Sources of Popular Support for Authoritarian Regimes
Author(s): Barbara Geddes and John Zaller
Source: American Journal of Political Science, Vol. 33, No. 2 (May, 1989), pp. 319-347
Published by: Midwest Political Science Association
Stable URL: http://www.jstor.org/stable/2111150
Accessed: 07-01-2016 21:16 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
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 support@jstor.org.
Midwest Political Science Association and Wiley are collaborating with JSTOR to digitize, preserve and extend access to
American Journal of Political Science.
http://www.jstor.org
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
Sources of Popular Supportfor
Regimes*
Authoritarian
of California,Los Angeles
Barbara Geddes, University
of California,Los Angeles
JohnZaller, University
to the
All authoritarian
attemptto controltheflowof newsand information
governments
public-but withwhateffect?To answerthisquestion,we adapt an existingmodel ofopinion
formationto conditionsin authoritariancountries,validate thatmodel on opinion data colperiod,and, finally,use the model to deriveexpectalected in Brazil duringits authoritarian
kindsof authoritariansystems.
tipnsabout patternsof regimesupportthatexistin different
The paper showsthatsupportforregimepoliciesdependsheavilyon citizens’level of politicalawareness.In general,highlyaware personsare more heavilyexposed to governmentdominatedcommunications
media, but are also betterable to resistthe propagandatheyencounter.As a result,people in the broad middle ranges of awareness-who pay enough
attentionto be exposed but are not sophisticatedenoughto resist-typicallyare mostsusceptible to governmentinfluence.
All authoritarian
attemptto managetheflowofnewsand
governments
to the public. They seek, on one side, to fillthe mass
politicalinformation
media witha steadystreamof progovernment
messagesand, on the other,
to stifleindependentcriticismand analysis.By these means authoritarian
governmentsattemptto shape the politicalattitudesof theircitizens.
Since thesegovernments
rarelypermittheconductof independentsurtheyindoctrinatetheir
vey research,no one reallyknowshow successfully
citizens. Can people who have been fed a steady diet of governmentmaintaincriticalattitudestowardtheirgovernment?
controlledinformation
What kindsof citizensare mostsusceptibleto governmentinfluence,and
Whattheoriescan accountfortheobserved
whatkindsare leastsus_eptible?
patternsof susceptibility?
Using opiniondata collectedin Brazil at the heightof its authoritarian
period,we develop answersto thesequestions.We findthatsupportforthe
authoritarian
policiestendsto be greatestamongcitizenswho
government’s
*We gratefully
acknowledgemanyhelpfulcommentson thepaper byChrisAchen, Larry
Bartels,HenryBrady, David Collier, Amauryde Souza, and several anonymousreviewers.
We especiallythankHerbertMcClosky,whose insightson the learningof politicalattitudes
thispaper seeks to extend.The researchwas supportedin partby an Academic Senate grant
at UCLA. The data reportedin thispaperwereoriginallycollectedundera grantfromtheFord
ConsortiumforPoliticaland
Foundationand weremade availablethroughtheInter-University
forany errorsof
Social Research at the Universityof Michigan.We alone bear responsibility
the paper may contain.
factor interpretation
AmericanJournalofPoliticalScience, Vol. 33, No. 2, May 1989, Pp. 319-347
? 1989 by the Universityof Texas Press, P.O. Box 7819, Austin,TX 78713
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
320
Barbara Geddes and JohnZaller
are moderatelysophisticatedabout politics-people, that is, who pay
line butwho are
enoughattentionto be heavilyexposed to thegovernment
notsophisticatedenoughto be able to resistit. Those mostresistantto persuasionare people who are bothhighlyattentiveto politicsand who are predisposed(because ofeconomicinterest,priorpoliticization,or personalvalues) againstauthoritarianism.
Althoughwe have data fromonly one authoritariancountry,we develop and test propositionsthat are entirelygeneral. Indeed, the central
findingof the paper is thatopinionformationin Brazil conformsin detail
to the principlesthathave been foundto hold in democraciesand in labenables us bothto place
oratoryexperimentson persuasion.This generality
the Brazilian case in a broader contextthanis usual and to verifythatexistingtheoriesof opinionformationhold in authoritarianas well as democraticnations.It also enablesus to developa typologyofhowregimesupport
kindsof authoritariansystems.
is likelyto varyin several different
Past Researchand Theory
A commonargumentin studiesof publicopinionin the United States
is thatbetter-educatedand more politicallyaware citizensare more likely
to embraceprevailingregimenormsthanare theless educated and aware.
This argumenthas been applied to public attitudeson civilliberties(Key,
1961; McCloskyand Brill, 1983), foreignpolicy(Gamson and Modigliani,
1966; Mueller, 1973; Sigelmanand Conover, 1981), and capitalistideology
(Chong, McClosky,and Zaller, 1985). As Key writes,”Probablya major
consequence of educationforopinionconsistsin the bearingof education
on thekindsofinfluencesto whicha personis subjectedthroughout
hislife.
The moreextendedtheeducationalexperience,themoreprobableitis that
a personwill be exposed to the discussionsof issues as theyarise. When,
as so oftenoccurs,the currentdiscussionis heavilyloaded on one side, it
mightbe expectedthatthiseducationallyconditionedexposurewould have
some bearingon the directionof opinion” (1961, p. 341). Notingthateducationwas associatedwithgreatersupportforracialequality,toleranceof
and certainfreeenterpriseideas, Key arguedthat”formal
nonconformists,
officialpoeducationmayserveto indoctrinate
people intothemore-or-less
liticalvalues of the culture” (1961, p. 340).
The centralidea in these and otherstudiesis thatexposureto political
communications-whether
exposureis measuredby a respondent’slevel of
education,informationabout politics,or politicalinvolvement-tendsto
promotesupportforthe “mainstream”politicalnormsembedded in those
communications.Hereafter,we shallreferto thisas themainstreammodel.
The mainstreammodel, however,applies onlyin cases in whichmainstream norms exist. When political elites disagree, or when norms are
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
321
changing(as theywere,e.g., on Vietnampolicyin theUnited Statesin the
1960s), the effectsof exposureto publicdiscourseare quite different
from
the effectswhen a settledelite consensusexists.
Converse (1962) was the firstto propose a dynamicmodel of mass response to partisan(i.e., nonmainstream)communications.His model, initiallydeveloped to explainattitudechangein electioncampaignsbut since
extendedto othertypesof attitudechange, is based on two assumptions:
is positivelyassociatedwithgenthatexposureto politicalcommunications
eral levelsof politicalawarenessand thatuncriticalacceptanceof suchmessages is negativelyassociated withawareness. Given these assumptions,it
followsthatpeople in themiddlelevelsofawarenesswillbe moresusceptible
to changethanpeople at eitherextreme.The mostaware citizensare heavily
exposed to political communications,but, just because they are highly
aware, theyscrutinizethem in lightof priorbeliefsand may then reject
them.The leastaware,incontrast,payso littleattentionto politicsthatthey
are likelyto escape influence.Finally,people of moderateawarenessare
butare unableto subject
fairlyheavilyexposed to politicalcommunications,
to persuasionis greatest.
themto real scrutiny;hence, theirsusceptibility
The keyto themodel is theselectiveresistanceof highlyaware persons
to the messageswhichtheyencounter.Recent researchhas shownthatif
a new idea is consistentwiththeirgeneralvalues, highlyaware people may
not resistit at all (Zaller, 1987). For example, if a conservativepolitician
gives a speech, conservativeindividualsmay simplyaccept its message (if
theyhappen to hear it). Here the exposure-acceptancemodel reducesto a
singlefactor-effective
exposure;themostaware are thenmostpersuadable
because theyare mostlikelyto hear thespeech. Onlywithingroupspredisposed to resistthepartisanspeech (in thisexample,liberals)shouldwe expect thenonmonotonicpatternof responsivenessthatarisesfromtheresistance of the highlyaware to discrepantideas.
Variantsof thistwo-factormodel, whichwe hereafterreferto as the
exposure-acceptancemodel, have been used in laboratorystudiesof attitude change (McGuire, 1969), studiesof the effectsof mass electioncampaigns (Converse, 1962; Dryer, 1971; Zaller, 1989) and of news coverage
on the salience of issues in thepublicmind(MacKuen, 1984), and analysis
of mass attitudeformationon such issues as the nuclear freeze, defense
spending,school desegregation,the VietnamWar, and gay rights(Zaller,
1987). We can bestillustratethedynamicsof themodel byusingMcGuire’s
as follows:
(1969) formulation,
Pr(Support) = Pr(Exposure) x Pr(AcceptancelExposure)
where,
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
(1)
322
Barbara Geddes and JohnZaller
Pr(Support) = Probabilityan individualwill supporta policy
Pr(Exposure) = Probabilityof exposure to a message favoringthe
policy
Pr(Acc. IExp.) = Probabilityof acceptingthe message,givenexposure
to it
We furtherassume thatthe probabilityof exposureto a givenmessage increases withgenerallevels of awarenessand thatresistanceto a message
(givenexposure)willbe greatestwhena personhas enoughpoliticalawareness to thinkcriticallyabout it and is predisposed(by values, interest,or
priorbelief)againstthemessage.Neitherawarenessalone nordispositional
factorsalone can induce resistanceto persuasion.
formalimplications,as we illustrate
These simpleideas have interesting
in Table 1. Suppose again thata conservativepoliticiangives a speech favoringa newconservativeidea and thatwe examinesupportforthenewidea
separatelyin samplesof conservatives,centrists,and liberals.The firstrow
of Table 1 shows exposure to the conservativeidea among conservatives
intheiraveragelevelsofawareness.These exposureprobabilities
whodiffer
increaseas awarenessincreases(from27 percentoftheleast awareto 87 percent of the most aware). Meanwhile,row 2 of the table shows acceptance
probabilities,whichconformto themodel by decreasing(from100 percent
theexposureprobabilities
to 81 percent)as awarenessrises.By multiplying
(row 1) by the correspondingacceptance probabilities(row 2), we obtain
thesupportscoresforthemessagein row3. Multiplicationofthesetwosets
of probabilities,as specifiedin equation 1, is the centralfeatureof the
exposure-acceptancemodel used in thispaper.
If the readernow examinesthe threepanels in Table 1, he or she will
see that,at givenlevelsofawareness,exposureratesare identicalin all three
populations. Acceptance rates in the three populations, however, vary
greatly.Among conservatives(panel A), acceptance rates decline only
whereasamongliberals(panel C), acceptance
modestlywithinformation,
of theconservativemessage (givenexposureto it) declinessharplywithinformation.As a result,the”messagesupport”levels(showninrow3 ofeach
panel) also differ,thoughnot in a straightforward
way. Among conservatives,a strongpositiverelationshipexistsbetweenawarenessand support.
Thus, themainstreammodel of policysupportemergesas a special case of
theexposure-acceptancemodel. But where,as in thecase ofcentristsor liberals,acceptanceratesdeclinemoresharplyas awarenessrises,therelationship between awarenessand supportis nonmonotonic.
Iftheseideas, whichoriginatedinstudiesofpublicopinionintheUnited
States, are applied to the Brazilian case, what patternsof regimesupport
do theylead us to expect?On certainissues the militaryregimesimplycar
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
323
TABLE 1
ThreePatterns
ofPolicySupport
LevelofInformation
Middle High
Low
Panel A:
Lowresistance
pattern
Panel B:
pattern
Moderateresistance
Panel C:
Highresistance
pattern
.27 .41 .58 .75 .87
Prob.ofexposure
Prob.ofacceptance 1.0 .99 .98 .93 .81
.27 .40 .57 .70 .71
Prob.ofsupport
Prob.ofexposure
Prob.ofacceptance
Prob.ofsupport
.27 .41 .58 .75 .87
.99 .98 .95 .85 .66
.27 .40 .55 .64 .57
Prob.ofexposure
Prob.ofacceptance
Prob.ofsupport
.27 .41 .58 .75 .87
.83 .61 .34 .15 .05
.22 .25 .20 .11 .05
NOTE: Althoughthenumbersappearingin thistable are intendedonlyto illustratethemodel,
betweensupport
thoseinpanelsA and B are derivedfroman actualestimateoftherelationship
forauthoritarian
policiesand threeindependentvariables:politicalawareness,education,and
religiosity.These variables are describedin the textbelow; the model used to estimatethe
relationshipis equation2, also shownbelow. The figuresin panel A showestimatedexposure,
acceptance,and policysupportratesfora personwho has onlyfouryearsofeducationand who
attendschurchweekly;thosein panel B showestimatedratesfora personwhohas fouryearsof
educationand neverattendschurch.The figuresin panel C showthesame exposureratesas in
A and B, butacceptanceratessimilarto thoseofthe”resistant”typeofBrazilian,as described
in thetext.
ried forwardthe settledpoliciesof previousregimes(more on thisbelow);
in these cases, we would expect the mainstreampatternof policysupport
to occur (as in panel A). But formanyotherissues, the regimesoughtto
build supportforpolicies thatdepartedfromthose of Brazil’s democratic
model to apply.
past. Here we would expectthedynamicsof thetwo-factor
Withinsubgroupsof the Brazilianpublicpredisposedtowardconservative
values (e.g., theconventionally
religious),we would expectthe monotonic
patternof supportshownin panel A: the heavierthe exposureto the governmentline, the greaterthe supportforit. Withinsubgroupspredisposed
against conservativevalues (e.g., persons rejectingtraditionalreligious
belief),we would expect the nonmonotonicpatternsof policysupport,as
in panels B and C, thatarise fromthe resistanceof the highlyaware.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
324
Barbara Geddes and JohnZaller
Our expectation,then,is that opinion formationoccurs in much the
same way in authoritarian
systemsas in democracies.Exposure to an elite
discoursethat is “heavily loaded on one side” tends, howeverthat onesidedness arises, to induce mass supportforthe values embedded in the
discourse-except among well-informedpeople who are predisposed
againstthose values.
In orderto testthisexpectation,we mustestablishtheelementsofboth
Brazil’s mainstreamtraditionand the specificdeparturesfromit underauthoritarianism.
We mustalso describethe data used in the study.
Brazilian PoliticalTraditionsand the AuthoritarianIntrusion
For 20 yearsfollowingWorld War II, Brazilian democracywas based
on a multiparty
systemwithwidespreadbut not universalsuffrage.Unionization and workerparticipationin politics-thoughalwaysdependenton
statepatronageand leadership-were at least toleratedand oftenofficially
encouraged(A. Rodrigues, 1968; L. M. Rodrigues, 1970, 1974; Weffort,
1973; Erickson,1977). The pressand othermediawereuncensoredand vigorous in attackingpolitical opponents, includingincumbentauthorities.
During this time politicalleaders stressedeconomic developmentas the
country’sprimarynationaltask (Lafer, 1975; M. L. Cardoso, 1978). In social policyBrazilin the1950sespoused an officialideologyofracialequality.
Brazil was, to be sure, never an ideal democracy.De factoracial discrimination
was widespread,and illiterates,a large fractionof the population,were excluded fromthe franchise.Corruption,electoralfraud,state
paternalism,and clientelismregularlyinfluencedelectionoutcomes.Nonetheless,fromtheend ofWorldWar II to theearly1960s,Brazil was a nation
in whichcompetitiveelections,therightto criticizegovernment
policy,and
an officialcommitment
to racial equalitywere accepted by virtuallyeveryone in public life as mainstreamideals.
Aftertwodecades ofrapidifsomewhaterraticeconomicprogress,a severe economiccrisis,accompaniedby strikesand widespreadpoliticalagitation,reached a peak in 1964 duringthe administration
of Jodo Goulart
(Skidmore,1967). The military,
expressingfearsofa leftisttakeover,seized
power in a coup. Promisinga returnto democracyas soon as the political
situationstabilized,the new leadershipinitiallytolerateda certainamount
ofpressfreedomand organizedopposition.Afteritscandidatesfaredpoorly
in the 1965legislativeelections,however,themilitary
shiftedcourse. It canceled theupcomingpresidentialelection,bannedmostformsofopposition,
institutedstrictcensorshipof major media, and assertedfirmcontrolover
labor unions. The militarygovernmentalso replaced the existingpolitical
partieswithtwo new government-sponsored
parties,controlledthe selection of governorsand presidents,and sharplycurtailedor eliminatedthe
powers of the remainingelected officials(Lamounier and F. H. Cardoso,
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
325
1976; Skidmore,1977). The regimealso resortedto abductions,torture,and
murderto deal withoutspokenopponents.On onlyone major democratic
regimemake no publicefvalue-namely, racialequality-did themilitary
fortto reversemainstreampolicies.
On the economicside the militaryregimeaggressivelypursueddevelopmentalistpolicies,as had previousdemocraticregimes.The militarydid
initiateimportantchangesin economicpolicy,includingreductionsin real
wages and variousmeasuresto encourageexports.Yet thegeneralstrategy
of stateactivismand heavyrelianceon foreigninvestment-a strategythat
was thehallmarkof postwarBrazilianpolicy-remained unchanged(Baer,
1983).
media depictedthe milDuringthisperiod the government-controlled
itaryregimein glowingterms.The regimewas portrayedas leadingthenationto spectaculareconomicgrowth,providingforthelaboringpoor more
than the “demagogic” politiciansof the past, and copingresoeffectively
who threatenedthe nationwithchaos. The imlutelywithurbanterrorists
plicationof thiscoveragewas thatthe militaryand its sternpolicies were,
afterall, good forBrazil.
The media did not emphasize the antidemocraticside of the governpoliciesas theywerepromulment’sprogram.It announcedantidemocratic
gated,but did not carrya dailystreamof storieson the need to censorthe
news,stifledebate, and controlelectionoutcomes.Rather,thepresstrumpolicies
as proofthatthegovernment’s
petedtheregime’saccomplishments
wereworking,whiledenyingaccess to anyonewishingto disputethatview.
The effectsofthisone-sidedcoveragein a contextinwhichtheeconomywas
well will be apparentshortly.1
performing
Nearly nine years afterthe militarycame to power, independentrede Pesquisas do Rio de Janeiro
searchersfromthe InstitutoUniversitairio
of
conducted
the public opinion surveyon
and the University Michigan
was seekingto buildsupwhichthispaper is based. At thattimethemilitary
thatlargelycontinued
a
economic
for
policies
program:
two-pronged
port
and industrialgrowthand
thenation’smainstreamgoal ofheavyinvestment
politicalpolicies thatrepresenteda sharpbreak fromthe country’sdemocraticpast.
Data and Measures
The data used in thisstudyare drawnfroma publicopinionsurveyconductedin Brazil in late 1972and early1973.2The studyincludedmorethan
‘We knowof no detailedcontentstudiesof theBrazilianmedia in thisperiod. This characterizationis based on our informalanalysisof leading printmedia and on Dassin (1982).
on thestudywere Youssef Cohen, PhilipConverse,Amaury
2Theprincipalinvestigators
de Souza, and Peter McDonough. Conductedin 1972-73, the studydrew a sample of 1,314
samBrazil,themostdevelopedpartofBrazil,and a supplementary
personsfromsoutheastern
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
326
Barbara Geddes and JohnZaller
450 questionson a wide rangeof politicallysensitivetopicsand constitutes
perhapsthehighestqualitypublicopinionsurveyever conductedin an authoritariancountry.
Althoughit is naturalto wonderwhetherrespondentsto thissurvey
mighthave been afraidto expressantiregimeattitudesto thepoll takers,we
do not believe thispossibilityposes a seriousthreatto our analysis.First,
the interviewers
were college-agepeople who mainlyopposed the regime;
yet,in personalconversations,several of themtold us theybelieved their
interviewees(who mainlyexpressedsupportforthe regime)were sincere
in theirattitudes.Second, usingthe interviewers’ratingof each subject’s
level of apparentsincerity,we foundthat”sincere” respondentswere no
more likelyto oppose the regime(at comparablelevels of exposureto its
policies)thanwere”insincere”ones. Finally,and mostimportant,
ouranalysis focusesonpatternsofsupportforregimepoliciesratherthanon absolute
levelsof support.We are unable to see how the complexpatternof regime
supportto be describedbelow could be an artifact
oftheuntruthful
answers
of fearfulrespondents.3
As indicated,”politicalawareness” functionsin our model as a determinantbothofexposureto publicaffairsand ofcapacityforcriticalscrutiny
of the ideas to whichone is exposed. In testsof the model thatfollow,we
shall use respondents’levels of political informationas our measure of
awareness.We do thisbecause past researchindicatesthat,of the several
is the mostgenerally
possiblemeasuresof politicalawareness,information
valid and reliable.4
The Brazilian surveycontainedmore than a dozen itemsdesignedas
testsof politicalinformation.
They rangedfromrelativelyeasy-the name
of the currentpresident,which was known by about 75 percentof the
ple of 352 unionmembers.The formerinvolveda multistageprobabilitysample of dwellings;
the latterwas selectedfrommembershiplistsof the 12 largestunionsin the region.We have
pooled the two samplesin orderto maximizethe numberof cases available foranalysis;the
largersampleimprovesthestatisticalprecisionofestimatesbutdoes notaffectthesubstantive
conclusionswe reach. For furtherinformation
on sample characteristics,
see Cohen (1982).
3The possibilitythatfearmaydistortsurveyresultshas been a major concernof governmentpoll takersin communistcountries.In a Bulgarianstudyof thisproblem,researchers
foundthatwhenpartyofficialsand representatives
ofacademicinstitutions
conductedparallel
studies,the expressedlevelsof regimesupportwere,on average, 15 percenthigheramongrespondentswho had been interviewedbypartyofficials.These “interviewer
effects,”however,
wereroughlyequal acrossall categoriesofrespondents,so thatoverallpatternsof regimesupportwere the same in both studies(Welsh, 1981, p. 192).
4Education,politicalinterest,media exposure,and politicalparticipationhave been used
byresearchersas measuresofwhatwe call politicalawareness.An examination,based on U.S.
data, of the relativeperformanceof such measuresover a varietyof criterionvariablesfound
thatinformation
consistently
outperformed
all rivals(Zaller, 1990).
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN REGIMES
327
asked themeaning
One ofthemostdifficult
respondents-toratherdifficult.
of SUDENE, which was known by fewer than 10 percent of the respondents.5From such questionswe constructedan awarenessindex that
rangesfromzero points,a score achievedby about 5 percentof thesample,
to 16 points,a score achieved by about 2 percent.6
InitialTests of the Model
Our principal findingis that both the mainstreamand exposureacceptance models work well in theirrespectivedomains. We make that
pointfirstwithsome simpletabularresultsand laterby estimatinga multivariateversionof equation 1. The firsttwo itemsin Table 2 illustratethe
mainstreammodel. For theseissueson whichthemilitary
essentiallycarried
forward”mainstream”policies,politicalawarenessis monotonicallyassociated withpolicysupport.
policiesis
The basic patternof supportfortheregime’santidemocratic
illustrated
bythefinaltwoitemsinTable 2. As can be seen, moderatelywellinformedrespondents(i.e., people in the65ththroughthe94thpercentiles
on awareness)are morelikelythantheless informedto favorcensorshipof
this,support
thepressand the prohibitionof strikes.Yet, notwithstanding
forgovernment
policiesfallsoffin thetop 5 percentoftheawarenessindex.
These resultsconfirmthe expectationsof the exposure-acceptancemodel.
In observingthatmorepoliticallyaware respondentsare morelikelyto
supportgovernmentpolicies,we do not implythatthe less aware opposed
thesepolicies.As Table 2 shows,fewrespondentsat low levelsofawareness
opposed officialpolicies,and, in manycases, theyexpressedno attitudesat
all. We infer,therefore,thattheprincipaleffectof exposureto progovernis to persuade the politicallyapatheticto become at
mentcommunications
least passive (i.e., verbal) supportersof governmentpolicy.In some cases
increasesin awarenesslead bothto greaterpolicysupportand to greateropposition,but (except amongthe mostaware) increasesin governmentsupportare greaterthanthe increasesin opposition,usuallyby wide margins.
More rigoroustestsof theseclaimswillcome shortly.In themeantime
a greatdeal ofcreditforshapitis clearthatour argumentgivesthemilitary
fortheDevelopmentofNortheastBra5SUDENE is theacronymfortheSuperintendency
zil, an importantand widelypublicizedgovernmentagency.
of .88. Four of theitemsweredouble-weighted
because
6Thescale has an alpha reliability
theydealt directlywithnationalpoliticsor politicalopposition;these items,scored between
0 and 2, were v134 (president’sname), v141 (natureof InstitutionalActs), v352 (numberof
officialparties),v353(name theofficialoppositionparty),wherevariablenumbersreferto the
ICSPR codebook. The othereightitems,scoredbetween0 and 1, dealtwithmoregeneralgovernmentalmatters(e.g., theSUDENE item).Theyare v130,v132,v135to v140. We awarded
as vague.
halfcreditforanswersrated by interviewers
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
328
Barbara Geddes and JohnZaller
TABLE 2
Attitudes
onSelectedPolicyItems
LevelofPoliticalAwareness
Low
Middle
High
(92)a (488) (583) (421) (81)
Brazilianmainstream
issues:
It’snotproperfora black
womantomarry
a white
man.
Strongly
agree 36b
Slightly
agree 10
Slightly
disagree 8
Strongly
disagree 41
Unsure 4
33
9
6
43
6
27
6
8
50
7
24
6
6
57
6
7
4
4
74
10
Completely
oppose 21
Moreorlessoppose 12
Neutral 4
Moreorlessfavor 15
Completely
favor 33
Don’tknow 14
34
10
4
13
29
8
41
10
5
13
27
2
63
8
5
8
15
1
63
13
4
5
14
Completely
favor 14
Moreorlessfavor 8
Neutral Moreorlessoppose 5
Completely
oppose 11
Don’tknow 61
30
14
3
6
13
32
42
19
6
4
12
15
50
18
6
7
4
4
31
28
5
10
22
3
As compared
withthe
Muchmore 16
situation
inBraziltoday,do
A littlemore 15
itwouldbe better Keepsituation
youthink
as is 9
if.. . thegovernment
had
A littleless
muchmorecontrol
overthe
Muchless 1
unions?A littlemore
No interest59
control?
etc.
24
17
13
1
1
39
27
19
21
2
2
27
20
18
33
3
6
17
5
9
49
9
15
12
To whatextent
areyou
opposedtoorinfavorof
giving
illiterates
theright
tovote?
Departures
fromBrazilian
mainstream:
To whatextent
do youfavor
fornewspapers,
censorship
radio,andtelevision?
NOTES: aFiguresin parenthesesare numbersofrespondentsin each category.
bCell entriesare percentages,but owing to miscellaneousresponses (e.g., both, refuse),
columnsdo notalwayssumto 100 percent.
ingtheattitudesofordinaryBrazilians.Is itnotpossible,however,thatthe
militaryregime inheriteda public that already supported authoritarian
policies-so thatthe regimewas more the effectthanthe cause of mass attitudes?Beforecontinuingour analysis,we digressto assess thispossibility.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
329
Popular SupportforAuthoritarianism
Many accounts of Brazilian political culture stress its authoritarian
roots.Yet muchevidenceindicatesthatpopularsupportfordemocraticinwas widespreadin Brazilin themid-1960s.Justbeforethemilitary
stitutions
decided to cancel the presidentialelection,forexample, residentsof Belo
Horizontewere asked in a poll whetherthe election”ought to take place
… (or wouldit) be betteriftherewereno election?”Sixty-five
percentsaid
theywantedtheelectionto occur,and only22 percentfavoredcancelingit.7
In responseto a somewhatmore generalquestion,83 percentsaid that”it
is importantto hold elections.” Finally,whenasked “whattypeof governmentdo you regardas thebest to solve Brazilianproblems?”70 percentfavored “a governmentelected by the people,” and only 12 percentwanted
“a strongmilitarygovernment.”
sufferedbythemilitary
These data, combinedwiththeembarrassment
in the 1965legislativeelections,make itclear thatwhateverculturalpredismay have existedin Brazil priorto the
positiontowardauthoritarianism
coup did not manifestitselfin widespreadpublic supportforauthoritarian
Yet, by the timeof the 1972-73 survey,Brazilian atpoliticalinstitutions.
titudeshad changeddramatically.Althougha surveyimmediatelypriorto
the coup had shownthatBrazilianswere waryof nearlyall social and pothemilitary
stoodout in the1972includingthemilitary,8
liticalinstitutions,
in thecountrywhichenjoyed extraordinary
73 surveyas theone institution
publicconfidence.Some 59 percentsaid theycould “always trust”the militaryor “trust(it) in mostcases,” while only 12 percentalwaysor mostly
includingthe Roman Catholic
distrustedit. No othergroupor institution,
church,rated nearlyas high.9
More directlypoliticalquestionsreinforcethe impressionthatthe militaryand its policies had become widelypopular by 1972. At a timewhen
the military’sdominationof politicswas nearlycomplete,41 percentdeclared themselves”completelyin favor” of “militaryinvolvementin nationalpolitics,”while another17 percent”more or less” favoredsuch in7Theremaining22 percentwereundecided.These data are available throughthe ICPSR,
study7613,”PoliticalBehavior and Attitudesin a Brazilian City, 1965-66.”
8A surveyof residentsof Rio de Janeiro,”World War II: AttitudestowardDomestic and
ForeignAffairs,”ICPSR study7048,was conductedonlyweeksbeforethecoup. Asked about
shouldhave in nationalaffairs,11 percentsaid it had too
the amountof influencethe military
little;32 percentsaid therightamount;20 percentsaid too much;and therestwereundecided.
These ratings,whichare similarto thosegivenothermajor politicalactors,betraylittleindication of a populationpredisposedtowardauthoritarianism.
9Figuresfor”co-workers”are 45 percenttrustand 22 percentdistrust;forpoliticians,20
percentand 47 percent;forfriends,40 percentand 30 percent;forpriests,47 percentand 27
was “the government,”whichthemilpercent.The onlygroupmoretrustedthanthemilitary
itarythencontrolled.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
Barbara Geddes and JohnZaller
330
volvement,and only16 percentopposed it. Moreover,81 percentsaid they
were completelyor mainlysatisfiedwith”governmentpolicy,” and only5
percentindicatedany degree of dissatisfaction.10
Because the1972-73 surveyfailedto repeatanyoftheearlierquestions,
we cannot know exactlyhow much attitudeshad changed since the mid1960s. But thereis littledoubtthatmuchchangehad occurred.How can it
be accountedfor?
Sources of Resistanceto Persuasion
Accordingto our model,supportforregimepoliciesdependsjointlyon
messagesand
thechancesthatindividualswillbe exposed to progovernment
theprobabilitythattheywill accept them,givenexposure.We initiallyassumedthatgenerallevelsof politicalawarenesswould be a major determinantofbothexposureand acceptance,butthatotherfactors,especiallypersonal values, would also play a role. We now consider these additional
factorsin detail.
Exposure Factors
Frompriorresearchwe expectthe mosteffectivemeasureof exposure
to politicsto be politicalawareness(Zaller, 1990). The Braziliansurveyalso
communicacontainstwomoredirectmeasuresof exposureto government
tions:itemsaskingwhetherindividualswatchtelevisionnewsor read a daily
newspaper.(Both mediawereheavilycensoredin thisperiod.) The twomedia variablesand the awarenessscale willthereforeconstitutemeasuresof
exposureto authoritarianvalues.
ResistanceFactors: Economic Interest
As noted,the militaryseized powerduringa periodof economicchaos
and politicalunrest.As part of its effortto stabilizethe economy,it held
downwages and suppressedunionactivity.Duringthefirstyearsofmilitary
rule, these policies led to a 25 percentdecline in real wages. The benefits
of rapiddevelopment,whentheybegan to appear in 1968, accrueddisproportionatelyto the middleand upper classes. One mightthereforeexpect
that,at comparablelevels of exposureto regimepolicies,union members,
unskilledand semiskilledworkers,and low-incomepersonswould be most
conresistantto regimepoliciesand thatblacks,who are disproportionately
centratedin thesegroups,wouldalso be resistant.On theotherhand,members of the middleand upper occupationalclasses and people withhighincomes mightbe expected to be most receptive.
in thispe’0Otherleadingaccountsalso stressthe popularityof the militarygovernment
riod (Cohen, 1982; McDonough, 1982, 1984).
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
331
ResistanceFactors: Politicization
A centralassumptionof themodel is thatthegreaterone’s attentionto
politics(or any subject), the greaterone’s capacityforcriticalscrutinyof
ideas relatingto it. Politicalawarenessis thebest indicatorof suchcapacity
butanything
thatdenotesgreaterlevelsofpoliticization
forcriticalscrutiny,
in politicsis one such factor.Educasimilar
to
have
effects.
Interest
ought
tion,whichis widelythoughtto imparta generalizedcapacityforcritical
scrutiny,may also promotescrutinyof politicalcommunications.
The 1972-73 surveycontainsseveral othermeasuresof politicization.
One is activityas a union officer,whichmightmultiplythe simpleeffects
of unionmembership(de Souza, 1979). Since politicalactivityis more intensein urban areas thanruralones, one mightexpectthatgrowingup in
an urbanarea would be a resistancefactor.Finally,one mightexpectBrazilianswho had “grownup withdemocracy”(i.e., people who came to poin the democraticperiod between1945 and 1964) would be
liticalmaturity
more resistantto authoritariandeparturesfromdemocracy.In contrast,
before
(i.e., came to maturity
thosewho “grewup withauthoritarianism”
1945 or after1964) mightbe less resistantto the military’spolicies.
ResistanceFactors: Personal Values
The surveycarriedseveral measuresof personalvalues thatmaypreOne is churchattendance,
dispose people foror againstauthoritarianism.
whichroughlymeasuresthe extentto whichpeople hold conventionalretowardthekind
mayindicatepredispositions
ligiousbeliefs.Such religiosity
of conservativevalues favoredby the military.”
In a relatedvein it has been argued(Adorno et al., 1950) thatsome individualspossess personalitiesthatpredisposethemtowardauthoritarianas F-scale
ism.The Braziliansurveycarriedseveralitems,knowntechnically
The itemscoversuchmatitems,designedto measurethesepredispositions.
ters as obedience to parents, respect for elders, and contempt for
weakness.12By summingup each person’s answersto these questionspoliticalcontent-we obtaineda measure
noneofwhichcontainsmanifestly
“At the timeof the survey,the Roman Catholic churchhad not yetbecome a forceof
oppositionto the regime.Had the surveybeen taken later,our expectationsconcerningthe
It is interesting,
however,thateven
effectsof churchattendancemighthave been different.
in Poland, wherethe Roman Catholicchurchis a focalpointof resistanceto the government,
churchattendanceis correlatedwithsupportforregimepolicies (Ryszka, 1987).
‘2In additionto these items(v300, v311, and v305, respectively),the otheritemsin the
scale involvedassertionsthat”the worldis dividedbetweenthe weak and the strong”(v307)
and thatemployersshould be toughon theirworkers(v303).
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
332
Barbara Geddes and JohnZaller
whichcould be expectedto tap psychologicalpredispositionsto favoror to
oppose authoritarianpolicies (given exposure to them).13
Finally,respondentswho could recallthenames of thepoliticalparties
fromthe country’sdemocraticperiod were asked whichof thembest reflectedtheirviews. We expectedpeople who preferredthe moreconservativeparties,the UDN and the PSD, to be more receptiveto authoritarian
the principallabor party,the PTB.
policies than people preferring
PreliminaryTestingforResistanceto Persuasion
The exposure-acceptancemodel createsverystrongexpectationsabout
how these resistancefactorsought to affectsupportfor regimepolicies.
Withingroupspredisposedtowardconservatism
(e.g., thosescoringhighon
theF-scaleitems),we expecta positiverelationship
betweenpoliticalawareness and supportforauthoritarian
policies (as in panel A of Table 1). Only
withingroupswhosepropensityagainstauthoritarianism
is strong(e.g., low
scorerson theF-scale) shouldwe findnonmonotonicpatternsofpolicysupport (panels B and C).
To testthese expectationswe created a summaryscale of Supportfor
authoritarianpolicies. The scale has eightitemson such mattersas press
freedom,theneed to combatsubversion,and suppressionofunionactivity.
The items,each coded either1 forsupportor 0 fornonsupport,were averaged to create a 0-1 scale. Thus, a score of .625 would indicatesupport
forthe governmentpositionon fiveof eightitemsand nonsupport(either
neutral,opposed, or uninterested)on the otherthree.
We also createda scale measuringaverage levels of Oppositionto authoritarianpolicies. The scale includesthe same itemsas in the Support
scale, except thathere theyare coded 1 foroppositionand 0 otherwise.14
Thus, a personwho opposed the governmenton all policies would have a
score of 1.0 on theOppositionscale and 0.0 on theSupportscale. A person
on all eightitemswouldscore0.0 on boththeSupwho expressedneutrality
portand the Oppositionscales. As willbe seen, thetwo scales enable us to
have oftenbeen criticizedon thegroundsthat
“3Measuresofauthoritarian
predispositions
theysufferfroman acquiescenceresponsebias thatis correlatedwitheducationand thatthey
measuresubculturalvalues ratherthanindividualattitudes(see Altemeyer,1981). To insure
we used multipleregressionto removestatistically
againstthesedifficulties,
the effectsof edfromthe measure.Thus, our
ucation,age, race, place of residence,and politicalinformation
authoritarian
dispositionsmeasureis uncorrelatedwithmeasuresofindividuals’subculturalmilieu.
14Inadditionto theitemsinTable 2 (v145,v341), thesewereto allowstrikesor not(v313);
more/lessoppositionto government(v157); more/lesseffortto combatsubversion(v165); respectforindividualrights(v298); favor/oppose
militaryinvolvementin politics(v343); favor/
oppose indirectelectionsof stategovernors(v344); favor/oppose
partieshavingmore power
(v345); favor/opposea third(i.e., not government-sponsored)
party(v356).
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
333
tracktrendsin supportand oppositionseparately;because of neutraland
“no interest”responses,the two indicatorsare farfrommirrorimages of
one another.15
Note thatall itemsin thesetwoscales involvepolicieson whichthemilitarydepartedfromthemainstreamdemocraticpoliciesofthepriorregime.
Thus if,forexample,a personscored highon Oppositionand low on Support,itwouldindicatetenacioussupportfordemocraticideas and resistance
to authoritarianones.
Finally,we createda six-itemindexof supportformainstreampolicies
suchas economicdevelopmentand racialtolerance.16These are policieson
continuedpoliciesof previousregimes.The
government
whichthemilitary
indexmeasuresaverage levels of supportformainstreampolicieson a 0-1
scale.
Figure 1 showshow one resistancefactor,eduBy way of illustration,
cation,affectsregimesupport.In the top panel we see thatamongpersons
withfouror feweryearsof school-about 60 percentof the generalpopulation in the surveyarea-there is a strongpositiverelationshipbetween
policies. Supportforgovernment
awarenessand supportforauthoritarian
in thisgrouprisesfroman average level of about 25 percentto an average
of 68 percent.Meanwhile,among moderatelyto highlyeducated persons,
therelationshipbetweensupportand awarenessis, as expected,nonmonotonic.
The lowerpanel of Figure 1 showsOppositionscores. As can be seen,
awarenessis associatedwithoppositionto regimepolicies,butonlyamong
persons(i.e., thosewithsome capacityforcriticalscrutiny
better-educated
of politicalideas).
The effectof awarenesson supportformainstreampoliciesis shownin
Figure2. Here greaterawarenessis associatedwithsteadilygreatersupport
forofficialpolicies, and the highlyeducated are, if anything,less resistant
to persuasionthanthepoorlyeducated. Because thispatternof supportfor
mainstreampoliciesholdsno matterwhichresistancefactor-education,Fscale scores,religiosity-isused as thecontrolvariable,we shall devote no
attentionto mainstreampolicies. The storyhere is simplythatexfurther
posure acrossseveralregimesto a discoursethatis “heavilyloaded on one
side” tends,as in the United States,to induce supportforthe policies embedded in thatdiscourse.
“There is, however,a pricefordividingthe itemsinto two scales. The division,in conjunctionwithheavyskewson manyof the 0-1 items,yieldsan alpha reliabilityof .63 forthe
Supportscale and .55 forthe Oppositionscale. The scales may nonethelessfunctionwell as
measuresof average levels of supportfora range of governmentpolicies.
“6Themainstreamitemswere those in Table 2 (v312, v397) plus: encourageforeigninminimumwage (v288); allow black man to marrywhite
vestment(v149); build factories/raise
woman (v299); favor/opposedeath penaltyforordinarycrimes(v395).
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
FIGURE 1
Mean Levels of Supportand OppositionbyEducationand PoliticalAwareness
PolicySupport
0.75
Low Educ.
Mid. Educ.
0.50 –
High Educ.
Average
Policy
Support
0.25 –
PercentileScore on Awareness
0.75
Average
Policy
Opposition
PolicyOpposition
–
0.50_
High Educ.
Mid. Educ.
0.25
Low Educ.
0.00
PercentileScore on Awareness
N=
9 48 116 46
80 219 180 26
85 394 310 119 10
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
335
FIGURE 2
MeanSupport
forMainstream
PoliciesbyEducationandPoliticalAwareness
HighEduc.
Mid.Educ.
LowEduc.
0.75 –
Average
o 50 –
for
Support
Mainstream
Policies
0.25-
0.00’
0.00
0.25
0.50
0.75
1.00
See Figure1 forcell frequencies.
Attitudestowardnonmainstreampolicies, however,requiremore attention.Figure 1 shows the effectof just one factor,education. What is
needed is a wayto take intoaccounthow exposuresimultaneously
interacts
withseveralresistancefactors.For thiswe returnto the formalstatement
of our model.
A Fully SpecifiedModel
The basic model is summarizedin equation 1. To estimateit one must
specifya functionalformforthe exposureand acceptancefunctions.Since
the model deals in probabilitiesof exposureand acceptance,the functions
can varyonlybetween0 and 1. They mustalso be monotonic.The logistic
formis themosttractableofthecommonlyused functionalforms
functional
variables
meetingtheserequirements.Ifwe assumethattheindividual-level
*, xn) and thatthe individual-level
affecting
exposure are (x1,x2,
variables affectingacceptancelexposureare (Yl, Y2,
*, Yn), we can rewrite
equation 1 in termsof logisticfunctionsas follows:
SuPPort (
>(
1?+F
exp[AO?Axl ?+A2x2.
1
1-1[{BB’B
Rl, |Rn,
.
.+
Anxn])
.?Bnyn}])ll
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
(2)
336
Barbara Geddes and JohnZaller
The firstterm in equation 2 is a monotonicallyincreasingfunctionof
individual-level
exposurevariablesin X, whilethe second is a similarlydecreasingfunctionof resistancevariablesin Y.17
We have estimatedequation 2 forboth the Support and Opposition
scales.18 The results,givenin Table 3, providecomplementary
perspectives
on theregime’ssuccessinshapingmassattitudes.The resultsfortheSupport
scale capturethe process by whichindividualscame to supportsuch antidemocraticpoliciesas a censoredpress,indirectelections,and so forth.Resultsforthe Opposition scale describepersistenceof prodemocraticattitudes in the face of the regime’seffortsto induce change.
We beginwithan examinationof exposureeffects.(To facilitatecomall variableshave been coded to a 0-1 range.)
parisonsamongcoefficients,
Political awareness is clearly the dominantexposure variable,19having
roughlythe same effectin both the Supportand Oppositionmodels.20Althoughvariablesmeasuringexposureto televisionnewsand newspapersare
in theSupportmodel,theireffectsare modest.Each
statistically
significant
of thesedichotomousvariableshas about the same effectas a movementof
one step on the 17-stepawarenessscale.
We turnnow to resistanceeffects.Positiveresistancecoefficients
in the
Supportmodel indicatehighresistanceto messagesfavoringauthoritarian
in the Oppositionmodel indicateresistanceto
policies;positivecoefficients
the democraticideas of previousregimesand readinessto abandon themin
the authoritarianperiod.
170ne interpretation
of thefloorparameter,F, in equation 2 is thatit capturestheeffects
of guessing.Anotheris thatthe readingof a questionconstitutesa message in its own right
and thatthissortof exposure(understoodas bothencounteringand comprehending
the mesthe parametershould be
sage) is capturedby the floorparameter.On eitherinterpretation
themodelwe have imposed
equal in boththeSupportand Oppositionequations.In estimating
thisconstraint.
‘8We used the nonlinearregressionwithnumerically
calculatedderivatives.To checkrobustness,we estimatedthe model numeroustimes.Withstartingvalues in the neighborhood
of thosein Table 3, SAS reconvergedto the initialsolution.For arbitrary
values distantfrom
fit.In addition,
thoseinTable 3, SAS eitherfailedto convergeor convergedon a muchinferior
we successfully
replicatedtheprincipalsubstantiveresultsin Table 3 witha morecumbersome
interactiveregressionmodel.
in the Oppositionscale reflectsthe extentto whichpeople
‘9The awarenesscoefficient
were exposed to prodemocraticideas fromanysource. We suspectthatmuchof thisexposure
occurredin theprevious,democraticregime,butitmayhave occurredvia wordof mouth,the
schools, or uncensored,small-circulation
publicationsin the yearsin whichthe militarywas
in power.
betweenthesecoefficients
20Thesimilarity
does notimplythatoverallexposurelevelswere
the same in both models. All else equal, the interceptshiftdescribedat the bottomof Table
3 impliesthatexposureto democraticideas was muchlower (at least duringthisperiod) than
exposureto authoritarianones.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
TABLE 3
forExposure-AcceptanceModela
EstimatedCoefficients
Exposurevariables:
Politicalawarenessb
WatchTV news(v62)
Read dailynewspaper(v70)
Resistancefactors:politicizationvariables:c
Politicalawarenessb
Middle education(5-11 yearson v75)
High education(12 + years)
Growup in city(v418)
Interestin politicsd
Union officer(v195)
Politicalgeneration(18-28, 47 + yrs.)
Resistancefactors:variablesmeasuringpersonalvalues:
Religiosity(v440)
F-scaleb(liberalscoredhigh)
PreferUDN party(v359)
PreferPSD party(v359)
PreferPTB party(v359)
variables:
Resistancefactors:economicinterest
Income (logged) (v254)
Union member(v170)
Highoccupationalstatus(1, 2 on v443)
Low occupationalstatus(7, 8 on v443)
Nonwhite(v408)
Policy
Support
Policy
Opposition
4.25
(.49)
0.33
(.09)
0.29
(.09)
4.57
(1.3)
Inapp.
1.78
(.54)
0.18
(.18)
0.58
(.15)
0.22
(.11)
-0.27
(.16)
0.79
(.26)
0.04
(.10)
-0.49
(.16)
1.60
(.29)
-0.45
(.17)
-0.32
(.16)
0.09
(.12)
-0.75
Inapp.
2.68
(.76)
-0.47
(.17)
-0.96
(.25)
-0.35
(.14)
-0.62
(.24)
-0.92
(.53)
0.11
(.12)
0.88
(.29)
-1.34
(.42)
0.18
(.18)
0.13
(.18)
-0.60
(.21)
(.61)
-0.36
(.10)
0.01
(.13)
0.13
-1.03
(.78)
0.17
(.13)
0.09
(.18)
-0.48
(.12)
0.00
(.17)
-0.21
(.13)
(.17)
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
Barbara Geddes and JohnZaller
338
(Table3 continued)
Constants
(seeeq. 2):
Ao
-3.78
(.78)
-0.17
(.73)
.26
Floorparametere
-1.89
(.23)
-2.57
(.70)
.26
(.05)
(.05)
Varianceexplained(percentage)
errorofestimate
Standard
31.4
.21
12.3
.18
Bo
aAll variablesare coded to a 0-1 scale. Standarderrors,whichshouldbe considered
belowthecoefficients.
arereported
inparentheses
estimates,
approximate
intext.
isdescribed
bVariable
thatthevariableis associatedwith
on a resistance
variableindicates
CApositivecoefficient
tothegiventypeofpolicy.
resistance
interest”
scale:v107
dThefollowing
variablesweresummedandrecodedto a 0-1 “political
ingovernment).
inpolitical
issues),v347(veryinterested
v346(veryinterested
(talkpolitics),
tobe equalinbothmodels.
eConstrained
The big storyhereis thestrongeffectsof the “politicization”and “personal values” variablesin contrastto theweak showingof “economicinterest” variables.Amongthe politicizationvariables,awarenessis clearlythe
inbothequations
mostimportant.It has largepositiveresistancecoefficients
and is the onlyvariable to do so. Thus, the most aware personsare most
selectivein theideas theywillaccept,regardlessofideologicalcontent.This
is a strongand nonobviousfindingand indicatesthatawarenessis a valuefreeindicatorof criticalresponsesto politics.21
efIn contrast,the otherpoliticizationvariablestendto have different
fectsin the two models: education,socializationin an urban area, and activityas a unionofficerare associatedwithgreaterresistanceto authoritarto democraticones. In addition,political
ian ideas and greatersusceptibility
inis moresubtlethanitmayseem.Sincethemodelinequation2 is highly
21This
finding
theeffect
ofanyone variabledependson thevaluesofall others.Suppose,then,
teractive,
intheOpposition
model.Ifthisweretrue,itwouldimplythat
hadno effect
thatawareness
ininducing
refactors
etc.)wouldbe equallyeffective
religiosity,
otherresistance
(education,
though
finding,
Sucha counterfactual
todemocratic
valuesat alllevelsofawareness.
sistance
to
thatresistance
ofthemodel,namely,
wouldviolatea basicsupposition
easilyimaginable,
and
forcritical
byawareness)
thinking
(as measured
dependson boththecapacity
persuasion
one towardresistance.
Notealso thatthelargecoefficient
one or morefactors
predisposing
or auawarenecessarily
resisteitherdemocratic
doesnotmeanthatthehighly
on awareness
arealsopresent
factors
thoritarian
onlywhenoneormoreotherpredisposing
values;theyresist
(see Figure3).
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
339
interesthas an importanteffectin the Oppositionmodel. Thus, the politicizationexperiencecapturedbythesevariablesappears-unlike thatofthe
variables,
awarenessvariable-to be a value-ladenone. Of thepoliticization
only “politicalgeneration”entirelyfailsto performas expected.
The variablesmeasuring”values” also workwell. The performanceof
theF-scaleis especiallystrong,approachingeven thatofpoliticalawareness.
Religiosityis also important,and, like conservativescores on the F-scale,
to authoritarian
ideas and resistanceto democratic
it inducessusceptibility
ones. Preferencesfor a labor party(PTB) or a more conservativeparty
(UDN), whichwe take as indicatorsof generalideologicalorientation,also
have effectsin the expected directions.
Meanwhile,just two of the “economic interest”variableshave statiseffects,and one of thesehas thewrongsign.The variable
ticallysignificant
thatconformsto expectationsis low occupationalstatus,whichhas a modeffect(butnotan anti-authoritarian
one). The variable
erateprodemocratic
thatbafflesexpectationsis union membership,whichinduces greatersusto authoritarianism.
The effectsof theeconomicvariables,howceptibility
ever, tendto be small. The onlylarge effectis thatof income,but, despite
in an unexpecteddirection.
thateffectrunsnonsignificantly
ourbestefforts,
perThus,despitetheirsupposedvestedinterestin theregime,high-income
sonsin theOppositionmodelare moreprodemocraticthanlow-incomepersons.
These results,taken together,confirmthe generaltendencyof “symbolic” concerns-value attachmentsand politicizationexperiences-to
dominate narrowly”self-interested”concerns in most attituderesearch
(Sears et al., 1980; Kinder and Kiewiet, 1981).22
to interpret,especiallyin a
Raw coefficients
are, of course, difficult
this
For
one can see inFigure
such
as
one.
example,
model
highlyinteractive
1 thateducationhas a muchgreaterimpactat highthanat low awareness,
and the same is trueforeveryvariable in the model. Therefore,we have
of the results.
preparedFigure3 as an aid to interpretation
Figure3 presentsestimatedSupportscores and Oppositionscores for
a median surveyrespondent.It also presentsnet scores, whichhave been
estimatedOppositionscores fromestimatedSupobtainedby subtracting
portscores.In addition,Figure3 presentsestimatesfor”types”who are eitherespeciallysusceptibleor especiallyresistantto authoritarianpolicies;
thesetypesrepresentextremebut not unlikelycombinationsof important
ofthe
measuresofeconomicinterestfarepoorly,thestrength
22Although
individual-level
Brazilianeconomyin the early 1970s is, as we have emphasized,almostcertainlythe reason
enjoyedthestrongoverallsupportthatitdid. This argumentis consistentwith
thegovernment
the notionthat,as Kinderand Kiewiet (1981) maintain,the economyaffectsindividualattitudes sociotropicallyvia theirevaluationsof the nationaleconomy.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
FIGURE 3
EstimatedPatternsofPolicy
PolicySupport
PolicyOpposition
Low resist
,-,.8o
t14.80
)
0
0 .80
Median
.60
0-
High res
. .60o
0
.40~~~~~~
>
.20
_________________
High resist
.00 1
0
20
40
6
60
80
10
100
~~~~~~
~~~~~
.20
Low re
0
20
40
60
80
10
AwarenessPercentiles
AwarenessPercentiles
Lowresistance
type:
Lessthanfouryrs.school
Bottomthird
onF-scale
Unionmember
Clericaljob
Weekly
church
attendance
Prefers
UDN party
Raisedincountry
Medianrespondent:
Sixyrs.school
MedianonF-scale
Nonunion
member
Manualjob
church
Monthly
attendance
No party
preference
Raisedincountry
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
341
variables.The “low-resistance”typeis a personwho has no morethanfour
yearsofeducation,was raisedinthecountry,worksat a nonmanualbutroutineoccupation,attendschurchat leastonce a week, scoresconservativeon
type
the F-scale, and feelsclose to the UDN party.The “high-resistance”
is a personwho has gone to the twelfthgrade in school, grewup in a city,
holds a middle-statusjob, never attendschurch,scores liberal on the Fscale, and feelsclose to the PTB, the Labor party.
The mostobviouspointin Figure3 is thatthecumulativeeffectsof the
resistancefactorsare verylarge. Supportforauthoritarianpolicies in the
low-resistance
grouprisesfrom.35 at low awarenessto .83 at highawaregroup,bycontrast,thepattern
ness; amongmembersof thehigh-resistance
ofsupportis gentlynonmonotonicand neverexceeds about .20. The effects
of the resistancefactorson the net scores are stilllarger.
thevariousresistancefactorsinteractwith
Note, however,howstrongly
thereis littledifference
betweenthelowAt low information,
information.
thedifferences
are
types;at highinformation,
resistanceand high-resistance
dramatically
large. This interactionoccursbecause thepoliticallyunaware,
failto ensimilarto thoseofothers,typically
thoughhavingpredispositions
(or propaganda) necessaryto converttheprecounterthecommunications
dispositionsinto actual policypreferences.Only the highlyinformedare
intoconventionally
appropriable reliablyto translatetheirpredispositions
ate policypreferences.
The finalpointin Figure3 is how littleoppositionto regimepoliciesexisted in 1972-73. In only one small segmentof the Brazilian publicmembersof the high-resistance
groupwho scored in the upper halfof the
awarenessscale-does one findthatnet supportforgovernmentpoliciesis
negative.A medianBrazilianwas almostas supportiveof the government
as the “low-resistance”type.We concludethat,undertheprosperousconditionsin Brazil at the timeof the survey,mostBrazilianswere highlysusappeals-provided thattheypaid enoughattention
ceptibleto authoritarian
to politicsto findout whichparticularpolicies the governmentsupported.
Variationsin Popular SupportforAuthoritarianRegimes
We began thispaper by askingwhichkindsof people are most likely
to supportauthoritariangovernmentsand whichare least likely.We can
now suggestdefiniteanswersto thisquestion.Supportis likelyto be stronmembersof the working
gest among people, such as the better-informed
communicaclass,who are bothheavilyexposed to government-controlled
tionsand lack theeducationand otherinternalresourcesforresistingthem.
Supportis likelyto be weakestamongthosewho eitherpay no attentionat
all to politics(e.g., mostilliterates)or who pay a greatdeal of attentionand
are predisposedbyvirtueofpersonalvaluesor pastpoliticizationagainstauthoritarianism.
These generalizations,we believe, should applyin any au-
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
342
Barbara Geddes and JohnZaller
thoritarian
countryin whichthe governmentappears to performwell and
achievescontrolor near controlover the flowof politicalcommunications.
One should not, however,place too muchemphasison the particular
patternsof regimesupportfoundin Brazil. The model leads us to expect
variationin thesepatternsfromone countryto another.We set
significant
out theseexpectationsin Figure4, a typologyclaimingthatvariationsin regimesupportdepend on how vigorouslygovernments
seek to indoctrinate
theircitizensand how muchaccess to alternativevalues theircitizenspossess.
In countriesin whichthegovernment
makes energeticefforts
to indoctrinateits citizens (the governmentsof Nazi Germany and the Soviet
Union23are examples) even the least politicallyinformedmembersof societymay (in comparisonwiththe least-informed
Brazilians) exhibitmoderatelyhighlevels of supportforregimenorms(compare the cases in the
right-hand
columnof Figure4 to thosein the left-handcolumn). A second
sourceofvariationinpopularsupportforauthoritarian
regimesis theaccess
of citizensin different
countriesto alternativesourcesof values. Many citizens in authoritarianBrazil retainedaccess to books and othersmall circulationpublicationsthatcarriedalternativevalues; some also remembered
the democraticnormsof thepreviousregime.In otherauthoritarian
countries,however,(Albania comes to mind) scarcelyany segmentof the populationhas eitherpersonalmemoryof, or access to, sourcesof antiregime
values. In consequence,thedeclinein regimesupportthatwe foundamong
Braziliansshouldshow up moreweakly,or perhapsnot
the best-informed
at all, amongthebest-informed
citizensof countriessuch as Albania (compare cases in the top row of Figure4 to those in the bottomrow).
It is difficult
to thinkof modernauthoritarian
countriesthatfallclearly
into the lower rightcell of Figure4. However, one mightreasonablyconsiderthe United States an exampleof a nonmobilizingregimewhose elites
have achievedhighlevelsofvoluntaryagreementon certainnormsrelating
to capitalism,democracy,and, at times,foreignpolicy.The worksof Key
(1961), Mueller (1973), and Chong, McClosky,and Zaller (1985) indicate
thatwhensucheliteconsensusexists,U.S. publicopinionindeed conforms
to the patternin the lower rightcell.
In proposingthistypology,we have in mindcases in whichthegovernmentmaintainsat least the appearance of competenceand effectiveness.
Certainlythiswas trueoftheBrazilianregimeat thetimeofour survey.For
23Indevelopingthistypology,we assumethatauthoritarian
regimesespouse authoritarian
policies.The Gorbachevgovernmentin the Soviet Union, however,is a partialexception.Its
glasnostand perestroikapolicies mustbe viewed, in the contextof Soviet history,as nonauthoritarian.We would thereforeexpect patternsof supportforthese policies to differfrom
those forecastforothercases.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
343
FIGURE 4
A Typology
ofSupport
forMainstream
Policies
RegimeEffort
toMobilizeOpinion
HighEffort
Low Effort
Nazi Germany? SovietUnion?
Brazil
(fromFig. 2)
Low
Educ.Lo
CitizensHave
v
Some Access
to Alternative
PoliticalValues
Educ.
High
Educ.
Educ
o
X
PoliticalAwareness
PoliticalAwareness
Albania?
CitizensHave
No Real Access
to Alternative
PoliticalValues
;
m
UnitedStates
issuesonly)
(consensus
Low
Educ.
High
Educ.
Low
Educ.
High
Educ.
m
m
>
X~
X
PoliticalAwareness
PoliticalAwareness
cases in whichgovernmentsare markedlyless effective(or less luckywith
the economy),one would expect more resistanceto governmentpolicies.
The model readilyaccommodatestheeffectsof suchheightenedresistance.
Returningto Table 1, we see thatas resistanceincreases,patternsof policy
supportbecome nonmonotonicand perhapsnegative.Hence, forthe case
of an ineffective
government,we would expect, ceterisparibus, a gently
nonmonotonicpatternof supportamongthe moderateto poorlyeducated
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
344
Barbara Geddes and JohnZaller
and a nearlyflator negativerelationshipwithawarenessamongthe better
educated. (It is straightforward
to redrawFigure4 to illustratesuchcases.)24
The exposure-acceptancemodel thusspecifiesthe ways in whicha varietyoffactors-a nation’spriorexperiencewithalternativeideologies,the
intensity
oftheregime’sefforts
at publicmobilization,theaccess ofcitizens
to alternativesourcesof ideas, and the regime’sperformance-interactto
affectoverallpatternsof regimesupport.One would have greatdifficulty
developingsucha modelon thebasis ofthescantyopiniondata thatis available frommostauthoritarian
nations.But nowthattheexposure-acceptance
model has been testedon the excellentdata available fromauthoritarian
Brazil,itcan be a usefultoolforinterpreting
themorefragmentary
evidence
fromotherauthoritarian
countries.For example, PhilipRoeder (1985) reviews evidence showingthat,althoughoppositionto the Soviet regimeis
greatestwithinthe best-educatedsegmentsof society,education is positivelyassociatedwithindividualsusceptibility
to indoctrination
programsin
factorysettings.In otherwords,supportforthe regimeseems to rise with
education-except at theveryhighestlevelsofeducation,wheresupportdeclines somewhat.On the basis of thisand otherprovocativeevidence,he
suggeststhat “the relationshipbetween dissentand social mobilizationis
curvilinear,decliningwithearlysocial mobilizationunder Partytutelage,
but risingas education and urbanizationbecome still more advanced”
(Roeder, 1985,pp. 5-6). Such findings,
we believe,can readilyand usefully
be interpretedin lightof our model.
Two surveysconductedin Poland in the aftermathof the Solidarity
movementprovideadditionalevidenceconcerningour theory.A 1984study
by researchersat the Universityof Warsawfoundthatsupportformartial
law and opposition to Solidaritywere greatest among people at lowto-middlelevels of education.In 1985,whenthe government
had begunto
promotea returnto normalcy,manylow-educationrespondentshad lapsed
into no opinion, so thatsupportformartiallaw was thengreatestamong
people at middlelevels of education (Ryszka, 1987, p. 253).25
An importantcaveat is in order at thispoint. In some authoritarian
countries,oppositionactivistsdevelop unofficial
networksto challengeregimedominationofpoliticalideas. This happenedin thePhilippinesduring
24Itis interesting
to notethatwhentheBrazilianeconomyfaltered,supportfortheregime
collapsed firstin just those quartersin whichour analysiswould lead us to expect it: among
themosthighlyeducatedand politicallyinvolvedsegmentsofthepopulation(Alves, 1984; Lamounier,1980). See Zaller (1987, p. 823) forfurther
analysisof the patternsthatcan be generated by the model.
25Wewish to thankStanislawGebethnerforbringingthese data to our attention.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
345
theMarcos regimeand recentlyin SouthKorea.26It is obviouslyimportant
to understandhowand whenalternativepoliticalnetworksarise.Our paper,
however,has not addressedthisvitalproblem.It seeks onlyto accountfor
thepatternsof regimesupportthatoccurwhenregimesare successfulin assertingcontrolor nearcontrolover the disseminationof news and analysis.
ConcludingRemarks
This paper has attemptedto advance theunderstanding
of publicopinion in severalways.First,ithas shownthata setofmodels-the mainstream
model and theexposure-acceptancemodel-that originatedin U.S. politics
and U.S. social psychologycan be putto effective
use on mass opiniondata
nation.In so doing,it not onlyconfirms
froman authoritarian
but extends
these models. The public opinion fieldhas fewmodels thathave survived
testsundersuchdiverseconditions.Second, thepaper providesa morerigorous and generalaccountof thepatternsof mass supportfortheBrazilian
regimethanhas previouslybeen available; in particular,it has been able to
uncoverand explainpatternsof regimesupportthatwould seem likelyto
defeatmore standardclass-basedapproachesto the understanding
of Brazilianopinion.Third,thepaper providesa basis foranticipating
thepatterns
of popular supportthatexistin severaldifferent
kindsof authoritarian
regimes.These expectationsmayproveusefulto scholarsattempting
to make
sense of the limitedopiniondata available fromsuch regimes.Finally,the
paper advances an importantsubstantivefindingabout supportforauthoritarianregimesof the typethatexistedin Brazil: supportwillbe strongest
amongcitizensin the broad middlerangesof politicalawareness-among
people, thatis, whoare informedenoughto be fairlyheavilyexposed to governmentindoctrination
programsbutwho are not sufficiently
sophisticated
or motivatedto resistthem.
Manuscriptsubmitted14 May 1987
Final manuscriptreceived25 July1988
REFERENCES
Daniel Levinson,and R. Nevitt,Sanford.1950.
Adorno,TheodorW., Else Fenkel-Brunswik,
The authoritarian
personality.New York: Harper.
Alves, Maria Helena Moreira. 1984. Estado e oposi,co no Brasil [Stateand oppositionin Brazil]. Petr6polis:Vozes.
ofManitobaPress.
authoritarianism.
Altemeyer,Bob. 1981. Right-wing
Winnipeg:University
Baer, Werner. 1983. The Brazilian economy.2nd ed. New York: Praeger.
261Inthelattercase, thegovernment,
allowedthemassmedia a fair
althoughauthoritarian,
its policies.
amountof latitudeforcriticizing
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
346
Barbara Geddes and JohnZaller
Cardoso, MiriamLimoeiro. 1978. Ideologia do desenvolvimento,
Brasil: JK-JQ [Ideologyof
development,Brazil: JK-JQ]. 2nd ed. Rio de Janeiro:Pas e Terra.
Chong,Dennis, HerbertMcClosky,and JohnZaller. 1985. Social learningand theacquisition
ofpoliticalnorms.Ch. 8 in McCloskyand Zaller, eds., TheAmericanethos.Cambridge:
HarvardUniversityPress.
Cohen, Youssef. 1982.The benevolentleviathan:Politicalconsciousnessamongurbanworkers
understate corporatism.AmericanPoliticalScience Review,76:46-59.
flowand thestabilityofpartisanattitudes.Public Opinion
Converse,Philip.1962.Information
Quarterly,26:578-99.
Dassin, JoanR. 1982. Presspolicyin thesoutherncone: The case ofBrazil 1964to thepresent.
Presentedat theannual meetingof the AmericanPoliticalScience Association,Denver.
De Souza, Amaury.1979. The natureof corporatistrepresentation:
Leaders and membersof
organizedlabor in Brazil. Ph.D. diss., MassachusettsInstituteof Technology.
Dryer,Edward C. 1971. Media use and electoralchoices: Some consequencesof information
exposure. Public Opinion Quarterly,35:544-53.
Erickson,Kenneth.1977. The Braziliancorporative
stateand workingclasspolitics.Berkeley:
Universityof CaliforniaPress.
Gamson, WilliamA., and Andre Modigliani.1966. Knowledgeand foreignpolicyopinions:
Some models forconsideration.Public Opinion Quarterly,30:187-99.
Key, V. 0. 1961. Public opinion and Americandemocracy.New York: Knopf.
Kinder,Donald, and D. RoderickKiewiet. 1981. Sociotropicpolitics.BritishJournalof PoliticalScience, 11:129-61.
Lafer, Celso. 1975. 0 sistemapolitico Brasileiro[The Brazilianpoliticalsystem].Sao Paulo:
Editora Perspectiva.
Lamounier,Bolivar, ed. 1980. Voto de desconfianca[No confidencevote]. Sao Paulo: Vozesl
Cebrap.
Lamounier,Bolivar,and FernandoHenriqueCardoso, eds., 1976. Os partidose as eleicoesno
Brasil [Partiesand electionsin Brazil]. Rio de Janeiro:Paz e Terra.
MacKuen, Michael. 1984. Exposure to information,
beliefintegration,
and individualresponsivenessto agenda change. AmericanPoliticalScience Review,78:372-91.
McClosky,Herbert,and Alida Brill. 1983. Dimensionsof tolerance.New York: Basic.
McDonough, Peter. 1982. Repression and representationin Brazil. ComparativePolitics,
15:73-99.
.1984. “Let us make therevolutionbeforethepeople do”: Elite-massrelationsin Brazil.
In Moshe M. Czudnowski,ed., International
yearbookforstudiesofpoliticalleadersand
leadership.DeKalb: Universityof NorthernIllinoisPress.
McGuire, William. 1969. Personalityand susceptibility
to social influence.In Edgar F. Borgattaand WilliamW. Lambert,eds., Handbook ofpersonalitytheoryand research.Chicago: Rand McNally.
Mueller,John. 1973. War,presidents,and public opinion. New York: Wiley.
Rodrigues,Albertino.1968. Sindicatoe desenvolvimento
no Brasil [Unions and development
in Brazil]. Sao Paulo: Difusao Europeia do Livro.
Rodrigues,Le6ncio Martins.1970. Industrializacaoe atitudesoperarias[Industrialization
and
workerattitudes].Sao Paulo: Brasilense.
. 1974. Trabalhadores,sindicatose industrializacao[Workers,unions,and industrialization]. Sao Paulo: Brasilense.
Roeder, Philip. 1985. Dissent and electoralavoidance in the Soviet Union. Presentedat the
annual meetingof the AmericanPoliticalScience Association,New Orleans.
Ryszka,Francisek,ed. 1987.KulturapolitycznaspoleczenstwaPolskiego(1983-1985) [The politicalcultureof Polish society].Warsaw: Universityof Warsaw.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
POPULAR SUPPORT FOR AUTHORITARIAN
REGIMES
347
vs. symbolic
Sears,David O., RichardLau, Tom Tyler,and HarrisAllen,Jr.1980.Self-interest
politicsin policy attitudesand presidentialvoting.AmericanPolitical Science Review,
74:670-84.
Sigelman,Lee, and Pamela JohnsonConover. 1981. Knowledgeand opinionsabout the Iranian crisis:A reconsiderationof threemodels. Public Opinion Quarterly,45:477-491.
Skidmore,Thomas. 1967. Politicsin Brazil. London: OxfordUniversityPress.
Brazil, 1937-71. In AlfredStepan,ed.,
. 1977.Politicsand policymakingin authoritarian
Authoritarian
Brazil. New Haven: Yale UniversityPress.
Francisco.1973. Originsdo sindicalismopopulistano Brasil: A conjuncturado posWeffort,
guerra[Originsof populistunionismin Brazil: The postwarconjuncture].Estudos CEBRAP, 4:67-104.
in EasternEurope and theSoviet
Welsh,WilliamA. 1981. Surveyresearchand publicattitudes
Union. New York: Pergamon.
Zaller, John.1990. Politicalawarenessand eliteopinionleadership.Social Cognition,special
issue on politicalexpertise,in press.
flowin contestedelections.PoliticalAnalysis,vol. 1, in
information
. 1989. Differential
press.
ofpoliticalattitudes.Journalof Personalityand Social Psychology,
. 1987.The diffusion
53:821-33.
This content downloaded from 130.101.133.154 on Thu, 07 Jan 2016 21:16:42 UTC
All use subject to JSTOR Terms and Conditions
Purchase answer to see full
attachment
Why Choose Us
- 100% non-plagiarized Papers
- 24/7 /365 Service Available
- Affordable Prices
- Any Paper, Urgency, and Subject
- Will complete your papers in 6 hours
- On-time Delivery
- Money-back and Privacy guarantees
- Unlimited Amendments upon request
- Satisfaction guarantee
How it Works
- Click on the “Place Order” tab at the top menu or “Order Now” icon at the bottom and a new page will appear with an order form to be filled.
- Fill in your paper’s requirements in the "PAPER DETAILS" section.
- Fill in your paper’s academic level, deadline, and the required number of pages from the drop-down menus.
- Click “CREATE ACCOUNT & SIGN IN” to enter your registration details and get an account with us for record-keeping and then, click on “PROCEED TO CHECKOUT” at the bottom of the page.
- From there, the payment sections will show, follow the guided payment process and your order will be available for our writing team to work on it.