You are on page 1of 32

Models and Realities of Popular Collective Action


Source: Social Research, Vol. 52, No. 4, Social Movements (WINTER 1985), pp. 717-747
Published by: The New School
Stable URL:
Accessed: 17-12-2017 17:13 UTC

Linked references are available on JSTOR for this article:
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

The New School is collaborating with JSTOR to digitize, preserve and extend access to Social

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to
Models and
Realities of

D y now, the scholarly literature on popular collectiv

contains plenty of good descriptions and numerous
models but leaves a void between the two. Researchers who

seek to account for real-life events such as particular soci

movements and concrete revolutions find that the available
theoretical apparatus gives them little grip on those events
Theorists who begin with general models find themselves
selecting simplified conventional accounts of presumably
well-known events or classes of events and gaining little fresh
insight into those events. Although theories of collective actio
will always benefit from improved description, for the tim
being theory lags behind the available evidence. These seem t
be the major difficulties:
(1) Gaps within the available models. The models provide mor
or less credible accounts of why, in principle, a set of peopl
who share an interest might act together, but they state les
convincingly why an individual would join a collective action
and they fail to specify the connections between the individ
ual and the collective decision.

(2) Use of single-actor models. Since collective action actually

consists largely of strategic interaction within and among
groups, models designed to explain the behavior of a single

SOCIAL RESEARCH, Vol. 52, No. 4 (Winter 1985)

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

actor miss the mark, especially

unitary disposition to the actor.
(3) Use of static models. Since colle
since its outcomes depend very
interaction, static models that sim
characteristics or outcome to gr
entire process poorly.
(4) Emphasis on causal rather than
ior. Although it would in princip
causal accounts of collective action,
means to create dynamic multia
complex phenomena. For some ti
els will probably serve their purp
As we shall see, some theorists e
of these difficulties; we have some
the gaps, some promising multiact
general a theorist achieves a solu
by exaggerating another one - co
of collective action, for instance,
having unitary dispositions. N
difficulties at once.
These difficulties help explain one of the peculiarities of
recent writing about collective action: the contrast among the
taut vitality of many partial models dealing with one aspect or
another of collective action, the bright insight of many de-
scriptions of particular kinds of collective action, and the mis-
cellaneous flabbiness of most efforts to create comprehensive
accounts of collective action. That indictment includes my own
efforts to create comprehensive accounts.
This paper will not solve the problems of collective-action
theory. It aims merely to help clear the ground for that
effort in two ways: by examining some concrete instances of
collective action in which the connection between individual
and collective is salient and problematic, by looking carefully
at the difficulties theorists face in making the connection.
Examining a real stream of collective action rather than a

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

fabricated instance will make neat mode

will help identify the weaknesses of m
begin with a look at the France of 19
discussion of ways to account for the so
that are visible in the France of 1906.
My account will emphasize conflict and discontinuous col-
lective action - contention rather than collaboration. The evi-
dence comes from catalogs of strikes, violent events, and
"contentious gatherings": occasions on which a number of
people gather in a publicly accessible place and, by word or
deed, make claims on others, claims that would, if realized,
affect the interests of those others. Contentious gatherings
include almost all events that authorities, ruling classes, and
unwary investigators call "riots," "disturbances," "disorders,"
and similar stigmatizing terms, plus a number of meetings,
parades, and other gatherings that have enough political
standing to escape stigma. My collaborators and I have pre-
pared those catalogs from the Statistique des greves, corre-
spondence and reports of the ministries of justice and interior,
local administrative records, political yearbooks, and national
newspapers. They fall short of constituting a comprehensive
or even a strictly representative sample of French contention,
but they do provide an exceptionally broad and rich portrait
of the forms, issues, participants, and outcomes of popular
collective action at a critical moment in France's twentieth-
century experience.

An Anxious Spring

In 1906, spring found France agitated. There was plenty to

be nervous about. Government employees were demanding
the right to organize, national labor federations were becom-
ing ostentatiously muscular, and defenders of the Catholic
Church were up in arms. Legislative elections, furthermore,
were coming up on 6 and 14 May. For the first time, the newly

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

unified socialist party was runnin

where. A heated campaign raged
This time, too, organized Catholics had a real grievance.
The government had enacted official separation of church
and state at the end of 1905. All religions were now, at least
officially, equal. Duly registered "religious associations" were
henceforth to operate church institutions. The government had
chosen February to begin inventories of church property be-
fore transferring control to the new associations. On the first
day of February, opponents of the inventories barricaded
Paris's churches of Sainte-Clothilde and Saint-Pierre-du-Gros-
Caillou, fought off the police who came to assure the inven-
tories, and forced the unwanted visitors to chop their way in.
Among those arrested at Sainte-Clothilde were counts Louis
de Bourbon and Guy de la Rochefoucauld. Aristocratic
legitimists were joining the ordinary faithful in their resist-

The struggles at Sainte-Clothilde and Sa

Gros-Caillou began a bitter series of confron
and in the provinces. The most consequent
Boeschepe (Nord) on 6 March. There, the s
torying official defended his father by fatally
onstrator. The parliamentary debate on the
down the government. Ten days later, the ne
sent its agents instructions to suspend inven
resistance seemed likely. By April, toe-to-toe
had become less frequent but had not disapp
The spring of 1906 also brought widespread
conflict. In Toulon, for example, April began
strike. On Saturday, 31 March, striking waite
through the city's streets, smashing the windo
remained open. On Sunday, 1 April, waiters
again. That night the wine steward at the caf
once more surrounded and taunted by striker
stiletto and stabbed Jean Bruno, a waiter from

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

Bruno died on the spot. Fearing a junct

from the arsenal with the striking caf
with the refusal of the socialist mayo
strations, the prefect himself took ov
Toulon's strikers held hostage a police o
fect had sent as his delegate; the strik
sion to march. Toulon's struggles continued through the

Industrial conflict was thriving in northern France as well.

After the disaster of Courrier es (10 March), in which 1,101
miners died in a giant explosion and subsequent underground
fires, much of the coalfield of Pas-de-Calais roared out on strike.
Georges Clemenceau, becoming the new government's minis-
ter of the interior on 13 March, soon sent troops to guard the
mines; for roughly 60,000 strikers, Clemenceau put 20,000
soldiers into the field. The soldiers found themselves protect-
ing both mine property against destruction by miners and
nonstriking miners against the attacks of their fellows.
During April, coalfield battles continued. In the Pas-de-
Calais and the Nord, miners blocked pitheads, roughed up
comrades who stayed on the job, and stoned the trains bring-
ing scabs from Belgium. On 2 April, a thousand workers
showing red flags marched into Billy-Montigny to demand the
release of miners jailed during earlier set-tos. On 4 April, the
557 men and 40 boys digging coal underground at Ligny-
lez-Aire joined the thousands of miners in neighboring vil-
lages who were already on strike. On 5 April, women gathered
at the guarded Courrières pit no. 4 outside of Billy-Montigny,
stoned the personnel who entered or left, then tried to break
in and press the search for miners still trapped below. On 9
April the wives of miners at Billy-Montigny demonstrated with
black, red, and tricolor flags.
More confrontations, some involving detonations of dyna-
mite, occurred in the northern minefields almost every day
from then to the month's end. Clemenceau activated the
cavalry. Forty-odd union leaders went to jail. Thus Clemen-

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

ceau's reputation began to shift. From Clemenceau,

implacable Jacobin and scourge of governments, he became
Clemenceau, strikebreaker.
In addition to miners, many other workers of northern
France joined April's action. On 4 April, for example, a crowd
in Roubaix attacked the carriage of former minister Jules
Meline, who had come to confer with the city's industrialists.
On 6 April, strikers at Fressenneville, Somme, broke the win-
dows of the lock factory in which they worked, then went on
to shatter the windows of nonstrikers as well as to sack and
burn the house of the factory owner; they sang the Marseil-
laise and the Carmagnole as they smashed.
Nor were striking waiters, locksmiths, and miners alone.
Here are some more items from April's calendar:

I April 1906: Twenty-odd Jews, impatient to emigrate to

Canada, broke into the Paris offices of the Jewish Colonization
Association and tore them up.

8 April: The official who came to inventory the property of a

church in Saint-Amans (Lozère) met a hail of stones and a
barricaded door before retreating to try again some other day.

10 April: Truckers who were on strike in Grenoble tried to

unhitch the horses from the wagon carrying the effects of the
new chief of gendarmerie; gendarmes and police fought them

10, 11, 12 April: Striking pitmen in the southern minefield of

Aies (Gard) gathered in the city streets and fought with
nonstrikers; on the 12th, 250 of them marched with a red flag.

II April: An important branch of the postal service (the sous-

agents) went on strike. The government requisitioned troops to
replace them.

13 April: Thousands of Catholics marched through Montpellier,

battling the police as they protested the disestablishment of their

14 April: Workers of Limoges marched with red flags, and with

a black flag bearing the word Germinal; they went to the tomb of

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

Vardelle, a worker who had died in demon

year. Two hundred anarchists followed the
cemetery, sang revolutionary songs, stopp
jeer the local garrison, then moved on to t
seized their banner and broke them up.

15 April: Demonstrating textile strikers

Voiron, near Grenoble; when troops arr
much of the city's labor force struck to d

16 April: Supporters of conservative and

for the Chamber of Deputies bloodied ea
Nommée (Vienne).

17 April: In the course of a strike, ab

workers of Lavelanet (Ariège) gutted th
ployers. That same day, French construc
and Bruley attacked Italians who had com

18 April: The striking miners of Ales fought nonstrikers and

tried to overturn the carriage of the mine's superintendent.

21 April: Retail and business clerks on strike in Lorient (Morbi-

han) threw up barricades and fought off soldiers after attempt-
ing to close the city's businesses.

22 April: In Brest, a meeting of some 3,000 socialists ended in a

street battle with supporters of an antisocialist candidate.

23 April: Striking and nonstriking masons fought each other in

Chatelguyon (Puy-de-Dôme).

25 April: Marseille's printers massed at their labor exchange,

declared themselves on strike, and proceeded to nonstriking
printshops; the thousand workers who arrived at the Moullot
printing plant confronted police.

26 April: UHumanit'e published a manifesto of the Socialist In-

ternational calling for a strike-demonstration on the First of

27 April: Clemenceau called in 26,000 troops to reinforce Paris's

garrison of 15,000 men. A dynamite charge went off at the
Argenteuil viaduct but failed to destroy it. Demonstrators again
battled troops in Lorient.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

30 April: Paris's prefect of police,

troops and police throughout the c
uniform appeared at the Paris labor
a socialist, and said he would never f
he was arrested at the exit, and carte

Much of France seemed to be fig


The Importance of 1 906

At first, this array of conflicts gives the impr

geoning disorder. Looked at more closely, how
some striking regularities. First of all, as com
theoretical possibilities, the range and number
are small; over and over again, we find the sa
workers, political factions, militant Catholics, po
others. For the most part, those same actors mai
tial social ties outside the moments of contention
those social ties to sustain their contention. Secon
of action they employ are few indeed: strikes, m
onstrations, a few more. Third, the events in que
connected clusters, and those clusters concern
power-^in *-he minefields, over the church, a
tional Assembly - that reach far beyond any
counter. In this turbulent year of 1906, the po
seem to be occurring on a national scale.
The years from 1905 to 1907 brought France through
one of its larger political transitions of the nineteenth and
twentieth centuries: disestablishment of the church, resolution
of the Dreyfus case against the army, establishment of so-
cialists and organized workers in the national political arena,
development of national strike waves as a recurrent fact of
life, appearance of southern winegrowers as a strident political
voice, varied and extensive contention throughout the land.
This surge of struggle raises important questions. Given the

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

fact that with respect to workaday life m

interests visible in the struggle had existe
and why did they gird for battle in the
tury? To put the same question another
map of French social structure and of in
1900, what more would we have had to know in order to
predict which groups would contend publicly over the next
few years, what form that contention would take, and what
consequences it would have?
The question is not entirely fantastic or even unhistorical.
Toward the end of April 1906, a remarkable report to the
minister of the interior presented a forecast of May Day. This
May Day mattered more than most, since it fell in the midst of
an electoral campaign in which socialists hoped to win a
number of seats and soon after the recently failed revolution
in Russia.

The forecast built on numerous reports from police com-

missaires, who in their turn drew from a network of spies,
informants, and other officers. It stated the likely action of
each important group of workers during the holiday, on
which the great fight for the eight-hour day was supposed to
begin. The forecast says of the carriagemakers, for example,
that they "will stay away on the First and will go on strike.
They won't demand the eight-hour day, since they're sure
they won't get it, but they will ask for a pay raise."1 The report
estimates that about 25,000 of them will join the general
strike, 50,000 "if the wheelwrights and mechanics join the
The electricians, on the other hand, seem "too divided, and
they know it; they say they're not ready. They can only suc-
ceed if all the plants go out at once, which is currently impos-
sible. A few small movements may persuade a few of them to
strike, 500 at the most."2 The predictions continue: The nav-

1 Archives Nationales, Paris, F7 13267.

2 Ibid.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

vies will have 30,000 strikers, of

plans to dynamite tunnels to st
will stay away from the general s
The forecasts later proved to be
The whole set of reports reach
the possibility of a large movem
based in Paris but spread throug
strikes already underway in the n
sion. Clemenceau and his prefect
arrest of Griffuelhes, secretary o
du Travail (CGT); call of troops to Paris; deployment of
armed force across the capital. Then began the largest strike
wave to occur in France up to that time, and the first national
wave centered on Paris. But despite nearly 200,000 strikers, in
Paris the demonstrations and meetings remained rather calm.
Nothing like a general rebellion occurred.
Those dossiers of the Interior Ministry would have sur-
prised an eighteenth-century police officer. They reveal a
practice of continuous, effective, anticipatory surveillance. By
contrast with the sometimes brutal but almost always retro-
active repression of the old regime, the government of 1906
sought to predict, prevent, and channel popular collective
action. It often succeeded. That predictive, preventive repres-
sion took shape during the Revolution and the Empire. After
a century of application, it had become quite impressive.
What is more, the police analysis shows a certain under-
standing of the conditions and mechanisms of workers' action.
At a time when Le Bon was writing about the madness of
crowds and Sorel was pinning workers' hopes to the cata-
strophic consequences of a general strike, these commissars
and officials were coolly appraising workers' strategy and or-
ganization, and getting a clear enough picture of both to
produce accurate predictions and effective repression.
I do not mean that the police agents of 1906 had a theoreti-
cal understanding of working-class collective action. Instead,
they had knowledge gained through two kinds of practice: as
direct participants in many of the public actions of workers,

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

and as recipients of information from

transmitted the militants' plans and
there was and is a wide gap between "
with popular collective action and the t
the language of government official
edge of working-class settings, of the
action, of the effectiveness of different
workers' strategy; on the other, a lan
order, of attitudes, of good workers an
of the state, they adopt the vocabular
despite defective theoretical understand
the police of 1906 could follow and ev
class collective action. Can contemporar
In what ways is the power-holders'
essentially the same ways as contempor
collective action are defective. First,
leap from dispositions of individuals to
without providing plausible accounts
necting the two. Second, they charact
havior of a single actor, the person or g
rebel, demand, or attack; the underlyin
interaction. Third, they are commonly
cations of the means by which action
outcomes. Fourth, their general accoun
behavior are causal rather than purposiv
their detailed analyses repeatedly take
tion. These weaknesses are perhaps un
to an interaction whose primary conc
control the behavior of their interlocutors. In the work of
analysts dealing systematically with popular collective action,
they lay down insuperable barriers to understanding.

Gaps in Our Models

If we assume that any collective decision flows from multiple

individual decisions, how do thousands of individual decisions

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

accumulate into a large social mo

one deduce the forms of individ
character of collective actions? Ca
demands made in the group's name
for individual involvement? Does
itself predictably transform ind
questions link the individual and
between the two levels is slippery;
to scale it. Since the challenge of Mancur Olson,3 which
blocked any direct, simple, and immediate translation of indi-
vidual interests into collective action (and, even more so, any
direct, simple, and immediate translation of collective interests
into individual action), theorists have often addressed the
question. They have so far had little success.
The analysis of industrial conflict, for example, entails the
relationship of individual to collective objectives. On the indi-
vidual level, we can easily imagine the formation of workplace
grievances by everyday experience; by extension, we move
easily to the decision to join a strike. At the level of a firm or
even of a whole industry, we likewise find it easy to attribute
common grievances to most workers; given a certain intensity
of organization, we arrive at an explanation of strike activity.
Yet the correlation between the degree of individual discon-
tent and the extent of worker organization, on one side, and
the amount of industrial conflict, on the other, is quite imper-
fect. That very imperfection calls for an analysis of the process
linking an interest or a shared attitude to collective action. In
that regard, existing models of industrial conflict lack sub-

The same is true of general models of mobilization. S

dard models of mobilization divide roughly into two r
different lines of reasoning, cumulative and constructiv
the one hand, a reasoning we might call cumulative becaus
its initial conception of a set of individuals, each having in
3 Mancur Olson, The Logic of Collective Action (Cambridge: Harvard University

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

Figure 1


000 000 000

000 000 000

0 0 0 ONO 0 OVO 0

0 0 0 0 cNsO 0 0'0

ests, who decide more or less consciously

themselves to others with whom they s
which promotes the development of
and finally of concerted action. The p
like the form in Figure 1. In the first s
als; in the second, some of those indiv
common framework; in the third, shared consciousness and
collective action - except for those individuals who remain
outside the frame.
The idea of a development of consciousness often marks
this mode of analysis. In summarizing her studies of "mes-
sianic movements" in the Third World, for example, Maria
Isaura Pereira de Queiroz declares that:

Messianic belief and messianic movement, two distinct social

facts, are two aspects of messianism that one can not separate
entirely: the first can exist without the second, but the second
always requires the first. Without the inspiration of the myth,
the collectivity cannot organize; nevertheless, the myth can en-
dure a long time without the formation of any movement. The
movement depends on the formation of a group, on the
gathering together of individuals having some function to ful-

Thus we arrive at a sharp separation of the stages, a separa-

tion depending at once on the development of consciousness
4 Maria Isaura Pereira de Queiroz, Reforme et revolution dans les sociétés traditionelles:
Histoire et ethnologie des mouvements messianiques (Paris: Anthropos, 1968), p. 291.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

and the creation of social ties as a consequence of a new

shared belief.

Such reasoning immediately raises two serious objections:

first, that it does not explain how and why an individual
ignores his individual interest in security and minimization of
effort; and second, that it lacks a specification of the social
process that produces the movement from one stage to the
next. Alteration of consciousness coupled with some form of
social creativity? Spiral accumulation of beliefs and of social
ties embedded in those beliefs? Here, as is often the case, the
reasoning runs backward, from the fact of a movement to a
search for the necessary conditions of its appearance.
The second line of reasoning presumes the prior existence
of a social structure that already connects most individuals,
and that changes and becomes more elaborate as a result of
repeated communication among individuals. Thus we may call
it constructive reasoning. In that line of thought, ties may well
multiply and reinforce each other in the course of collective
action, but previously existing ties form the main base for
mobilization and collective action. Furthermore, the divisions
of interest that motivate collective action rest especially on social
cleavages that already inform workaday social life. Schemati-
cally, we might represent the constructive line of thought as
shown in Figure 2. The diagram portrays a mobilization pro-
cess activating existing social networks, extending those net-
works, forming coalitions, and generalizing established oppo-
sitions. Thus Michael Schwartz bases his analysis of the move-
ment pitting tenants and landowners in U.S. cotton regions
against each other from 1880 to 1890 directly on the social
Figure 2


o o- a o ft- # +- CONFLICT

o- A A - «- ^d ° - *^' a

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

Figure 3


0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

0 0 0 0 0 0

relations establish
creation of the S
fense organizatio
success) at once as
struggles that preceded and succeeded the SFA's days of
Often, it is true, analysts represent collective action as a
direct expression of the common interest of a fully formed
group - a social class, community, ethnic group, union, party,
or something else - without asking seriously what connection
exists between the acting group and the base population it
claims to represent. This theoretical short circuit suppresses
the first stage of the cumulative model (Figure 3). The short
circuit avoids the problem of individual mobilization without
resolving it. Only the full accumulative and constructive
arguments sketch continuous circuits from individual to collec-
tive action.

Single-Actor Models

Although constructive models, as I have sketched them,

accommodate relations of conflict as well as of cooperation, at
bottom they describe the change of state within a single actor.
What is more, they ordinarily assume that the actor has a
5 Michael Schwartz, Radical Protest and Social Structure: The Southern Farmers' Alliance
and Cotton Tenancy, 1880-1890 (New York: Academic Press, 1976).

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

unitary disposition - a single ut

vision, no significant change in
French strikers of April and May
us the means of explaining whi
more quickly and fully for confr
officials. They say almost noth
among actors, and especially abo
Most users of accumulative and constructive models, indeed,
set up their analyses on the analogy of epidemiology: they
define a population at risk to some form of collective action,
then try to discover the conditions determining susceptibility
to that condition. Strike propensities, riot propensities,
social-movement propensities, and other propensities to col-
lective action become the variables to explain. Studies of par-
ticipation have their value; without them, for example, riffraff
theories of involvement in rebellions and revolutions would no

doubt be even more prevalent than they are. But they

strengthen an illusion: that the explanations of such events lie
in the characteristics of rebels and revolutionaries.
Consider the treatment of industrial conflict, where formal
modeling has gone quite far. Analysts of industrial conflict
customarily treat strikes as workers' actions, representing the
action of management or of government officials as an influ-
ence external to the decision to stop work. Official collections
of strike data reinforce such an interpretation; they emphasize
characteristics and actions of the workers, and neglect both
characteristics of employers and relations between workers
and employers.
To be sure, such authors as Orley Ashenfelter and
George Johnson, Walter Korpi, Douglas Hibbs, and Paul Ed-
wards have laid out different kinds of power-struggle in-
terpretations for strike activity.6 Business-cycle analysts such as

6 Orley Ashenfelter and George Johnson, "Bargaining Theory, Trade Unions and
Industrial Strike Activity," American Economic Review 40 (1969): 35-49; Walter Korpi,
The Working Class in Welfare Capitalism: Work, Unions and Politics in Sweden (London:

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

Albert Rees have interpreted the ge

unemployment on strike activity as
in the relative bargaining positions of
Still, their formal modeling, measu
have characteristically taken the for
scribing the frequency, size, and/or
implicitly conceiving of those featu
behavior. By concentrating on the inc
furthermore, they have ruled out a
interaction between workers and em
real life, the beginning, the develo
strike always take place within a co
gaining that includes a continuous, m
relative strength of the parties, and o
and tactical considerations that are dis
that the workers will gain or lose the
model would take direct account of that mutual interaction.
Eric Batstone, Ian Boraston, and Stephen Frenkel stand out
among analysts of industrial conflict for their attempt to fill
the gap between individual and collective action by looking
directly at the mobilization process. Their results are promis-
ing. Yet notice how they set up the problem:

There are a number of factors which have to be taken into

account in any satisfactory and total explanation of strikes. First,
it is essential to recognize that strikes, as an expression of in-
dustrial conflict, reflect the subordination of workers within
industry and, indeed, society more generally. Second, the in-
stitutions of collective bargaining, forms of social and political
integration more generally, and management and trade union
organization all have some relevance to the probability of strike

Routledge & Kegan Paul, 1978) and The Democratic Class Struggle (London: Routled
& Kegan Paul, 1983); Douglas Hibbs, Jr., "Industrial Conflict in Advanced Industr
Societies," Ameúcan Political Science Review 70 (1976): 1033-58, and "On the Politic
Economy of Long-Run Trends in Strike Activity," British Journal of Political Scienc
(1978): 153-175; Paul K. Edwards, Strikes in the United States, 1881-1974 (Oxford:
Blackwell, 1981).
7 Albert Rees, "Industrial Conflict and Business Fluctuations," Journal of Politic
Economy 60 (1952): 371-382.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

action. However, the extent to wh

will be most immediately determin
among workers themselves. For thi
the distribution of power among
cabularies employed is crucial to a

Thus the problem of explaining

problem of accounting for the be
concessions to management and
model remains that of a single

The same theoretical habits also appear in other domains o

collective action. Consider something so mundane as a demo
stration. We have the habit of representing it as expressing th
unified will of a fairly well-defined group: a demonstration o
veterans, students, local residents, or something of the sor
But that representation obscures the reality in two important
ways: first, because - as every organizer of demonstration
knows - the action of demonstrators results from an effort at
construction, often painful, that itself often involves the strik-
ing of bargains; second, because every demonstration involves
not one but at least four parties: the poeple who gather in the
street, the object of their attention (ordinarily a symbol, an
organization, or an important person), the immediate
spectators, and the social base the demonstrators claim to
represent. Furthermore, authorities ordinarily pay attention to
demonstrations, even when the demonstrators are attacking or
supporting someone else.
Several of those parties are commonly present at the action;
in MacCannell's study of "protest demonstrations" in the
United States around 1970, "protest targets are encountered
by demonstrators 60% of the time . . . bystanders are present
in 94%, police in 89%, media in 54%, and organized counter-
demonstrators in 36% of the demonstrations."9 The interac-

8 Eric Batstone, Ian Boraston, and Stephen Frenkel, The Social Organization of Strikes
(Oxford: Blackwell, 1978), p. 4.
9 Clark McPhail and Ronald T. Wohlstein, "Individual and Collective Behaviors
within Gatherings, Demonstrations, and Riots," Annual Review of Sociology 9 (1983):

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

tion is not incidental but integral to t

again we need models with multiple
Current models of social movemen
lems. Available models almost unifor
as the expressions of particular gro
sure, assign those groups unusual st
posite, and fluid. More sophisticate
divide the participants into several co
base from which the movement dr
the movement but remain inactive, m
ers, and so on. Nevertheless, the norm
consists in identifying a more or less
and then explaining the movement
of the social characteristics of thos
classic sociological procedure. When
what he regards as a profound transformation of social
movements in the postindustrial era, for example, he sticks to
the same kind of argument; he characterizes the new social
movements as:

segmented, reticulated, polycephalic structures. The mo

is composed of diverse, autonomous units that expe
portant part of their resources on internal solidarity.
of communication and exchange keeps the cells in cont
each other; information, persons, and models of beh
culate in the network, moving from one unit to an
thus promoting a certain homogeneity of the whole st
Leadership is not concentrated but diffuse; it is limited
cific aims, and different people can assume leaders
depending on the function to be fulfilled.10

The image of a special sort of group, but a group

less, informs Melucci's description. His descripti
makes an implicit contrast with a party or a labo
There lies the problem. The proper analogy to a social
movement is neither a party nor a union but a political cam-
paign. What we call a social movement actually consists

10 Alberto Melucci, "Mouvements sociaux, mouvements post-politiques," Revue in-

ternationale d'action communautaire 10 (1983): 14.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

in a series of demands or challe

name of a social category that
position. Although we sometimes
conflicts, the social movement
doing collective business in We
nineteenth century. Its deman
groups, as electoral campaigns d
cases the interaction among actor
unity of the movement.
The analysis of the workers' m
Touraine and his collaborators h
require, say Touraine and comp
and political institutions be com
by an analysis of working-class
class as an actor in the central co
that is, of syndicalism as a soci
add the obligation "to consider
social actor defined at once by
positive reference, shared with
orientations of industrial society.
erences to actor and action, man
analysis, which turns out later
relations among workers, manage
cials, and the public.
The reality of the social move
mystification, shared by both sid
tifies the current actors with a
very moment when the self-st
movement rush to create coalitions, eliminate rival leaders,
solidify their own bases, avoid visible breaks, and organize
public displays of unitary will. That preparatory work does not
belie the sincerity, seriousness, popular support, or efficacy of
many social movements. It builds the structure of the social

11 Alain Touraine, Michel Wieviorka, and François Dubet, Le mouvement owner

(Paris: Fayard, 1984), p. 22.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

phenomenon bearing that name. Su

quires a model of interaction with mu
a single-group model.

Static Models

If the theoretical disarray with respect to connections be-

tween individual and collectivity in the course of collective
actions results in part from the application of models of
unitary action to realities involving multiple actors, it also
results from the static character of current models of collective
action. Having generally abandoned natural-history models
postulating stages for revolutions, social movements, or other
forms of collective action, investigators have not been able to
substitute other models giving the same sense of internal
change. Even Manuel Castells, for all his knowledge of and
sympathy with urban social movements, proposes a model that
serves mainly to match the aims and composition of a social
movement with the social structure surrounding it. As for the
process that produces movements, he settles for the declara-
tion that "the production of the structural formula leading to
urban social movements is specific to each national-cultural
context, and any attempt to find a general formulation is to
resort to metaphysics."12 In trying to avoid metaphysics, how-
ever, Castells sends us to cultural history, without telling us
what features of cultural history to explore.
Castells's abdication resembles that of other analysts.
Neither familiarity with concrete collective action, readiness to
make comparisons, nor theoretical imagination suffices to
overcome the obstacles to dynamic representations of popular
collective action. To the extent that we adopt single-actor
models with unitary dispositions, and lodge the explanation of

12 Manuel Castells, The City and the Grassroots: A Cross-Cultura I Theon of Urban Social
Movements (Berkeley: University of California Press, 1983), p. 324.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

action in those dispositions, we c

models in which changes of disp
phase of the action. To the extent
accounts of those changes - for
each successive action in the actor's material interests - we take
on a task that is impossible in the present state of knowledge.
Yet we do have some clues about dynamics. At the level of
micromobilization, Mark Granovetter, Clark McPhail, John
Lofland, and other researchers have formulated useful models
of the communications processes that transform a passive
aggregate into an acting group.13 William Gamson's research
team, for example, conducted a series of experiments on re-
sistance to unjust authorities.14 The central experiment ex-
posed its subjects to a supposed researcher who systematically
and progressively violated the group's initial agreement by
seeking to influence their testimony. According to Gamson's
analysis, open resistance to that violation, when it occurred,
resulted from the coalescence of three kinds of actions. Or-

ganizing actions increase the group's collective capacity, di-

vesting actions neutralize ties to the authority, and reframing
actions create a new context for the interpretation of the au-
thority's actions; in this case the new frame labels the authority
as unjust.
A successful rebellion against unjust authority, in this
model, results from a sequence of organizing, divesting, and
reframing acts. But the experiment simplifies its analytic
problem by giving the authority very little room for maneu-
ver; he reveals and reinforces his injustice without really being

13 Mark Granovetter, "Threshold Models of Collective Behavior," American Journal

of Sociology 83 (1978): 1420-43; Clark McPhail and David L. Miller, "The Assembling
Process: A Theoretical and Empirical Examination," American Sociological Review 38
(1973): 721-735; McPhail and Wohlstein, "Individual and Collective Behaviors"; John
Lofland, "Collective Behavior: The Elementary Forms," in Morris Rosenberg and
Ralph Turner, eds., Social Psychology: Sociological Perspectives (New York: Basic Books,
14 William Gamson, Bruce Fireman, and Steve Rytina, Encounters with Unjust Au-
thoiity (Homewood, 111.: Dorsey, 1982).

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

able to bargain or change strategy.

mains poorly represented. Therefo
best a specification of the necessary
consequences of interaction. It is not
of the interaction itself.
Mark Granovetter has edged in on the dynamics of mi-
cromobilization from another angle. His "threshold models of
collective behavior" postulate a distribution of actors each
having his own calculation of costs and benefits for participa-
tion in a particular action, a calculation that depends strongly
on the proportion of others who are already acting or com-
mitted to act. Activation of the entire group, if it occurs,
depends on the successive arrival of different actors at their
thresholds - for example, 20 percent of the rest, 40 percent of
the rest, 90 percent of the rest - as others join the action. In
these models, two groups with identical average propensities
to act (e.g., two groups in which the average member is pre-
pared to join the action when 40 percent of the rest are
already involved) can differ significantly in their collective
propensity to act, depending on the distribution of individual
thresholds. Granovetter's models lend insight into the
information-gathering about other people's commitment that
commonly precedes a risky action:- milling, reviews of tactics,
recollections of previous encounters, appeals to solidarity,
striking of bargains among pairs of participants, and so on.
Granovetter achieves that much dynamism and verisi-
militude by deliberately adopting a purposive frame. As he
puts it:

These models treat the aggregation of individual preferences;

they do not consider how individuals happen to have the prefer-
ences they do. . . . Most existing literature, by contrast, channels
its main effort into determining how norms, motives, and pref-
erences are caused and assumes that nothing more need be
done to explain collective behavior. I maintain instead that once
these are known, there is still a great deal to be done, and that
outcomes cannot be determined by any simple counting of pref-
erences. This will be particularly clear in cases where a very

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

small change in the distribution of

difference in the outcome. Analys
nation of preferences could not e

It is only a step from this reaso

responses to antagonists operate
sponses to potential allies that Gr
Nevertheless, it is two steps or m
actors are not simply deciding
also which of any array of acti
antagonists likewise make choices. The models of micro-
mobilization lack at least two elements of a truly dynamic
analysis: (1) a description of the transition from one stage to
the next; (2) a representation of the interaction of parties in
opposition or coalition. For these purposes it would be useful
to construct models of strategic interaction in which each dis-
placement of one party incites more or less calculated re-
sponses on the part of the others. We therefore need to adapt
and improve models of rational action.

Causal vs. Purposive Models

Ideally, it would be useful to create models taking as their

starting point the structure of social relations characterizing a
set of actors, cataloging the means of action open to individu-
als and groups, allowing for the creation of new social ties, and
centering on strategic interaction. In the present state of
knowledge, there is no way of making such an analysis
strongly causal. It will have to be purposive; that means taking
the actors and decision rules as given at the start of the
interaction. Most likely the analytic program will have to break
into two parts: (1) a causal analysis determining what actors
are available for collective action, what decision rules their
action will follow, and within what constraints they will oper-

15 Granovetter, "Threshold Models," p. 1421.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

ate; (2) a purposive analysis taking a

constraints as given. For the latter, w
rational action but of rational interaction.
Models of rational interaction generally assume a fixed set
of actors with a specified amount and type of information
about each other's identities, actions, and interests. They usu-
ally require specifications of (1) the actors, (2) their interests,
(3) the decision rules adopted by each actor, (4) current values
of the elements of those decision rules. Those elements, in
their turn, typically concern (5) probable costs of the various
sequences of action that are available to each actor, (6) proba-
ble benefits of each of those sequences, (7) the capacity of each
actor to sustain the costs of each sequence. Thus, crudely
speaking, we explain a single actor's participation in a social
movement as a function of a choice among multiple alterna-
tives whose relative attraction depends on a product:

(expected benefits - expected costs) x (capacity to act)

A single actor may, of course, consist in an individual or a

whole set of individuals.
In order to convert such a model of rational action into a
model of rational interaction, we make the expected benefit
and costs (and possibly some of the other variables listed
above) for each actor depend on the actions of other actors,
and institute communication among the actors. Within th
framework, specifying relevant actors, interests, decision rules,
costs, benefits, and capacities - not to mention the relation
among them - sets the theoretical challenge.
To see the nature of the challenge, take the intelligent
synthesis of Guy Caire, La greve ouvrière. After giving a com-
prehensive summary of research on strike activity, Caire find
himself at a loss when he arrives at the actual initiation and
internal development of strikes. He makes an essential dis-
tinction himself: "Cost-benefit analysis has an ex post inspira-
tion; in contrast, gaming analysis takes an ex ante position. In

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

that domain, alas, almost everyth

He resorts to the fifty-year-old m
sents the crossing of a concession
agement and a resistance curve on t
a function of a strike's duration- which comes down, as Caire
himself remarks, to a static model of the threat of conflict
rather than a dynamic model of battle or bargaining.
Furthermore, Hicks's model assumes a group of workers
who are already organized and decided; it thereby finesses the
essential problem instead of resolving it. When he returns to
the problem of strike strategy, Caire borrows a military anal-
ogy and proposes the stages of outbreak- battle- settlement;
although he has intelligent comments to make on each stage, he
is unable to order them according to an explicit logic. From a
theoretical point of view, the analysis reaches a dead end.
There is nothing unusual about that dead end. That is the
point. Caire finds himself in the normal position of students of
collective action. Two formidable obstacles block the road: the
complexity of strategic interaction and the difficulty of
dynamic modeling.

Games and Strategic Interaction

Yet we have at our disposal two related analytic traditions

that permit dynamic modeling of strategic interaction: game
theory and simultaneous-equation modeling of mutual-
influence processes. To adopt either one, we must be pre-
pared to use purposive models, at least to the extent of at-
tributing decision rules to each actor, and must specify actors,
interests, decision rules, costs, benefits, and capacities to act.
(We may still, to be sure, derive each of these from empirical
description and/or causal analysis.)
Robert Axelrod's studies of the prisoner's dilemma show

16 Guy Caire, La greve ouvrière (Paris: Editions Ouvrières, 1978), p. 156.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

that the specifications may be crude a

results.17 In its elementary form, the
interaction in which the self-interest
leads to undesirable outcomes for bo
action leads to more desirable outcome
nation of self-interested and cooperat
the two participants leads to an even
for the self-interested party and an
the other. Many real-life situations
dilemma: environmental pollution, a
gaining, and even the natural encount
the possibility, but not the certain
course of a single interaction, both pa
to avoid cooperation and to pursue t
without regarding the interests of th
If, however, the parties enter into f
situation changes. During repeated e
egoistic parties tend to gain from str
cooperation with sharp discriminat
pending on whether the other part
to serve his individual interest. The st
start by cooperating in our first en
your response faithfully - tends to w
tic strategy. The advantage of a strate
furthermore, grows with (a) the pr
counters, (b) the sharpness of discrimination among re-
sponses, (c) the certainty of the other party's identity, his
actions, and their consequences. Even in the midst of a popu-
lation of irremediable egoists, a cluster of tit-for-tat players
tends to win. So the analysis demonstrates, among other
things, the advantages of coalition.
In that regard, Axelrod's findings recall Mancur Olson's
analysis, in The Rise and Decline of Nations, of the probability
that small groups and groups having access to selective incen-

17 Robert Axelrod, The Evolution of Cooperation (New York: Basic Books, 1984).

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

tives will form "distributional coalitions."18 Industrial em-

ployers, craft unions, and producers' associations serve as
examples. Such groups, according to Olson's scheme, gain
from their organizational advantage by influencing the pro-
duction and distribution of goods. In the long run, that influ-
ence leads to sclerosis, or at least to pronounced deviation
from classic market rationality. Whence a cycle following each
great national struggle: first a relatively open expansion dur-
ing which those who have the capacity to form distributional
coalitions start to do so, then slowdown as a result of the action
of the coalitions.

There exist, according to Olson, two ways of escaping from

that silting up: either smash the coalitions from time to time,
or assure the formation of global coalitions whose particular
advantages also serve the general interest. In the schemes of
Axelrod and Olson, the certainty and continuity of social rela-
tions facilitate the formation of stable coalitions serving the
mutual interest of actors who continue to follow their particu-
lar interests and global coalitions serving the general interest.
Axelrod's theoretical and experimental results immediately
suggest analogies to legislative bargaining, military and diplo-
matic alliances, and collusion among industrial firms. These
analogies, in their turn, suggest the possibility of generalizing
the foundations of game theory to the level of large-scale
structural processes. That is, in fact, the project proposed
recently by Jon Elster, and pursued (in rather a different
direction from Elster's) by Andrew Schotter.19
Some assurance that the effort is worth it comes from two
sources: from analyses of international conflict, in which in-
teresting and crudely plausible models based on game theory
and simultaneous-equation modeling are proliferating, and

18 Mancur Olson, The Rise and Decline of Nations (New Haven: Yale University Press,
19 Jon Elster, "Marxism, Functionalism, and Game Theory: The Case for Method-
ological Individualism," Theory and Society 11 (1982): 453-540; Andrew Schotter, The
Economic Theory of Social Institutions (Cambridge: Cambridge University Press, 1981).

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

from scattered but promising ap

proaches to domestic conflict. One
Richard Berk's application of a gam
1972 confrontation at Northwestern
Barbara Salert and John Sprague's
simulation of the conditions for "p
quences of violent encounter given a

Nevertheless, game theory alone will not suffice. Eventually

we will have to place the character of ties among persons at
the center of the analysis. Many ties that constitute and shape
social life involve so little strategic interaction as to require
other modes of analysis. Communications networks, daily re-
lations among bosses and workers, circulation of tax money,
paths of disease, movements of capital, chain migrations, and
promotion ladders all sometimes involve strategic interaction.
But their crystallization in durable structures and their in-
cessant transformation call for a specifically structural analysis.
Even within the zone of strategic interaction, none of this is
easy. In popular collective action, the number, identity, and
boundaries of relevant actors change frequently. Identifying
the actors and the stakes, not to mention the rules of the
game, often requires a profound analysis of class structure
and political process. Confronting such complexity, we will
either have simplifications or build very complex models.
Thus we return to constructive models of collective action.
Reflecting on rational interaction adds two elements to con-
structive models. First, we must consider each of the social ties,
and especially those between antagonists, as a site of more or
less continuous communication and negotiation following a
strategic logic whose consequences transform the structure
and the content of its ties. Thus a dynamic model of strikes

20 Richard Berk, "A Gaming Approach to Crowd Behavior," American Sociological

Review 39 (1974): 355-373.
21 Barbara Salert and John Sprague, The Dynamics of Riots (Ann Arbor: Inter-
University Consortium for Political and Social Research, 1980).

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

should take into account at leas

within these pairs:

workers/government officials
unions/government officials
bosses/government officials

Often a fifth participant enters the picture: some sort of pub-

lic. The problem is complex, but at least it is explicit and draws
attention to stakes, forms, and consequences of each of these
Finally, reflection on models of rational interaction suggests
the value of a separate analysis of the established forms of
interaction. For each pair of interlocutors adopts and mod-
ifies a rather limited set of means of negotiation - what one
may call the characteristic repertoire of the pair. In the case of
the pair workers/bosses, for example, the strike, the lockout,
the layoff, the firing, the plant meeting, sabotage, absen-
teeism, and a few other forms of action which vary by region,
industry, and period, have constituted the standard repertoire
in many Western countries for close to a century. In our
French conflicts of 1906, we saw some of these routines re-
playing, with strategic variations, in encounter after en-
counter. As the events of 1906 suggest, a limited repertoire is
quite compatible with maneuver and change. While there are
many variations and combinations, and while each party seeks
to gain the advantage by innovating within each established
form and by insisting on its own interpretation of custom and
law, in general the parties know and support the rules of the
Sometimes an important innovation, such as the occupation
of workplaces, comes into being despite the resistance of at
least one of the parties. Most of the time the repertoire re-

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to

mains in place and serves as a well-def

tion. Recognizing that frame simplifi
what seemed at first an unlimited ran
to a constrained choice among relati
of action. Our next theoretical step co
of models of strategic interaction
within the framework of action repe
each pair of interlocutors.
At first glance, this program looks
historical concreteness for theoretical abstraction. At second
glance, however, it brings us back to history. For it forces us to
ask precisely the questions entailed by a sound historical
understanding of such events as the conflicts of 1906: who
were the actors, how were they organized, what relations of
power, conflict, and solidarity did they maintain with other
actors, what were their interests and strategies, with whom did
they interact, within what limits, with what outcomes. Histo-
rians, too, need to turn from static, strictly causal, single-actor
models that leap uncritically from individual interests to col-
lective action.

* This paper draws heavily on my "Action collective et mobilisation individuelle,"

forthcoming in Sur l'individualisme, edited by Pierre Birnbaum and Jean Leca (Paris:
Presses de la Fondation Nationale des Sciences Politiques, probably 1985). It also
borrows some material from my "Fights and Festivals in Twentieth-Century Ile de
France," Working Paper 305, Center for Research on Social Organization, University
of Michigan, December 1983. The descriptions and enumerations of events in the
paper come from the chronological files of the study of social change and collective
action in France now going on at the New School for Social Research, after residing
for many years at the University of Michigan. A brief description of those files
appears in my "Social Change and Collective Action in France and Britain. Summary
of Research Completed and Data Available at the End of October 1983;" CRSO
Working Paper 304, December 1983. A bibliography of research reports using the
files appears in "Selected Papers, 1963-1984 from the Study of Social Change and
Collective Action (3/84)," CRSO Working Paper 313, March 1984. The most extended
analysis from those files appears in The Contentious French (Cambridge: Belknap Press
of Harvard University Press, forthcoming around January 1986). The National Sci-
ence Foundation supports the research reported here.

This content downloaded from on Sun, 17 Dec 2017 17:13:59 UTC
All use subject to