Cornell University ILR School

DigitalCommons@ILR
Articles and Chapters

ILR Collection

3-1-1983

Institutional Sources of Change in the Formal

Structure of Organizations: The Diffusion of Civil
Service Reform, 1880-1935
Pamela S. Tolbert
Cornell University, [email protected]

Lynne G. Zucker
University of California, Los Angeles

Follow this and additional works at: http://digitalcommons.ilr.cornell.edu/articles

Part of the Civil Engineering Commons
Thank you for downloading an article from DigitalCommons@ILR.
Support this valuable resource today!
This Article is brought to you for free and open access by the ILR Collection at DigitalCommons@ILR. It has been accepted for inclusion in Articles
and Chapters by an authorized administrator of DigitalCommons@ILR. For more information, please contact [email protected].

Institutional Sources of Change in the Formal Structure of Organizations:

The Diffusion of Civil Service Reform, 1880-1935
Abstract

This paper investigates the diffusion and institutionalization of change in formal organization structure, using
data on the adoption of civil service reform by cities. It is shown that when civil service procedures are
required by the state, they diffuse rapidly and directly from the state to each city. When the procedures are not
so legitimated, they diffuse gradually and the underlying sources of adoption change overtime. In the latter
case, early adoption of civil service by cities is related to internal organizational requirements, with city
characteristics predicting adoption, while late adoption is related to institutional definitions of legitimate
structural form, so that city characteristics no longer predict the adoption decision. Overall, the findings
provide strong support for the argument that the adoption of a policy or program by an organization is
importantly determined by the extent to which the measure is institutionalized whether by law or by
gradual legitimation.
Keywords

civil service reform, institutional change, organizational change

Disciplines

Civil Engineering
Comments

Suggested Citation
Tolbert, P. S., & Zucker, L. G. (1983). Institutional sources of change in the formal structure of organizations:
The diffusion of civil service reform, 1880-1935 [Electronic version]. Administrative Science Quarterly, 28,
22-39.
http://digitalcommons.ilr.cornell.edu/articles/131/
Required Publisher Statement
Reprinted from Institutional Sources of Change in the Formal Structure of Organizations: The Diffusion of
Civil Service Reform, 1880-1935 by Pamela S. Tolbert and Lynne G. Zucker, published in Administrative
Science Quarterly Volume 28 Issue 1 by permission of Administrative Science Quarterly. 1983 by Cornell
University.

This article is available at DigitalCommons@ILR: http://digitalcommons.ilr.cornell.edu/articles/131

Institutional Sources of
Change in the Formal
Structure of Organizations: The Diffusion of
Civil Service Reform,
1880-1935
Pamela S. Tolbert and
Lynne G. Zucker

This paper investigates the diffusion and institutionalization of change in formal organization structure, using data
on the adoption of civil service reform by cities. It is shown
that when civil service procedures are required by the state,
they diffuse rapidly and directly from the state to each city.
When the procedures are not so legitimated, they diffuse
gradually and the underlying sources of adoption change
overtime. In the latter case, early adoption of civil service by
cities is related to internal organizational requirements, with
city characteristics predicting adoption, while late adoption
is related to institutional definitions of legitimate structural
form, so that city characteristics no longer predict the
adoption decision. Overall, the findings provide strong
support for the argument that the adoption of a policy or
program by an organization is importantly determined by
the extent to which the measure is institutionalized
whether by law or by gradual legitimation.*
Explanations of formal structure in organizations are as divergent as the current approaches to organization theory. Two
approaches in particular have generated strong debate: one
views organizations as rational actors, albeit in a complex
environment (Thompson, 1967; Blau and Schoenherr, 1971),
while the other views organizations as captives of the institutional environment in which they exist (Meyer and Rowan,
1977; Zucker, 1982, 1983). Both approaches have important
implications for the processes underlying diffusion of an innovation in the formal structure of organizations, the first pointing
to the need for effectiveness or efficiency that may follow
adoption, the latter pointing to the need for legitimacy of the
organization in the wider social structure. Though these two
approaches are not necessarily incompatible, since organizations may adopt innovations for different reasons, they are
seldom both investigated in the same empirical study.

1983 by Cornell University

0001 -8392/83/2801-0022/S00.75
The authors are jointly responsible for the
theoreticalargumentand analysis. M. Craig
Brown suggested the topic of civil service
reform. Maureen J. McConaghy, Nancy
Brandon Tuma, Glenn R. Carroll, and P. Y.
Liu provided methodological advice;
Sharon Stevens aided early computational
work. Both of us are grateful to Phillip
Bonacich for his advice throughout the
research, and to Marshall W. Meyer, John
W. Meyer, William G. Roy, Herman Turk,
Richard A. Berk, David McFarland, Oscar
Grusky, and Jeffrey Pfefferfortheir helpful
comments on an earlier draft.

Here, one important point of convergence is explored by using

both perspectives to explain the adoption of civil service
procedures by municipal governments from 1880 to 1935.
Adoption of formal structure can be investigated in diverse
organizational contexts; here we examine the process in cities,
often used as the context for such research (e.g., Schnore and
Alford, 1963; Knoke, 1982). Early adoption of civil service
reform before 1915 appears to reflect efforts to resolve
specific problems confronting municipal administrations, while
later adoption is rooted instead in the growing legitimacy of civil
service procedures, with the diffusion of societal norms serving
to define local structure (Parsons, 1951). We first turn to a brief
history of civil service reform, leading to a discussion of the
institutionalization of reform. We then examine basic assumptions in the organization literature about the sources of change
in formal structure to establish the basis for the analysis of
adoption patterns of civil service procedures.
HISTORY OF MUNICIPAL CIVIL SERVICE
Civil service reform represents one of the earliest attempts to
"rationalize" local administration by instituting a system of
written examinations for municipal appointees and by insulating
administrative personnel from political influence through tenure (White, 1949; Griffith, 1974). This entailed legally invest22/Administrative Science Quarterly, 28(1983): 2 2 - 3 9
Copyright 1983. All rights reserved.

Diffusion of Civil Service

ing responsibility for personnel appointments in a central

agency or commission. For the most part, civil service procedures were not required by law or other regulation from the
wider environment. The only piece of national civil service
legislation during this period, the Pendleton Act of 1882, dealt
exclusively with federal government organizations and did not
mention local or state government (Thelen, 1972). It was also a
relatively "weak law that effectively allowed each administration to classify public offices as it chose" (Wiebe, 1967: 61).
Only three states New York, Massachusetts, and Ohio
adopted statewide measures for civil service reform during the
time period considered here. Most city governments were not
required to adopt civil service reform because they were
relatively autonomous of higher level organizations, state or
federal (Griffith, 1974; Gelfand, 1975).
Why, then, didcitiesadoptcivilserviceprocedures? Depending
on the particular historian consulted, there are a number of
different answers to this question. The most common answer
identifies both the rampant corruption in the political machines
common during this period and dissatisfaction with governmental performance, especially in service delivery (Wiebe,
1967: 4-5): "Corrupt bargains, crude force, and extralegal
expedients had become the new standard.... The inability of
city government to provide even minimum services.. .added its
measure to the chaos." Other historians have stressed the role
of political cleavage, in which dominant social groups maintained, or won, their position (Hays, 1964; Weinstein, 1968). In
some cases, civil service reform provided the basis on which
these groups could reassert their dominance over immigrantrun machines (Hofstadter, 1956; Wolfingerand Field, 1966).
With growth of the Progressive Movement, and its emphasis
on scientific management (Griffith, 1974: 15), came a basis for
government reform. Basically, the reform movement attempted to change the conception of the city from that of a
political body to that of a business corporation, with the city "a
joint stock affair in which the taxpayers are the stockholders"
(Clinton, 1886; Crandon, 1886-1887: 524). Reformers engaged
in a series of highly publicized struggles to promote municipal
reforms in almost every major city (Wiebe, 1967: 168). The first
city passed legislation requiring civil service in 1884; by 1935,
over 450 cities across the United States had enacted some type
of civil service legislation (Van Riper, 1951). Thus, by 1935 the
transformation of city government from a politically based
system to a bureaucratically based system was well underway
(Hays, 1972: 9).
INSTITUTIONALIZATION OF REFORM
Civil service reform thus took place during a period of general
ferment over the role and shape of government organizations.
Some of the reforms proposed during this period were adopted
by very few governments; other reforms, such as the commission form of government, were widely adopted but were rapidly
eclipsed by new innovations. Such reforms are more typical of
those previously investigated in diffusion studies; these traditional explanations rest largely on investigation of innovations
that failed to be institutionalized. Two principal characteristics
that indicate the relatively high degree of institutionalization of
civil service reform set it apart from the other reforms proposed
23/ASQ, March 1983
Copyright 1983. All rights reserved.

and implemented during this period: (1) its most rapid spread
occurred after the initial ferment subsided, indicating that it was
taken for granted, and (2) it is the most permanent and
widespread of all reforms accomplished during this period.
Both of these characteristics are independent of the major
dependent variable, adoption, which we used in our analysis.
Turning to the first point, the most acrimonious debates over
civil service procedures occurred before 1900; after 1910 the
procedures were generally discussed without much conflict,
and by 1920 civil service procedures were accepted as the
properwaytoconductcitybusiness(Schiesl, 1977: 187).Table
1 summarizes data coded from the major publication of the
National Municipal League, a federation of local organizations
active in the promotion of civil service reform. A content
analysis of the proceedings of the League's annual meeting and
of papers delivered at each meeting was carried out for five
years -1894,1900,1905,1909, and 1915. These records were
coded in two ways. First, the total number of times a topic
related to civil service reform was mentioned was counted in
each yearly report. Such mentions were divided into two
categories: criticisms of existing arrangements (e.g., references to spoils, machines, or political patronage), and procedures forimplementing the reform (e.g., merit systems, promotion systems, and performance evaluation). Second, the total
amount of space devoted to discussing any of the topics was
measured, yielding a summary number of pages for each year.
In order to control for the length of each report, which varied
considerably from year to year, each summary measure was
divided by the total number of pages in the report for that year.
Table 1
Content Analysis of Civil Service Discussion over Time, 1894-1915*
Publication
date
1894
1900
1905
1909
1915

Stimulus for
reformt
Mean (S.D.)
5.77
9.87
2.60
2.07
.20

(1.08)
(2.99)
(1.45)
(0.67)
(0.17)

Implementation*
Mean (S.D.)
1.43
3.87
2.77
1.97
.60

(2.23)
(1.70)
(2.80)
(2.00)
(0.60)

Space devoted to
civil service reform
.041
.060
.020
.016
.018

See Appendix for sources. All data taken as a proportion of the total number of
pages.
tCoded as mentions of spoils system, patronage, machine politics.
Coded asi mentions of merit system, promotion system, performance evaluation.

Exhortations concerning the evils of the former governmental

systems, providing the stimulus for reform, showed a dramatic
decrease between 1900 and 1905. Discussion of the more
technical aspects of implementation (merit systems, promotion, and performance evaluation) followed a similar pattern,
though neither the rise from 1894 to 1900 nor the decline from
1900 to 1905 was as sharp. The proportion of pages devoted to
general discussion of civil service reform showed a similar,
though less marked, decline after 1900. These data indicate,
then, that direct agitation for civil service adoption peaked in
1900 and declined considerably thereafter. Since the rapid rise
24/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

in adoption occurred after 1915, it appears to have occurred in

the context of the recognized need for new procedures, rather
than as a result of extensive direct pressure (or of regional
diffusion, which had only a weak effect). 1 In general, then, the
reduction in the amount of debate and discussion of the
stimulus and implementation of the reforms in the League's
proceedings indicates that the reforms had increasing legitimacy and were more taken for granted.
Turning now to the question of permanence, regardless of
whether the procedures were transmitted through state regulation or gradual adoption by municipal governments, civil
service reform caused changes in city government organization
that became enduring elements of structure. Cities seldom
discarded civil service once it was adopted, and federal and
state legislation after 1935 increasingly required the use of local
civil service procedures. The rapid institutionalization of the
reform rested on the assumed isomorphism between it and the
ideal rational bureaucratic form (Zucker, 1983). Government
organizations, increasingly oriented to service delivery, were
modeled after the business corporation, where personnel
selection and promotion were presumably based on merit, not
familial or other personal ties. Other municipal reforms promoted during this period, such as the city manager form of
government, nonpartisan ballots, and city-wide elections, also
reflected the drive to "rationalize" municipal administration.
A hundred years after their initial introduction, civil service
structure and procedures are nearly universal. Hence, one of
the key defining elements of institutions "establishment of
relative permanence of a distinctly social sort" (Hughes, 1936:
180) is present in the case of the civil service. Local
government structure became clearly patterned by the wider
culture over time; civil service procedures became ubiquitous.
CHANGE IN THE FORMAL STRUCTURE OF
ORGANIZATIONS

Of course, not all variables related to adoption can be explored here. We focus on
those identified by historians of this period
as most significant. Unlike most other
studies of reform processes (e.g., Knoke,
1982), regional differences were nonsignificant except in the case of the South
which, as expected from prior research,
lagged behind the other regions. Tables
showing this data are available from
Pamela S. Tolbert. Although the southern
states were tardy in adopting civil service
reform, excluding them from the analysis
below had virtually no impact on the results; they were therefore included.

Institutionalization refers to the process through which components of formal structure become widely accepted, as both
appropriate and necessary, and serve to legitimate organizations. Most fundamentally, the process is one of social change.
This process may occur in different ways (Hemes, 1976): (1)
initial endogenous change may take place when the process is
gradual and not required and/or (2) exogenous change may take
place later in the process or when the process is required. That
the different processes of change are not incompatible can be
seen in their mutual influences over the course of civil service
reform. Before examining this in more detail, some general
perspectives on sources of organizational structure need to be
considered.
For the most part, organizational theorists have analyzed formal
structure as if it were static, focusing on its sources at one point
in time. Radically different views of these sources have
emerged. In one view, formal structure arises from internal
sources, either directly (Scott, 1975) through problems of
coordination and control (e.g., Anderson and Warkov, 1961;
Woodward, 1965; Blau, 1970) or indirectly (Aldrich and Pfeffer,
1976) through power, leadership, and socialization to specific
organizational roles, often mediating environmental effects
25/ASQ, March 1983
Copyright 1983. All rights reserved.

(e.g., Child, 1972; Thornton and Nardi, 1975; Pfefferand

Salancik, 1978). From the other viewpoint, formal structure
arises from external sources, from the direct effects of the
institutional environment. In order to survive, organizations
conform to what is societally defined as appropriate and
efficient, largely disregarding the actual impact on organizational performance (Meyer and Rowan, 1977: 353; Zucker,
1982).
These perspectives are not incompatible, but rather point to
conditions under which changes in the formal structure of
organizations will derive from internal or institutional sources.
Before changes in formal structure become societally legitimated and/or required, they are a d o p t e d much as is any other
innovation through a process of diffusion depending in large
part on the value of the changes for the internal functioning of
the organization (Utterback, 1971; Aldrich, 1979). It is not
always easy to assess this, of course, especially w h e n the
outputs of the organization are difficult to evaluate. Under these
conditions, the need for changes will often be determined by
the lack of consensus, or degree of conflict, within the organization (Cyert and March, 1963; March and Olsen, 1976; Pfeffer, 1981). Fundamentally, existing structure comes to be
viewed as problematic, and the "logic of good f a i t h " (Meyer
and Rowan, 1977) is disrupted. Therefore, to the extent that an
organization is an early adopter of an innovation in formal
structure, its decision to adopt will depend on the degree to
which the change improves internal process (for example, by
streamlining procedures or reducing conflict).
In contrast, once historical continuity has established their
importance (Bergerand Luckmann, 1967; Zucker, 1977),
changes in formal structure are adopted because of their
societal legitimacy, regardless of their value for the internal
functioning of the organization. When some organizational
elements become institutionalized, that is, w h e n they are
widely understood to be appropriate and necessary components of efficient, rational organizations, organizations are
under considerable pressure to incorporate these elements into
their formal structure in order to maintain their legitimacy. By
doing so, "an organization demonstrates that it is acting on
collectively valued purposes in a proper and adequate manner"
(Meyer and Rowan, 1977: 319). This may be necessary to
ensure access to various resources that the organization needs
for survival (in the case of cities, favorable bond ratings,
membership in some national organizations, state or federal
funding, etc.). It is assumed that the adoption of an innovative
measure may have little or no effect on the actual efficiency of
organizational operations; its adoption fulfills symbolic rather
than task-related requirements. Hence, to the extent that an
organization is a late adopter of an innovation in formal structure, its decision to adopt will depend on the degree to which
there is a common understanding that the change is necessary
for efficient organizational performance (Walker, 1969).
Two basic problems are confronted here: What is the effect of
explicit hierarchical legitimation of a reform, and what is the
effect of rapid and widespread legitimation of a reform on its
subsequent adoption? The first analysis compares the effect of
hierarchical control by the state w i t h the effect of nonmandated spread of reform on the rate of adoption. In the second
26/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

analysis, changes in the ability to predict adoption on the basis

of particular organizational characteristics from the early
periods to the later periods are explored in detail. It is expected
that when innovations are rapidly institutionalized, early adoption can rest on either the force of the hierarchical structure
that requires the reform or on particular characteristics that
make it appropriate to adopt the reform; however, while
particular characteristics may predict early adoption, they lose
their explanatory power rapidly.2 In contrast, if the adoption fails
to become legitimated, the same characteristics that predict
early adoption continue to predict which units are more likely to
adopt throughout the time period (Hamblin, Jacobsen, and
Miller, 1973: Ch. 7). In such cases, regional effects and
particular sets of characteristics, such as high status, that
identify a set of "innovators" emerge as central predictors of
adoption (Rogers, 1962).
LAW AND HIERARCHICAL LEGITIMATION
In organizational networks in which the control of resources and
authority is centralized in a few powerful organizations, the
institutionalization of an element of formal structure is largely
dependent on its legitimation by those organizations (Benson,
1975). Once legitimated by higher level organizations, through
legal mandate or other formal means, dependent organizations
generally respond by rapidly incorporating the element into their
formal structure. This adoption is seldom problematic when the
elements have high face validity and there is common agreement concerning their overall utility. However, under certain
conditions, strong resistance can develop. For example, lack of
consensus on the value of an innovation, such as curriculum
reform in public schools (Rowan, 1982), can lead to failure to
adopt or to early rejection of the innovation. In addition, strong
coalitions or interest groups can block a hierarchically mandated
change, as happened with busing to integrate schools.

Work to date on the adoption process has

been largely descriptive, providing few predictions that generalize beyond a particular
case. In this paper we do not attempt to
construct an explanation for all types of
adoption processes but to specify a particular class of adoption situations to which our
generalizations are expected to apply.
There are a number of important relations
that we do not consider here. First, different modes of gradual adoption are not
considered. The innovation may produce a
striking increase in effectiveness or efficiency and be rapidly adopted on "rational"
grounds, or as in the case of a technological
innovation, the value may be assessed as
superior and the innovation may be rapidly
adopted simply on the basis of professional
opinion. Second, we do not explore sources
of resistance. Some adoptions, even when
supported by professional opinion, meet
with much more resistance than others
(e.g., computerized accounting systems in
small businesses or fluoridation of city
water supplies). Finally, we explore only a
limited set of factors that may lead to
adoption, though we have focused on
those identified by historians as most
significant.

We do not explore the conditions underlying such resistance; it

is important simply to note that legal requirements do not
always ensure adoption. In the case of civil service, states, such
as Wisconsin, with strong interest groups that opposed adoption failed to pass statewide measures. However, in states
where there was little organized opposition, and consensus
concerning the potential value of the innovation was high, the
speed and pattern of diffusion among cities under state mandate are expected to reflect the degree of hierarchical legitimation; diffusion will occur from the state to each city, ratherthan
among cities. When states pass laws requiring municipal civil
service, it is expected that there will be a landslide effect in the
first year of implementation of the law, with the state influencing the remaining cities to adopt in subsequent years. These
effects should be apparent in the three states adopting such
laws during the time period investigated here.
Sample
Adoption was defined as the passage of any legal requirement
for the institution of civil service procedures. In most cases, civil
service requirements affected only parts of city government;
fire and police departments were frequently the first affected.
Data on adoption of civil service procedures, 1880 to 1935, were
collected for 167 cities. Of these 167 cities, 74 are located in the
three states that adopted civil service reform requirements for all
27/ASQ, March 1983
Copyright 1983. All rights reserved.

cities prior to 1930 and 93 are in states that had no such

requirement.
Statewide laws requiring municipal civil service were passed by
New York in 1883 and by Massachusetts in 1884. Ohio made
civil service mandatory for fire and police departments under its
municipal code in 1902; this code was extended in 1908, and in
1912 municipal civil service was given constitutional status
(Griffith, 1974). In the analysis of the three states mandating
civil service, the 74 cities having a population of 50,000 or
greater in 1930 were included. Data were aggregated for the
three states requiring adoption both because of the small
sample size and because initial analyses indicated no significant
differences in the patterns of adoption between states.
The 93 cities located in states that did not mandate civil service
were drawn from a sample of 150 cities randomly selected from
a sampling frame of all cities having a population over 25,000 in
1930, stratified by size. The cities were selected from the 150
by using two criteria: (1) cities of less than 50,000 were
excluded, because data on smaller cities proved extremely
scanty, particularly in the earlierdecades, and (2) cities in states
mandating civil service were excluded.
The data sources for all the analyses reported in this paper are
listed in the Appendix. Charles N. Halaby and M. Craig Brown
initially conceived of a longitudinal quantitative study of the
adoption of civil service systems by American city governments. The Civil Service Reform League reports that make
such a study possible were brought to our attention in 1977 by
them.
Findings
In the Figure, which plots the actual time of adoption of civil
service procedures, 1880 to 1930, we can compare the differences in the rate of adoption of civil service reforms between
cities in states that required it and those that had no requirements. When adoption was required by the state, the rate of
adoption was rapid. Within the first ten years, over 60 percent
of all cities adopted civil service procedures; then the rate
slowed, but all cities have adopted within 37 years. In sharp
contrast, cities with no such requirement initially adopted much
more gradually. In the first fifteen years the rate of adoption
was low. After this period, however, the rate of adoption
progressively accelerated over time. At the end of the time
period considered here, a little over 60 percent of these cities
had adopted civil service procedures, their rate of adoption
about equal to that reached after ten years by cities required to
adopt.
There is also some evidence that the underlying process of
adoption is different for these two groups of cities. Two
models, incorporating different assumptions about the sources
of diffusion, were estimated for both groups of cities (Coleman,
1964: 495-505). Both models were operationalized using algorithms developed inZucker(1975). Cities that adopted civil
service procedures when no statewide requirement existed can
be treated as sets of small separate groups, with full communication between some but no communication between others.
Using a diffusion model of decentralized influence, the overall
fit was acceptable, though the estimated values diverged from
28/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

100 T

90

80-3
ui
O

>

70

'

>
o
o 60-i
z
Ia.
O
Q

<

50

UJ

t
o

U.

40

UJ

fS
zUJ
O

30 -i

DC
UJ

a.
20

10-i

T"
10

"
T
30
20
YEARS FROM INITIAL ADOPTION

40

50

Adoption by cities in states mandating adoption (/?2 = .99, Error = .038)

Adoption by cities in states not mandating adoption {R2 = .89, Error = .408)
Figure. Rate of adoption of civil service by cities over 50-year time period.

the actual values in both the initial adoption period (overestimated) and the adoption after 1912 (underestimated). In contrast, in those cities that adopted civil service procedures when
the state required it, diffusion from the source was almost
immediate, so that adoption by cities in the first year does not fit
a diffusion model, even one that assumes centralized influence.
In our case, over a third of the sample adopted procedures in
that first year, creating a landslide rather than a diffusion effect.
However, the remaining years showed a pattern of single
source diffusion that yielded a close fit between the actual and
estimated values. Hence, as these results demonstrate, two
fundamentally different patterns of adoption, resting on different processes, occurred as cities adopted civil service
procedures.3
GRADUAL ADOPTION PROCESS
Full results, including estimation of the
different models of adoption proposed by
Coleman (19641, can be obtained from
Pamela S. Tolbert

We argue that, when not mandated by state government, civil

service was adopted at first in response to conflict generated by
different conceptions of the appropriate role and function of
29/ASQ, March 1983
Copyright 1983. All rights reserved.

municipal government held by older, established groups and/or

community business leaders, and those held by lower status
groups in the community, particularly the politically organized
immigrants. In line w i t h work by historians on civil service
adoption and by students of municipal structure, w e expect
early adopters of civil service reform to have a relatively larger
foreign-born population, more middle class members (smaller
proportion of manufacturing wage earners and illiterates), and a
narrow scope of administration (lower municipal expenditures).
It is also expected that they will tend to be larger and younger
than nonadopters, although the overall effects of age and size
are somewhat unclear since they are apt to be inversely related
to adoption. These variables are discussed in more detail below.
Our prediction, in contrast to the earlier research, is that while
these variables are important determinants of the adoption of
an innovation early in the process of its diffusion, they become
relatively poorer predictors as the reform measure becomes
more institutionalized. Overtime, adoption is expected to
become independent of internal factors, as external definitions
of modern municipal administration become more significant.
In line w i t h earlier empirical work on institutionalization (Zucker,
1977), it is expected that as a reform measure is increasingly
taken for granted because of social legitimation, cities will begin
to adopt it as a "social fact," regardless of any particular city
characteristics. Hence, the ability of these city variables, taken
as a whole, to differentiate between adopters and nonadopters
should progressively decline. It is also possible that the effectiveness of a particular variable may change, such that it
becomes a relatively better or poorer predictor at different
points in time; our primary concern here, however, lies not in
tracing the effects of specific city characteristics on adoption,
but in assessing the effects of overall differences between
adopters and nonadopters over time.
We do not assume that a particular set of characteristics is
always or usually related to adoption of innovation, but rather
that those characteristics that make it more "rational" to adopt
will be important early in the diffusion process. This may explain
w h y no consistent set of characteristics predisposing individuals or organizations to adopt innovations have been discovered,
despite considerable research aimed specifically at characteristics as explanatory variables (Downs and Mohr, 1976). Each
specific innovation should be related to a set of adopter
characteristics, only some of which will overlap w i t h other
innovations unless they are linked to a common justification.
City Characteristics and Reform: Measures
Before turning to the analysis, the characteristics that have
previously been identified as predictors of adoption of civil
service reform need to be more fully discussed. While civil
service reform was promoted by the Progressive Movement as
increasing the efficiency and effectiveness of the administration of local government (Woodruff, 1903; Thelen, 1972), some
historians have argued that civil service reform was used as a
political weapon by social groups (the industrialists or the
middle class) to gain or maintain their political dominance (e.g.,
allowing them to define administrative positions in such a way
as to virtually ensure the appointment of the group's members
to municipal office).
30/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

Immigrants and reform. Local reform efforts have often been

viewed as a response to t h e growth of machine politics
associated w i t h the tremendous influx of immigrants into
American cities during the late nineteenth and early twentieth
centuries (Hofstadter, 1956; Wolfingerand Field, 1966). Since
civil service reform eliminated political patronage, it represented a particularly effective way for the traditional AngloAmerican elites to attack immigrant-dominated machines by:
(1) establishing criteria for appointment to office, including
standards of education or literacy that were difficult for immigrants to meet, and (2) establishing rules governing tenure, to
protect appointees from patronage-induced turnover in city
government (Hays, 1964; Gordon, 1968). According to this
argument, reform would be most common among cities w i t h
large foreign-born populations, where tensions between older
Anglo-American groups and newly arrived immigrants would be
highest. This is measured here as the percentage of the
population that is foreign-born.
Socioeconomic bases of reform. Receptivity to municipal
reform has also been linked to the socioeconomic composition
of a city, w i t h the supporters of reform coming from the
educated and professionalized middle class. Two different
interpretations both suggest that municipal reform received its
most enthusiastic support in cities with a large proportion of
educated, white-collar citizens. On the one hand, the basic
values of the newly emerging middle class efficiency,
impartiality, rationality were most consistent w i t h those
underlying reform (Wiebe, 1967); on the other hand, however,
middle class reformers were motivated by pragmatic concerns
of securing representation of their political and economic
interests in local government (Hays, 1964; Weinstein, 1968).
The socioeconomic composition of a city was measured in t w o
ways: the percentage of illiterates was used as a measure of
the level of education in a city, and the number of manufacturing wage earners (residualized on city size) 4 was used as a rough
index of the concentration of members in blue-collar occupations. Both were expected to be inversely related to the
adoption of civil service reform.

To reduce problems of multicollinearity,

both the number of manufacturing wage
earners and municipal expenditures (discussed under "Scope") were linearly regressed on the log of city size (Maddala,
1977).
S
Liebert's (1976: 33) research indicates that
this measure is strongly correlated with the
number of functions performed by municipal governments (r = .72).

Scope. A third factor that has been suggested to have affected

the adoption of reform is scope, or the number of functions
performed by local government (Liebert, 1976; Turk, 1977).
According to this argument, cities w i t h broader scopes encouraged the development of competing special interests and
higher levels of political activity by offering greater opportunity
for political influence. These cities tended to have higher
resistance to reform, because reform measures frequently
limited accessibility to formal leadership positions. In contrast,
cities in which "narrow scope presumably limited the relevance
of the government and of its leaders hip vis-a-vis many types of
possible interests" had higher rates of reform (Liebert, 1976:
97). Total municipal expenditures (residualized on city size) were
used as an indicator of governmental scope; 5 this variable was
expected to be inversely related to the adoption of civil service
reform.
Age. Another possible determinant of the adoption of reform is
city age. According to Stinchcombe (1965) and several recent
empirical studies (Kimberly, 1975; Liebert, 1976; Meyer and
Brown, 1977), the formal structure of an organization tends to
31/ASQ, M a r c h 1983
Copyright 1983. All rights reserved.

reflect the historical era in which it originated, since organizations generally adopt and retain the form that was predominant
at that time. By this reasoning, it would be expected that
younger cities, those that were just beginning to develop w h e n
the municipal reform movement swept the country, would be
more likely to adopt reforms than older cities whose municipal
structures were already well entrenched and often supported
by vested interests (see Williamson and Swanson, 1966, on age
of city and adoption of industrial innovations). Age was measured as the year in which a city became incorporated; age
therefore remains the same for each city throughout all the
time periods. Given this measurement, age is expected to be
positively related to adoption of civil service reform.
Size. A final factor that has also been linked to the adoption of
municipal reform is city size. Although studies of different
types of reform measures have found size to have a varying
impact on adoption (e.g., Kessel, 1962; Schnore and Alford,
1963), studies that have specifically examined civil service
reform have found a simple positive relationship between the
adoption of the reform and city size (cf., Wolfinger and Field,
1966). City size was logged to normalize its distribution.
Data and Analysis
The effect of these variables on cities' adoption of civil service
measures was analyzed using a proportional hazards regression
model (Cox, 1972). First developed in the biological sciences,
models of this type have been adopted by social scientists to
explore a variety of phenomena (cf., Hannan, Tuma, and
Groeneveld, 1978; DiPrete, 1981; Carroll and Delacroix, 1982).
A central advantage of these models over other cross-sectional
approaches, such as logit or probit, is the explicit incorporation
of the timing of changes in a qualitative dependent variable
(Carroll, 1982). Essentially, the objective is to model the instantaneous transition rate, or the transition probability of moving
from one discrete state to another over an infinitesimally small
unit of time. Thus, the transition rate between state/ and state
k, where p is the probability of such a transition, is defined as

At-0

At

This hazard rate may depend both on time and on a set of

exogenous variables.
Here w e use partial likelihood estimation procedures, based on
work by Cox (1972). This approach requires less information for
estimation and makes weaker parametric assumption than full
likelihood methods. With partial likelihood, the likelihood function has t w o components, one that rests on the order of events
and another that rests on the exact timing of the events.
Maximization of the likelihood function is based only on the first
component; thus, correct ordering of the events is required.
Estimators obtained w i t h this procedure, like full likeli hood
estimators, have excellent asymptotic properties (Tuma, 1980).
The general form of the model is
hllt(t\X) = h0{t)exp(pX)
or
\n{hik(t\X)lh0(t)}=pX,
32/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

where h(t) is the hazard function, or the rate of leaving a

particular state among a set of units, in this case, moving from
nonadoption of reform (0) to adoption (1). X is a vector of
covariates, (3 is a vector of regression coefficients and h0(t) is a
hazard function for a unit with X = 0 (Hopkins, 1981).
Data. The data sources for adoption have already been discussed and are fully presented in the Appendix. Also listed in
the Appendix are the sources of data on city characteristics,
gathered from the decennial censuses, from 1890 to 1930.
Missing data were estimated using regression to generate
predicted values (Maddala, 1977). Only in one case, the percentage of illiterates in the population in the first time period,
was estimation unreliable; the variable was excluded from the
analysis in that period.
Unfortunately, the definitions of some of the independent
variables changed over time and were noncomparable. This
problem is frequently encountered in the use of early census
data. As a consequence, pooling of the data is not possible.
When these data are treated as comparable, serious estimation
errors undoubtedly occur, since some central measures, such
as the basic definition of manufacturing industries, changed
dramatically during the fifty-year period. While a single analysis
can be used if the changed definitions are entered as new
variables, interpretation is problematic. Therefore, four separate successive analyses of the adoption of municipal civil
service reform were conducted (see Williamson, and Swanson,
1966, for a similar resolution of the problem).
In the first analysis, the effects of the independent variables, as
measured in the 1890 Census, on the transition rates for cities
adopting civil service measures between 1885 and 1904 were
examined. During this timespan only a small proportion of cities
(about 11 percent) formally contracted with a commission or
board to set standardized personnel requirements for municipal
employees. Similarly, in the second analysis, the effects of city
characteristics (measured ten years later) on the rate of adoption between 1905 and 1914 were again examined; cities
adopting the measure in the previous period were excluded
from the analysis. This procedure was repeated for the third
and fourth analyses, of rates of adoption between 1915 and
1924, and between 1925 and 1934. In each analysis, city
characteristics as measured in the decade just preceding the
adoption period were used as predictors. Our objective, then,
was to assess the continued effectiveness of these characteristics in predicting the rate of adoption over time.
Findings
The results of the analyses are presented in Table 2. For each
independent variable, the top row shows the parameter estimates, the second row the standard errors of the coefficients,
and the third row the exponential raised to the power of the
coefficient. This last row indicates the proportion of change in
the adoption rate induced by a change of the predictor variable
by one unit. When a variable has no effect, this value is 1.0. A
value greater than unity indicates a positive effect; less than
unity indicates a negative effect. For example, 1.1 indicates a 10
percent increase in the adoption rate, while .9 indicates a 10
percent decrease per unit change.
33/ASQ, March 1983
Copyright 1983. All rights reserved.

Table 2
Proportional Hazards M o d e l of Civil Service Adoption over Time, 1885-1935

Time
period

1885-1904
[N = 83)

B
SE
exp(B)
1905-1914 B
(W = 74)
SE
exp(B)
1915-1924 B
(A/ = 52)
SE
exp(B)
1925-1934 B
SE
{N = 39)
exp (B)

Percentage
Manufacturing
foreignPercentage wage
Municipal
Log
born
illiterate
earnerst
expenditurest size

.114'"
.040
1.120

.056"
.027
1.058

-.069
.079
.933
-.279"
.127
.757
-.006
.171
.994

.029
.030
1.029
.021
.040
1.020

-.406*"
.888
.666
-.070
.550
.932
.569
.863
1.766
-.296
.742
.744

-1,657
.910
.191
.034
.661
1.035
.299
1.267
1.349
-1.361
1.734
.256

Age

Model
-2 log 1 chi-square

63.27 15.52
.487* .009
.326
(p < .008)
.016
1.627
1.009
.383* .0001 175.67 15.77
.249
.008
(p < .02)
1.467
1.000
.016 104.07 10.38
.325
.013
.428
(p < .11)
1.384 1.016
57.25 2.04
-.766 -.010
.651
.016
(p < .50)
.465
.990

Model
D
.51

.35

.35

.17

p < . 0 1 ; " p < . 0 5 ; * " p <.10

Excessive missing data; no reliable estimate possible.
tResidualized on city size.

However, w e are less interested in the effects of specific

variables than in the overall fit of the model containing the
variables in combination, since historians generally identify
these as a cluster and since they might be expected to interact
w i t h each other. Harrell (1979) has argued that the D statistic
can be interpreted analogously as the more familiar/? 2 , but this
interpretation is not widespread. Consequently, w e focus on
the chi-square statistic and its significance level; both statistics,
however, lead to the same conclusion.
The data provide strong support for our predictions. In the first
period, the overall model chi-square is significant beyond the
.01 level (p < .008). In the second period, the effects appear to
be weaker, as the overall significance level is lower and fewer
of the independent variables have significant coefficients. The
predictive power of the variables as a group continues to
weaken through the third period and drops very sharply in the
fourth period. The initial good explanation and the steep decline
in explanatory power in the last time period are consistent with
our expectations. As the process of adoption continues, the
characteristics of cities become increasingly less relevant to the
adoption process. 6

Some suggestive results were obtained

using a continuous version of the dependent variable (year of adoption) that
showed an even more striking decrease
over time in the proportion of variance
explained after correction for sample size.
For further information, contact Pamela S.
Tolbert.

It is clear from the effects of the individual city variables

reported in Table 2 that the decision to adopt in the early time
periods was based to a significant degree on those characteristics historians have described as important predictors of adoption of civil service procedures, those related to reducing
conflict and to streamlining the internal functioning of city
government. In both the first and second time periods, the
percentage of foreign-born and city size exerted significant
influence on the adoption process in the expected direction.
Also as predicted, in the first time period middle-class cities
(those with a smaller proportion of blue-collar workers), were
more likely to adopt the reform. However, as the process of
adoption continued, the characteristics of cities were less
frequently significant predictors of adoption. Only one variable,
the percentage of illiterates, emerged as significant in the third
time period; no variables were significant in the last time period.
34/ASQ, M a r c h 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

Table 3
Means and Standard Deviations of City Characteristics over Time, 1885-1935*
Time
period

Percentage
foreignborn

Manufacturing
wage
earnerst

Municipal
expenditurest

Log
size

Age

Proportion
adopting

1885-1904
(A/ = 83)

2172
( 1173)

-.0075
(.4487)

-.0000
(.3818)

3.8581
(1.072)

1849.88
(29.1994)

.0968
(.2973)

1905-1914
(N = 74)

1767
( 1066)

6.9025
(4.8043)

-.0096
(.4614)

-.0116
(.3968)

3.9973
(1.1049)

1849.71
(30.4510)

.2976
(.4600)

1915-1924
(W = 52)

1577
( 1192)

5.4500
(3.2914)

.0399
(.5637)

-.0653
(.3165)

4.1660 1850.19
(.9712)
(31.4503)

.2881
(.4568)

1925-1934
(N = 39)

1332
( 1216)

4.8809
(2.7061)

-.0288
(.6031)

-.0417
(.3085)

4.5989 1847.76
(.6721)
(33.1720)

.1905
(.3974)

Percentage
illiterate

Standard devia tions in parentheses.

tResidualizedor I city size.
Missing data.

A major competing interpretation of the results can be eliminated by examining the changes in variance overtime in city
characteristics. It is clear from Table 3 that the variance does
not decrease systematically over time.
CONCLUSIONS AND IMPLICATIONS
Our hypotheses concerning the changing sources of formal
structure received considerable support in all analyses carried
out here. Civil service procedures were adopted much more
rapidly by cities when the state mandated them and the
process of adoption was directed by a single source. In contrast,
when no state-level legitimation occurred, civil service procedures were adopted gradually, diffusing largely through social
influence among cities. Most important for organizational
theory, however, are the findings that internal organizational
factors predicted adoption of civil service procedures at the
beginning of the diffusion process, but did not predict adoption
once the process was well underway. As an increasing number
of organizations adopt a program or policy, it becomes progressively institutionalized, or widely understood to be a necessary
component of rationalized organizational structure. The legitimacy of the procedures themselves serves as the impetus for
the later adopters. These findings permit a partial integration of
the generally conflicting approaches focusing on the internal or
the institutional sources of formal structure. In addition, they
reassert the critical role of history for understanding organizational structure and its change (Stinchcombe, 1965; Meyerand
Brown, 1977).
The results reported here also have implications for two major
areas of research that we did not directly address. First,
treatment of the spread of innovation in a general theoretical
framework permits the researcher both to gain more insights
into the processes at work and to obtain more precise specification of expected differences in patterns, rates, and correlates of
diffusion. The ad hoc quality of most diffusion studies (e.g.,
Brown and Philliber, 1977) has made cumulative development
nearly impossible except when the substantive diffusion is
exactly the same, like the diffusion of hybrid corn, as discussed
by Feller (1967). In contrast, we expect that our model can be
35/ASQ, March 1983
Copyright 1983. All rights reserved.

applied to a wide range of phenomena. For example, w h e n

diffusion patterns are suddenly truncated early in the process,
w e expect that there was a failure to legitimate the change and
the characteristics that initially predicted adoption will remain
good predictors throughout the process. In fact, this may well
explain the pattern of the diffusion of some innovations in
education, though the available data are not sufficient fora test
(Rowan, 1982).
Second, the results have implications for methodology. The
use of cross-sectional data and the measurement of city
characteristics with available data, not from the historical period
under investigation, has been customary in research on the
adoption of municipal reforms by city governments (cf., Sherbenou, 1961; Kessel, 1962; Schnoreand Alford, 1963; Wolfingerand Field, 1966; Lineberry and Fowler, 1967). The results
reported here should make it clear that such methodological
shortcomings may introduce serious biases into the data: (1)
these studies of municipal reform neglect the fact that contemporary data may not accurately reflect the city's standing on
various characteristics at the time the reform was actually
adopted and, thus, it is difficult to be certain whether presentday differences between " r e f o r m e d " and " u n r e f o r m e d " cities
are causally or consequentially related to the adoption of the
reforms, and (2) these studies, by relying on cross-sectional
designs, ignore the fact that adoption of reforms by cities
occurred over time and that factors influencing adoption may
have varied from one point in time to the next. A city characteristic important in predicting early adoption may be irrelevant
twenty years later.
Thus, the approach and results presented here have implications, both theoretical and methodological, for studies of
change in the formal structure of organizations, for studies of
innovation and diffusion, and for studies of adoption of reforms
by municipal governments. But, most significant in terms of the
goals of our research, the boundaries between the rational and
the institutional approaches to organizations have been more
clearly specified and the central role of history in understanding
organizations confirmed. An adoption process rooted in the
internal needs of the organization can become over time a
process rooted in conformity to institutional definition.
REFERENCES
Aldrich, Howard
1979 Organizations and Environments. Englewood Cliffs, NJ:
Prentice-Hall.
Aldrich, Howard, and Jeffrey Pfeffer
1976 "Environments of Organizations." In A. Inkeles, J. Coleman, and N. Smelser (eds.),
Annual Review of Sociology, 2:
79-105. Palo Alto, CA: Annual
Reviews.
Anderson, Theodore, and Seymour
Warkov
1961 "Organizational size and functional complexity: A study of
administration in hospitals."
American Sociological Review,
26: 23-28.

Benson, J. Kenneth
1975 "The interorganizational network as a political economy."
Administrative Science Quarterly, 20: 229-249.
Berger, Peter, and Thomas
Luckmann
1967 The Social Construction of Reality. New York: Doubleday.
Blau, Peter M.
1970 "A formal theory of differentiation in organization." American
Sociological Review, 35:201 218.

Brown, Lawrence A., and Susan G.

Philliber
1977 "The diffusion of a populationrelated innovation: The planned
parenthood affiliate." Social
Science Quarterly, 58: 2 1 5 228.
Carroll, Glenn R.
1982 Dynamic Analysis of Discrete
Dependent Variables: A Didactic Essay. Mannheim, Germany: Zentrum fur Umfragen,
Methoden und Analysen.

Blau, Peter M., and Richard A.

Schoenherr
1971 The Structure of Organizations.
New York: Basic Books.
36/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

Carroll, Glenn R., and Jacques
Delacroix
1982 "Organizational mortality in the
newspaper industries of Argentina and Ireland: An ecological
approach." Administrative Science Quarterly, 27: 169-198.
Child, J.
1972 "Organization structure, environment, and performance
The role of strategic choice."
Sociology, 6: 1-22.
Clinton, Spencer
1886 "Reform in municipal government." Papers read before The
Cleveland Democracy of Buffalo (Buffalo): 101-116.
Coleman, James S.
1964 Introduction to Mathematical
Sociology. New York: Free
Press.
Cox, D. R.
1972 "Regression models and life
tables." Journal of the Royal
Statistical Society, 34: 187220.
Crandon, Frank P.
1886 -1887 "Misgovemment of
great cities." Popular
Science Monthly, 30:
521-524.

Hamblin, Robert, I . B. Jacobsen,

a n d J . L L Miller
1973 A Mathematical Theory of Social Change. New York: Wiley.

Cyert, Richard, and James G. March

1963 Behavioral Theory of the Firm.
New York: Free Press.

Hofstadter, Richard
1956 Age of Reform. New York:
Knopf.

DiPrete, Thomas
1981 "Unemployment over the life
cycle: Racial differences and
the effects of changing economic conditions." American
Journal of Sociology, 87: 2 8 6 307.

Hopkins, Alan
1981 "Regression with incomplete
survival data." Biomedical
Statistical Software. Berkeley,
CA: University of California
Press.

Downs, George, and Laurence

Mohr
1976 "Conceptual issues in the study
of innovation." Administrative
Science Quarterly, 21:
700-714.
Feller, Irwin
1967 "Approaches to the diffusion
of innovations." Explorations in
Entrepreneurial History, 4:
232-244.
Gelfand, Mark
1975 A National Compendium of
Cities. New York: Oxford University Press.
Gordon, Daniel
1968 "Immigrants and urban governmental form in American
cities, 1933-60." American
Journal of Sociology, 74: 158
171.
Griffith, Ernest
1974 A History of American City
Government: The Progressive
Years and Their Aftermath,
1900-1920. New York:
Praeger

Hannan, Michael, Nancy B. Tuma,

and LyleGroeneveld
1978 "Income and independence effects on marital dissolution:
Results from the Seattle and
Denver income-maintenance
experiments." American Journal of Sociology, 84: 611 -633.
Harrell, Frank
1979 "PHGLM Procedure." SAS
Technical Report. Raleigh, NC:
SAS Institute.
Hays, Samuel
1964 "Politics of reform in the Progressive Era." Pacific Northwest Quarterly, 55: 157-169.
1972 "The new organizational society." In Jerry Israel (ed.), Building the Organizational Society:
Associational Tendencies in
Early Twentieth-Century
America: 1-9. New York: Free
Press.
Hemes, Gudmund
1976 "Structural change in social
processes." American Journal
of Sociology, 82: 513-547.

Hughes, Everett C.
1936 ' The ecological aspects of institutions." American Sociological Review, 1: 180-189.
Kessel, John
1962 "Governmental structure and
political environment: A statistical note about American
cities." American Political Science Review, 56: 615-620.
Kimberly, John
1975 "Environmental constraints and
organizational structure: A
comparative analysis of rehabilitation organizations."
Administrative Science Quarterly, 20: 1-9.
Knoke, David
1982 "The spread of municipal reform: Temporal, spatial, and
social dynamics." American
Journal of Sociology, 87:
1314-1339.
Liebert, Roland
1976 Disintegration and Political Action. New York: Academic
Press.
37/ASQ, March 1983
Copyright 1983. All rights reserved.

Lineberry, R. L , and E. P. Fowler

1967 "Reformism and public policies
in American cities." American
Political Science Review, 6 1 :
701-716.
Maddala, G. S.
1977 Econometrics. New York:
McGraw-Hill.
March, James G., and Johan P.
Olsen
1976 Ambiguity and Choice in Organizations. Oslo, Norway:
Universitetforlaget.
Meyer, John W., and Brian Rowan
1977 "Institutionalized organizations: Formal structure as myth
and ceremony." American
Journal of Sociology, 83: 3 4 0 363.
Meyer, Marshall, and M. Craig
Brown
1977 ' 'The process of bureaucratization." American Journal of
Sociology, 83: 364-385.
Parsons, Talcott
1951 The Social System. New York:
Free Press.
Pfeffer, Jeffrey
1981 Power in Organizations.
Marshfield, MA: Pitman.
Pfeffer, Jeffrey, and Gerald R.
Salancik
1978 The External Control of Organizations. New York: Harper and
Row.
Rogers, Everett M.
1962 Diffusion of Innovations. New
York: Free Press.
Rowan, Brian
1982 "Organizational structure and
the institutional environment:
The case of public schools."
Administrative Science Quarterly, 27: 259-279.
Schiesl, Martin
1977 The Politics of Efficiency.
Berkeley, CA: University of
California Press.
Schnore, Leo, and Robert Alford
1963 "Forms of government and
socioeconomic characteristics
of suburbs." Administrative
Science Quarterly, 8: 1 -17.
Scott, W. Richard
1975 "Organizational structure." In
Alex Inkeles (ed.), Annual Review of Sociology, 1: 1 -25.
Palo Alto, CA: Annual Reviews.
Sherbenou, Edgar
1961 "Class, participation and the
council-manager plan." Public
Administration Review, 21:
131-135.

Stinchcombe, Arthur L.
1965 "Social structure and organizations," In J. G. March (ed),
Handbook of Organizations:
142-193. Chicago: Rand McNally.
Thelen, David P.
1972 The New Citizenship: Origins
of Progressivism in Wisconsin,
1885-1900. Columbia, MO:
University of Missouri Press.
Thompson, James D.
1967 Organizations in Action. New
York: McGraw-Hill.
Thornton, R., and P. M. Nardi
1975 "The dynamics of role acquisition." American Journal of
Sociology, 80:870-885.
Tuma, Nancy B.
1980 Invoking Rate. Stanford, CA:
Stanford University.
Turk, Herman
1977 Organizations in Modern Life.
San Francisco, CA: JosseyBass.
Utterback, James
1971 "The process of technological
innovation in the firm."
Adademy of Management
Journal, 14:75-88.

Van Riper, Paul

1951 History of the United States
Civil Service. Evanston, IL:
Row, Peterson.

Woodruff, Clinton
1903 "An American municipal program." Political Science Quarterly, 18: 47-58.

Walker, Jack
1969 "Diffusion of innovations
among the American states."
American Political Science Review, 63: 880-899.

Woodward, Joan
1965 Industrial Organization: Theory
and Practice. London: Oxford
University Press.

Weinstein, James
1968 The Corporate Ideal in the Liberal State. Boston: Beacon
Press.
White, Morton G.
1949 SocialThoughtin America: The
Revolt Against Formalism. New
York: McGraw-Hill.
Wiebe, Robert
1967 The Search for Order, 18771920. New York: Hill and Wang.
Williamson, Jeffrey G., and Joseph
A. Swanson
1966 "The growth of cities in the
American Northeast, 18201870." Explorations in Entrepreneurial History, 11 (Supplement). 1-101.
Wolfinger, Raymond, and John
Field
1966 "Political ethos and the structure of city governments."
American Political Science Review, 60: 159-195.

Zucker, Lynne G.
1977 "The role of institutionalization
in cultural persistence." American Sociological Review, 42:
726-743.
1982 "Institutional structure and organizational processes: The
role of evaluation units in
schools." In Adrianne Bank and
Richard C. Williams (eds.),
Studies in Evaluation and Decision Making. CSE Monograph
Series in Evaluation, No. 10:
69-89. Los Angeles, CA: Center for the Study of Evaluation.
1983 "Organizations as institutions."
In Samuel B. Bacharach (ed),
Perspectives in Organizational
Sociology: Theory and Research. ASA Series, Vol. 2.
Greenwich, CT: JAI Press
(forthcoming).
Zucker, Joel Steven
1975 Pattern Discrimination Based
on an Exponential Error Criterion. Santa Monica, CA: Rand
Corporation.

APPENDIX: Data Sources

Civil Service Reform League. Civil Service Agencies in the United States: A
7940 Census. Chicago: The Assembly, 1942.
Columbia Lippincott Gazetteer of the World. New York: Lippincott, 1954.
Department of Labor. U.S. Bureau of Labor Statistics Bulletin 30 (Table XVII):
997-999. Washington, DC: U.S. Department of Labor, 1900.
National Municipal League. Proceedings of the Philadelphia Conference for
Good City Government and the First Annual Meeting of the National Municipal
League. Philadelphia: National Municipal League, 1894.
National Municipal League. Proceedings of the Milwaukee Conference for
Good City Government and the Sixth Annual Meeting of the National Municipal
League. Philadelphia: National Municipal League, 1900.
National Municipal League. Proceedings of the New York Conference for Good
City Government and the Eleventh Annual Meeting of the National Municipal
League. Philadelphia: National Municipal League, 1905.
National Municipal League. Proceedings of the Cincinnati Conference for Good
City Government and the Fifteenth Annual Meeting of the National Municipal
League. Philadelphia: National Municipal League, 1909.
National Municipal League. National Municipal Review. Concord, NH: Rumford
Press, 1915.
U.S. Bureau of the Census. Compendium of the Tenth Census. Part 1, Table
XXVI: 452^163; Table XXXVI: 860-909. Washington, DC: U.S. Bureau of the
Census, 1880.
38/ASQ, March 1983
Copyright 1983. All rights reserved.

Diffusion of Civil Service

U.S. Bureau of the Census. Report on Valuation, Taxation and Public Indebtedness. Table IV: 248-261. Washington, DC: U.S. Bureau of the Census. 1880.
U.S. Bureau of the Census. Compendium of the Eleventh Census. Part 3, Table
3: 317-318; Table 95: 570-571. Washington, DC: U.S. Bureau of the Census,
1890.
U.S. Bureau of the Census. Report on Wealth, Debt and Taxation. Part 2, Table
12: 556-557; Table 14: 580-599. Washington, DC: U.S. Bureau of the Census,
1890.
U.S. Bureau of the Census Abstractor the Twelfth Census. Table85:115-117;
Table 88: 124-129. Washington, DC: U.S. Bureau of the Census, 1900.
U.S. Bureau of the Census. Manufacturers. Part 2, Table 2: 992-1005.
Washington, DC: U.S. Bureau of the Census, 1900.
U.S. Bureau of the Census. Population. Part 1, Table L: CIX-CX. Washington,
DC: U.S. Bureau of the Census, 1900.
U.S. Bureau of the Census Abstract of the Thirteenth Census. Table 19:
95-96; Tables 33-34: 250-253. Washington, DC: U.S. Bureau of the Census,
1910.
U.S. Bureau of the Census. Financial Statistics of Cities Having a Population of
over 30,000. Table 3: 118-119. Washington, DC: U.S. Bureau of the Census,
1910.
U.S. Bureau of the Census. Population. Tables III and IV: 152-291. Washington,
DC: U.S. Bureau of the Census, 1910.
U.S. Bureau of the Census. Financial Statistics of Cities Having a Population of
over 30,000. Table 3: 96-111. Washington, DC: U.S. Bureau of the Census,
1919.
U.S. Bureau of the Census. Abstract of the Fourteenth Census. Tables 31 -43:
110-117; Tables 139-141: 441 -445. Washington, DC: U.S. Bureau of the
Census, 1920.
U.S. Bureau of the Census. Manufacturers. Tables by state. Washington, DC:
U.S. Bureau of the Census, 1920.
U.S. Bureau of the Census. Population. Table 18: 130-131; Table 20: 240-335.
Washington, DC: U.S. Bureau of the Census, 1920.
U.S. Bureau of the Census. Manufacturers. Part 3, Table 2 (for each state).
Washington, DC: U.S. Bureau of the Census, 1929.
U.S. Bureau of the Census. Abstract of the Fifteenth Census. Tables 43-48:
101 -112; Tables 144-146: 284-291. Washington, DC: U.S. Bureau of the
Census, 1930.
U.S. Bureau of the Census. Financial Statistics of Cities Having a Population of
over30,000. Table4: 199-225. Washington, DC: U.S. Bureau of the Census,
1930.
U.S. Bureau of the Census. Population. Table 8: 21; Table 57: 100-103.
Washington, DC; U.S. Bureau of the Census, 1930.
U.S. Bureau of the Census. Population. Part 1, Table 13: 34-36; Table 64:
160-165. Washington, DC: U.S. Bureau of the Census, 1940.

39/ASQ, March 1983

Copyright 1983. All rights reserved.