Policy Sciences (2006) 39: 1–18

DOI: 10.1007/s11077-005-9004-1 
C Springer 2006

Understanding the historical turn in the policy sciences:

A critique of stochastic, narrative, path dependency and
process-sequencing models of policy-making over time

MICHAEL HOWLETT & JEREMY RAYNER

Simon Fraser University and Malaspina University College

Abstract. This article evaluates four general models of historical change processes which have emerged in
various fields in the social sciences – namely stochastic, historical narrative, path dependency and process
sequencing – and their application to the study of public policy-making. The article sets out and assesses
the merits and evidence for each, both in general social research and in the policy sciences. The article
suggests that more work needs to be done examining the assumptions and presuppositions of each model
before it can be concluded that any represents the general case in policy processes. However, since neither
the irreversible linear reality assumed by narrative models, nor the random and chaotic world assumed
by stochastic models, nor the contingent turning points and irreversible trajectories required of the path
dependency model are found very often in policy-making, these models are likely to remain less significant
than process-sequencing models in describing the overall general pattern of policy dynamics.

Introduction: The historic turn in the policy sciences

History is a difficult subject for students of policy-making. Although most policy

studies focus on changes which occur in government actions over time, the need to
carefully examine the often implicit theories of history behind identified patterns of
policy development has only been recently recognized (Mahoney and Rueschemeyer,
2003; Saldana, 2003; Pierson, 2004). Sociologists and others at the onset of the 1990s
turned to this same historical question and generated an excellent corpus of conceptual
and methodological work on the subject (Somers, 1996; Abbott, 2001).1 However
these insights are only now slowly penetrating into the evaluation and critique of the
concepts and methods used in the policy sciences.
For many years, policy studies, and especially formal policy analysis, often pro-
ceeded along the unspecified and implicit assumption that policy-making followed
the precepts of what Andrew Abbott has termed a “general linear reality” (Abbott,
1988). That is, that policy causes and effects could be ascertained empirically and
that a general set of social forces drove policy-making, with individual deviations
from deterministic outcomes existing as “noise” or random error (Aminzade, 1992;
Griffin, 1992; Stinchcombe, 1968). Outcomes from such processes, such as policy
decisions, were seen as the realization of stochastic processes, in which some underly-
ing set of factors (independent variables) with certain kinds of parameters combined
to “determine” a result. In this “general linear model,” as Abbott has termed it, there
is an implicit notion that history, per se, doesn’t matter. That is, time is thought of
as a discrete, infinitely divisible entity and “history” is simply the assemblage of
“moments” on a temporal continuum. Which time period is chosen for analysis is im-
material as each outcome is more or less precisely determined at that moment by the
existing configuration of variables. As Abbott put it, in this model, “the social world
2

is made up of fixed, given entities with variable properties” – cases and variables – in
which outcomes consist of “the succession of the values of a dependent property or
properties over time” (Abbott, 1990).
In most social sciences, including policy science, however, this general conception
of the ahistorical temporality of social processes has recently been challenged by in-
vestigators who have argued that “sequence matters.” That is, that the determination
of policy outcomes is not deterministic, in a stochastic sense, but much more con-
tingent than previously assumed; with the sources of contingency being not merely
individual actions in a given environment, but also more structural factors such as
historical timing or the “ordering” of policy-relevant events (Pierson, 2000b; Abbott,
1990).
This emphasis on structured sequencing, of course, is a significant aspect of many
recent neo-institutional approaches to the study of public policy-making (Steinmo
et al., 1992; Hall and Taylor, 1996; Kato, 1996). However, as this article will show,
the basic model of historical sequencing found to date in many neo-institutionalist
studies has tended to focus only upon a single alternative model of historical processes
– path dependency – that is only one of several possible alternatives to the stochastic
model. Most neo-institutional studies have generally not considered (a) the alternative
possibilities and models which exist (Thelen, 2003; Lieberman, 2001) and, more
surprisingly, (b) whether or not the specific attributes of the path dependent model in
fact fit the reality of most policy-making situations (Kay, 2005; Greener, 2005).
This article addresses both these issues, assessing the merits and demerits of three
general alternative models of historical sequencing to the stochastic model – path
dependency, historical narratives and process sequencing – both in their application
to social phenomena in general and as specifically applied to the study of public
policy-making.

Three alternate policy-relevant conceptions of historical sequencing

to the stochastic model

Inevitable sequence models: Historical narratives

While path dependency may have emerged in recent neo-institutional accounts of

policy-making as a serious competitor to the “general linear model,” in general in the
social sciences the principal alternative to the stochastic model of historical reality
has been one which is much more profoundly historical in nature but often equally
un-reflective of its conceptual foundations. This is the reading of history as a causal
narrative, a model and methodology prevalent in academic History Departments and
many other areas of social, cultural and political studies (Abbott, 1992; Stinchcombe,
1968: pp. 101–29) In this model, an entire sequence of events is read, retroactively,
as a single entity. The challenge for the analyst is to uncover the “plot” or “narrative”
which can “explain” the development of the process from start to finish, or “birth”
to “death” (Abbott, 1990). In this view, “causality” is much more complicated than
usually assumed in the stochastic model, in that the sequencing of events is highly
significant in terms of gauging the effects of causative agents or variables. At one
 3

point in a narrative story, for example, the presence or absence of a certain variable
may have one effect, while later on it may have none, or a different one altogether.2
Recent work in the social-sciences by authors such as Czarniawska, Abbott, and
others have attempted to develop the outlines of more formal models of narratives,
specifying the key assumptions of different narrative models and suggesting their
utility and disutilities (Abbott, 1992; Czarniawska, 1998; Ospina and Dodge, 2005).
We distinguish between two general approaches, narrative positivism and narrative
postmodernism. Abbott, for example, specified three major “story properties” which
are methodologically significant in narrative analysis. These are (1) enchainment –
or the “narrative analogue of causality,” in which it is assumed that there are links
between one point in the narrative and another; (2) order – in which the narrative
must proceed in a strict sequence if the observed outcome is to be explained; and (3)
convergence – in which a narrative sequence might achieve a “steady state;” a special
case allowing stochastic analysis to be an appropriate and effective tool for analyzing
developments occurring within a narrative chain (Abbott, 1992).
Abbott has also set out some of the major problems which face this kind of analysis.
These include problems establishing “endpoints” in the analysis (for example, when
the “birth” and “death” of a phenomenon are not easily identified); problems of
multiple overlapping and intersecting “plots” which can make a central narrative
storyline difficult to discern; and problems with too many characters obscuring the
plotlines of a story (Abbott, 1992). All of these problems, of course, emphasize the
significance for this model of the role played by the analyst as interpreter of events,
who may by force of circumstances choose to simplify a story in order to reveal
more clearly its fundamental plot, but who does so at the risk of imposing their own
interpretation of events on history (Buthe, 2002).
The question of validating narrative interpretations, of course, has long been an
issue in the hermeneutic tradition (Gadamer, 1989; Ihde, 1974). When postmodern
social theory embraced its “literary turn,” for example, a development that overlapped
with but has remained distinct from the narrative interests of historians, it drew on this
rich tradition (Lyotard, 1984). The idea that actors make sense of social phenomena
by telling stories about them, stories that they use to orient themselves and to guide (or
misguide) other actors, is both intuitively plausible and directs theoretical attention to
the much-studied hermeneutic mechanisms of enplotment and narratology. However,
as Barbara Czarniawska points out, the epistemological status of narratology itself
as a method of social inquiry remains an open question and her discussion of the
methodological alternatives presented by this technique, which draws heavily on the
hermeneutic tradition, pushes the model further than does Abbott (Czarniawska,
2004a).
The first option Czarniawska presents is to treat actors’ narratives as partial and
sometimes deliberately misleading accounts of “what is really going on” in the social
world. The business of the theorist is to unmask the partial and motivated character of
actual narratives, sometimes by telling an alternative story but more often in policy
studies by appealing to other methodologies to demonstrate the gap between the
stories and an underlying reality (see also Buthe, 2002). The second, equally extreme,
option is to undermine the very idea that there are other methodologies that are not
also forms of story telling so that there is no underlying reality against which the truth
4

or falsity of particular stories can be checked. The first option is close to the position
Buthe, Czarniawska and Abbott have termed “narrative positivism.” The theorist is
trying to “get the story right” by correctly identifying the causal mechanisms that are
actually operating in a world that exists independently of narratives.
The second option, what might be termed “narrative post-modernism” is less well-
adapted to social studies, although it should be noted that it is not necessarily anti-
realist and confined to the analysis of the imaginary worlds of literature or philosophy
and religion (Edelman, 1964; 1988). This is the model built on the observation that
there is a world outside of the narrative itself but the moment we try to describe it
we are again telling stories (Foucault, 1972). The shock value of the second option
has been much exploited in such areas as sociologically-inspired accounts of science
and technology, where scientists are analyzed as (mere) story-tellers (Traweek, 1992;
Law, 2000).
More often, however, social theorists have sought some intermediate position be-
tween the two extremes, one which brackets metaphysical questions about the exis-
tence of a world outside stories and focuses instead on the relations between different
stories and between story tellers. In Czarniawska’s version of narrative postmod-
ernism, which draws on the popular literary concept of intertextuality, for example,
texts speak to other texts. The effect is to create patterns and regularities in interpre-
tation and action which Czarniawska calls “institutions.” As she puts it: “the reader
is able to see how a text was made. . .because reader and writer are both producers
and consumers of the same set of human institutions” (Czarniawska, 2004a). In re-
cent work, she has focused particularly on the mechanism of framing and the role
played by the existing stock of discursive elements that go to make up such a frame
(Czarniawska, 2002, 2004c).
Unlike the case with the stochastic model, in narrative positivism one would expect
to find lock-step and irreversible patterns of historical development, ones with specific
causal patterns related to historical sequencing. In narrative postmodernism, the role
of sequencing is much less clear. Presumably, it matters when framing and reframing
take place because of the limited stock of “ideas in good currency” that are available
at any particular time (Fischer, 2003). However, in spite of Czarniawska’s depiction
of frames as “iron cages” (Czarniawska, 2004b), her characteristically postmodern
emphases on reframing and free play tend to undermine any sense that history moves
in any kind of specific direction meaning, ironically, that this approach in the end
shares much in common with the stochastic model.

The contingent sequence model: Path dependency

Although less predominant in the social sciences than the stochastic model, as a
result of earlier struggles between critical theorists and advocates of more scientistic
‘behavioural’ methods of social inquiry (Adorno et al., 1976; Almond and Genco,
1977) narrative models remain very popular in fields such as history and literary
studies, as well as in the policy sciences (Yanow, 1996, 1999). These are not the only
models of historical sequencing available to help analyze historical processes such
as policy-making, however. For a variety of reasons related to high profile disputes
between institutional and other types of economists, one of the best known recent
 5

alternatives to stochastic and narrative modes of analysis in the social sciences is the
“path dependency” model (Greener, 2002b, 2005).
The contours of the development of the path dependence model in the social
sciences are now well known, especially the influence of debates in the economics lit-
erature on whether or not it is possible for market transactions to result in sub-optimal
outcomes as inferior technologies come to be “locked-in” to specific economic “tra-
jectories” (Arthur, 1988, 1989; David, 1985, 1986; Liebowitz and Margolis, 1990,
1995). Path dependency in this sense represents a kind of failure to achieve a techni-
cally efficient solution that is attributed to any one of a number of factors: to “network
effects” or the ability of inferior technologies to spread and block the adoption of more
efficient ones; to “increasing returns” or the historical accident of the timing of the
entry of new technologies into the market place; or to (premature) “standardization”
which can also block the spread of superior technologies. Debates in economics tend
to turn on issues of what Liebowitz has called the “degree” of path dependency, that is,
how hard it is to “turnaround” a sub-optimal process once it is underway (Liebowitz
and Margolis, 1995). For some authors turnaround is almost impossible, for others,
it is somewhat less difficult to accomplish.
In the social and political realm, the use of the concept of path dependency is
less specific than is found in economics and applies to the description of historical
processes which observers have found to be highly contingent in origin and inertial in
nature. As mentioned above, it is commonly associated with neo-institutional forms
of social and political analysis.
Mahoney outlines the three principal elements of a path dependent model of his-
torical evolution as variations on general narrative precepts. That is: (1) only early
events in a sequence matter; (2) these early events are contingent; and (3) later events
are inertial (Mahoney, 2000). These elements highlight the crucial aspects of path
dependent models of historical development that separate this model from narrative
analyses and from other models – like process sequencing – which are discussed
below: that initial conditions are chance-like, and have a significant influence over
the irreversible course of events followed later in the sequence.
Identifying these “turning points” or “conjunctures” is thus critical to path de-
pendency analyses of historical processes, although there is significant debate in
the literature over exactly what is meant by characterizing an event as “contingent”
(Wilsford, 1985; Abbott, 1997). At its simplest, contingency implies that, although
the sequence of events is not a strictly necessary one, predictable from the conditions
of the starting point according to general laws, there is nonetheless an explicable
pattern which relates one point to another, especially in the early part of the sequence.
While a random sequence implies that any event has an equal probability of following
from any other, in a contingent sequence each turning point renders the occurrence
of the next point more likely until, finally, “lock in” occurs and a general explanatory
principle, such as increasing returns, takes over the work of explanation.
Adherents of stochastic methods, like Herman Schwartz, ridicule this conception
of contingency as involving a redundant distinction between small causes and big
causes. In this view, big causes always supply the ultimate explanation for social
outcomes, so why bother with the analysis of little ones? (Schwartz, n.d.). While
Schwartz’s objection rests on his demand that there be only a single explanation of
6

historical sequencing that subsumes all the others, combined with an hostility to the
idea of structural overdetermination, more positively, his objections raise the impor-
tant question of the nature of, and reasons for, the “embeddedness” of sequences or
“trajectories” in path dependent models. It is important to be reminded that the tri-
umphs of the QWERTY keyboard, VHS standards, or the Windows operating system
described and debated by Lane and Margolis and Liebowitz and others, for example,
took place in the larger context of the development of large-scale social processes
such as bureaucratization. But it is not a fair objection to path dependent explana-
tions to conclude that, therefore, “QWERTY, or more precisely typing, represented
not a completely chance outcome, but rather a deliberate choice by designers. . .”
(Schwartz, n.d.: p. 7). Contingent does mean random in this model, even if only with
respect to initial starting conditions. So while events located later in a trajectory may
be less random, they do not approach the status of “designed,” unless their ad hoc
starting point is ignored. Path dependency attributes outcomes to an overall situation
in which microcausation and sequence matter and, hence, “deliberate choices” cannot
be assumed but require detailed analysis and explanation (List, 2004).

Punctuated equilibrium models – process sequencing

As this discussion has shown, as the extent of contextual embeddedness of social actors
increases to the point where we are concerned with apparently irreversible sequences
that could not have been other than they were, we find ourselves in the world of the
narrative. At the other extreme, where contextual embeddedness is deemed irrelevant
and only chance-like conditions prevail, the stochastic model prevails (along with
its post-modern narrative equivalent). In between these two extremes lies the path
dependency model, combining narrative’s attention to sequence in understanding later
trajectories of events, but focusing on contingency and randomness in understanding
the causal dynamics of the “critical junctures” which start those trajectories in motion.
Conceptions of history and the analytical methodologies which are associated with
them, however, do not line up neatly on a simple spectrum from context-bound to
random. This is made clear by the elements and assumptions of a fourth model of
history found in the present-day social sciences which provides an alternative con-
ception of the nature of social processes to both narratives and path dependency.
In this model, unlike in the stochastic or narrative postmodernist models, sequence
does matter. However, unlike the narrative positivist model, it is not concerned with
irreversible sequences and, unlike the path dependency model, it does not rely upon
random or purely contingent initial conditions to set trajectories in motion. This is
the “process sequencing” model which conceives of social processes as “the con-
nections between events in different time periods as reiterated problem solving”
(Haydu, 1998). Proponents of this model among sociologists, such as Jeffrey Haydu,
argue that it has advantages over both the narrative and path dependency models as it
“provides a plausible way to represent and account for historical trajectories; it builds
social actors and multiple causal timelines into explanatory accounts; and it offers a
richer sense of how earlier outcomes shape later ones” (p. 341).
Rather than connect historical events through stories or paths, Haydu argues, events
can be demarcated on the basis of “contrasting solutions for recurring problems”
 7

(p. 354). That is, “continuities across temporal cases can be traced in part to enduring
problems, while more or less contingent solutions to those problems are seen as
reflecting and regenerating the historical individuality of each period” (p. 354).
Based in part on work in evolutionary biology which suggested evolutionary pro-
cesses proceeded in a stepped or “punctuated equilibrium” fashion (Gersick, 1991;
Eldridge and Gould, 1972; Gould and Eldridge, 1977), this model looks at first blush
somewhat like path dependency in its emphasis on turning points and trajectories,
and its combination of elements of the stochastic and narrative models. Significantly,
however, the model lacks path dependency’s emphasis on randomness in the start-
ing points of trajectories and is not wedded to the idea of irreversible trajectories
found in narrative positivism. That is, process sequencing stresses not how outcomes
at historical switch points are accidents, but how they are firmly based or rooted in
previous events and thinking as related structural processes of negative and positive
feedback affect actor behaviour (Baumgartner and Jones, 2002). Changes in trajecto-
ries in this model are not random or chaotic, but are outgrowths of earlier trajectories.
Hence, although process sequencing shares some of the characteristics of the path
dependency model, it is not the same. The idea of change occurring as a result of an
embedded “crisis,” for example, in the process sequencing model, is not the same as
that focusing on random critical junctures found in a typical path dependency expla-
nation. Moreover, this model does not require a uni-directional trajectory following
an initial conjuncture, but allows for the kinds of reversals in trajectories identified in
the narrative postmodern model as the development of ideas and discourses proceeds
apace.
This model has become increasingly popular in fields such as political science as
an alternative to path dependent models, providing a better explanation of phenomena
such as the creation and development of national and sectoral political institutions as
well as political ideas, discourses and paradigms (Lieberman, 2002; Lindner, 2003;
Lindner and Rittberger, 2003; Pierson 2000c), and appears to be more consistent with
the actual empirical record of changes found in many countries and sectors than is the
path dependency model (Dobrowolsky and Saint-Martin, 2005; Morgan and Kubo,
2005; Rico and Costa-Font, 2005).

Narratives, path dependency and process sequencing as alternate models

in the historic turn in the policy sciences

Before examining their specific application in the policy sciences, it is important to

reiterate the several key features identified above which differentiate these several
general models of historical sequencing from each other. That is, the key aspect of
the stochastic model is its ahistoricism, the notion that sequences don’t matter, a
point shared with post-modern narratology in the sense that the origin and order of
stories is unimportant. The narrative positivist model, on the other hand, emphasizes
adaptation within a sequence and the irreversibility of events and trajectories which
could not be otherwise than they have been described. In the path dependency model,
the key features are the random aspects of critical junctures and hence the possibility
of alternative trajectories at early points in a sequence, along with, as in the narrative
8

positivist model, the irreversibility of sequences once “locked-in.” These contrast with
the process sequencing model’s emphasis on reiterated problem solving as embedding
a new trajectory in a previous one, and the ability of trajectories to shift, and even
reverse, direction.
Figure 1 below illustrates the principal differences between these general models
in terms of their assumptions about the direction and origins of historical sequences.
The existence of these competing conceptions of historical explanation, given the
historical development of policies, poses problems for the policy sciences. Since
many preliminary methodological and conceptual issues relating to the design of pol-
icy studies and the assimilation of their findings are related to the temporal models
used, which, if any, is the correct or most appropriate alternative to the stochastic
model, is a serious question for those studying public policy. In what follows below,
examples of the application to policy-making of each alternative conception of his-
torical sequencing are provided and analyzed in terms of their merits and demerits as
a model of historical sequencing appropriate to the policy sciences.

Inevitable policy narratives: Thick descriptions and policy process tracing

The narrative positivist approach has lain largely unnoticed and unacknowledged at
the heart of many efforts to analyze policy change and development. Many works
which have examined policy development over time, from undergraduate student
essays to the autobiographical memoirs of key policy actors, simply adopt a narrative
positivist approach, working backwards from ultimate outcomes to trace causative
agents and pivotal moments in the historical record of what are viewed as inevitable
sequences of policy-making events (Raadschelders, 1998).
This type of policy analysis has often been castigated by adherents of the stochastic
model as providing little more than idiosyncratic, non-cumulative “thick description”
(Calhoun, 1998). And it is certainly the case that as applied to the policy sciences
this model has many methodological difficulties related to its ex-post facto character,
difficulties involved in the replication of the analysis, and problems with generalizing
results across policy “stories” (Griffin, 1993). However, the difficulties do not mean
that policy narratives are not, a priori, an alternative model of policy dynamics, merely
that this approach has usually not been systematically analyzed as a methodology for
policy analysis (Thompson, 2001; George and Bennett, 2005).

Fig. 1. Conceptions of historical change compared.

 9

Nevertheless, as described above, both the positivist and post-modern variants of

the narrative approach have many weaknesses which are evident when they are applied
to the study of policy-making. Although narrative postmodernism dispenses with the
idea of a master narrative, its conception of stability as maintained by shared frames
of meaning that constrain actors’ preferences has the same contours and weaknesses
as its stochastic counterpart as a model of policy change. In Czarniawska’s account
of urban policy in three European cities, for example, she creates a complex narra-
tive of shared frames interpreted in local contexts where reflexivity further ensures
that the construction of a city identity is, at least in part, an attempt to construct an
“alterity” from other cities’ own efforts at identity construction using the same frames
(Czarniawska, 2002). Contemporary embeddedness, in other words, can take place at
a number of different temporal and spatial scales simultaneously and change is con-
ceptualized as a sequence of reframing through mechanisms such as frame reflection
(Schön and Rein, 1994).
The problem with the application of this post-modern narrative model to pol-
icy issues is that already identified above as the general problem with this class of
explanations of social phenomena. That is, the policy narratives generated through
application of either positivist or post-modern methods always tend to be idiosyncratic,
retroactive explications of events which have already occurred, where inevitability is
observable only in hindsight. This approach, especially, encounters great difficulty in
developing periodization strategies which are not idiosyncratic or “pre-determined”
by the narrative structure attributed to the story by the narrator (Lieberman, 2001;
Lustick, 1996; Clemens, 1999). Again, this is not to say that narrative methods and
thick descriptions of process tracing are always inappropriate in the policy sciences,
only that they raise special problems in the identification of common patterns and
in adherence to methodological guidelines that can generate convincing, replicable,
results. Also, and more significantly, this method only “works” when there is a match
between the method and the underlying empirical reality of the policy context be-
ing examined. That is, while for narrative post-modernism the policy-making world
that is being narrated must be relatively chaotic so that fluctuation in frames may be
apparent, in its positivistic variant the reverse is true. That is, policy-making must
be highly stable, so that the inevitability portrayed in the narrative mirrors the sense
of inevitability found in the narratives of the policy actors themselves (Bernstein,
2000).
These problems have led to the need to combine policy narratives with other forms
of analysis more suitable to the mixed record of stability and instability apparent in
many empirical policy circumstances. Efforts have been made by authors such as
Tim Buthe to rectify the inherent problems of narrative methods by, for example,
proposing the use of narratives only to provide empirical support for causal models
constructed along familiar stochastic principles (Buthe, 2002). Arguing that historical
analysis should be focused on understanding processes, he notes that the key challenge
for models of history is not to understand policy change which, from the stochastic
perspective, is a given. He argues that narrative methods can help explain the policy
stability that is missed in purely stochastic formulations.
On the one hand, this analysis proposes a solution much like path dependency;
which, in a sense, as mentioned above, can be seen as a synthesis of elements of
10

the stochastic and narrative positivist models. However, Buthe also assumes that
policy processes are driven by actor preferences and that stability emerges from static
preferences – defined as “those interests of an actor that determine how an actor rank-
orders possible outcomes” (p. 484). Since Buthe argues that changes in preferences
arise from learning – that is from changes in ideas and knowledge about preferred
and/or likely outcomes – this approach also moves narrative methods far from their
origins and distinctly in the direction of process sequencing models. Sequences of
events, in Buthe’s formulation, can be thought of as sequences of moments of changes
in ideas and knowledge that alter actor interests and preferences. Thus, the overall
master narrative of policy-making is one of problem-solving and learning, just as it
is in the process-sequencing model.

The contingent model: Path dependency in public policy-making

What is there to choose, then, between the path dependency and process sequencing
models as a preferred alternative to purely stochastic or purely narrative approaches
to policy history? Which, if either, better accords with the empirical record of general
policy dynamics?
First, in evaluating path dependency models it must be noted that many policy stud-
ies claiming to apply path dependent models are relatively unsystematic and may only
be using the term to assert that “history matters” in the development and implemen-
tation of policies (Cox, 2004).3 However, there are examples of the more systematic
application of the general model of path dependency to policy-making. Probably the
most well known are the analyses of welfare state development in the United States
conducted by Pierson and Hacker, among others (Pierson, 2000a; Hacker, 2002). Pier-
son’s version of path dependency uses the concept of increasing returns to explain
why a particular path is taken and ultimately “locked in.” Hacker also defines the
explanatory mechanism of path dependency as increasing returns, although he argues
that whether increasing returns actually occur or not cannot be predicted in advance.
While both authors suggest that two versions – broad (non-inevitable lock-in) path
dependency and narrow (inevitable lock-in) path dependency – are possible, they ul-
timately opt for the narrow version in their studies and argue that without the concept
of sub-optimal lock-in, path dependent analysis simply reverts to the weak causation
characteristic of historical narratology.
Pierson sets out an explicit model of path dependency processes in political life
which is worth discussing in some detail since it is a model often cited if seldom
subjected to detailed analysis.4 His key hypothesis is that since political life is one
involving (1) collective action; (2) institutions; (3) political authority; and (4) com-
plexity, it will generate increasing returns to key players, leading to lock-in and narrow
path dependency. Hacker makes a similar argument, stating that the possibilities of
path dependence processes are great in politics and policy-making since:

first, policy creates or encourages the creation of large scale organizations

with substantial set-up costs; second, a policy directly or indirectly benefits
sizable organized groups or constituencies; third, a policy embodies long-lived
commitments upon which beneficiaries and those around them premise crucial
 11

life and organizational decisions; fourth, the institutions and expectations a

policy creates are of necessity densely interwoven with the broader features
of the economy and society, creating interlocking networks of complementary
institutions; and fifth, features of the environment within which a policy is
formulated and implemented make it harder to recognize or respond to policy
outcomes that are unanticipated or undesired (Hacker, 2002: p. 55).
There are several problems with this formulation, however, which raise questions
about its ability to serve as a superior model of historical explanation in the policy
realm. First, there is the discussion of the collective nature of politics. Here, Pierson
uses the analysis of public goods and pluralist arguments to make his case that larger
groups are more powerful than smaller ones (Olson, 1965; Skocpol et al., 2000), but
ignores the actual reasons for group formation and membership growth (Nownes and
Neeley, 1996; Nownes, 1995; 2000; Nownes and Cigler, 1995) and the impact of
discourses rather than size of groups on policy influence (Burt, 1990). Taking the em-
pirical evidence on these factors into account it is apparent that size of groups, alone, is
not a factor leading to path dependence. Second, Pierson follows Bachrach and Baratz
(1970) in arguing that actors may use power to reinforce their other advantages – that
power asymmetries are an important source of increasing returns. However, as Deeg
(2005) has noted, this does not necessarily lead to lock in if powerful actors wish
change to occur. Third, Pierson argues that learning leads to paradigmatic lock-in.
However, as Buthe (2002) and others have noted, learning can also involve exogenous
lesson-drawing, i.e. bringing new idea into existing subsystems, and undermining
existing paradigms (Dobrowolsky and Saint-Martin, 2005, Kay, 2005). Finally, Pier-
son’s argument that institutional densities are a key factor leading to increasing returns
since “the cost of exit from institutions is very high” is simply asserted with no proof
offered. Although, some studies have found evidence that structures constrain change
(Rayner et al., 2001), other studies have found that institutions can also promote and
facilitate change, so that the one-way dynamics suggested by Pierson are not present
(Gains et al., 2005).
Similarly, the emphasis on contingency in creating a trajectory is assumed but
not systematically developed in most works purporting to apply path dependency
models to policy events (Greener, 2002a; Thelen, 2003). Studies of policy windows,
for example, have found that these only rarely open and close at random (Kingdon,
1995; Howlett, 1998; Keeler, 1993), implying that the path dependency model, while
alluring in its possible blend of narrative and stochastic elements may be, in fact, very
much like those other two models, only relevant in very limited and specific policy
circumstances.

Embedded evolution – process sequencing models in the policy sciences

Probably the best developed empirical evidence supporting a particular model of

historical sequencing in the policy sciences is that supporting the ‘punctuated
equilibrium’ form of the general process sequencing model of policy change (Baum-
gartner and Jones, 2002; True et al., 1999). Much of this evidence has been put forward
by Baumgartner and Jones in their many works on US public policy-making (Jones
et al., 1998; 2003; Jones and Baumgartner, 2005; True, 2000) although other authors
12

have also found evidence of such processes at work in other countries as well (John
and Margetts, 2003).
In such models, “normal” or marginal, incremental, change and atypical policy
dynamics or “paradigmatic” change are linked together to form a particular overall
“stepped” or punctuated equilibrium pattern of policy evolution through broken tra-
jectories. That is, change occurs as an irregular, stepped, function in which relatively
long periods of policy stability are interspersed with infrequent periods of substantial
change. Baumgartner and Jones discovery of leptokurtotic distributions in U.S. fed-
eral government annual budgetary allocations provides strong empirical evidence of
the expected pattern of policy punctuations occurring in this area (Jones et al., 1998;
2003).
In the policy realm, punctuated equilibrium describes a situation whereby normal
policy-making involves fairly common, routine, non-innovative changes at the mar-
gin of existing policies utilizing existing policy processes, institutions, and regimes.
Atypical, paradigmatic or non-incremental change then involves new policies which
represent a sharp break from how policies were developed, conceived, and imple-
mented in the past but are still rooted in the same general concerns and problems
(Berry, 1990; Cox, 1992). Frequently cited examples of such changes include shifts
in fiscal and monetary policy in most western countries from a balanced-budget or-
thodoxy to Keynesian demand-management principles and practices in the 1930s and
1940s and a subsequent shift away from Keynesianism to forms of neoliberalism in
the 1970s and 1980s (Hall, 1989, 1992).
In these kind of process sequencing models, change is seen as occurring in a series
of steps as policy paradigms are constructed and destroyed (Hall, 1990; 1993), a pro-
cess linked by authors such as Sabatier and Baumgartner and Jones, to the construction
and destruction of belief systems in policy subsystems (Sabatier, 1988; Baumgartner
and Jones, 1991; 1993). As theorists have come to a better understanding of how be-
lief systems persist and change, the original analogies that the punctuated equilibrium
model made between the extended periods of incremental policy adjustment and the
Kuhnian idea of “normal science” and, hence, also between punctuations and scientific
revolutions, have been replaced with more grounded accounts of the relations between
value shifts, policy learning, and change (McNamara, 1998; Coleman et al., 1996).
Because of this work, earlier general criticisms of the punctuated equilibrium
model to the effect that, while it assumed long periods of policy stasis, it provided
no way of accounting for critical punctuations, which were explained only through
narrative positivist or post-modernist methods – that is, in retrospect –, are no longer
valid (Schlager, 1999). This is not to say that only this model is appropriate in the
policy sciences, however since instances of extreme persistence, on the one hand, and
contingent change, on the other, even though rare at the systemic level, do exist and are
amenable to treatment using narrative, stochastic and, sometimes, path dependency
methods. Moreover, a significant challenge to punctuated equilibrium explanations
noted by some of its original exponents is the relation to different levels or orders
of change (True, 2000). That is, changes can occur both at the systemic and sub-
systemic level and between policy elements versus overall policy areas or fields,
leading to some consequent difficulties in discriminating the scale and significance
of a change at different levels of policy-making (Cashore and Howlett, 2006).
 13

As True (2000) and Cashore and Howlett (2006) have noted, punctuated equi-
librium theory suggests that some policy punctuation is always happening in some
subsystem or other, or within some element of a policy regime, but that at the sys-
temic or regime level the overall picture may remain one of stability, posing a continual
problem for punctuated equilibrium notions of policy dynamics. That is, extreme per-
sistence and contingent change are more likely to be found at the sub-systemic and
sub-regime levels rather than the systemic or regime ones and detailed explanations
of change at the subsystem or sub-regime level are correspondingly more likely to
need other kinds of historical explanation than process sequencing. Explanations of
stability and change in particular issue areas within policy subsystems and among
the component parts of a policy regime thus require a careful matching of modes of
historical explanation to empirical circumstances on the part of the analyst.

Conclusion: Reconciling multiple, competing models of historical

sequencing in the policy sciences

At the present time the policy literature has taken a definite historical turn, with
the stochastic models favored by many analysts searching for causal determinants in
the early years of the policy sciences largely falling by the wayside as contemporary
analysts grapple with the dynamics and the phenomena of policy change and stability.
While the stranglehold on analysis of the stochastic model may have been broken,
however, the contours of the best alternative model have not been clarified. Although
there has been some thought given to systematizing the older tradition of thick descrip-
tions of policy processes through the development of the postmodern and positivist
streams of policy narratology, the most interesting new work has a hybrid character
that incorporates narrative and stochastic methods in the course of developing models
such as path dependency and process sequencing.
Most attention, hitherto, has been paid to the potential for path dependency to syn-
thesize elements of both the narrative and stochastic models into a powerful “mixed”
model of historical processes. However, at present, in the policy sciences, “path depen-
dence” remains a much used, and abused, model of historical sequencing. Although
it has been applied to such diverse cases of policy-making and political re-structuring
as European, Danish, and post-Soviet regime transitions (Holzinger and Knill, 2002;
Nee and Cao, 1999; Rona-Tas, 1998; Torfing, 2001) and environmental, industrial and
health policy-making (Kline, 2001; Rahnema and Howlett, 2002; Bevan and Robin-
son, 2005; Courchene, 1993; Wilsford, 1994), most of the works which employ it in
the policy sphere have tended to apply it unsystematically, or to somewhat uncriti-
cally accept analogies from the economics literature where it developed. As has been
shown above, even the work of its most prominent exponents in the policy sciences,
while distinguishable from the accounts provided in the economics literature, rests on
many unsubstantiated theoretical assertions and incorrect empirics.
As the discussion above has shown, process sequencing is, in fact, a more promis-
ing synthetic approach, incorporating elements of the path dependent, narrative and
stochastic models in a policy-relevant way with an impressive array of empirical
studies behind it. In its punctuated equilibrium formulation, process sequencing has
received a great deal of empirical support to back its candidacy for being a superior
14

general model of the most common type of policy dynamics. There is an increasing
weight of evidence from cases studies of process sequencing in policy studies; that
is, of trajectories changing while being embedded in previous policy legacies so that
their new form is not random or contingent, but thoroughly embedded in the old. The
challenge for the theorist remains, however, to correctly identify the circumstances in
which “deep embeddedness” lends itself to explication through the methods of nar-
rative, and those where disembedding requires that the narrative be supplemented or
even supplanted by alternative methods such as path dependency and process sequenc-
ing. Empirical observation suggests that both sets of circumstances will be present at
different times and at different levels of policy specificity as policies change. More
work remains to be done to determine the best match of model to empirical evidence
at the sub-systemic and sub-regime levels. However, at present, at a general level, the
kind of hybrid model represented by process sequencing, one that draws creatively
on the methods of the literary critic or the economist without seeking to imitate them
in a slavish way, has the most potential to provide an historically-sensitive alternative
transcending the ahistorical “general linear” model in the study of policy-making over
time.

Notes

1. Much of this work appeared in the American Journal of Sociology. For a selection of reprinted articles
by a major figure in this discussion see Abbott, 2001.
2. A good example of such analyses is the picture of leadership attributes of the American president put
forward in Robert Caro’s (2002) monumental study of the career and administration of Lyndon Baines
Johnson. At certain points in his career LBJ’s aggressive leadership style was said to greatly aid the
successful adoption of certain policies while at others it is argued to contribute to their failure or had
no discernable impact at all upon policy outcomes.
3. For a critique of such studies see Dobrowolsky and Saint-Martin, 2005.
4. For detailed critiques see Deeg, 2001 and Kay, 2005.

References

Abbott, A. (1988). ‘Transcending general linear reality,’ Sociological Theory 6: 169–186.

Abbott, A. (1990). ‘Conceptions of time and events in social science methods,’ Historical Methods 23:
140–151.
Abbott, A. (1992). ‘From causes to events: Notes on narrative positivism,’ Sociological Methods and
Research 20: 428–455.
Abbott, A. (1997). ‘On the concept of turning point,’ Comparative Social Research 16: 85–105.
Abbott, A. (2001). Time Matters: On Theory and Method. Chicago: University of Chicago Press.
Adorno, T. W., et al. (1976). The Positivist Dispute in German Sociology. London: Heinemann.
Almond, G. A. and S. J. Genco (1997). ‘Clouds, Clocks and the Study of Politics,’ World Politics 29:
489–522.
Aminzade, R. (1992). ‘Historical sociology and time,’ Sociological Methods and Research 20: 456–480.
Arthur, W. B. (1988). ‘Self-Reinforcing – Mechanisms in Economics,’ in P. W. Anderson, K. J. Arrow and
D. Pines, eds., The Economy as an Evolving Complex System. Redwood City: Addison Wesley.
Arthur, W. B. (1989). ‘Competing Technologies, Increasing Returns, and Lock-In by Historical Events,’
The Economic Journal 99: 116–131.
Bachrach, P. and M. S. Baratz (1970). Power and poverty: Theory and Practice. New York, Oxford
University Press.
 15

Baumgartner, F. R. and B. D. Jones (1991). ‘Agenda Dynamics and Policy Subsystems,’ Journal of Politics
53: 1044–1074.
Baumgartner, F. R. and B. D. Jones (1993). Agendas and Instability in American Politics. Chicago: Uni-
versity of Chicago Press.
Baumgartner, F. R. and B. D. Jones (2002). ‘Positive and negative feedback in politics,’ in F. R. Baumgartner
and B. D. Jones, eds., Policy Dynamics. Chicago: University of Chicago Press.
Bernstein, S., et al. (2000). ‘God gave physics the easy problems: Adapting social science to an unpredictable
world,’ European Journal of International Relations 6: 43–76.
Berry, W. T. (1990). ‘The confusing case of budgetary incrementalism: Too many meanings for a single
concept,’ Journal of Politics 52: 167–196.
Bevan, G. and R. Robinson (2005). ‘The interplay between economic and political logics: Path dependency
in health care in England,’ Journal of Health Politics, Policy and Law 30: 53–78.
Burt, S. (1990). ‘Canadian women’s groups in the 1980s: Organizational development and policy influence,’
Canadian Public Policy 16: 17–28.
Buthe, T. (2002). ‘Taking temporarility seriously: Modeling history and the use of narratives as evidence,’
American Political Science Review 96: 481–493.
Calhoun, C. (1998). ‘Explanation in historical sociology: Narrative, general theory, and historically specific
theory,’ American Journal of Sociology 104: 846–871.
Caro, R. A. (2002). Master of the Senate: The Years of Lyndon Johnson. New York: Alfred A. Knopf.
Cashore, B. and M. Howlett (2006). ‘Behavioural Thresholds and Institutional Rigidities As Explanations
of Punctuated Equilibrium Processes in Pacific Northwest Forest Policy Dynamics,’ in R. Repetto, eds.,
Punctuated Equilibrium and the Dynamics of U.S. Environmental Policy. New Haven: Yale University
Press.
Clemens, E. S. (1999). ‘Continuity and Coherence: Periodization and the Problem of Institutional Change,’
in F. Engelstad and R. Kalleberg, eds., Social Time and Social Change. Oslo: Scandinavian University
Press.
Coleman, W. D., G. D. Skogstad, and M. Atkinson. (1996). ‘Paradigm shifts and policy networks: Cumu-
lative change in agriculture,’ Journal of Public Policy 16: 273–302.
Courchene, T. J. (1993). ‘Path-Dependency, Positive Feedback and Paradigm Warp: A Schumpeterian
Approach to the Social Order,’ in E. B. Reynolds, eds. Income Security in Canada: Changing Needs,
Changing Means. Montreal: Institute for Research on Public Policy, pp. 43–82.
Cox, R. H. (1992). ‘Can welfare states grow in leaps and bounds? Non-incremental policymaking in the
Netherlands,’ Governance 5: 68–87.
Cox, R. H. (2004). ‘The path-dependency of an idea: Why Scandinavian welfare states remain distinct,’
Social Policy and Administration 38: 204–219.
Czarniawska, B. (2004a). ‘The Uses of Narrative in Social Science Research,’ in Melissa Hardy and Alan
Bryman, eds. Handbook of Data Analysis. London: Sage Publications, pp. 649–666.
Czarniawska, B. (1998). A Narrative Approach to Organization Studies. London: Sage Publications.
Czarniawska, B. (2002). A Tale of Three Cities: On the Glocalization of City Management. Oxford: Oxford
University Press.
Czarniawska, B. (2004b). Narratives in Social Science Research. London: Sage Publications.
Czarniawska, B. (2004c). ‘On time, space, and action nets,’ Organization 11: 773–791.
David, P. A. (1985). ‘Clio and the economics of QWERTY,’ The American Economic Review 75: 332–
337.
David, P. A. (1986). ‘Understanding the Economics of QWERTY: The necessity of history,’ in W. N. Parker,
ed. Economic History and the Modern Economist. London, Basil Blackwell, pp. 30–49.
Deeg, R. (2005). ‘Path Dependency, Institutional Complementarity and Change in National Business
Systems,’ in G. Morgan, R. Whitley and E. Moen, eds., Changing Capitalisms? Internationalization,
Institutional Change, and Systems of Economic Organization. Oxford: Oxford University Press, pp.
21–52.
Deeg, R. (2001). ‘Institutional Change and the Uses and Limits of Path Dependency: The Case of German
Finance,’ Koln: Max Planck Institute fur Gesellschaftsforschung MPIfG, Discussion Paper January 6,
2001.
Dobrowolsky, A. and D. Saint-Martin (2005). ‘Agency, actors and change in a child-focused future: ‘Path
dependency’ problematised,’ Commonwealth and Comparative Politics 43: 1–33.
16

Edelman, M. J. (1964). The Symbolic Uses of Politics. Chicago: University of Illinois Press.
Edelman, M. J. (1988). Constructing the Political Spectacle. Chicago: University of Chicago Press.
Eldredge, N. and S. J. Gould (1972). ‘Punctuated Equilibria: An Alternative to Phyletic Gradualism,’ in T.
J. M. Schopf, eds. Paleobiology. San Francisco: Freeman, Cooper, pp. 82–115.
Fischer, F. (2003). Reframing Public Policy: Discursive Politics and Deliberative Practices. Oxford: Oxford
University Press.
Foucault, M. (1972). ‘The Discourse on Language,’ in: The Archaeology of Knowledge New York: Pantheon,
pp. 215–235.
Gadamer, H. (1989). Truth and Method. New York: Crossroad.
Gains, F., P. C. John, and G. Stoker (2005). ‘Path dependency and the reform of English local government,’
Public Administration 83: 25–45.
George, A. and A. Bennett (2005). ‘Process Tracing and Historical Explanation,’ in A. George and A.
Bennett, eds. Case Studies and Theory Development in the Social Sciences. Cambridge: MIT Press.
Gersick, C. J. (1991). ‘Revolutionary change theories: A multilevel exploration of the punctuated equilib-
rium paradigm,’ Academy of Management Review 16: 10–36.
Gould, S. J. and N. Eldredge (1977). ‘Punctuated equilibria: The tempo and mode of evolution reconsidered,’
Paleobiology 3: 115–151.
Greener, I. (2002a). ‘Theorizing path-dependency: How does history come to matter in organizations?’
Management Decision 40: 614–619.
Greener, I. (2002b). ‘Understanding NHS reform: The policy-transfer, social learning and path dependency
perspectives,’ Governance 15: 161–183.
Greener, I. (2005). ‘The potential of path dependence in political studies,’ Politics 25: 62–72.
Griffin, L. J. (1992). ‘Temporality, events and explanation in historical sociology,’ Sociological Methods
and Research 20: 403–427.
Griffin, L. J. (1993). ‘Narrative, event-structure analysis, and causal interpretation in historical sociology,’
American Journal of Sociology 98: 1094–1133.
Hacker, J. S. (2002). The Divided Welfare State. New York: Cambridge University Press.
Hall, P. A., ed. (1989). The Political Power of Economic Ideas: Keynesianism Across Nations. Princeton:
Princeton University Press.
Hall, P. A. (1990). ‘Policy Paradigms, Experts, and the State: The Case of Macroeconomic Policy-Making
in Britain,’ in S. Brooks and A. G. Gagnon, eds. Social Scientists, Policy, and the State. New York:
Praeger, pp. 53–78.
Hall, P. A. (1992). ‘The Change from Keynesianism to Monetarism: Institutional Analysis and British
Economic Policy in the 1970s,’ in S. Steinmo, K. Thelen and F. Longstreth, eds. Structuring Pol-
itics: Historical Institutionalism in Comparative Analysis. Cambridge: Cambridge University Press,
pp. 90–114.
Hall, P. A. (1993). ‘Policy paradigms, social learning and the state: The case of economic policy making
in Britain,’ Comparative Politics 25: 275–296.
Hall, P. A. and R. C. R. Taylor (1996). ‘Political science and the three new institutionalisms,’ Political
Studies 44: 936–957.
Haydu, J. (1998). ‘Making use of the past: Time periods as cases to compare and as sequences of problem
solving,’ American Journal of Sociology 104: 339–371.
Holzinger, K. and C. Knill (2002). ‘Path dependencies in European integration,’ Pubic Administration 80:
125–152.
Howlett, M. (1998). ‘Predictable and unpredictable policy windows: Issue, institutional and exoge-
nous correlates of Canadian federal agenda-setting,’ Canadian Journal of Political Science 31: 495–
524.
Ihde, D. (1974). The Conflict of Interpretations: Essays in Hermeneutics. Evanston: Northwestern
University Press.
John, P. and H. Margetts (2003). ‘Policy punctuations in the UK: Fluctuations and equilibria in central
government expenditure since 1951,’ Public Administration 81: 411–432.
Jones, B. D., F. R. Baumgartner, and J. L. True (1998). ‘Policy punctuations: U.S. budget authority, 1947–
1995,’ The Journal of Economic Literature 60: 1–33.
Jones, B. D. and F. R. Baumgartner (2005). ‘A model of choice for public policy,’ Journal of Public
Administration Research and Theory 15: 325–351.
 17

Jones, B. D., T. Sulkin, and H. A. Larsen (2003). ‘Policy punctuations in American political institutions,’
American Political Science Review 97: 151–169.
Kato, J (1996). ‘Review article: Institutions and rationality in politics – three varieties of neo-
institutionalists,’ British Journal of Political Science 26: 553–582.
Kay, A (2005). ‘A critique of the use of path dependency in policy studies,’ Public Administration 83:
553–571.
Keeler, J. T. S. (1993). ‘Opening the window for reform: Mandates, crises and extraordinary policy-making,’
Comparative Political Studies 25: 433–486.
Kingdon, J. W. (1995). Agendas, Alternatives and Public Policies. Boston: HarperCollins College
Publishers.
Kline, D. (2001). ‘Positive feedback, lock-in and environmental policy,’ Policy Sciences 34: 95–107.
Law, J. (2000). ‘On the subject of the object: Narrative, technology, and interpellation,’ Configurations 8:
1–29.
Lieberman, E. S. (2001). ‘Causal inference in historical institutional analysis: A specification of periodiza-
tion strategies,’ Comparative Political Studies 34: 1011–1035.
Lieberman, R. C. (2002). ‘Ideas, institutions and political order: Explaining political change,’ American
Political Science Review 96: 697–712.
Liebowitz, S. J. and S. E. Margolis (1990). ‘The fable of the keys,’ Journal of Law and Economics 33:
1–25.
Liebowitz, S. J. and S. E. Margolis (1995). ‘Policy and path dependence: From QWERTY to Windows 95,’
Regulation 18: 33–41.
Lindner, J. and B. Rittberger (2003). ‘The creation, interpretation and contestation of institutions – revisiting
historical institutionalism,’ Journal of Common Market Studies 41: 445–473.
Lindner, J. (2003). ‘Institutional stability and change: Two sides of the same coin,’ Journal of European
Public Policy 10: 912–935.
List, C. (2004). ‘A model of path-dependence in decisions over multiple propositions,’ American Political
Science Review 98: 495–513.
Lustick, I. S. (1996). ‘History, historiography, and political science: Multiple historical records and the
problem of selection bias,’ American Political Science Review 90: 605–618.
Lyotard, J. (1984). The Postmodern Condition: A Report on Knowledge. Minneapolis: University of Min-
nesota Press.
Mahoney, J. (2000). ‘Path dependence in historical sociology,’ Theory and Society 29: 507–548.
Mahoney, J. and D. Rueschemeyer, ed. (2003). Comparative Historical Analysis in the Social Sciences.
Cambridge: Cambridge University Press.
McNamara, K. R. (1998). The Currency of Ideas: Monetary Politics in the European Union. Ithaca: Cornell
University Press.
Morgan, G. and I. Kubo (2005). ‘Beyond path dependency? Constructing new models for institutional
change: The case of capital markets in Japan,’ Socio-Economic Review 3: 55–82.
Nee, V. and Y. Cao (1999). ‘Path dependent societal transformation: Stratification in hybrid mixed
economies,’ Theory and Society 28: 799–834.
Nownes, A. J. (1995). ‘The other exchange: Public interest groups, patrons, and benefits,’ Social Science
Quarterly 76: 381–401.
Nownes, A. J. (2000). ‘Policy conflict and the structure of interest communities,’ American Politics Quar-
terly 28: 309–327.
Nownes, A. J. and A. J. Cigler (1995). ‘Public interest groups and the road to survival,’ Polity 27: 380–
404.
Nownes, A. J. and G. Neeley (1996). ‘Toward an explanation for public interest group formation and
proliferation: “Seed Money,” disturbances, entrepreneurship, and patronage,’ Policy Studies Journal 24:
74–92.
Olson, M. (1965). The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge,
Harvard University Press.
Ospina, S. M. and J. Dodge (2005). ‘It’s about time: Catching method up to meaning – the usefulness of
narrative inquiry in public administration research,’ Public Administration Review 65: 143–157.
Pierson, P. (2000a). ‘Increasing returns, path dependence, and the study of politics,’ American Political
Science Review 94: 251–267.
18

Pierson, P. (2000b). ‘Not just what, but when: Timing and sequence in political processes,’ Studies in
American Political Development 14: 72–92.
Pierson, P. (2000c). ‘The limits of design: Explaining institutional origins and change,’ Governance 13:
475–499.
Pierson, P. (2004). Politics in Time: History, Institutions, and Social Analysis. Princeton: Princeton Uni-
versity Press.
Raadschelders, J. C. N. (1998). ‘Evolution, institutional analysis and path dependency: An administrative-
history perspective on fashionable approaches and concepts,’ International Review of Administrative
Sciences 64: 565–582.
Rahnema, S. and M. Howlett (2002). ‘Impediments to industrial policy: Overcoming path dependency in
Canada’s post staples transition,’ Journal of Australian Political Economy 49: 114–135.
Rayner, J. and M. Howlett, et al. (2001). ‘Privileging the sub-sector: Critical sub-sectors and sectoral
relationships in forest policy-making,’ Forest Policy and Economics 2: 319–332.
Rico, A. and J. Costa-Font (2005). ‘Power rather than path dependency? The dynamics of institutional
change under health care federalism,’ Journal of Health Politics, Policy and Law 30: 231–252.
Rona-Tas, A. (1998). ‘Path dependence and capital theory: Sociology of the post-communist economic
transformation,’ East European Politics and Societies 12: 107–131.
Sabatier, P. A. (1988). ‘An advocacy coalition framework of policy change and the role of policy-oriented
learning therein,’ Policy Sciences 21: 129–160.
Saldana, J. (2003). Longitudinal qualitative research: analyzing change through time. New York: Altamira,
2003.
Schlager, E. (1999). ‘A Comparison of Frameworks, Theories, and Models of Policy Processes,’ in P.
Sabatier ed. Theories of the Policy Process. Boulder: Westview Press, p.252.
Schön, D. and M. Rein (1994). Frame Reflection: Toward the Resolution of Intractable Controversies. New
York: Basic Books.
Schwartz, H. (no date). ‘Down the Wrong Path: Path Dependence, Increasing Returns, and Historical
Institutionalism,’ Available at http://www.people.virginia.edu/∼hms2f/Path.pdf
Skocpol, T., M. Ganz, et al. (2000). ‘A nation of organizers: The institutional origins of civic voluntarism
in the United States,’ American Political Science Review 94: 527–546.
Somers, M. R. (1996). ‘Where is Sociology after the Historic Turn? Knowledge Cultures, Narrativity,
and Historical Epistemologies,’ in T. J. McDonald, eds. The Historic Turn in the Human Sciences. Ann
Arbor: The University of Michigan Press, pp. 53–89.
Steinmo, S., K. Thelen, and F. Longstreth, ed. (1992). Structuring Politics: Historical Institutionalism in
Comparative Analysis. Cambridge: Cambridge University Press.
Stinchcombe, A. (1968). Constructing Social Theories. New York: Harcourt, Brace, and World.
Thelen, K. (2003). ‘How Institutions Evolve: Insights from Comparative Historical Analysis,’ in J. Ma-
honey and D. Rueschemeyer, eds. Comparative Historical Analysis in the Social Sciences. Cambridge:
Cambridge University Press, pp. 208–240.
Thompson, W. B. (2001). ‘Policy making through thick and thin: Thick description as a methodology for
communications and democracy,’ Policy Sciences 34: 63–77.
Torfing, J. (2001). ‘Path-dependent Danish welfare reforms: The contribution of the new institutionalisms
to understanding evolutionary change,’ Scandinavian Political Studies 24: 277–309.
Traweek, S. (1992). ‘Border Crossings: Narrative Strategies in Science Studies and among Physicists
in Tsukuba Science City, Japan,’ in Andrew Pickering ed. Science as Practice and Culture. Chicago:
University of Chicago Press, pp. 429–466.
True, J. L. (2000). ‘Avalanches and incrementalism: Making policy and budgets in the United States,’
American Review of Public Administration 30: 3–18.
True, J. L., B. D. Jones, and F. R. Baumgartner (1999). ‘Punctuated-Equilibrium Theory: Explaining
Stability and Change in American Policymaking,’ in P. A. Sabatier, ed. Theories of the Policy Process.
Boulder: Westview Press, pp. 97–115.
Wilsford, D. (1994). ‘Path dependency, or why history makes it difficult but not impossible to reform health
care systems in a big way,’ Journal of Public Policy 14: 251–284.
Yanow, D. (1996). How Does a Policy Mean? Interpreting Policy and Organizational Actions. Washington
DC: Georgetown University Press.
Yanow, D. (1999). Conducting Interpretive Policy Analysis. Thousand Oaks: Sage.