Sie sind auf Seite 1von 13

Party Identification, Realignment, and Party Voting: Back to the Basics

Author(s): Warren E. Miller


Source: The American Political Science Review, Vol. 85, No. 2 (Jun., 1991), pp. 557-568
Published by: American Political Science Association
Stable URL: http://www.jstor.org/stable/1963175
Accessed: 17-03-2015 19:47 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.

American Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to The American
Political Science Review.

http://www.jstor.org

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

PARTYIDENTIFICATION,
REALIGNMENT,AND PARTY
VOTING: BACK TO THE BASICS
WARREN E. MILLER
Arizona State University
he argumentis presentedfor definingparty identificationby the root question,
"Generallyspeaking,do you usually think of yourselfas a Republican,a Democrat,an
independent,or what?"Withthis definitionalbase, the partisanbalancebetweenDemocratsand Republicansbetween1952 and 1980shows no evidenceof realignmentoutside
the South, belying the implicationsof the Markus-Converseand Fiorinaanalyses that
suggest volatility in responseto short-terminfluences.It also appearsthat the correlation betweenparty identificationand voter choicesfor presidentare very constantover
time in the South as well as outside the South. Party line voting by party identifiers
varies by region and party but did not decreasebetween1952 and 1988.

I argue the utility of distinguishing


between the overlappingconceptsof partisanshipand party identification.I do so
by presentingsome of the consequencesof
limitingthe measurementof party identification to the responses evoked by the
classic root question, "Generallyspeaking, do you usually think of yourself as a
Republican,a Democrat,an independent,
or what?"In presentingthis set of findings
as a researchnote, I will not attempt to
add to the literatureon the differencesbetween strong identifiersand weak identifiers, nor will I reexaminethe many interpretations of the partisan sympathies of
"independentleaners."
Even without undertakingsuch tasks,
my mode of presentingdata on the historical record of party identification, narrowly defined, calls into questionat least
a portion of the current conventional
wisdom about the nature of party identification and about the responsivenessof
party identification to economic and
social events in the lives of individual
voters. I shallquestionthe conclusionthat

a nationalparty realignmentprecededthe
election of 1984. I shall also question the
conclusion that dealignmenthas reduced
the relevance of party identificationfor
the vote choice. Finally,in pursuingthese
conclusions I shall question some revisionist argumentsconcerningthe impact
of short-terminfluenceson party identification.

Party Identification and


Its Operational Measure
My approachto encouragingreconsideration of some of the conventionalwisdom about party identificationhas three
major structural components. First, I
employ a conceptualdefinitionof Democrats and Republicansthat rests entirely
on answersto the root question, "Generally speaking, do you usually think of
yourself as a Republican,a Democrat,an
independent, or what?" People who
answer "independent"or "no preference"
or "some other party"are not treatedas

AMERICAN POLITICAL SCIENCE REVIEW


VOL. 85 NO. 2 JUNE 1991

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

American Political Science Review Vol. 85


Republicansor Democrats, even though
they may subsequently admit to being
closer to one of the two parties. Moreover, I will not distinguishbetweenstrong
or weak partisans. Thus my basic measure is restrictedto separatingparty identifiers (Democratsand Republicans)from
nonidentifiers.
The reasonsfor this strictinterpretation
of the meaningof party identificationare
more firmly groundedin theory than in
data. There is little question that variations in the "strengthof partisanship"
have reflectedvariationsin the short-term
fortunes of the respective parties, and
have led to changesin the sheerintensity
of partisanenthusiasms.It has also been
well documentedthat the sameshort-term
forces have both attracted and repelled
many citizenswho, while not majorparty
identifiers, have on different occasions
seen themselvesto be closer to one of the
two major party alternatives.However,
neither of these considerations speaks
directly to the question of individuals
respondingto the same short-terminfluences by self-consciouslymoving across
the boundaryseparatingidentifiersfrom
nonidentifiers. The question is not
whether"independentleaners"may, from
time to time, be more partisan in their
voting or their issue preferencesthan are
weak identifiers.And the questionis not
whether independentleaners are covert
partisans; they are demonstrably and
overtly partisan. The question is the
stability(andthe meaningfulness)of one's
self-identification as a Democrat, a
Republican,or as somethingelse.
In searchingfor an answerto this question I "returnto the basics"as I reconsider
the originaltreatmentof the concept and
the measurementproceduresreportedin
The American Voter (Campbell, et al.
1960, 121-28). A return to the source
places the originalmeasurementsin context. It is clear that the effort in The
AmericanVoterwas to build on the concept of group (party) identification,but

also to createan indicatorthat would differentiate among degrees of "partisanship"or "partisancoloration."Withhindsight, it now seemsthat the effortto maximize the versatility of an operational
measure blurred the clarity of the basic
concept of identificationwith a political
party. On the one hand, going beyondthe
root question to differentiateadditional
degrees of partisanship muddled the
dimensionalityof the resultingmeasure.It
introducedintransitivityinto a presumed
continuum.Most importanthere, it also
created indicators of partisanship that
were reflectiveof short-terminfluencesof
preferencesfor issues or for candidatesas
well, perhaps, as variations in the relatively enduringsense of partisanpolitical
self that is the explicitheartof the concept
of identification(Brody1978;Keithet al.
1986; Miller1991).
The significance of attending to
"details"of measurementwhen analyzing
party identification has recently been
forcefullyarguedby Converseand Pierce.
They emphasizethat there are "two elementswhich have been absolutelycentral
to the whole notion of party identification: an extendedtime horizon and some
engagementof partisanfeelingswith selfidentity. . . . These two elements ...
imply ... that numerousforms of partisan feelingsmay be experiencedby an individual, and reportedupon to investigators, which do not constitutethe possession of a partyidentificationas such"(emphasis added; Converseand Pierce 1987,
143). In this exercisewe have set aside the
differentiationof independentleaners.We
have done so in part becausein the original interview sequence the independent
leaners clearly deny a "temporallyextended self-identity" as Democrat or
Republican.It is also truethat the followup question, "Do you think of yourselfas
closer to the Republicanor Democratic
party?"does not attemptto elicit a qualified or limited sense of an "enduring
engagementof partisanfeelingswith self-

558

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

Stability of Party Identification


identity";the questionis askedonly in the
presenttense, and it calls only for a cognitive assessmentof currentcircumstance.1
The answers may indicate partisanship
but they do not reflect a sense of party
identification.I agree with Converseand
Pierce when they note that there may be
no "right"way to measure partisanship
but "it is of great importancenot to treat
diversemeasuresof partisanshipas functional equivalentsof one another. Partisanship has multiple facets, and keeping
clear which facet is being measured,is a
basic investigatorresponsibility."2
One result of the common practice of
attending to variations in strength of
party identification (strong or not so
strong), and variations in the partisan
sympathiesof nonidentifiers(theso-called
independentleaners)has been to obscure
the relative stability of the basic sense of
the political self, elementalparty identification. Short-termenthusiasmsfor a Lyndon Johnson, Democraticdismay with a
George McGovern, and Republicandistress with Watergateare clearly reflected
in abruptchanges in the now traditional
seven point measure of partisanship;I
shall documentthe very limitedimpactof
such phenomenaon answersto the basic
identificationquestion, "Generallyspeaking, do you usually think of yourself as a
Republicanor a Democrat?"(Brody1978;
Shively 1979; Claggett1981).3
In a related development,attention to
variations on the strength dimension of
partisanshiphad focusedintereston what
has come to be called dealignment.At
some point in the reconsiderationof party
identification,the role of microlevel dealignmentas a forerunnerto systemicrealignment must be taken up anew.
Becauseof the limitedgoal of this research
note I shallnot present,nor follow the implications of, data connecting dealignment to realignment.It is enough to be
concerned with the historical record of
party alignments, e.g., the empirical
record of the numericalbalance between

those who are self-identifiedRepublicans


and Democrats.Thatrecordwill be based
on the classic definitionof party identification rather than on reflections of the
broaderconstructof partisanship(Miller
1991).

The HistoricalRecord
The second major structural component of my analysis is the decisionto exploit the full 36-year time series array of
National Election Studies presidential
election study data, 1952-88. Examining
the full sweep of the period covering 10
elections provides a historical context
essential to the analysis of party realignment, as well as to the simplestudy of aggregateindicatorsof stabilityand change
throughtime.

SubgroupDifferences
The thirdelementin my strategyof inquiry is to consider, more or less in
tandem, several strandsof evidence that
are usually presented in isolation, one
from the other, in the literature.To this
end I shall "disaggregate"the electorate
and examinesuch constituentsegmentin
the presence of all other segments; the
parts will sum to the whole, but I will be
able to assess the contribution of each
part to the whole.
ElectoralParticipation
My firstdisaggregationseparatesvoters
fromnonvoters.I shallnote that the party
identifications of nonvoters among
various subgroupsin the electoratediffer
from the party identificationsof voters in
the same subgroups.I shall also note that
combining the two often obscures patterns that characterizevoters alone. This
must certainlymeanthat analysesrelating
aggregatenational distributionsof party

559

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

American Political Science Review Vol. 85


identificationin the total electorateto aggregate national election outcomes have
missed the mark insofar as the divergent
distributionsof nonvotershave been permitted to intrudeon the interpretationof
the electoraldivisions among voters.
Race
My second disaggregationcomes from
separatingthe electorateon the dimension
of race. Ideally I would like to examine
each of the contributionsto the nation's
growing ethnic diversity. Becauseof the
limited numbers of Hispanics, Asians,
and other minorities in the samples, I
trace only the black citizenry over the
past four decades. However, separateattention to the partisanshipof black citizens is crucial because of the dramatic
changes in their contributionto the nation'spolitics. Evenso, the numbersin the
national samples are too small to disaggregate blacks into categoriesother than
voters and nonvoters.
Region
The third focus for disaggregationof
nonblacksin my historicalreconstruction
of stabilityand changein party identification adds the theme of regional differences. At least since Converse's discussion, party realignmentin the South has
been an acknowledged topic of importance in contemporarypolitical analysis
(Converse 1963; Wolfinger and Hagen
1985). Earland Merle Blackadded to the
work of Converse, Campbell, Beck,
Petrocik, Wolfinger, and others and
analyzed the continuation of change in
the South duringthe Reaganyears (Black
and Black 1987). Curiously, however,
there was no immediatefollow-up to the
Blacks' work to ask what the analytic
removalof the South did to the remaining
nationalestimatesof party identification.
I shall ask and answerthe questionimmediately by adding a South/non-South

comparisonto our comparisonsof voters


and nonvoters among nonblacks.4
Gender
In my "returnto basics,"as I examine
the recordof party identificationdistributions over the past 10 elections,I shalladd
one more dimension, gender. Genderrelateddifferencesusually drew comment
in the 1950s to explainhow widows were
responsiblefor the slightly pro-Republican cast to the femalevote of the 1950s.In
more recent years, the persistent proDemocratic, presumablyliberal, cast of
women's votes (when comparedto male
votes) has been labeled "The Gender
Gap."As an empiricalmatter,it is real. In
all of the recent elections the female vote
has been more Democraticthan has the
malevote, and this contrastsrathersharply with both the 1950sand 1960s. Our last
question, therefore is: "Is there now a
'gender gap' among nonblack voters
South or non-South-that constitutesan
element of party realignment?"
Still other dimensions of interest to
both practical politics and political
theory, such as religionand age, could be
added to this list. However, it is not
necessary to go beyond the set I have
selectedin orderto make the point that a
reassessmentof the historical record of
the natureand role of party identification
is needed. Among the dimensionsof disaggregationI have specified, the distinction between voters and nonvoters is the
most important(Epstein1985; Wolfinger
and Hagen 1985; Kelley 1988).

The Distribution of
Party Identification
Somewhat arbitrarily,I first draw attention to evidence related to party
realignment and the regional contrast,
South and North, depicted among white
voters in Table1. Betweenthe electionsof
1952 and 1980, outside the South, neither

560

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

Stability of Party Identification


men nor women voters revealedany significantchangein the net balanceof their
partisansentimentsacross the 30-yearinterval. With genderdifferencesexceeding
two percentagepoints during this period
only in 1964 and 1976, this is a remarkable demonstration of stability. For
Northernwhites, the "steadystate"period
of relativelyunchangingparty identification apparentlylastednot 12 years-1952
to 1964-but a full 30 yearsand was finally interruptedonly after the first Reagan
election.5

This extraordinarydisplay of persistence in the net party balanceamong the


Northernwhite voters, who made up between 75% and 80% of all white voters
over the past four decades, provides a
strikingimplicitcommentaryon the literature on the stability of party identification and party alignment.It does not, of
course, necessarily negate analyses of
short-termfluctuationsbetween the quadrennial readings, although both Green
and Palmquist(1990)and Abramsonand
Ostrom (1991) have recently spoken to

Table 1. Partisan Balance of Party Identification Within Selected Groups of Voters


and Nonvoters, a 1952-88
Non-South
Men
Women

ElectionYear
Votersb
1952
1956
1960
1964
1968
1972
1976
1980
1984
1988
% Distribution
1952
1988
Nonvoters
1952
1956
1960
1964
1968
1972
1976
1980
1984
1988
% Distribution
1952
1988

Men

South
Women

Nation
Blacks

All

4
3
1
10
3
1
3
3
-10
-9

3
2
3
6
4
0
-3
5
-3
-7

57
57
58
39
28
21
25
3
5
-3

54
43
43
39
39
23
14
21
19
23

52
36
30
75
87
68
70
80
73
80

15
12
15
32
22
13
12
15
5
4

41
30

38
36

8
11

9
13

4
10

100
100

24
14
10
31
8
12
-4
0
7
-2

22
-4
20
47
18
16
12
24
8
-1

49
53
44
49
21
19
16
28
20
8

50
46
59
35
27
29
48
18
13
26

43
41
30
69
71
55
52
60
52
43

39
24
34
48
32
24
24
24
17
14

19
21

27
29

9
14

24
20

20
17

100
100

aEachentryis the proportionof Democraticidentifiers(strongplusweak)minus(-) theproportionof Republicansidentified(strongplus weak). A negativesign indicatesa Republicanplurality.
bVotersdefinedas "validatedvoters"in 1964, 1976, 1980, 1984, 1988;in all otheryears the definitionis provided by the respondents'self-reportsin the postelectioninterview.
561

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

American Political Science Review Vol. 85


this point. However, the evidenceof pervasive, long-term,aggregatestabilityoutside the Southis so dramaticthat it would
seem to call for a reexaminationof many
conclusionsabout the originsof changein
party identification.It at least calls into
questionanalysesthat have, for example,
used changes in national economic indicators to explainnationalchangesin party identification, when electorally relevant changein party identificationapparently did not takeplace outsidethe South.
It raisesmoredirectquestionsabout the
implicationsof the thesis that party identification changes incrementally as the
consequence of prior voting behavior.
From1952 to 1980, accordingto National
ElectionStudies data, the Northernvote
division among white voters averaged
57% Republican. Despite very large
Republicanpluralities in virtually every
year except 1964, the Republicanshareof
party identifiersdid not increaseas a simple extrapolationof the work of Markus
and Converse (1979) might have suggested.
The same evidence of stability in the
partisan balance of party identification
calls attentionto the absenceof any cumulative impact of a series of "running
tallies" that should have produced a
Republicanincrementbetween 1952 and
1980 (Fiorina1981). The aggregatestabilities may, of course, conceal compensatory changesthat have offset a driftaway
from the Democratsand into the Republicans' camp. No reasons for, or evidence
of, such compensatory, pro-Democratic
changescome immediatelyto mind. Consequently, it seems fair to conclude that
the evidenceof aggregatestabilityamong
nonblackvoters outside the South should
prompt furtherstudy of the dynamicsof
microlevelchangein party identification.6

Realignment in the South


Equallyclearevidenceof the mutability
of partisanloyalties is providedby Table

1 and its description of change among


Southern white voters in general, and
white Southernmalesin particular,across
the sametime span. Apparentlythe beginning of the end of single-partydominance
among Southern white male voters
startedshortly after the Kennedyelection
of 1960. By the time of the first Reagan
presidency,20 years later, a virtual80-20
division favoring the Democratic party
had been replacedby near parity for the
Republicanparty. This would seem to be
evidence of a classic version of the realignment of partisanship. It was a
realignmentof massive proportions, involving a Democratic-to-Republican
switch of at least 3 out of every 10
Southern nonblack male voters. It was
apparently a secular realignmentas defined by V. 0. Key, Jr. (1959) and introduced in the currentdiscussionby Converse (1976).
Furtheranalysis is neededto determine
the relative importanceof contributions
from conversion, mobilization, and
cohort replacement among Southern
white voters over the 36 years includedin
Table 1, but the net effect is unmistakable. While conditions outside the South
did not provoke any net change in the
party alignment of nonblack voters between 1952 and 1980, changeswithin the
South produced a virtual revolution. A
closerexaminationof both sets of circumstance should tell us more about party
identifications. Earl and Merle Black
(1987)have suggestedmajorthemesto be
explored in the analysis of the South.
Outside the South suggestionsof cohort
replacementin the changingcomposition
of the electorate offer a promising first
line of inquiry.

The GenderGap
The possible unitarynatureof regional
factors capable of producing such party
realignmentamong Southernwhite male
voters is initially reflectedin the parallel

562

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

Stability of Party Identification


shift among nonblackSouthernfemale of the electoratehighlightsthe recentconvotersacrossthe sevenelectionsbetween tributionsof black voters to partisanna1952and 1976.A complicating
anomaly tional elections. The apparentimpact of
thenappears.Changesin partyidentifica- the Kennedy-Johnson era was more
tion amongSouthernfemalevotersafter dramaticamong black voters than non1976 do not match the patternof any voters. And, given their mobilization
othersetof nonblackvoters.Thisis strik- beginningin the mid-sixties,black voters
ing becausethe otherthreegroupingsall across the nation were ultimately only
reveala shiftto smallRepublican
plural- slightly fewer in number than were
itiesat theconclusionof theReaganerain Southern,white male voters. As a conse1988,whilewhiteSouthernwomencon- quence, the fact that black voters across
tinuouslyexhibit a set of clearlypro- the nation almost tripledtheir marginof
Democraticpreferencesthroughoutthe supportfor the Democraticpartybetween
1980s.Moreover,theirDemocraticplu- 1960 and 1968 by itself more than offset,
rality in 1988 matchesthe figuresfrom in sheer numbers,the 50% declinein the
1972andresultsfroma clearincrease,not Democratic plurality among Southern
a decrease, in relative Democratic white males.
The countervailingmoves within these
strengthafterthe electionof 1976.Why
two politicallysignificantsegmentsof the
thisshouldbe so is not obvious.
Theimportance
of explaining
themale- electorateunderlinethe hazardsof drawfemale differences among nonblack ing conclusionsbased on national aggreSouthernvoters is accentuatedby the gations. National totals did not suggest
realizationthat those differencesin the any net changein the nationalbalanceof
1980s are primarilyresponsiblefor the party identifications between 1956 and
much discussed"gendergap" for those 1980. This is true, in general,becausethe
years(Baxterand Lansing1983;Franko- precipitouslyreal pro-Republicanshift in
vic 1982).Evenwithout"understanding"the white (male) South was counterbalthesegenderdifferences,
it seemspossible anced by the large growth in Democratic
thatthe appearance
of the gendergap in pluralitiesamong newly mobilized black
theReaganyearswasnot as mucha func- citizens.7
tion of a liberal,pro-Democratic
growth
in thepartisansentimentsof womenas a
functionof the sharplyconservative
proVoters and Nonvoters
Republicanmove among men (Wirls
1986). The Republicans
have not had a
A suggestion that variations in the
newproblemwithwomen;theDemocrats nature and meaning of party identificahave had a continuingproblemamong tion may be associatedwith variationsin
men. And once regionaldata are sorted political awareness and involvement is
out, the specification
of thegenderprob- provided by the comparison of voters
lemis largelyconfinedto the South.
with nonvoters among blacks. Although
changes within each group were similar
between 1960 and 1964, sharp differBlack Voters
ences appear after 1968. Those differences are accentuated by the contrary
Turningawayfromregionandgender movements between 1984 and 1988. It
differenceswithin white voters to con- was black nonvoters, but not black
siderthepartisansympathiesof theblack voters, who contributedto the aggregate
citizenry,the separationof blackvoters evidence suggestiveof party realignment
and blacknonvotersfromthe remainder duringthe Reaganyears.8
563

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

American Political Science Review Vol. 85


More generally,my analyticseparation Republicanparty. In neitherstage of the
of votersfromnonvotersexposesrelative- realignmentafter 1980 were race, region,
ly well-ordered evidence of stability or gender as relevant as age, ideology,
among Northern white voters, stability and political involvement (Miller 1990).
that contrasts with greater volatility The narrowestpoint to be drawnfrommy
among Northern nonvoters. Both in the introductory analysis is, quite simply,
unevennessof patterns of change across that systematic inspection of disaggretime and in the variabilityof genderdif- gated time series data may have been a
ferences,party identificationamong non- good startingpoint for questioningmuch
voters seems less a matterof stable, long- conventionalwisdomaboutpartyrealigntermpredispositionsand morea matterof
ment and the stabilityof party identificaresponsivenessto short-term,election-by- tion, but it does not provide answers to
election fluctuationof circumstancethan many of the questionsthat it raisesor that
is true for voters. A similarcontrast ap- I have posed. The broaderpoints of theopears both among black citizens and retical interest in the analysis are the
amongSouthernwhites, althoughit is less many implicationsfor shapingfuture instrikingon first inspectionbecauseof the quiry into the nature of party identificapervasivepatternsof changewithin these tion and into the conditionsthat facilitate
groups. While pondering the reason for its stability, or provoke change.
these differencesbetweenvoters and nonvoters, it should also be recognizedthat
Party Identificationand the Vote
the differencesmay contain the basis for
reconciling other scholars' conclusions
Just as conventionalwisdom about the
about the apparentgeneralresponsiveness
of party identificationto short-terminflu- stability of Democratic and Republican
ences with the new evidence of interelec- partyidentificationshas encouragedsome
tion stability among Northern white dubious interpretations of changing
distributionsin party identification,quesvoters.
As a final commenton Table 1, it may tions have also been raisedconcerningthe
be noted that the particulardisaggrega- meaningfulnessof party identificationas
tions that are so revealingof differentpat- a determinantof the vote duringthe 1970s
ternsof stabilityand changebetween1952 and 1980s. It seemsreasonableto presume
that two notabledeviationsin the correlaand 1980 do not serveparticularlywell to
cast new light on the morerecentchanges tion of party identificationwith preferin the distributionsof party identifica- ences for presidentialcandidates,first in
tions that reflect a limited national re- 1964 and then again in 1972, have been
alignment during the Reagan era. The the rememberedevidencefor presuminga
1980-84 and 1984-88 changes in party generallydiminishedimportancefor parbalancewere primarilya Northern,white ty following the elections of the 1950s
(Miller, et al. 1976). A systematicexamphenomenonand were not apparentat all
among black voters or Southern white ination of the simple national bivariate
women voters. A detailedanalysis of the relationshipsbetweenparty identification
realignmentof the Reagan years, devel- and vote choice over time, presentedin
oped elsewhere, indicates two quite Table 2, documents both episodes of
distinct stages of change in which declining correlation;it also documents
the atypicalityof the 1964 and 1972 elecReagan'spersonalleadershipand then his
tions.
first
moved
the
partisan ideology
The left hand columnof Table2 reveals
younger, less politicizedvoters and then
the older, more politicized into the only two elections in which the unstan564

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

Stability of Party Identification


Table 2. Correlations of Party
Identification with the Presidential
Vote Choice, 1952-88
ElectionYear

Bivariate

Partiala

1952
1956
1960
1964
1968
1972
1976
1980

.69
.67
.68
.57
.69
.52
.69
.73

.64
.65
.60
.48
.55
.43
.64
.63

1984

.74

.65

1988
Mean

.75
.67

.68
.60

aPartial correlationwith controls on race, education, gender, religion, income, and union
membership.

dardized bivariate correlation falls well


below the 10-election average of .67.
They were the Johnson-Goldwatercontest, in which Republicansvoted against
their conservative senator from the
desert, and the Democrats'disastrousexperiment with the radical liberal McGovern challenge to an immensely
popular Republicanincumbent, Richard
Nixon. After having risen steadily from
the 1972 low, the party identification/
vote choice correlationwas up to postwar
highs by 1984 and 1988. Thereis no indication from any recentelectionthat party
identificationis less relevant to the vote
decision in the 1980s than it was three
decadesearlier.
As a further test of the constancy of
correlationswith the vote, the secondcolumn of Table 2 presents the partial corre-

lations of party identificationsand vote


choice after the simultaneousimposition
of controlson race, gender,religion,education, income, and union membership.
The dual message of the partial correlations is clear:(1) year afteryear very little
of the party-vote correlationcan be consideredthe spuriousconsequenceof their
sharing these common antecedents;and

(2) the passage of time has seen no


diminution in the total effect of party
identificationon vote choice.
The causal role of party identification
in the shapingof the attitudesand perceptions that are the immediateor proximate
causesof a voter'spreferencefor one candidate over the other had been a continuous topic of inquiry among students of
electoral behavior. Evidence that the
causal role of party identificationas an
antecedentto the vote is largely indirect
through its influence on policy preferences and appraisalsof presidentialperformanceand candidatetraitsis provided
by the Shanksand Miller (1990)analyses
of the Reagan elections. Such inquiries
can continue to be motivated, at least in
part, by the knowledgethat over time the
correlational evidence is stronger, not
weaker, for assigning a major explanatory/causalrole for partyidentificationin
the context of presidentialelections(Beck
1977, 1982; Fiorina 1977, 1981; Markus
and Converse1979; Page and Jones1979;
Petrocik 1981; Wattenberg 1984;
Whiteley 1988; Jacoby1988).

Party Voting
The more generalconclusionthat there
has been no across-timedecreasein the
extent to which the national presidential
vote is a party vote is neitherchallenged
nor furtherilluminatedby our disaggregation of voters by gender, region, or
race. Yearin and year out, women have
been no more likely than men to cast a
party vote nor to defect and cross party
lines to vote for a president.After 1960,
the black vote was as unwaveringly
Democraticas were blackvoters'partisan
loyalties. And within regional comparisons, as with the others, party voting in
the 1980s was every bit as common-or
uncommon-as it had been in the 1950s.
Votingin line with one'spartyin 1984and
1988 was as common as it had been in
1952 and 1956.

565

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

American Political Science Review Vol. 85


This conclusion holds for partisanson
time, Republicansincreased their numboth sides of the aisle. Republicansin the
bers among party identifierswithout sufSouth have an almostperfectrecord;they
feringany dilution of party loyalty at the
have reportedvoting for Republicancan- polls. Even in the changing South, the
didates an averageof at least 95% of the
party identification/vote choice relationtime from the days of Eisenhowerand ship has been stable for more than three
Nixon to the era of Reagan and Bush. decadesfor both Democraticand RepubOutside the South the Republicanvote
lican identifiers. Thus it would appear
has been slightly more variable, but
that thereas elsewheretypicallyless-thanNorthern Republicans have reported faithful Democrats have converted to
more than 90% support for their party's become typically faithful Republicans.
presidentialcandidates.The most notice- Because Republican identifiers are subable, if still minor, occasions for defec- stantiallymore faithfulparty voters than
tions came in 1964 and 1976, but the
are Democraticidentifiers,both elements
Republican figures for 1984 and 1988, in the exchangearguea greaterincreasein
both South and non-South, and among the election day strength of Republicans
both men and women, were fully equalto
than implied by the simple distributional
their reported party votes in 1952 and shifts in party identifications.The poten1956.
tially good news for the Republicanparty
The recordof Democraticidentifiersat
rests on the validity of my deductions
the polls is a persistentrecordof substan- based on a return to the basics. Even
tially less party support than given by
though a massive regional realignment,
Republicans, and it is somewhat more followed by a limited national realignvaried than the record of Republicans. ment, has cost the Democratstheir oneNevertheless,the Democraticpatternsof
time advantagein the distributionof parregional party voting do not appear to
ty identification,the implicationsof party
have changed at all over three decades. identification for the presidential vote
Democrats,regardlessof either region or
have not changed materially with the
sex, were just as faithful-or unfaithfulpassageof time and the changeof political
to party in their1984and 1988votes as in
circumstance.And in 1984 and 1988 the
earlierdecades. Thereis no evidenceper- Republicanticket benefited doubly from
taining to either party in eitherregion of
the leveling of the partisanplaying field.
any differencein the level of party voting
amongparty identifiersthat would distinguish the 1980s from the 1950s.
Notes
There are real world consequencesof
I am indebtedto supportextendedto RegentsProthis observationthat extend well beyond
fessors
by ArizonaStateUniversity.Dataprocessing
it
what says to political scientists about was done
by Tao Wu; manuscriptpreparationwas
the more or less enduringnatureand role done by Linda Coddingtonand Julie Verrill. The
of party identification. The data pre- data were createdby the Centerfor PoliticalStudies
sented in Table 1, and my earlierdiscus- of the Institutefor SocialResearchat the University
of Michiganwith grantsfrom the National Science
sion, indicate that Democrats quite apFoundation, and made available by the Interparentlydeclinedin numericalstrengthin
University Consortium for Political and Social
the South between 1960 and 1980, and
Research.I also thank Paul Beck, MorrisFiorina,
acrossthe entirenation duringthe Reagan JohnGeer, RuthJones, KimKahn,Donald Kinder,
PatrickKenney,and RaymondWolfingerfor critical
era, 1980-88. This happenedwithout the
and advice.
emergence of greater loyalty or more comments
1. The emphasison an extendedtime horizon is
faithful voting performancefrom those properly the center of the recent Abramsonand
who remainedas Democrats.At the same Ostrom critique of "Macropartisanship" by
566

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

Stability of Party Identification


MacKuen, Erikson, and Stimson (1989).
2. It should be noted that the authors of The
American Voter have, both early and late, contributed to the conventional wisdom against which this
note is directed. For example, in The Dynamics of
Party Support (Converse 1976) weak identifiers
were often combined with independent leaners in
the measurement of "strength";and in the American
National Election Studies Data Sourcebook (Miller
and Traugott 1989) summary measures of the partisan balance of party preference combine partisan
leaners with party identifiers, as in Table 2.32, pp.
103-104. It might also be noted, however, that the
definition of party identification groups displayed in
tables in The American Voter usually followed the
orthodoxy being promoted in this note, and the
Data Sourcebook treats party identification as
separate and distinct from party preference, as in
Table 2.34, pp. 105-106.
3. In this regard my conclusion is on all fours with
the analysis of Abramson and Ostrom who demonstrate that the temporal stability of party identifications can be missed-and often has been missedwhen Gallup-like questions that emphasize the current moment ("In politics as of today . . .") are used
as indicators of party identification (Abramson and
Ostrom, 1991).
4. It should be noted that for simplicity of expression I shall, at times, refer to the non-South as the
North, even though I mean Westerners, MidWesterners as well as "Northerners"; and, taking
somewhat greater liberties, I shall refer to nonblacks as whites even though this group includes
Hispanics and other ethnic minorities in very small
numbers.
5. My earlier comment on measurement comes
into play here. If one focuses on "strength of partisanship" and takes variations in the proportions of
"strong"identifiers and the proportions with no partisan preference into account, one sees the aggregate
evidence that prompted discussions of dealignment.
These variations are not reflected in root selfidentification for nonblack voters outside the South.
For an elaboration of this discussion, see Miller
1990.
6. Further work on the responsiveness of party
identification to short-terms forces should also take
into account the analyses by Donald P. Green and
Bradley Palmquist (1990). Their analyses suggest the
importance of taking measurement error into
account when looking for evidence of partisan
instability in nonrecursive models.
7. I am indebted to David Leege for the observation that, given the dispersal of black voters
throughout the nation, their mobilization did not
redress the potential shift in electoral votes resulting
from the regional concentration of Democratic
losses and Republican gains in the Southern states.
8. One consequence of our separation of voters
and nonvoters should be a reinvigoration of interest

in citizen turnout. It is quite possible, for example,


that some portion of the apparent pro-Republican
swing among Southern, nonblack, male voters in the
1980s may have been caused by declining turnout
among Southern, nonblack Democratic males rather
than by pro-Republican conversions within a constant, unchanging population of voters.

References
Abramson, Paul, and Charles W. Ostrom, Jr. 1991.
"Macropartisanship: An Empirical Assessment."
American Political Science Review 85:181-92.
Baxter, Sandra, and Marjorie Lansing. 1983.
Women and Politics. Ann Arbor: University of
Michigan Press.
Beck, Paul A. 1977. "Partisan Development in the
Postwar South." American Political Science
Review 71:477-96.
Beck, Paul A. 1982. "Realignment Begins? The
Republican Surge in Florida." American Politics
Quarterly 10:421-38.
Black, Earl, and Merle Black. 1987. Politics and
Society in the South. Cambridge: Harvard
University Press.
Brody, Richard A. 1978. "Change and Stability in
the Components of Partisan Identification."
Prepared for the NES Conference on Party Identification, Tallahassee, Florida.
Campbell, Angus, Philip E. Converse, Warren E.
Miller, and Donald E. Stokes. 1960. The
American Voter. New York: John Wiley & Sons.
Campbell, Bruce A. 1977a. "Change in the Southern
Electorate." American Journal of Political
Science 21:37-64.
Campbell, Bruce A. 1977b. "Patterns of Change in
the Partisan Loyalties of Native Southerners:
1952-1972." Journal of Politics 39:730-61.
Clagget, William. 1981. "Partisan Acquisition
Versus Partisan Intensity: Life Cycle, Generation
and Period Effects, 1952-1976." American Journal of Political Science 25:193-214.
Converse, Philip E. 1963. "On the Possibility of
Major Political Realignment in the South." In
Change in the Contemporary South, ed. Allan P.
Sindler. Durham: Duke University Press.
Converse, Philip E. 1976. The Dynamics of Party
Support: Cohort Analyzing Party Identification.
Beverly Hills: Sage.
Converse, Philip E. and Roy Pierce. 1987. "Measuring Partisanship." Political Methodology
11:143-66.
Epstein, Laurily K. 1985. "The Changing Structure
of Party Identification." PS 13:48-52.
Fiorina, Morris P. 1977. "An Outline for a Model of
Party Choice." American Journal of Political
Science 21:601-25.
Fiorina, Morris P. 1981. Retrospective Voting in
American National Elections. New Haven: Yale

567

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

American Political Science Review Vol. 85


University Press.
Frankovic, Kathleen. 1982. "Sex and Politics-New
Alignments, Old Issues." PS 3:439-49.
Green, Donald P. and Bradley Palmquist. 1990. "Of
Artifacts and Partisan Instability." American
Journal of Political Science 34:872-902.
Jacoby, William G. 1988. "The Impact of Party
Identification on Issue Attitudes." American
Journal of Political Science 32:643-61.
Keith, E. Bruce, David B. Magleby, Candice J.
Nelson, Elizabeth Orr, Mark K. Westlye, and
Raymond E. Wolfinger. 1986. "The Partisan Affinities of Independent Leaners." British Journal
of Political Science 16:155-84.
Kelley, Stanley. 1988. "Democracy and the New
Deal Party System. In Democracy and the
Welfare State, ed. Amy Gutmann. Princeton:
Princeton University Press.
Key, V. 0. Jr. 1959. "Secular Realignment and the
Party System." Journal of Politics 17:3-18.
MacKuen, Michael B., Robert S. Erikson, and James
A. Stimson. 1989. "Macropartisanship."
American Political Science Review 83:1125-43.
Markus, Gregory B. and Philip E. Converse. 1979.
"A Dynamic Simultaneous Equation Model of
Electoral Choice." American Political Science
Review 73:1055-70.
Miller, Arthur H., Warren E. Miller, Alden E.
Raine, and Thad D. Brown. 1976. "A Majority
Party in Disarray: Policy Polarization in the
1972 Election." American Political Science
Review 70:753-78.
Miller, Warren E. 1990. "The Electorate's View of
the Parties." In The Parties Respond, ed. Sandy
Maisel. Boulder: Westview Press.
Miller, Warren E. 1991. "Party Identfication." In
Encyclopedia of American Political Parties and
Elections, ed. Sandy Maisel. New York: New

York GarlandPublishing,Inc.
Miller, WarrenE., and Santa M. Traugott. 1989.
American National Elections Studies Data
Sourcebook. Cambridge:Harvard University
Press.
Page, BenjaminI., and Calvin C. Jones. 1979.
"ReciprocalEffectsof Policy Performance,Party
Loyalties, and the Vote." American Political
ScienceReview 73:1071-89.
Petrocik,John R. 1981. Party Coalition. Chicago:
Universityof ChicagoPress.
The South,
Petrocik,John R. 1987. "Realignment:
New Party Coalitionsand the Electionsof 1984
and 1986." Where's The Party? Washington:
Centerfor NationalPolicy.
Shanks, J. Merrill, and Warren E. Miller. 1990.
"PolicyDirectionand PerformanceEvaluation:
ComplementaryExplanationsof the Reagan
Election."British Journal of Political Science
20:143-235.
Shively, E. Phillips. 1979. "The Developmentof
PartyIdentificationAmong Adults:Exploration
of a Functional Model." American Political
ScienceReview 73:1039-54.
Wattenberg,Martin P. 1984. The Decline of the
American Political Parties 1952-1980. Cambridge:HarvardUniversityPress.
Whiteley,Paul F. 1988. "TheCausalRelationships
Between Issues, Candidate Evaluations,Party
Identification,and Vote Choice-the View from
'Rolling Thunder.' " Journal of Politics
50:961-84.
The Gender
Wirls, Daniel. 1986. "Reinterpreting
Gap."Public OpinionQuarterly50:316-30.
Wolfinger,Raymond,and MichaelC. Hagen.1985.
"Republican Prospects: Southern Comfort."
PublicOpinionOct./Nov.: 8-13.

WarrenE. Milleris a RegentsProfessorof PoliticalScience,ArizonaStateUniversity,


Tempe, AZ 85287.

568

This content downloaded from 128.205.172.127 on Tue, 17 Mar 2015 19:47:40 UTC
All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen