12 Sympathizers (Oegema, Klandermans) PDF

Why Social Movement Sympathizers Don't Participate: Erosion and Nonconversion of
Support
Author(s): Dirk Oegema and Bert Klandermans
Source: American Sociological Review, Vol. 59, No. 5 (Oct., 1994), pp. 703-722
Published by: American Sociological Association
Stable URL: https://www.jstor.org/stable/2096444
Accessed: 09-06-2019 15:33 UTC
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 [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Sociological Association is collaborating with JSTOR to digitize, preserve and

extend access to American Sociological Review
This content downloaded from 172.73.47.201 on Sun, 09 Jun 2019 15:33:13 UTC
All use subject to https://about.jstor.org/terms
WHY SOCIAL MOVEMENT SYMPATHIZERS DON'T PARTICIPATE:
EROSION AND NONCONVERSION OF SUPPORT*
DIRK OEGEMA BERT KLANDERMANS

Free University, Amsterdam Free University, Amsterdam
Social movement organizations face the challenge of converting action preparednes

into action participation, and accordingly, they must deal with factors that lead to
nonparticipation. We identify two routes to nonparticipation- "nonconversion" and
"erosion." Nonconversion refers to a movement's failure to transform sympathizers
into active participants. Erosion refers to the loss of sympathizers. We provide a theo-
retical framework for explaining nonconversion and erosion and apply it to a longi-
tudinal study of a mobilization campaign of the Dutch Peace Movement: the People's
Petition against cruise missiles. Data came from telephone interviews we conducted
among random samples (N = 224) from the populations of four communities before
and after the campaign (in May and November 1985). We use binomial logistic re-
gression analyses to test several explanations of nonconversion and erosion. Factors
that led to erosion were moderate action preparedness at the beginning of the cam-
paign, declining preparedness during the campaign, and a social environment per-
ceived by participants as becoming less and less supportive. Factors that led to
nonconversion were moderate but stable action preparedness combined with the pres-
ence of barriers to action and an indifferent social environment. Results are discussed
and related to social movement literature.
Activating individuals who are already There is every reason to assume that the
sympathetic to a movement-or, in proportion of sympathizers that potentially
Klandermans's (1984) words, action mobili- can be activated varies over time, by the
zation-is more difficult than one might situation, and by the activity. For instance,
imagine. Movement participation, we have Klandermans (1984) observed that, within a
previously argued, evolves in four steps: (1) single union, union members varied consid-
one becomes a sympathizer of the move- erably in their willingness to participate in
ment, then (2) a target of mobilization at- union action depending on the specific ac-
tempts; next (3) one becomes motivated to tion proposed. Moreover, in different com-
participate, and finally, (4) one overcomes panies the proportion of the union member-
the barriers to participation (Klandermans ship willing to participate in industrial ac-
and Oegema 1987). Action mobilization in- tion also varied considerably. Studies of
volves the last three steps of this sequence. many social movements record a similar
Our analysis of a large peace demonstration variability in participation. To cite only a
in the Netherlands reveals what every orga- few examples: Walsh (1988) observed dif-
nizer knows from experience: There is a big ferent levels of participation in the various
difference between being a sympathizer and communities surrounding Three Mile Is-
becoming an active participant, and many a land; Henig (1982) found variation among
sympathizer fails to become active. different neighborhoods in Boston during
the anti-busing protests; and Bridt,
* Direct all correspondence to Dirk Oegema, Klandermans, and Kroon (1987), in a study
Dept. of Political Science, Free University,
of the women's movement, found substan-
Amsterdam, the Netherlands. We thank the Edi-
tial differences in levels of participation, de-
tor and reviewers of ASR for their useful com-
ments on previous versions of this paper. [Re- pending on the particular activity, among a
viewers acknowledged by the authors include population of women in a Dutch town. Such
Benigno Aguirre, Robert Kleidman, Nelson differences in levels of participation within
Pichardo, Suzanne Staggenborg. -ED.] (or between subsets of) the same population
American Sociological Review, 1994, Vol. 59 (October:703-722) 703
704 AMERICAN SOCIOLOGICAL REVIEW
can originate at each transition point in the THEORY

four-step model of movement participation,
Generalized Action Preparedness,
and the relative contribution of each step to
Specific Action Preparedness, and
the final number of participants points to the
Action Participation
strengths and weaknesses of a mobilization
campaign. To further define and explain nonconversion
In this paper we depart from the usual ap- and erosion, we distinguish and describe
proach to studying movement campaigns three stages in the mobilization process: gen-
and focus on a campaign's failure to acti- eralized action preparedness, specific action
vate sympathizers. The literature on social preparedness, and action participation. Gen-
movements offers little information on eralized action preparedness describes an
nonparticipation in response to action mobi- individual's expressed willingness to support
lization campaigns. Mass mobilizations a movement, to take part in different types
have generally been studied in terms of of collective action the movement might
what they achieve rather than what they fail stage. Adherents of the movement possess
to achieve. Consequently, we know more this disposition to a greater or lesser degree.
about participants than about nonpartici- A movement's mobilization potential in a so-
pants (see McAdam 1986 for a similar criti- ciety can be thought of as the proportion of
cism). There might be a methodological rea- individuals within that society that is in-
son for this neglect of the dynamics of clined to support the movement (Klander-
nonparticipation-a lack of proper quantita- mans and Oegema 1987). This proportion is
tive longitudinal data. Without such data, relatively stable over time and defines the
which help us map preexisting levels of participation limits of mobilization cam-
sympathy, it is difficult to make valid esti- paigns. The mobilization potential of a
mates of the proportion of movement sym- movement, then, theoretically encompasses
pathizers who actually end up being inac- those individuals with a generalized action
tive. preparedness greater than zero.
We propose two terms to identify two dif- Generalized action preparedness for a par-
ferent forms that nonparticipation can take. ticular movement can be seen as a function
First, a campaign can fail to transform sym- of the existence and magnitude of grievances
pathizers into active participants, an outcome and the existence and appeal of a movement
we call nonconversion. Second, people who addressing these grievances (Oegema and
initially support the movement may change Klandermans 1992). The more individuals
their minds and become unwilling to become who believe that a specific movement can ef-
active. In this case, the problem isn't that fectively mobilize to redress shared griev-
sympathy is not converted into action, but ances, and the more serious these grievances
rather that sympathy disappears-hence our are, the higher the generalized action pre-
use of the term erosion. Nonconversion and paredness for this movement (Schwartz and
erosion of support are two measures of a mo- Paul 1992).
bilization campaign's effectiveness or inef- But ideological commitment alone does
fectiveness. From the standpoint of move- not guarantee participation in concrete ac-
ment organizers, of course, neither is desir- tion: Action mobilization campaigns must
able: Organizers must convert support into transform generalized action preparedness
action. into actual participation. In successful mobi-
We assume that rates of nonconversion and lization campaigns, generalized action pre-
erosion vary over time and across move- paredness is successfully converted into the
ments, movement organizations, campaigns, preparedness to participate in specific ac-
actions, communities, and subsets of populations, and this specific action preparedness
tions. We elaborate here on these assump- materializes as actual action participation.1 If
tions and offer explanations for this varia-
tion. We test these explanations on data we
1 Although our tripartite distinction is related
collected from a study of participation in the to McCarthy and Zald's (1976) distinction be-
Dutch peace movement. tween movement adherents and movement con-
EROSION AND NONCONVERSION OF SOCIAL MOVEMENT SUPPORT 705
GENERALIZED ACTION ~.SPECIFIC ACTION -4 PARTICIPATION

PREPAREDNESS ' PREPAREDNESS -
Figure 1. Process of Action Mobilization
individuals are not being targeted by mobili- individuals perceive the ratio of costs to ben-
zation attempts, however, it is unlikely that efits as becoming less favorable over time,
they will participate (Klandermans and Oeg- and/or their grievances are no longer press-
ema 1987). Incentives contingent on the par- ing, and/or their sympathy for the movement
ticular action and the circumstances of the wanes.
individual further influence an individual's
motivation to participate. For any individual,
The Antecedents of Nonconversion and
the lower the perceived costs of participation
Erosion
and the higher the benefits, the more moti-
vated he or she will be to participate (Klan- Nonconversion, then, seems to be related to
dermans 1984; Opp 1989a). Whether moti- circumstantial factors that make the conver-
vation-that is, specific action prepared- sion of action preparedness (general or spe-
ness-is converted into actual participation cific) into actual participation less likely.
depends on the presence of barriers, for ex- Mobilization attempts, incentives, and barri-
ample sickness or lack of transportation ers do not occur randomly throughout a
(Klandermans and Oegema 1987). The more population, but coincide with characteristics
barriers that exist, the less likely it is that an of political opportunity structures, character-
individual will participate. The process of istics of movement organizations, campaign
action mobilization is represented by the characteristics, specific actions, characteris-
simple schema shown in Figure 1. tics of individual communities, and social
We can now define nonconversion and ero- categories. The more favorable the political
sion in terms of these concepts. Noncon- opportunity structures, the more likely it is
version is the nonparticipation of individuals that action preparedness will be converted
who are prepared to participate but somehow into participation (Duyvendak forthcoming;
fail to convert their preparedness into actual Koopmans forthcoming). The more extensive
participation. Erosion is the nonparticipation a movement's networks, the more likely it is
of individuals who, though once prepared to that individuals will be targeted (Snow, Zur-
participate, have changed their minds and cher, and Ekland-Olson 1980). Sophisticated
lost their preparedness to take action. mobilization techniques increase the likeli-
We hypothesize that nonconversion occurs hood of conversion (Oliver and Marwell
when individuals have not been targeted by 1992). Strategic choices of specific actions
mobilization attempts, and/or they find that influence the balance of costs and benefits
the ratio of costs to benefits is unfavorable, (Oberschall 1973; McAdam 1986). Charac-
and/or barriers prevent them from participat- teristics of communities, such as their social
ing. Erosion, we hypothesize, occurs when and political composition, help determine in-
centives for participation (Opp 1988, 1989b).
And persons from different social catego-
stituents, the two distinctions are not identical: ries-such as male versus female or young
The latter concerns attitudinal support, whereas versus old-traditionally differ in their ac-
generalized action preparedness, specific action
tion participation (Barnes and Kaase 1979).
preparedness, and actual participation all concern
Erosion depends on changing circumstan-
action participation, be it preparedness to act or
ces that undermine existing action prepared-
actual participation, and the three stages com-
bined can be conceived of as a funneling process ness, be it general or specific. Although ero-
in which sympathizers face increasingly intense sion of support occurs not only in the context
demands. of action mobilization-other contexts might
be movement decline, changes in public opin- DATA AND METHODS

ion, and issue attention cycles-we limit our
inquiry to the erosion of support during ac- The data come from a longitudinal study of
tion mobilization campaigns. We hypothesize the mobilization campaign for the People's
that action mobilization can backfire and pro- Petition against cruise missiles, organized in
duce a reverse effect because it generally po- 1985 by the Dutch peace movement. The
larizes a population-cognitively and so- People's Petition was the final act of a cam-
cially. Cognitive polarization takes place be- paign that had lasted almost 10 years and
cause, in the context of action mobilization, was able to orchestrate the two largest dem-
the mobilizing organization's features be- onstrations the Netherlands had ever experi-
come especially distinct: Goals and means enced (Oegema 1991; Rochon 1988). The
become pronounced, rhetoric changes, and Petition was an attempt to demonstrate once
interactions with opponents become confron- again the movement's maximum strength. In
tational. Opposing parties argue, previously the months preceding November 1, 1985-
mild debates sharpen, and latitudes of indif- the day the Dutch government had commit-
ference become smaller and smaller. Oppo- ted itself to decide on deployment-the
nents and countermovement organizations are movement tried to have a petition against de-
often extremely skilled in creating caricatures ployment signed by as large a proportion of
of the movement and sowing doubt in the its constituency as possible. Its organizers
hearts of halfhearted sympathizers (Conover estimated-and the polls indicated-that
and Gray 1983; Mansbridge 1986). Often a they would be able to win the support of
movement organization bears within itself the close to 50 percent of the Dutch population
ammunition for a countercampaign (Chafetz 15 years old and older. If they could achieve
and Dworkin 1987). this goal, the government would be in trouble
In other words, action mobilization forces because a large proportion of its own con-
a shift in public discourse: In the media and stituency would be among the signers.
in informal conversations among citizens, To realize a maximum response, organiz-
public discourse becomes increasingly fo- ers chose an action that made it easy for even
cused on campaign issues. As a result, indi- marginal sympathizers to support the move-
vidual citizens and societal actors are forced ment: signing a petition directed at the par-
to take sides. liament. Assuming that merely signing a pe-
Action mobilization thus implies social tition required little or no effort, the organiz-
polarization, for it splits multi-organizational ers supposed that all they had to do was to
fields (Klandermans 1989; Rucht and Della reach everybody who was prepared to sup-
Porta forthcoming) into the movement orga- port the peace movement. Accordingly, they
nization's alliance and conflict systems. An invested a great deal of effort in a campaign
individual's social environment becomes re- designed to guarantee that no potential sup-
arranged into proponents and opponents of porter would be overlooked. They mailed
the movement. Most individuals live in a signature cards to every postal address in the
fairly homogeneous social environment and country and made sure that as many homes
will find themselves unambiguously in one as possible were visited by activists collect-
camp or the other, but some may discover ing the cards. A single card could carry up to
that groups, organizations, or parties with five signatures. Signed cards could be mailed
which they identify are suddenly in their in or handed to the collectors going from
enemy's camp or that groups and people with door to door. Five-and-one-half million sig-
whom they feel little affinity have become nature cards were distributed, and an esti-
allies. If they don't like the social identity mated 50,000 volunteers were involved in
implied by these new arrangements, they door-to-door card collection (this number, by
may choose to detach themselves from the the way, was 30,000 short of what the move-
movement. In the context of election cam- ment had calculated would be necessary
paigns, Lazarsfeld, Berelson, and Gaudet [Kriesi and Van Praag 1988]). The campaign
(1948:56-64) referred to this process of con- proper started on August 31, but from May
flicting identifications as "cross pressure" onward local peace groups were involved in
(see also Lane 1964:197-203). organizing. In many communities these
groups made use of the canvassing experi- Those people who in May and June stated
ence of local political party activists. that they would sign the petition were inter-
In the course of the campaign, the cruise viewed for a second time between November
missile issue was hotly debated within and 9 and November 28 (after the campaign was
between political parties. The government- over) so we could ascertain whether they did
composed of the Conservatives and the in fact sign. Unless otherwise specified, all
Christian Democrats-feverishly attempted analyses are based on sample 1. However, we
to formulate a compromise. During the cam- also selected a second sample that consisted
paign it became clear that, over the preced- of individuals who were interviewed only in
ing two to three years, the peace movement November (sample 2). We use this second
had become an antigovernment coalition, sample occasionally to control for the effects
welcoming everyone opposed to the govern- of repeated measurement.
ment and the parties in office. As we will see, For both samples 1 and 2, 100 addresses
this development had a significant impact on were randomly drawn from the most recent
sympathizers who identified with one of telephone directories for each of the four se-
those parties. Meanwhile the movement be- lected communities. A letter was mailed to
came more controversial and, occasionally, each address explaining that the residents
countermobilization occurred. would be contacted by telephone for an in-
Eventually, 3.75 million people signed sig- terview regarding peace, disarmament, and
nature cards. This number constituted 30 per- the peace movement. As it turned out, 16
cent of the Dutch population 15 years old cases in sample 1 (May/November) and 9 in
and older. Although this was an impressive sample 2 (November only) no longer be-
percentage that was possible only because of longed to the population (they had moved to
sophisticated organizing, it was a disappoint- another town or their phone had been discon-
ment to the movement. Thirty percent was nected); thus our net sample sizes were 384
too far below their initial goal and also was and 391 respectively. We achieved response
too low a turnout to unsettle the government. rates of 58 percent (224 cases) and 61 per-
On November 1, 1985 the government did cent (231 cases) respectively for the two
indeed decide to deploy cruise missiles in the samples; for both samples we could not ar-
Netherlands. range interviews for 160 addresses. Of these,
11 percent in the first sample (8 percent in
the second) proved impossible to contact. We
Samples
were able to reach the remaining 89 percent
Between May 23 and June 13, 1985 (at the (92 percent), but for various reasons they
start of the campaign) we conducted tele- were not able or willing to interviewed: 17
phone surveys among random samples of the percent (15 percent) were not in good health
population in four communities from differ- or were too old; 12 percent (12 percent) had
ent parts of the Netherlands (sample 1).2 no time; 11 percent (11 percent) were op-
posed to interviews in general; 4 percent (15
2 We applied the same interview strategy we percent) disliked the topic; 30 percent (20
had used in previous studies (Klandermans and
percent) mentioned some other reason; and
Oegema 1987). In the regions where we con-
15 percent (19 percent) refused without giv-
ducted our field work, telephone density was
above 95 percent, so we did not risk much bias ing any clear reason. The resulting samples
by restricting ourselves to addresses that could be are, of course, not random samples of the en-
reached by phone. To increase response rates, we tire Dutch population. But we were inter-
adopted a strategy suggested by Frey (1983): We ested, not in assessing the extent of popular
sent an introductory letter informing respondents support of the peace movement in the Neth-
that they would be contacted for an interview. erlands (see Oegema 1991; and Rochon 1988
Because we needed the combination of telephone
for such estimates), but in the rate of peti-
numbers and addresses for this strategy, we used
tion-signing among supporters of the move-
telephone books to determine our samples. This
approach had the disadvantage of missing people
ment and in opinion changes that occurred
with unlisted numbers, but the advantage that pri- over time among supporters. A comparison
vate numbers could be distinguished from busi- of the two resulting samples revealed that
ness numbers. they did not differ significantly in terms of
such variables as age, level of education, and scale indicates an individual's willingness to
party identification. There was, however, a participate in peace movement activities on
significant difference in gender composition a continuum from modest action (signing a
(49 percent male in May, 59 percent in No- petition) to violent action (sabotage). (2) As
vember, F = 4.84, p = .03). We adjusted our noted earlier, theoretically, generalized ac-
repeated measurement tests for this sample tion preparedness is a function of grievances
bias. and the individual's evaluation of the move-
The interviews took place between 6:30 ment. In the case of the peace movement,
P.M. and 10-00 P.M. At each address the inter- concerns about the nuclear arms race rather
viewer asked to speak to the person whose than firsthand experiences or instances of
birthday was closest to the interview date and personal harm generated the grievances the
who was older than 17 years. If necessary, movement wanted to redress. Therefore we
the interviewer made an appointment to call included a scale to measure concerns about
again later. If no contact was established af- the nuclear arms race constructed from three
ter four attempts, the address was given up. statements about nuclear armament: "Nucle-
Interviews lasted 20 minutes on average, and ar arms are needed to guarantee peace"; "I
we followed a structured questionnaire for- am really concerned about the arms race";
mat. "Would you favor a decision to deploy cruise
The 154 respondents from sample 1 who missiles?" (Cronbach's alpha = .71). (3) The
in May/June announced that they planned to individual's evaluation of the peace move-
sign the petition were approached again in ment was measured by a scale consisting of
November for a follow-up interview. Eighty- four questions, (a) about the movement ("I
six percent (132) responded. In 22 cases a have a very positive attitude toward the peace
second interview could not be arranged: 12 movement"); (b) the movement's goals ("I
of the subjects could not be reached; 10 re- have a very positive attitude toward the goals
fused. The nonrespondents did not differ from of the peace movement"); (c) the
the respondents on any of the key variables. movement's activities ("I am fully endorsing
the activities of the peace movement"); and
(d) the people in the movement ("I feel
Variables
strongly akin to the people in the peace
The following sets of variables were in- movement") (Cronbach's alpha = .86).
cluded in our analyses: Specific action preparedness. From Klan-
Participation. (1) May/June preparedness dermans's (1984) participation model we in-
to sign a petition against the deployment of ferred the following variables related to pre-
cruise missiles. Two groups were distin- paredness to sign the People's Petition: (1)
guished: (a) individuals who were not pre- agreement with the petition's goal (prevent-
pared to sign, and (b) individuals who said ing deployment of cruise missiles); the reac-
they were prepared to sign. (2) Signing be- tion of significant others-(2) in May/June,
havior in November. Here we distinguished the expected reactions if one were to sign
three groups: (a) individuals who did not ("How would people who are important to
sign and had indicated that they had not you, such as members of your family and
wanted to sign, (b) individuals who did not friends, respond if you signed the petition?"
sign but emphasized that they had wanted to [positive, negative, indifferent]), and (3) in
sign, (c) individuals who did sign. November, the experienced reactions to
Generalized action preparedness. (1) one's signing or not signing ("How did
Adapting Barnes and Kaase's (1979) action people who are important to you, such as
potential scale, we constructed a measure of members of your family and friends, respond
generalized preparedness to participate in to your signing of the petition?" [positive,
peace movement activities. Unlike Barnes negative, indifferent]).
and Kaase, however, we employed partial The presence of barriers. In each commu-
credit modeling (Masters 1982) rather than nity, organizers offered the population a
Gutman-scaling to construct a generalized number of opportunities to sign the petition.
action preparedness scale (GAP-scale) for Signature cards were distributed by mail,
the peace movement (Prins 1990). The GAP- cards were available at stands in shopping
areas, a card collector could call at the door, Democrats favored deployment, the left-
and so on. If by chance none of these oppor- wing opposed it).
tunities were available (the person was not a Demographic variables. Demographic
target of the mobilization attempt), a person measures included age, level of education,
had no choice but to try to find his or her own and sex.
way to sign. In a campaign that required as
modest an effort as signing a petition, this
Controlling for Repeated Measurement
was the only barrier we could think of.
Perceived social environment. Although Because our research question concerned
the characteristics of one's social environ- changes over time it was crucial that we
ment are reflected in the expected and expe- eliminate repeated measurement as an alter-
rienced reactions of significant others, we in- native explanation for observed differences
cluded three additional perceived character- between May and November (Campbell and
istics in our analyses, as it was our assump- Stanley 1963). We did this by comparing
tion that characteristics of the environment samples 1 and 2. None of our tests suggested
and changes in the environment are impor- any effect of repeated measurement: Levels
tant factors in nonconversion and erosion. of signing were virtually the same in the two
These three characteristics are: (1) the per- samples, and those individuals in the May
ceived opinion of the peace movement among sample who signed did not differ from the
people in one's environment, (2) the expected signers in the November-only sample with
number of one's acquaintances who would respect to any of the key variables; similar
sign (in May/June), and the perceived num- proportions of the two samples indicated
ber of acquaintances who did sign (in No- their intention to sign, and those from the
vember), 3 and (3) the extent to which cruise May sample who had wanted to sign did not
missiles were discussed within one's environ- differ from those who had wanted to sign in
ment by others and by oneself. the November sample with respect to any of
Left party identification. In the Nether- the key variables. Moreover, logistic regres-
lands the political parties can be placed on a sion analyses of the November correlates to
left-right continuum. For our analyses we signing and nonsigning revealed identical
have combined the small rightist and leftist determinants in the May sample and Novem-
parties in two clusters. The result is the fol- ber sample. We thus concluded that repeated
lowing continuum (from the political right to measurement could be ruled out as an alter-
the political left): radical rightist parties, native explanation for any of our findings.
Conservatives (VVD), Christian Democrats Accordingly, we could safely conduct our
(CDA), Centre Democrats (D'66), Social study of nonconversion and erosion using the
Democrats (PVDA), and radical leftist par- interviews in May/June and November using
ties. In some analyses the first two parties are sample 1 alone.
combined into "right-wing parties" and the
last three are combined into "left-wing par-
RESULTS
ties." All political parties had clear opinions
on the cruise issue (right-wing and Christian We present our results from two different
angles. First, we describe the respondents'
3 The question about the number of acquain- preparedness to sign in May/June and their
tances who signed produced a high proportion of subsequent reports of signing in November
"don't knows" (28.6 percent). Because SPSS lo- and try to distinguish erosion from noncon-
gistic regression analysis works with listwise de- version. Next we explain why some of those
letion of missing values, we were forced to relate individuals who in May/June said they would
the "don't knows" to our scale from 0 = nobody sign ultimately failed to do so.
through 4 = almost everybody, rather than elimi-
nate these respondents. Assigning them to 0 = no-
body seriously influenced the outcomes of the Preparedness to Sign and Actual Signing
analyses. Therefore we chose a more conserva-
tive strategy by giving them the mean of those Figure 2 presents the basic parameters of our
who did know how many in their environment results. In May/June, at the start of the cam-
signed. paign, over 68.7 percent of our respondents
FSPECIFIC ACTION PREPAREDNESS PARTICIPATION

Interviewed in May/June Interviewed in November Petition Signing
(Total = 224) (N = 132)a (N = 108)
Prepared to sign Still prepared to sign Signed

68.7%, N = 154 81.8%, N = 108 85.2%, N = 92
Not prepared to sign EROSION NONCONVERSION

31.3%, N = 70 No longer prepared to sign Did not sign
(Switchers) 14.8%, N = 16
18.2%, N = 24
Figure 2. Preparedness to Sign a Petition Against the Deploy

Respondents to Telephone Interviews, the Netherlands, 1985
a Of the 154 respondents in May/June, 22 were not available for interviews in November.
(N = 224) were prepared to sign; among the have refrained from signing-an outcome
respondents from this group that were inter- that would have been in line with the litera-
viewed in November (N = 132), 18.2 percent ture on action participation (Barnes and
had changed their minds and indicated that Kaase 1979). But due to a low level of
they no longer wanted to sign (we refer to nonconversion-very few women who
these respondents as "switchers"). The other wanted to sign failed to do so-the reverse
81.8 percent remained prepared to sign happened: More males than females failed to
("nonswitchers"). Of those who maintained sign.
their preparedness (N = 108), 14.8 percent Age. Consistent with the general finding
failed to sign ("nonsigners"); the remaining that young people are more likely to take part
85.2 percent did indeed sign ("signers"). In in collective action than are old people
other words, 69.6 percent of those who were (Barnes and Kaase 1979), we found a highly
prepared to sign in May/June and were later significant correlation between age and pre-
interviewed in November (N = 132) did in paredness to sign. But because each of the
fact sign the petition. The remaining 30.3 four age cohorts in Table 1 manifested a dif-
percent, who despite their initial prepared- ferent pattern of erosion and nonconversion,
ness to sign did not do so, consists of indi- this correlation did not remain when it came
viduals representing the two processes: non- to actual signing. Although among the sec-
conversion (those who said they wanted to ond group (ages 31 to 39) we found the ex-
sign but failed to do so-12.1 percent of pected low percentage of nonsigning, the
those prepared to sign and who were re-in- high levels of erosion and nonconversion in
terviewed in November) and erosion (switch- the youngest age group, and the very low
ers who deliberately did not sign- 18.2 per- level of nonconversion in the oldest age
cent). group compensated for the differences in
When we introduced the concepts of ero- preparedness to sign reported in May/June.
sion and nonconversion we hypothesized that Community. The four selected communi-
different subpopulations would manifest dif- ties followed strikingly different routes to
ferent rates of erosion and of nonconversion. nonsigning. This pattern coincides remark-
In Table 1 we compare several subpopula- ably well with our knowledge of the local
tions. As expected, nonsigning in these sub- campaigns. The most successful campaign
populations could be attributed to strikingly was that in Zuiderstad. An average degree of
different factors. erosion combined with an extremely low
Sex. In May/June the number of male and level of nonconversion produced the lowest
female respondents who were not prepared proportion of nonsigning respondents in any
to sign was virtually the same, but in the of the four communities. Indeed, Zuiderstad
course of the campaign the proportions had the highest number of card collectors
changed. Had the change been caused only relative to its population size, a fact that ac-
by erosion, more females than males would counts for the very low level of noncon-
Table 1. Not Prepared, Erosion, and Nonconversion: Three Forms of Nonparticipation in Signing Among Re-
spondents to Telephone Interviews, the Netherlands, 1985
Percent
Independent Variable N Not Prepared + Erosion + Nonconversion = Nonsigning
Total 224 31.2 + 12.5 + 8.3 = 52.0
Sex
Male 109 30.3 + 10.9 + 13.1 = 54.3
Female 113 31.8 + 14.3 + 4.0 = 50.1
Age
< 31 51 21.6 + 16.5 + 12.4 = 50.5
31-39 66 18.2 + 12.8 + 7.2 = 38.2
40-56 58 37.9 + 10.4 + 10.3 = 58.6
> 56 48 50.0 + 10.6 + 2.6 = 63.2
Communities
Zuiderstad 62 24.8 + 12.1 + 2.8 = 39.7
Randstad 63 27.0 + 10.4 + 8.4 = 45.8
Kleinoord 50 30.0 + 16.5 + 12.3 = 58.8
Grootland 49 46.9 + 11.2 + 11.2 = 69.3
Party Identification
Conservatives 38 69.7 + 12.1 + 9.1 = 90.9
Christian Democrats 49 47.1 + 20.7 + 10.7 = 78.5
Left-wing 93 5.4 + 5.0 + 8.1 = 18.5
Note: For these computations, the 22 respondents who could not be interviewed again are presumed to have the
same distributions as those interviewed twice.
version there. Comparing Zuiderstad with the four communities: Two-thirds of the re-
Kleinoord is illuminating. In May/June the spondents ended up not signing the petition.
proportions of the samples in the two com- From a theoretical standpoint, these four
munities which were not prepared to sign communities illustrate the separate processes
were similar. In November, however, almost of erosion and nonconversion. First, the pro-
60 percent of the original sample in portion of respondents prepared to sign in
Kleinoord did not sign, as compared to 40 May/June differed significantly from one
percent in Zuiderstad. The low support in community to another, largely because the
Kleinoord was the result of a high level of political composition of the population of the
erosion combined with a relatively high level four communities differed. Second, the one
of nonconversion. The high degree of erosion community in which a vigorous counter-
'in Kleinoord is undoubtedly related to the campaign emerged had the highest degree of
intense countercampaign conducted in that erosion. Third, the community with the most
community. In Grootland, the community sophisticated and elaborate campaign had the
with the highest proportion of Conservatives lowest level of nonconversion. In other
and Christian Democrats of the four commu- words, different characteristics of the four
nities, the number of people not prepared to communities accounted for different effects
sign was initially high. In addition, the anti- of the process of action mobilization.
cruise missile campaign conducted there was Party identification. As expected, we
relatively weak. Grootland, in fact, had the found widely divergent patterns among the
lowest number of card collectors relative to constituencies of the main political currents
population size of the four communities. in the country-the right, the Christian
Thus, Grootland had the lowest turnout of Democrats, and the left. Not surprisingly, the
turnout among the three constituencies var- Table 2. Logit Coefficients for Regression of Pre-
paredness to Sign a Petition in May/June on
ied dramatically: 90 percent of the respon-
selected Independent Variables: Respondents
dents from the right and more than 75 per-
to Telephone Interviews, the Netherlands,
cent of the Christian Democrats did not sign 1985
the petition. Among the respondents on the
Independent Variables Model 1 Model 2
left, however, fewer than 20 percent failed to
sign. Clearly, these results have their origins
Constant -5.24*** 1.94
in the beginning of the campaign: In May, (1.28) (2.00)
two-thirds of the right, almost one-half of the
Christian Democrats, but only one-twentieth Demographic Variables
of those on the left were not prepared to sign. Female .02 -.18
Then, during the campaign, erosion and (.19) (.26)
nonconversion combined to produce the di- Age -.04*** -.02

vergent patterns reflected in the November (.01) (.02)
results. Most striking are the differences in Education .07 -.08

(.08) (.11)
erosion among the constituencies of the three
political groups. Twenty-one percent of the Left Party identification 1.20*** .31
Christian Democrats who initially supported (.23) (.28)
the petition withdrew their support, com- Generalized action 2.07***
preparedness scale (.45)
pared to only 5 percent of the left. Propor-
tionally, loss of support due to nonconversion Agreement with goal 12.24***
was about equal for all three political groups. (.30)

Likelihood ratio 180.73 106.24
Degrees of freedom 214 212
Determinants of Preparedness, Erosion, and Change in likelihood ratio 90.61 74.49
Nonconversion Degrees of freedom 4 2
P-value of likelihood test .000 .000
The remainder of our analysis concentrates
Number of observations 219 219
on three questions: (1) Why were some
people prepared to sign the petition while *** p < .001 (two-tailed tests)
others were not? (2) Why did some of the re- Note: Numbers in parentheses are standard errors.
spondents who, in May, were willing to sign, Preparedness is coded so that positive coefficients in-
eventually changed their minds? (3) Why did dicate that a higher level of the independent variable is
associated with increased odds of being prepared to
some of those respondents who in November
sign.
maintained their willingness to sign the peti-
tion ultimately fail to sign it? Thus framed,
our research questions are akin to the mul- sign. Model 2, which includes the attitudinal
tiple kinds of events Allison (1985) has dis- determinants, not only improves the fit con-
cussed as being appropriate for modeling siderably but renders both age and party
with a sequence of binomial logit analyses. identification insignificant. In other words,
Prepared to sign versus not prepared. age and party identification are related to
Table 2 presents the results of two logistic signing because people who were generally
regressions of preparedness versus nonpre- prepared to take action on behalf of the peace
paredness to sign the petition on two catego- movement and who were against deployment
ries of independent variables. Model 1 con- of cruise missiles were younger and more of-
firms our previous results. ten identified with the political left. Alto-
The respondents who were not prepared to gether, the results of the final model are
sign were older than those who were pre- fairly straightforward: People were not will-
pared to sign. With respect to political affili- ing to sign the petition because they were in
ation, the further to the right the respondents general not prepared to take action on behalf
were, the less inclined they were to want to of the peace movement and because they
sign. The fit of the model indicates that on were not against the deployment of cruise
the basis of age and political affiliation missiles. Note that attitude toward deploy-
alone, we can predict fairly accurately the ment contributes to the explanation of pre-
odds of preparedness and nonpreparedness to paredness independent of generalized action
Table 3. Logit Coefficients for Regression of Nonswitching and Signing on Selected Independent Variables: Re-
spondents to Telephone Interviews, the Netherlands, 1985
Nonswitching Signing (Among Nonswitchers)
Independent Variables Model 1 Model 2 Model 3 Model 1 Model 2 Model 3
Constant -3.71 * .75 2.04 -4.97* -3.33 -8.37*

(1.71) (2.23) (2.82 (2.44) (2.96) (4.22)
Demographic Variables
Female .01 -.43 -.05 .65 .44 1.51*

(.26) (.34) (.41) (.35) (.37) (.67)
Age .01 .02 .02 .03 .04 .05
(.02) (.02) (.02) (.03) (.03) (.04)
Education .11 -.03 .04 .16 .10 .25
(.11) (.15) (.19) (.13) (.16) (.22)
Party identification (left) .76*** -.14 .17 .83** .49 .97

(.24) (.32) (.41) (.32) (.37) (.52)
May/June
Generalized action - 2.32*** .89

preparedness scale (.72) (.72)
Expected positive .80** .30

reactions of others (.32) (.35)
Agreement with goals .56 .65

(.43) (.50)
November
Generalized action 2.15* .78

preparedness scale (.90) (.72)
Experienced positive 1.52** .77
reactions of others (.52) (.60)
Agreement with goals .91 - .29
(.55) (.84)
Presence of barriers -.45 -2.75**

(.82) (.88)
Likelihood ratio 95.64 66.69 46.84 65.57 60.22 37.19

Degrees of freedom 116 113 112 96 93 92
Change in likelihood ratio 12.86 28.95 19.79 11.97 5.37 22.06

Degrees of freedom 4 3 1 4 3 1
P-value of likelihood test .01 .000 .000 .02 .15 .000
Number of observations 121 121 121 101 101 101
<.05 ** < .01 p < .001 (two-tailed tests)

Note: Numbers in parentheses are standard errors. Switching and signing are code
indicate that a higher level of the independent variable is associated with increased o
ing.
preparedness. This finding is important be- dividuals who were not prepared to sign in
cause it indicates that it was not just general May/June changed their minds and ended up
commitment to a movement that made signing.
people willing to sign the petition, but also Switching versus nonswitching. Our next
dedication to a specific cause. set of logistic analyses explores the determi-
As the campaign was designed to activate nants of erosion. The first three columns in
sympathizers rather than persuade oppo- Table 3 displays the logit coefficients for re-
nents, the canvassers made no attempt to gression of switching versus nonswitching
change peoples' minds. Indeed, only two in- on the independent variables as measured in
May/June and November. Model 1 indicates been the target of mobilization attempts, or,
that erosion was more likely among people by implication, has encountered barriers to
who identified with parties on the right. If participation.
we Interestingly, when controlling
refer to our observations in Table 1, we find for the other parameters, respondent's sex
that the respondents primarily responsible makes a difference: The degree of noncon-
for this outcome are those who identify with version was lower among women than
the Christian Democratic Party. Model 2, among men-not, as one might hypothesize,
however, indicates that the differences in because women were targeted more often
generalized action preparedness and the ex- than men. Our results indicate that, regard-
pected reactions of significant others related less of whether they were targeted, women
to party identification account for most of the who were prepared to sign did so more often
observed impact of party identification. than men who said they were prepared to
Model 2 improves on the first model, imply- sign. Among male respondents, however,
ing that individuals who indicated a lower those who were not targeted failed to sign
level of generalized action preparedness in more often than those who were. The effect
May/June and also expected less supportive of gender that we already encountered in
reactions from significant others in May/June Table 1, then, appears to hold in multivariate
were more likely to switch. analysis.
In Model 3 the May/June parameters are Finally, a comparison of nonswitchers and
replaced by the November parameters. This signing nonswitchers in Table 3 reveals theo-
model represents a considerable improve- retically meaningful differences between the
ment over Model 2. The basic pattern yielded two configurations of determinants. The first
by the November data is similar to that pro- three columns display an expanding attitudi-
duced by the data collected in May/June: In nal gap between switchers and nonswitchers
terms of generalized action preparedness, together with their increasingly divergent
switchers and nonswitchers are dispersed, perceptions of the social environment. Noth-
whereas in terms of the reaction of signifi- ing of the kind is apparent in the second two
cant others, their experiences fit their May/ columns. To be sure, nonconversion was
June expectations. Note that switching is not more likely among respondents who identi-
related to the presence of barriers. fied with right-wing parties, but it was pre-
Signing versus nonsigning. Not everyone dominantly the lack of mobilization attempts
who was prepared to sign the petition did among these groups that determined their
eventually sign. Nonconversion occurred nonsigning. This conclusion is supported by
among 15 percent of those who remained two analyses not reported in Table 3. A
willing to sign. The last three columns in model (not shown) that included both the
Table 3 examine the determinants of such May/June and November parameters re-
nonconversion. Model 1 reveals the signifi- vealed a fit significantly better than that pro-
cance of party identification. The odds of duced by Model 3 in the case of switching
nonconversion were higher among people versus nonswitching (change in likelihood
who identified with parties on the right. ratio = 15.76, d.f. = 3, p < .001). The fit,
Model 2 does not improve the fit; that is, in however, does not change in the case of
May/June nonsigners and signers did not dif- nonsigning versus signing. In other words,
fer in terms of generalized action prepared- the odds of switching-unlike the odds of
ness, expected reactions of significant others, signing-were affected by changing circum-
or attitudes toward deployment. Model 3, stances between May/June and November.4
however, implies a substantial improvement
over Model 2. It is theoretically important 4 Further confirmation is provided by uni-
that neither generalized action preparedness variate analyses of the changes in these param-
eters. Whereas the GAP-scale declined for all cat-
nor attitude toward deployment nor the ex-
egories, it dropped most strongly among switch-
pected or experienced reactions of significant
ers, and only switchers experienced more reac-
others affected petition signing. Hence, it is tions from significant others supportive of non-
not changing attitudes or environments that signing than they had expected. All others expe-
produce nonconversion. The most important rienced more supportive reactions to signing than
determinant is whether an individual has they had expected.
In sum, for respondents who, at the start Declining Generalized Action Preparedness
of the campaign, intended to sign the peti- and Unsupportive Social Environments
tion we registered reports of actual signing
among those who still intended to sign at A decline in generalized action preparedness
the end of the campaign. Our objective was, and an unsupportive social environment ap-
of course, to determine the extent to which pear to correlate with erosion; for noncon-
intentions expressed in May/June could pre- version, however, the correlate was an ab-
dict actual signing. In the absence of ero- sence of mobilization attempts.
sion and nonconversion, intentions ex- Declining generalized action prepared-
pressed in May would be perfect predictors ness. The decline in generalized action pre-
of actual signing and other variables mea- paredness is elaborated further in Table 4.
sured at later times would fail to improve Theoretically, generalized action prepared-
our models. In fact, our assessment that al- ness for a movement is a function of the
most one-third of those who intended to population's appreciation of the movement
sign in May/June (and who were also inter- and the intensity of the grievances the move-
viewed in November; N = 132) failed to do ment seeks to redress. We translated "griev-
so because of erosion or nonconversion (see ances" into concern over the nuclear arms
Figure 2), already implies less than perfect race and divided "appreciation for the peace
prediction. Indeed, the results demonstrate movement" into three elements: appreciation
that determinants measured in May/June did of goals, of activities, and of people. In Table
predict signing and nonsigning. But, the re- 4, switchers and nonswitchers and signers
sults reveal also that determinants as mea- and nonsigners are once again compared,
sured in May/June are not sufficient predic- here in terms of these antecedents of general
tors of the eventual outcomes-not in the action preparedness.
case of erosion and definitely not in the case These logistic regression analyses confirm
of nonconversion. Interestingly, erosion and our argument and provide some important
nonconversion each have its own configura- supplementary details. Even in May/June,
tion of predictors: for erosion, generalized switchers were less concerned about the
action preparedness as measured in May/ nuclear arms race and less positive about the
June and November, and expected and expe- peace movement than were nonswitchers
rienced reactions of significant others; for (Model 1). Interestingly, the sign of the logit
nonconversion, respondent's sex and the ab- for the respondents' appreciation of the
sence of mobilization attempts. In other people in the movement is negative. Indeed,
words, changes in the determinants of sign- in May/June switchers and nonswitchers
ing during the campaign produced a rate of were more alike in their attitudes toward the
signing different from what one would have people in the movement than in their appre-
expected given the values of these determi- ciation of any other aspects of the movement.
nants in May/June. In Model 2 the measures for November are
For switchers the results confirm an inter- entered in the equation, and they produce a
pretation that nonsigning occurred because considerable improvement of fit, which can
of erosion of support. A generalized action be attributed to concerns over the nuclear
preparedness that was no more than moder- arms race and the appreciation of the people
ate at the start of the campaign eroded; more- in the movement, but this time with a posi-
over, these respondents changed their minds tive sign. Apparently, during the campaign
in the context of a social environment they the initially relatively positive feelings
perceived as becoming less supportive of among switchers toward the people in the
signing the petition. movement changed into strongly negative
Nonconversion is the most plausible inter- ones: In November they no. longer felt any
pretation of nonsigning for those who main- sympathy for the people in the movement.
tained their willingness to sign. Unlike the This factor, together with their already less
switchers, the nonswitchers failed to sign be-than positive feelings toward other aspects of
cause of an absence of mobilization attempts the peace movement, contributed to their
rather than a change of mind. change of mind.
Table 4. Logit Coefficients for Regression of Nonswitching and Signing on Concerns Over Nuclear Arms Race
and Four Measures of Appreciation of the Peace Movement: Respondents to Telephone Interviews, the
Netherlands, 1985
Independent Variable Model 1 Model 2 Model 1 Model 2
Constant -2.01 * 1.34 -.69 .99

(.82) (.65) (1.17) (.65)
May/June
Concerns about nuclear arms 1.42** 1.75**

(.46) (.55)
Evaluation of Peace Movement
Appreciation of peace movement 1.47 .77

(.50) (.54)
of goals .76 - .07

(.42) (.60)
of activities .33 - .47

(.33) (.36)
of people -1.08** - -.49

(.41) (.40)
November
Concerns about nuclear arms 1.13* 1.34*

(.48) (.58)
Evaluation of Peace Movement
Appreciation of peace movement .35 -.38

(.36) (.49)
of goals -.20 -.02
(.36) (.43)
of activities -.23 .66

(.34) (.40)
of people - 1.38** - .02
(.47) (.40)
Likelihood ratio 81.67 66.43 62.96 64.69
Degrees of freedom 114 114 92 92
Change in likelihood ratio 32.67 47.91 17.42 -15.42

P-value of likelihood test .000 .000 .004 .01
Number of observations 120 120 98 98
*p <.05 **p < .01 (two-tailed tests)

Note: Numbers in parentheses are standar
indicate that a higher level of the indepen
ing.
The second two columns of Table 4 reveal underlines the distinction between erosion
a completely different picture, which under- and nonconversion as two theoretically sepa-
scores our argument regarding noncon- rate forms of nonparticipation.
version. In May/June signers were more con- Unsupportive social environments. As in-
cerned about the nuclear arms race than dicated, unlike nonswitchers, switchers per-
nonsigners, and in November that was still ceived their social environment as increas-
the case. Except for this one difference the ingly unsupportive. Table 5 refines this ob-
two group are similar in all respects in May/ servation by specifying particular character-
June as well as in November. This finding istics of the respondents' social environment:
Table 5. Logit Coefficients for Regression of Nonswitching and Signing on Five Measures of Perceived Social
Environment
Independent Variable Model 1 Model 2 Model 1 Model 2
Constant -23 .83 3.92 -4.35

(1.30) (1.56) (1.69) (1.56)
May/June
Expected positive reactions .76* .72a

of others (.32) (.37)
Perceived positive opinion of .36 -.45
peace movement (.35) (.42)
Expected number of acquaintances .62 -.61

who would sign (.48) (.56)
Cruise missiles discussed in -.22 .05

environment (.54) (.67)
Did discuss cruise missiles recently .82 .07
(.59) (.59)
November
Expected positive reactions 1.34* .11

of others (.50) (.52)
Perceived positive opinion of .14 -1.15
peace movement (.42) (.74)
Number of acquaintances - 1.03 -2.80**
who signed (.66) (.93)
Number of acquaintances who - -1.05 - -2.32
signed unknown (.80) (1.03)
Cruise missiles discussed in -.59 - -1.28
environment (.84) (.96)
Did discuss cruise missiles 1.29 2.09*
recently (.84) (1.10)
Likelihood ratio 92.03 70.05 71.63 49.56

Change in likelihood ratio 17.70 38.92 4.22 26.28

P-value of likelihood test .003 .000 .519 .000
Number of observations 116 116 95 95
*p <.05 **p < .01 a.Io >p > .05 (two-tailed tests)
Note: Numbers in parentheses are standard errors. Switching and signing are coded so that positive coefficients
indicate that a higher level of the independent variable is associated with increased odds of nonswitching or sign-
ing.
perceived opinions in their environment, ex- two groups experienced similar social envi-
pected and reported signing by acquaintan- ronments in May/June. Five months later the
ces, and discussions on cruise missiles with situation changed. The social environments
individuals in their environment. of switchers and nonswitchers and of non-
In May/June nonswitchers expected more signers and signers became increasingly dis-
supportive reactions for signing than did parate. For switchers and nonswitchers the
switchers. Similarly, at that time signers ex- difference materialized in changes in experi-
pected more supportive reactions than non- enced reactions of significant others-they
signers. As for the other characteristics, the became even less supportive of signing than
had been expected a few months before .13 for the remaining respondents; F = 3.52,
(means for switchers in May/June was -.73 P< .01).
and in November it was -1.05, as compared In summary, erosion occurred predomi-
to .12 and .24 for nonswitchers). nantly among sympathizers who identified
Among the nonswitchers, signers and non- with one of the two political parties in power
signers experienced different changes in their in the government. These individuals re-
social environments. In November perceived ported that their social environments became
reactions of significant others no longer in- less supportive during the antimissile cam-
fluenced the individual's actions. The actual paign, and because their support was only
signing by acquaintances and awareness of lukewarm to begin with, the increasingly
actual signing became more important, as did negative environment undermined their mo-
the extent to which cruise missiles were dis- tivation to sign (as shown by lower values on
cussed in one's environment. As for actual such variables as positive evaluation of the
signing, signers knew the rate of signing peace movement and rejection of cruise mis-
among their acquaintances and reported that siles). Interestingly, these individuals re-
most of their acquaintances signed. Non- ported an increase in the number of their per-
signers, however, were either not aware of sonal conversations about cruise missiles.
the rate of signing by their acquaintances or This increase and the fact that their decision
reported low rates of signing. As for discus- not to sign evoked supportive reactions from
sions on cruise missiles, signers more often significant others are evidence of social pres-
discussed cruise missiles with people in their sure. We emphasize, however, that it is the
environment than did nonsigners. Indeed, if combined impact of these factors that ac-
we compare the means, nonsigners were in counts for erosion. Identification with one of
the only groups that did not report an in- the two parties in government was not in it-
crease in discussions (means for switchers: self sufficient to produce erosion-after all,
.83 [May/June] and 1.17 [November] for dis- three-fifths of those who identified with one
cussions in their environments and .79 [May/ of the parties had wanted to sign or did in-
June] and .92 [November] for discuss them- deed sign. Rather, the turnabout from sym-
selves; means for signers: 1.15 [May/June] pathizer to switcher was the result of a com-
and 1.65 [November] and 1.24 [May/June] bination of factors: identification with one of
and 1.65 [November]; means for nonsigners: the two parties; the perception that one's en-
1.06 [May/June] and 1.06 [November] and vironment did not support the movement;
1.06 [May/June] and.88 [November]). plus, initially, the expectation, and later, the
Not included in Table 5 are two additional experience, of negative reactions from sig-
findings. Among the respondents who nificant others, which weakened an already
signed, two-thirds could cite at least one ac- halfhearted motivation.
tive member of the peace movement among Nonconversion, too, can be linked to char-
his or her acquaintances, whereas among acteristics of the individual's social environ-
those who switched positions the proportion ment, but in a different way. It was not so-
was a little less than and among those who cial pressure but the lack of it that was re-
planned to sign but didn't, the proportions sponsible for the failure to sign of "non-
was one-third. converts." These individuals were only mod-
Party identification is the last element in erately supportive throughout the campaign,
the social environment we consider. Three- thus the minor barrier of not having been of-
quarters of those who turned away from the fered the opportunity to sign sufficed to dis-
peace movement identified with the Conser- courage them from signing. Their social en-
vatives or Christian Democrats, the two par- vironment was conducive to such an out-
ties that were in office at the time and which come: For instance, only one-third of these
constituted a government determined to de- individuals could identify one or more
ploy cruise missiles. It was among support- movement activists among their acquaintan-
ers of these two parties in particular that ini- ces, and contrary to the general trend they
tial sympathy for the peace movement even- reported a decline in the number of conver-
tually evaporated (difference between May/ sations about cruise missiles. They were not
June and November was -.49 as compared to affected by the campaign and did little to
put themselves in situations where they isting levels of preparedness are high, even
would be. poor campaigns can be effective. Low levels
of action preparedness, however, require
elaborate campaigns, because prospective
DISCUSSION
supporters need more incentives to join and
Social movement organizations repeatedly are more susceptible to countercampaigns.
face the challenge of converting action pre- Effective mobilization campaigns have low
paredness into active participation. But will- rates of nonconversion, especially in the con-
ingness to participate, as we found, is no text of a supportive social environment; ef-
guarantee of actual involvement: The ab- fective countercampaigns produce high de-
sence of mobilization attempts, the presence grees of erosion, and are especially success-
of barriers, and/or an unsupportive social en- ful in a context of a nonsupportive social en-
vironment may ultimately prevent a person vironment.
from participating (for similar findings, see Paradoxically, these conditions make cam-
Ajzen and Fishbein [1980] and Granberg and paigns designed to mobilize low-risk support
Holmberg [1988]). (Ennis and Schreuer 1987) more vulnerable
Patterns of nonconversion and erosion ap- to erosion than campaigns for high-risk ac-
pear to vary across subpopulations. Counter tivities. Because the strength of low-risk
to the general finding that women participate campaigns is in the potentially high numbers
in collective action less often than men of participants, they must appeal to as large
(Barnes and Kaase 1979, but see Wallace and a proportion of the population as possible,
Jenkins forthcoming), we found that women and include marginal sympathizers. The fact
signed the peace petition more frequently that only a low-risk, undemanding act of sup-
than men. As the campaign proceeded port is requested does not necessarily make
women seemed to become more sensitive to erosion less likely, because the public debate
the controversial tone that gradually marked can still be intense, as this petition campaign
public discourse, as witnessed by a higher demonstrated.
degree of erosion. But, this trend was more Generalized action preparedness could ac-
than offset by a high degree of conversion count for a considerable proportion of the
among women. In addition, men who were differences in the probability of nonsigning
not targeted by the campaign failed more of- versus signing even four months later. But by
ten than women to overcome this barrier. incorporating specific factors we improved
Similarly, the combined impact of erosion our models substantially. This finding may
and conversion nullified the traditionally help resolve the classic question of whether
negative correlation between age and active movement participation is a function of rela-
participation. And with respect to political tively stable ideological commitment to a
party identification, cross pressure (Lazars- movement or the relatively fluctuating incen-
feld et al. 1948) produced high degrees of tives associated with a specific activity (Opp
erosion among adherents of the Christian 1989a).
Democratic Party. Finally, our comparison of Our findings question once again the im-
four communities emphasizes the differential portance of the free-rider problem. Recent
impact of the intensity of campaigns and social movement theory argues that the free-
countercampaigns: An intense mobilization rider problem is not, in fact, a major obstacle
campaign in one community produced high for movements that mobilize through com-
levels of conversion; an intense counter- munity networks and rely on purposive com-
campaign in another community produced mitment and solidarity (Fireman and Gam-
high levels of erosion. son 1979; Klandermans 1988; Oliver and
Our results reveal mobilization to be a Marwell 1992; Ferree 1992). Even in the
complicated process in which preexisting case of low-risk forms of action, such as
levels of action preparedness, characteristics signing a peace petition, levels of action pre-
of mobilization campaigns and counter- paredness, friendship networks, weak mobi-
campaigns, and characteristics of the indi- lization attempts, and the presence of barri-
vidual's social environment interact to deter- ers seem to be more important than free-rider
mine movement participation. When preex- logic. Indeed, shortfalls occur not so much
because sympathizers take a free ride, but DIRK OEGEMA is Lecturer in the Department of
because people with moderate levels of ac- Political Science at the Free University in
tion preparedness either lose sympathy for Amsterdam, the Netherlands. He recently com-
pleted his dissertation on the decline of the Dutch
the movement or are embedded in social net-
peace movement. Currently he is conducting re-
works that fail to put their principles into
search in the field of communication and politi-
practice (that is, their networks do not impel
cal participation, more specifically with regard
them to act). to voting and participation in social movement
To what extent can we generalize from organizations.
these findings? Signing a petition is a very
minimal act. Although we can expect non- BERT KLANDERMANS is Professor of Applied Social
Psychology at the Free University in Amsterdam,
conversion and erosion to occur in most mo-
the Netherlands. His research focuses on mobili-
bilization campaigns, we can assume that
zation and participation in social movements. He
different types of action will balance the is currently studying the Farmer's Protest in the
various factors that lead to nonconversion or Netherlands and Spain, and (with Johan Olivier,
erosion in different ways. There was every University of Pretoria) the responses of move-
reason to expect that the petition campaign ment and countermovement supporters to the so-
as a whole would produce extremely low lev- cial and political transitions in South Africa. He
els of nonconversion. First, it involved a low- is Editor of Social Movements, Protest and Con-
tention (University of Minnesota), a series on so-
risk activity, and one would expect low lev-
cial movements. His is author of Social Construc-
els of nonconversion in this case. However,
tion of Protest: Social Psychological Principles of
our finding that high expected and experi- Movement Participation (Basil Blackwell, forth-
enced rates of signing in an individual's per- coming), and with Craig Jenkins has edited The
sonal environment increased the odds that he Politics of Social Protest: Comparative Perspec-
or she too would sign is similar to what tives on States and Social Movements (Univer-
McAdam (1986) discovered in his study of sity of Minnesota, forthcoming).
applicants for the Mississippi Freedom Sum-
mer. McAdam explicitly placed his observa- REFERENCES
tion in the context of high-risk participation.
Ajzen, Icek and Martin Fishbein. 1980. Under-
Our study demonstrates that the same mecha-
standing Attitudes and Predicting Social Be-
nisms work at the opposite end of the scale,
havior. Englewood Cliffs, NJ: Prentice-Hall.
in the context of low-risk participation. Sec-
Allison, Paul D. 1984. Event History Analysis:
ond, the petition campaign was extremely Regression for Longitudinal Event Data.
well organized and was designed to reach as Beverly Hills, CA: Sage.
high a level of conversion as possible. Less Barnes, Samuel H. and Max Kaase. 1979. Politi-
well organized campaigns will not reach cal Action: Mass Participation in Five West-
such high levels, as illustrated by the diver- ern Democracies. London, England: Sage.
gent results of the campaigns in our four Bridt, Martien, Bert Klandermans, and Frederike
communities. Third, the petition campaign Kroon. 1987. "How Women Become Involved
in the Women's Movement." Pp. 44-64 in The
was one phase in an extended campaign that
Women's Movements in the U.S. and Western
had been going on for several years. Conse-
Europe: Consciousness, Political Opportunity
quently, social networks that in other circum-
and Public Policy, edited by M. Katzenstein
stances might have remained indifferent had and C. Mueller. Philadelphia, PA: Temple Uni-
already been co-opted by the movement. The versity Press.
activity of these networks helped to lower the Campbell, Donald T. and Julian C. Stanley. 1963.
level of nonconversion. Experimental and Quasi-Experimental Designs
Note, however, that factors that decrease for Research. Chicago, IL: Rand McNally.
nonconversion do not necessarily produce Chafetz, Janet Saltzman and Anthony Dworkin.
1987. "Action and Reaction: An Integrated,
lower levels of erosion. We assumed that ero-
Comparative Perspective on Feminist and An-
sion results from the sharpening of loosely
tifeminist Movements." Paper presented at the
defined goals, countercampaigns, and polar-
annual meeting of the American Sociological
ization. Because these conditions may ac- Association, Aug., Chicago, IL.
company even highly effective campaigns, Conover, Pamela Johnston and Virginia Gray.
some of the factors that reduce noncon- 1983. Feminism and the New Right. The Con-
version may well foster erosion. flict Over the American Family. New York:
Praeger. Van Gorcum.

Duyvendak, Jan Willem. Forthcoming. The Lane, Robert E. 1964. Political Life. New York:
Power of Politics. New Social Movements in an Free Press.
Old Polity, France 1965-1989. Boulder, CO: Lazarsfeld, Paul F., Bernard B. Berelson, and Ha-
Westview. zel Gaudet. 1948. The People's Choice. New
Ennis, James G. and Richard Schreuer. 1987. York: Columbia University Press.
"Mobilizing Weak Support for Social Move- Mansbridge, Jane L. 1986. Why We Lost the ERA.
ments: The Role of Grievance, Efficacy and Chicago, IL: University of Chicago Press.
Cost." Social Forces 66:390-409. Masters, G. N. 1982. "A Rasch Model for Partial
Ferree, Myra Marx. 1992. "The Political Context Credit Scoring." Psychometrika 47: 149-74.
of Rationality: Rational Choice Theory and McAdam, Doug. 1986. "Recruitment to High-
Resource Mobilization." Pp. 29-52 in Fron- Risk Activism: The Case of Freedom Sum-
tiers in Social Movement Theory, edited by A. mer." American Journal of Sociology 92:64-
Morris and C. Mueller. New Haven, CT: Yale 90.
University Press. McCarthy, John and Mayer Zald. 1976. "Re-
Fireman, Bruce and William A. Gamson. 1979. source Mobilization and Social Movements: A
"Utilitarian Logic in the Resource Mobilization Partial Theory," American Journal of Sociol-
Perspective." Pp. 8-45 in The Dynamics of So- ogy 82:1212-41.
cial Movements, edited by M. N. Zald and J. Oberschall, Anthony. 1973. Social Conflict and
D. McCarthy. Cambridge, MA: Winthrop. Social Movements, Englewood Cliffs, NJ:
Frey, J. H. 1983. Survey Research by Telephone. Prentice-Hall.
London, England: Sage. Oegema, Dirk. 1991. "The Dutch Peace Move-
Granberg, Donald and Soren Holmberg. 1988. ment, 1977 to 1987." Pp. 93-142 in Interna-
The Political System Matters: Social Psychol- tional Social Movement Research. Vol. 3:
ogy and Voting Behavior in Sweden and the Peace Movements in Western Europe and the
United States. Cambridge, England: Cambridge United States, edited by B. Klandermans.
University Press. Greenwich, CT: JAI.
Henig, Jeffrey R. 1982. Neighborhood Mobiliza- Oegema, Dirk and Bert Klandermans. 1992. "The
tion, Redevelopment and Response. New Erosion of Support: Perceived Political
Brunswick, NJ: Rutgers University Press. Changes and Participation in the Peace Move-
Klandermans, Bert. 1984. "Mobilization and Par- ment in the Netherlands." Paper presented at
ticipation: Social Psychological Expansions of the First European Conference on Social
Resource Mobilization Theory." American So- Movements, 29 Oct.-I Nov., Berlin, Germany.
ciological Review 49:583-600. Oliver, Pamela E. and Gerald Marwell. 1992.
. 1988. "Union Action and the Free "Mobilizing Technologies for Collective Ac-
Rider Dilemma." Pp. 77-92 in Research in So- tion." Pp. 251-73 in Frontiers in Social Move-
cial Movements, Conflict and Change. Vol. 10: ment Theory, edited by A. Morris and C.
Social Movements as a Factor of Change in the Mueller. New Haven, CT: Yale University
Contemporary World, edited by L. Kriesberg Press.
and B. Misztal. Greenwich, CT: JAI. Opp, Karl-Dieter. 1988. "Community Integration
. 1989. "Interorganizational Networks." and Incentives for Political Protest." Pp. 83-
Pp. 301-15 in International Social Movement 103 in From Structure to Action: Comparing
Research. Vol.2: Organizing for Change: So- Movement Participation Across Cultures, ed-
cial Movement Organizations in Europe and ited by B. Klandermans, H. Kriesi, and S.
the United States, edited by B. Klandermans. Tarrow. Greenwich, CT: JAI.
Greenwich, CT: JAI. . 1989a. The Rationality of Political
Klandermans, Bert and Dirk Oegema. 1987. "Po- Protest. Boulder, CO: Westview.
tentials, Networks, Motivations, and Barriers: . 1989b. "Social Integration Into Volun-
Steps Toward Participation in Social Move- tary Associations and Incentives for Legal and
ments." American Sociological Review 52: Illegal Protest." Pp. 345-63 in International
519-31. Social Movement Research. Vol. 2: Organizing
Koopmans, Ruud. Forthcoming. Democracy for Change: Social Movement Organizations in
From Below and the Political Systems in West Europe and the United States, edited by B.
Germany. Boulder, CO: Westview. Klandermans. Greenwich, CT: JAI.
Kriesi, Hanspeter and Philip Van Praag, Jr. 1988. Prins, Peter. 1990. "Actie Potentiaal: Ra, Ra?!"
"De beweging en haar campagne" (The Move- (Action Potential: Rah-Rah?!). Masters thesis.
ment and Its Campaign). Pp. 13-53 in Tekenen Department of Social Psychology, Free Univer-
voor de vrede. Portret van een campagne sity, Amsterdam, the Netherlands.
(Signing for Peace. Portrait of a Campaign), Rochon, Thomas R. 1988. Mobilizing for Peace:
edited by B. Klandermans. Assen, Netherlands: The Antinuclear Movements in Western Eu-
rope. Princeton, NJ: Princeton University People: Why Consensus Movements Cannot
Press. Be Instruments of Social Change." Pp. 205-24
Rucht, Dieter and Donatella Della Porta. Forth- in Frontiers in Social Movement Theory, edited
coming. "Left-Libertarian Movements in Con- by A. Morris and C. Mueller. New Haven, CT:
text: A Comparison of Italy and West Ger- Yale University Press.
many, 1965-1990." In The Politics of Social Wallace, Michael and J. Craig Jenkins. Forthcom-
Protest: Comparative Perspectives on States ing. "The New Class and the Ideological Bases
and Social Movements, edited by J. C. Jenkins of Political Protest: A Comparison of Eight
and B. Klandermans. Minneapolis, MN: Uni- Western Democracies." In The Politics of So-
versity of Minnesota Press. cial Protest: Comparative Perspectives on
Snow, David A., Louis A. Zurcher and Sheldon States and Social Movements, edited by J. C.
Ekland-Olson. 1980. "Social Networks and So- Jenkins and B. Klandermans. Minneapolis,
cial Movements: A Microstructural Approach MN: University of Minnesota Press.
to Differential Recruitment." American Socio- Walsh, Edward J. 1988. Democracy in the Shad-
logical Review 45:787-801. ows: Citizen Mobilization in the Wake of the
Schwartz, Michael and Shuva Paul. 1992. "Re- Accident at Three Mile Island. Westport, CT:
source Mobilization Versus the Mobilization of Greenwood.

12 Sympathizers (Oegema, Klandermans) PDF

Uploaded by

Copyright:

Available Formats

12 Sympathizers (Oegema, Klandermans) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

12 Sympathizers (Oegema, Klandermans) PDF

Uploaded by

Copyright:

Available Formats

Why Social Movement Sympathizers Don't Participate: Erosion and Nonconversion of

American Sociological Association is collaborating with JSTOR to digitize, preserve and

DIRK OEGEMA BERT KLANDERMANS

Social movement organizations face the challenge of converting action preparednes

American Sociological Review, 1994, Vol. 59 (October:703-722) 703

can originate at each transition point in the THEORY

GENERALIZED ACTION ~.SPECIFIC ACTION -4 PARTICIPATION

Figure 1. Process of Action Mobilization

be movement decline, changes in public opin- DATA AND METHODS

FSPECIFIC ACTION PREPAREDNESS PARTICIPATION

Prepared to sign Still prepared to sign Signed

Not prepared to sign EROSION NONCONVERSION

Figure 2. Preparedness to Sign a Petition Against the Deploy

Independent Variable N Not Prepared + Erosion + Nonconversion = Nonsigning

Total 224 31.2 + 12.5 + 8.3 = 52.0

Male 109 30.3 + 10.9 + 13.1 = 54.3

Female 113 31.8 + 14.3 + 4.0 = 50.1

< 31 51 21.6 + 16.5 + 12.4 = 50.5

31-39 66 18.2 + 12.8 + 7.2 = 38.2

40-56 58 37.9 + 10.4 + 10.3 = 58.6

> 56 48 50.0 + 10.6 + 2.6 = 63.2

Zuiderstad 62 24.8 + 12.1 + 2.8 = 39.7

Randstad 63 27.0 + 10.4 + 8.4 = 45.8

Kleinoord 50 30.0 + 16.5 + 12.3 = 58.8

Grootland 49 46.9 + 11.2 + 11.2 = 69.3

Conservatives 38 69.7 + 12.1 + 9.1 = 90.9

Christian Democrats 49 47.1 + 20.7 + 10.7 = 78.5

Left-wing 93 5.4 + 5.0 + 8.1 = 18.5

nonconversion combined to produce the di- Age -.04*** -.02

results. Most striking are the differences in Education .07 -.08

was about equal for all three political groups. (.30)

Nonswitching Signing (Among Nonswitchers)

Independent Variables Model 1 Model 2 Model 3 Model 1 Model 2 Model 3

Constant -3.71 * .75 2.04 -4.97* -3.33 -8.37*

Female .01 -.43 -.05 .65 .44 1.51*

Party identification (left) .76*** -.14 .17 .83** .49 .97

Generalized action - 2.32*** .89

Expected positive .80** .30

Agreement with goals .56 .65

Generalized action 2.15* .78

Presence of barriers -.45 -2.75**

Likelihood ratio 95.64 66.69 46.84 65.57 60.22 37.19

Change in likelihood ratio 12.86 28.95 19.79 11.97 5.37 22.06

Number of observations 121 121 121 101 101 101

<.05 ** < .01 p < .001 (two-tailed tests)

Nonswitching Signing (Among Nonswitchers)

Independent Variable Model 1 Model 2 Model 1 Model 2

Constant -2.01 * 1.34 -.69 .99

Concerns about nuclear arms 1.42** 1.75**

Evaluation of Peace Movement

Appreciation of peace movement 1.47 .77

of goals .76 - .07

of activities .33 - .47

of people -1.08** - -.49

Concerns about nuclear arms 1.13* 1.34*

Evaluation of Peace Movement

Appreciation of peace movement .35 -.38

of activities -.23 .66

Change in likelihood ratio 32.67 47.91 17.42 -15.42

Party identification (left) .76* -.14 .17 .83 .49 .97

Concerns about nuclear arms 1.42 1.75