationship between
the context of analysis and the kind of role analysis is likely to play on the
political process. The three models can be synthesized using three key ele-
ments in the policy change framework, including (1) the focus on interaction
between opposing advocacy coalitions, (2) the analysis of factors explaining
policy change, and (3) the explication of belief systems of opposing coali-
tions.
The framework's focus on interaction between opposing coalitions incor-
porates many of the elements put foreward by theorists of the interest group
school and carried over into the study of ideological conflict. This interaction
includes competition for control of resources and coalitions' efforts to gain
implementation of favorable policies. Interaction takes place through many
 267

of the political and govenunental processes identified by interest group the-

ories as the mechanisms of negotiation, bargaining and comproniise. The
interaction between coalitions is continual and typically leads to incremental
policy changes. This interaction is also the process for working out subjective
political values in the progressive model or ideological conflicts in the ideo-
logical model.
The framework is focused on factors that can explain how policy changes
occur within a particular policy subsystem. Such factors include forces and
events extemal to the subsystem such as general economic conditions and
electoral outcomes. But more importantly, the framework offers the possibil-
ity of explaining the evolution of policy as the result of policy oriented leam-
ing in which analytical debate is a key process. In such debate, the results of
policy analysis can provide the primary substantive arguments and analysts
may directly or indirectly play the role of advocates for their coalition's posi-
tion. Within the framework, the nature and extent of policy oriented learning
are to be explained by reference to characteristics of the policy context.
These include the analytical tractability of key issues, the nature of the fomm
in which the debate occurs, and the level and intensity of conflict over beliefs.
The explication of coalitions' belief systems as a way of characterizing pol-
icy debate is an important synthesis of aspects of political systems that have
often been treated as separate, or even incompatible, in the progressive, in-
terest group and ideological models. A belief system incorporates ideology in
its core beliefs about the causal relationships that affect material interests in
its secondary aspect. Thus the debate between opposed coalitions is moti-
vated by, and conducted in terms of, ideology, material interests and informa-
tion on causes of the problem eind consequences of policy options. The belief
systems concept thus allows explicit recognition that these three elements
are, in general, combined in most policy debates while the proportions of the
mixture and its effects on policy change differ across subsystems and policy
issues.
The synthesis of the progressive, interest group and ideological models
achieved by the policy change framework has significant implications for the
issue of the legitimacy of the role of policy analysis in a representative politi-
cal system. Critics of policy analysis have charged that policy analysis con-
tributes to a drift toward centralized, technocratic control (EUul 1964:258
259). Others have argued that policy analysis serves to distort the expression
of individual political preferences by subsuming them into a narrow benefit-
cost framework (Kelman 1981).^ The policy change frjunework makes clear
that analysis rarely speaks with a single voice, and even more rarely deter-
mines the outcome of a policy debate. Contrary to the fears of its critics,
users and providers of analysis have proliferated among interests groups and
public agencies, increasing the range and number of policy preferences that
268

are bolstered by analysis (Leman 1982). Only in rare policy contexts those
characterized by highly analytically tractable issues subject to low levels of
conflict in a 'professionalized' forum will aneilysis be likely to contribute to
centralized, technocratic control.
The policy change framework also suggests that analysis is rarely likely to
foist a distorted picture of public preferences onto decision makers. As noted
above, analysis tends to be mobilized by all active interests in the subsystem.
The framework makes clear that, on major policy debates, advocacy analysis
will persist over an extended period of time up to a decade or more. Thus,
should policy analysis bias public expression, those slighted can anticipate
subsequent analytical fora in which to press for redress.
As a final point regarding the legitimacy of policy analysis, the framework
indicates that analyses can stimulate genuine leaming among policy sub-
system members. At worst, analysis stimulates intra-coalition leaming, an
elaboration of the secondary aspect related to an advocacy coalition's policy
core. At its best policy analysis stimulates cross-coalition leaming, leading to
modification of the policy core of one or more advocacy coalitions. In recent
years this process has been clearly evident in the airline and tmcking deregu-
lation issues (Derthick and Quirk 1985).
The advocacy role of policy analysis in the democratic policy making proc-
ess has similarities to the advocacy role of attomeys in the judicial process
and, indeed, to the advocacy role of scientists in debating opposing theories.
Such advocates abide by a set of mles appropriate to their endeavors, mles of
evidence, relevance and logic by which opponents can criticize and judge
each proposition. When policy analysis is used to advocate positions in the
policy process, similar mles stmcture the debate. The policy change frame-
work provides a way to determine the nature of such mles by examining the
content of the debate. It also provides the potential for understanding how
the context of the debate effects the role of policy analysis and other factors
in leading to policy oriented leaming and policy change. We now tum to an
examination of the context of policy debate and its implications for the prac-
tice of policy analysis.

3. Policy analysis in context

What are the implications of the framework for the practice of policy anal-
ysis? In broad terms, the hypotheses drawn from the framework can be used
to depict those regions of the policy context that are hostile or hospitable to
the effective use of policy analysis in policy change. The relevant variables
underlying the policy context, as noted in the first two articles of this
volume, are (1) the intensity of conflict over the political issue as derived
from the level of conflict between belief systems of competing subsystem
 269

coalitions, (2) the analytical tractability of the issue, and (3) the nature of the
forum in which the analytical debate takes place. These variables in combina-
tion pro\dde the context in which analysts attempt to shape the dimensions and
weights ofthe analytical policy space, as described in Jenkins-Smith (above).
Within subsystems, policy learning can take place within and/or across
advocacy coalitions. In the former case, a given coalition elaborates upon the
causal connections and relations within its belief system, or perhaps finds
more persuasive ways to defend its policy core and secondary aspect. In the
case of cross-coalition learning, however, one or both coalitions add to or
modify their policy cores or secondary aspects in response to the analytical
debate. Thus while both kinds of leaming involve adjustment of belief sys-
tems, intra-coalition leaming primarily involves reinforcement or elaboration
of preexisting beliefs, while cross-coalition learing involves genuine alteration
of beliefs. An important aspect of cross-coalition learning is the change in
beliefs of people who were not coalition members. Strong arguments can
convince the undecided to support a coalition's policy, perhaps even to join
and contribute resources.
To the extent that the policy context enhances prospects for cross-coali-
tion leaming, then, the context can be termed hospitible to policy analysis. A
hostile context, on the other hand, will inhibit cross-coalition learning,
although it may instead foster intra-coalition leaming. What then are the fac-
tors that lead to hostile or hospitable context for the application of policy
analysis?
Given the hypothesis developed in Sabatier (above), policy analysis can be
expected to generate cross coalition leaming under conditions in which:
1. conflict between subsystem coalitions is moderate. Moderate conflict typi-
cally stems from conflict between the policy core of one coalition and the
secondary aspect of another. The result is conflict sufficient to result in the
allocation of analytical resources, but conflict moderate enough to permit
both coalitions to contemplate change in their belief systems.
2. the policy problem at issue ranks fairly high on the dimension of analytical
tractability. Thus when the issue under debate has well developed theory,
is well conceptualized and operationalized, and adequate data exists, the
probability that analysis will contribute to cross-coalition leaming is
increased. Issues that are relatively new are tractable if these analytical
tools can be developed in a reasonable period, or carried over from a
related policy issue area. As a general rule, we would expect analytical
tractability to be highest among natural resource issues, and lowest in such
social issues as welfare and health care.
3. the forum in which the policy debate is conducted approximates the 'profes-
sionalized' forum. Thus when the participants of the debate share a com-
mon einalytical background, such that common bases for assessing the
270

validity of analytical claims exists, analysis has a better chance to effect

cross-coalition learning. A professionalized forum also provides some
assurance of an undecided audience that will apply the same rules and
criteria in judging the analysis and arguments made by the opposing coali-
tions. Thus the hospitality of the policy context will be affected by factors
that determine who participates in the forum.
Of course, there will also be combinations of characteristics of the policy
context that severely inhibit the ability of policy analysis to foster cross-coali-
tion learning. In some (but not all) respects, these are the converse of the con-
ditions leading to hospitable contexts noted above. They include the follow-
ing:
1. either intense conflict or the absence ofcoriflict across advocacy coalitions
within the policy subsystem. When intense conflict exists, considerable
analytical resources may be committed to the political fray, but neither
coalition is likely to evidence much willingness to modify its belief system.
On the other hand, where conflict is absent or insignificant, it is unlikely
that either coalition will commit sufficient analytical resources to effect
any possible leaming.
2. the policy issue under consideration is highly analytically intractable. If
Sabatier's hypotheses 8 is correct, most analytical debates concerning
social policy wherein the objects of the analysis are themselves advo-
cates in the debate will be subject to a hostile policy context.
3. the issue is addressed in an open political forum. In other words, the partic-
ipants in the political debate are highly heterogeneous in their training and
in their bases for acceptance of analytical claims.
Thus the framework of policy leaming provides us with extreme points, indi-
cative of those policy contexts that are most hostile and most hospitable to
cross-coalition leaming. Of course we would expect many policy contexts to
fall someplace between these extremes, perhaps with moderate conflict com-
bined with a relatively intractable analytical issue. A degree of pessimism
may be in order, however, in light of the fact that professionalized fora are
relatively rare in the American political system.^
Even when the policy context is hostile to prospects for cross-coalition
learning, analysis may result in policy change. In the case of high levels of
conflict, wherein advocacy coalitions in the subsystem rigidly adhere to the
existing belief system, repeated attempts to ignore or refute compelling ana-
lytical criticisms results in the loss of analytical credibility. The loss of analyti-
cal credibility forces the coalition to expend more scarce political resources
in support of translation of its belief system into policy via other avenues
such as lobbying (see Hypothesis 5, Sabatier, above). In the extreme, a con-
vincing and sustained analytical assualt on the belief system of the dominant
coalition within a subsystem like the extensive body of research produced
 271

in the tobacco policy subsystem indicating that significant health problems

result from the use of tobacco products can result in a loss of support from
the system-wide governing coalition, in tum resulting in forced changed from
outside the subsystem. Although coalition members may not change their
beliefs, previously undecided participants may become convinced and lend
support to the opposing coalition to bring about policy change.
The development of the policy change framework provides an opportunity
to examine in detail a variety of cases having different policy contexts
and different analysis roles. This volume presents only a few of the possibili-
ties, giving preliminary indications of the usefulness of the framework. Of
particular interest to practicing policy analysts are questions about the selec-
tion of policy issues as targets of analysis and the relationship of the style or
role of the analyst to the policy context.

4. Managerial implications: allocating analytical resources

Given the effect of the policy context on the role that analysis can be expect-
ed to play, how can (usually scarce) analytical resources be allocated in a
manner that enhances prospects for cross-coalition leaming? In answering
such a question, one must remember that analysts must respond to their pol-
icy making clients, usually officials in agencies or organized interest groups: it
is the client, after all, who has political legitimacy and authority (Foster
1980). Policy analysts usually have a weaker claim to represent the beliefs of
a policy coalition, except perhaps in professionalized fora devoted to anal-
ysis.
Because analysis must be responsive to the client's needs, the overall allo-
cation of analyticil resources frequently must take into account objectives
other than maximizing cross-coalition learning. Like it or not, analysis will
sometimes be called upon to delay decisions ('paralysis by analysis') or to
legitimize a preconceived policy position of the client (Marver 1979). In
essence, the range of free choice of over which analytical tasks to undertake
can vary significantly (Jenkins-Smith, 1982). Acknowledging that such
demands may need to be met to 'pay one's dues' within an organization, there
is often some degree of discretion over selection of issues and allocation of
analytical resources. What guides for allocation of analytical resources can be
discerned from the policy change framework?
Managers of policy analysis are faced with a variety of choices as they par-
ticipate in an extended analytical debate. They must decide which issues or
beliefs to analyze, whether to focus primarily on the beliefs of their own coa-
lition that are imder attack by the opposition or to focus on the beliefs of the
opposing coalition. The framework suggests that managers should, in part
272

identify opportunities in hospitable policy contexts, and in part try to manip-

ulate the policy context to make it more amenable to analysis. In the former
case, managers will select those issues for which the context is characterized
by moderate conflict, a relatively tractable subject matter, and a forum in
which participants can be expected to evidence a modicum of agreement
over the bases for assessing analytical claims. Cases scoring high on all those
factors will be rare, however, so the task will ordinarily be one of eliminating
the 'hopeless cases' where analysis is most likely to be ineffectual. Where the
discretion of the broker permits such screening, the benefits could be sub-
stantial: the release of analytical resources, including staff time and the cost of
consultants, would permit a concentration of effort on issues where analysis
is more likely to make a genuine contribution.
The possible contribution of analysis is not always obvious. In some cases,
the policy context can be manipulated by judicious tailoring of the analytical
issue. For example, an issue of importance, like natural gas deregulation, may
be subject to very high levels of conflict (Rep. John Dingel, chairman of the
House Energy Committee, was quoted in the press as saying full deregulation
of natural gas would be achieved over "my dead body"), but some important
sub-issues (like the 'old gas' supply response issue) of the debate may be less
exposed to the political cross-fire. Furthermore, tailoring analysis may permit
focus on an aspect of the larger policy issue that is relatively tractable. This is
precisely the manner in which analysis was deployed in the 1983 natural gas
deregulation debates, recounted in Jenkins-Smith (above), with the result
that all participants in the debate adjusted their belief systems to accoimt for
the benefits of increased supply under deregulation.
Another conceivable mechanism for manipulating the policy context
one that is less likely to be under the control of the policy manager is the
selection of the policy forum. Usually the forum will be established in ac-
cordance with the distribution of authority over the policy issue in question.
Where the broker might be able to infiuence the forum will be in cases where
a client e.g., a Cabinet level secretary has the power to convene a forum,
invite participants and set the agenda. In the end, of course, issues wiU be
debated in many fora, only some of which can be structured to facilitate pol-
icy oriented learning.
The optimal choice in selection of an analytical forum will be based in part
on how persuasive the analysis is, and on the sponsoring coalition's political
resources. If the analysis is compelling, the manager should attempt to steer
the issue into a professionalized forum, perhaps by soliciting highly respected
bodies like the National Academy of Sciences to sponsor studies and/or con-
ferences on the issue. If the analysis is less compelling, but the coalition has
abundant political resources, the manager is more likely to meet with success
by routing the issue to an open political forum, like the U.S. Congress.
 273

Whatever the strategy, the manager should realize that changing the core
beliefs of opposing advocacy coalitions will take a considerable period of
time. But it can happen, as illustrated by the airline deregulation (Derthick
and Quirk 1985) and tobacco regulation (Fritschler 1983) cases. On the
other hand, when the debate is focused on secondary aspects, as illustrated
by the gas deregulation case recoimted above in Jenkins-Smith, results can be
expected to be evident much more quickly.

5. Implications for the individual analyst

The policy framework also provides important insights for individual ana-
lysts. First and foremost, it is a mistake to view policy analysis as a discrete
act in which a problem is analyzed, options reviewed, ind advice given.
Viewed in isolation from the interactions of advocacy coalitions within policy
subsystems, 'success' and failure' of analysis are too often seen as wholly
contingent on the technical perfection of its content, and on whether policy
makers accept or reject its conclusions. Such a view fails to recognize that
analysis is more often one part of an extended, many-sided exchange about
how issues should be perceived, and what policy options merit consideration
by policy elites. Such a view may also fail to recognize that what counts as the
success of a particular analysis is at least partly contingent on the prevailing
policy context: to expect scrupulously objective, technically precise analysis
to make dent in a highly conflictual, intractable policy debate is to invite dis-
appointment and frustration.' On the other hand, to write a partial, even
adversarial, analysis that makes the case for an important but neglected pol-
icy perspective can be a significant corrective when the debate is conducted
in a closed and/or politicized forum.
More broadly, the framework sheds light on the question of the appro-
priate professional role for the policy analyst in democratic politics. A survey
of the literature reveals that three dominant roles for policy analysis are of-
fered as appropriate guides for professional behavior. (For a summary of that
literature, see Jenkins-Smith 1982.) The objective technician, operating in the
progressive mode, holds the primary objective to be provision of neutral,
objective and comprehensive analysis and then to retire from the field.*
Advocacy is not the game. The primary professional value is analytical integ-
rity. The issue advocate, on the other hand, does not object to joining the
political fray, and will use analysis to pursue some conception of the 'good
society.' It is probably the case that most issue advocates, by dint of training
in the fundamentals of policy analysis, are advocates for economic efficiency.
Alternatively, the client's advocate uses analysis to make the best case for his
political client's preferred option. For this analyst, it is the client who has
274

political legitimacy not the analyst and the analyst is obliged to sub-
ordinate values derived from a personal view of the good society or technical
integrity to service for the client (Foster 1980). Often, of course, such advo-
cates come to share their client's beliefs, either through self-selection or
socialization within the coalition.
The framework suggests that each of these roles has legitimacy, strengths
in particular kinds of policy contexts, and fatal weaknesses in others. The
objective technician may do well with regard to a low conflict issue debated
in a professionalized forum, but would be utterly at sea in an analytical brawl
over a high conflict-intractable issue in an open forum. Furthennore, the
advocate styles of analysis may be appropriate for analytical debates in which
analytical conflict is sufficient to assure that all relevant viewpoints are repre-
sented, but they may actually undermine the credibility of analysis when
employed in contexts of less (or uneven) conflict. Thus there is an inescap-
able and uncomfortable tension here: none of the styles of analysis seems
adequate in and of itself, and yet, applied indiscriminately, the various styles
of analysis could well undermine one another.
Of course, tailoring analytical practice to context would not be easy. Diffi-
culty may arise over determination of which context actually exists; the bulk
of the analytical contexts fall somewhere between extremes on dimensions of
analytical conflict, tractability, and types of fora. Different analysts could thus
quite conceivably perceive these factors differently, and therefore adopt and
apply incompatible styles of analysis in the same debate. Altematively, the
context may change as the analytical debate wears on, forcing periodic
assessment of which style of analysis is appropriate. It is nonetheless highly
likely that most analysts could, particularly in extreme cases, discern the
major characteristics of the analytical forum. Furthennore, as illustrated in
the case studies presented in Jenkins-Smith (above), analysts do as a matter
of course modify their perceptions of the policy context through experience
and adjust their styles of analysis accordingly.^ The problem is to determine
what the appropriate style would be in a given context.
On the basis of the framework of policy leaming, we offer a tentative reso-
lution to the problem. The three styles of analysis appear to best correspond
with three extremes in the policy context. The resolution calls for an explicit
adoption of a combination of styles of analysis, flexibly adapted to the policy
context. In a context in which the issue draws little analytical conflict, when
the forum is closed or not widely visible and approximates a professionalized
forum, participants should tend toward the objective technician style. Be-
cause of low levels of conflict, participants cannot count on the interests of
those aggrieved to readily correct for analysis that neglects or overempha-
sizes particular dimensions of value or beliefs. Furthermore, the closed or
low-profile forum may serve to screen participants who may otherwise have
 275

added important dimensions and interests to the debate. The objective tech-
nicians' penchant for comprehensive neutral analysis, though unlikely to be
fully realized in practice, will serve to increase the likelihood that the most
important factors ait included.
Sometimes, however, the closed or low-visibility forum will already have a
distinctive pattem of expressed interests represented in analysis; analysis may
reflect the existence of 'client politics' (Wilson, 1980), in which a small group
of directly affected beneficiaries of a policy have incentives to join the debate
while the larger, more diffuse population of those bearing the costs of the
policy do not. In this context, objective analysis using the general concepts of
applied welfare economics could be expected to produce information con-
flicting with the positions of particular interests. If efficiency-oriented anal-
ysis is to be presented in such instances, an issue advocate role is hard to
escape. Stressing the costs of a special interest policy to the broader popula-
tion, especially within a coalition in which the special interest has substantial
power, requires a more assertive advocate style, but may entail some risks.
The issue advocate role may be appropriate as well in high conflict debates
waged in open, politicized fora. In those instances, a variety of interests to the
debate can be expected to be well represented by analysis. Because this mul-
tiplicity of interests is mobilized, each hammering home its own grievances,
rights and expectations, the prudent analyst could well assume the role of
partisan efficiency advocate without concem that the structure of values
represented to policy elites is unduly distorted.
The practicability of the role of the partisan efficiency advocate will often
be contingent, of course, on the analyst working for an uncommitted (or like-
minded) client. Because the latter cannot be counted on in high conflict pol-
icy issues, analysts employed by a client intensely committed to a particular
policy outcome will be pressed to adopt the role of client's advocate. For high
conflict issues debated in open fora, such a result is not as bad as critics have
suggested. The tendency of the analytic contribution to such policy contexts
to be pluralistic in character would serve to ameliorate advocacy for any par-
ticular partisan or position in the aggregate analysis produced. It is well for
the analyst to remember, too, that the publicly elected or appointed policy
maker has at least an indirect legitimacy in the positions taken by virtue of
representative electoral politics. Finally, service of the client's interests, con-
sonant with the prevailing analytical tractability of the issue at hand, may be
necessary in order to maintain the client's sympathetic ear on other issues.
For these reasons, the prudent analyst may on occasion assume the role of
the client's advocate.
Demands for service of the client's interest may become unreasonable,
however. In particular, the demand that analysis be shaded in the client's in-
terest to a degree that departs from the prevailing analytical consensus on the
276

question cannot be met in good conscience. In those instances, should the

exercise of voice not modify the demand, the analyst may be compelled to
employ exit.*
The appropriate application of the range of styles of analysis knowing
when each is appropriate and what are their limits would require consider-
able judgment on the part of the analyst. Nonetheless, the policy change
framework suggests that a flexible combination of styles of analysis is both
legitimate and practical. It is appropriate for the role of policy analysis to
evolve by adapting to the characteristics of various policy contexts. In some
contexts a detached, objective scientific role is appropriate while in others
advocacy is called for. Experienced and perceptive policy analysts, using the
advocacy coalition framework, should be able to adapt their analyses and
their style to the policy context and thus contribute more effectively to the
injection of shared knowledge into the policy making process.

Notes

1. To be sure, there are many uses of policy analysis some of which are overtly political
uses designed to legitimize preconceived policy positions. Furthermore, clients may seek
out analysts who will agree with the policy maker's policy perspective (often called 'hired
guns') to assure 'compatability' of analysis and the client's policy position. Marver (1979)
provides a provocative list of the full range of functions' of analysis. Nonetheless such uses
are typically perceived as abuses of analysis, and if widely employed will undermine the
credibility of all analysis.
2. For an extensive discussion of these criticisms of policy analysis, see Jenkins-Smith (1985),
Chapters 3 and 7.
3. Professionalized fora may be more common in European political systems. See, for exam-
ple, Brickman's (1984) discussion of the differences between pattems of formulation of
toxic waste policy in the United States and in European countries.
4. See Frischler (1983). Another case of analysis leading to forced change from outside the
policy subsystem concems airline regulatory policy. See Derthick and Quirk (1985).
5. Amold Meltsner quotes an analyst frustrated by the reception of his work at the U.S. Office
of Economic Opportunity (OEO) as saying: "That's what drove me out of OEO; the
leadership simply didn't understand the role of evaluation" (Meltsner 1976:23).
6. See Meltsner (1980) for a description of the technician model of analysis. Also see Jenkins-
Smith (1982) for a development ofthe 'objective technician' style of analysis, that typifies
the dominant paradigm in policy analysis.
7. In the Alaskan oil export case, it was after confronting advocacy for the maritime interests
on the part of MarAd analysts that the DOE analyst raised more uncertain issues that
would tug the analysis results back in the other direction. And in the Strategic Petroleum
Reserve case, it was after repeated experience with OMB analysts doggedly pursuing ana-
lytical results that would show a large SPR to be unjustified that analysts from DOE dug in
their heels and extracted a concession that permitted them to use their own, improved ana-
lytical approach.
8. See Weimer and Vining (1986), Chapter 2, for a very useful way of understanding the ethi-
cal dilemmas confronted by a policy analyst whose client demands analysis in support of a
preconceived policy position.
 277

References

Banfield, Edward, 'Policy Sciences as Metaphysical Madness.' Bureaucrats, Policy Analysts,

Statesmen: Who Leads! Edited by Robert Goldwin, Washington DC: American Enterprise
Institute, 1980.
Brickman, Ronald. 'Science and the Politics of Toxic Chemical Regulation: U.S. and European
Contrasts,' Science, Technology, and Human Values, Vol. 9, No. 1, (Winter 1984), 3950.
Derthick, Martha, md Quirk, Paul. The Politics of Deregulation Washington, DC: Brookings
Institute, 1985.
Ellul, Jacques. The Technological Society. New York, NY: Vintage Books, 1964.
Foster, John L. 'An Advocate Role Model for Policy Analysis,' Policy Studies Joumal, Vol. 8,
No. 6 (Summer 1980), 958-964.
Fritschler, A. Lee. Smoking and Politics. 3rd Ed. Englewood Qiffs, NJ: Prentice-Hall, 1983.
Horowitz, Irving. 'Social Science Mandarins: Policymaking as a Political Formula,' Policy Sci-
ences, Vol. 1 (1970), 339-360.
Jenkins-Smith, Hank. Politics and Policy Analysis: Efficiency and Democratic Norms in Tension.
Rochester, NY: University of Rochester, Ph.D. dissertation, 1985.
Jenkins-Smith, Hank. 'Professional Roles for Policy Analysts: A Critical Assessment,' Joumal
of Policy Analysis and Management, Vol. 2, No. 1, (Fall 1982), 88100.
Kelman, Steven. 'Cost-Benefit Analysis: An Ethical Critique,' Regulation, 5(1) (Jan/Feb) 1981,
34-42.
Leman, Christopher. Some Benefits and Costs of the Proliferation of Analysis in Natural Re-
source Budgeting. Discussion Paper D-101. Washington, DC: Resources for the Future,
December 1982.
Marver, James. Consultants Can Help: The Use of Outside Experts in the U.S. Office of Child
Development. Lexington, MA: Lexington Books, 1979.
Melisner, Amold. Policy Analysts in the Bureaucracy. Berkeley, CA: University of Califomia
Press, 1976.
Moore, Mark. 'Statesmanship in a World of Particular Substantive Choices.' Bureaucrats, Policy
Analysts, Statesmen: Who Leads? Edited by Robert Goldwin. Washington, DC: American
Enterprise Institute, 1980.
Nelson, Robert. 'The Economics Profession and the Making of Public Policy.' Joumal of Eco-
nomic Literature, Vol. 25 (March 1987), 4284.
Rein, Martin, and White, Sheldon. 'Policy Research: Beliefs and Doubt,' Policy Analysis, Vol. 3,
No. 2, (Spring 1977), 239-271.
Rein, Martin, and White, Sheldon. 'Can Policy Research Help Policy?' The Public Interest, Vol.
49,(Fall 1977), 119-136.
Schultze, Charles. The Politics and Economics of Public Spending. Washington, DC: Brookings
Institute, 1968.
Weimer, David, and Vining, Aidan. Policy Analysis: Introduction to an Emerging Profession.
Draft manuscript, 1986.
Weiss, Carol. 'Research for Policy's Sake: The Enlightenment Function of Social Research,' Pol-
icy Analysis, Vol. 3, No. 4, (Fall 1977), 531545.
Wilson, James O. Politics of Regulation. New York, NY: Basic Books, 1980.
Wilson, Woodrow. 'The Study of Administration.' Political Science Quarterly, (June 1887),
197-222.