Вы находитесь на странице: 1из 19

Ideological Representativeness of Presidential Primary Voters Author(s): Barbara Norrander Source: American Journal of Political Science, Vol.

33, No. 3 (Aug., 1989), pp. 570-587 Published by: Midwest Political Science Association Stable URL: http://www.jstor.org/stable/2111063 Accessed: 06/09/2010 12:00
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=mpsa. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. 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 is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Political Science.

http://www.jstor.org

Ideological Representativeness of Presidential PrimaryVoters*


Barbara Norrander, San Jose State University
littleevidence Contrary conventional to wisdomand previousresearch,thisarticlefinds thatpresidential this primary votersare ideologically unrepresentative. drawing conclusion, In groupforprimary twoassertions made. First, article are the arguestheappropriate comparison nonvoters. The votersis generalelectionvoterswho failto vote in primaries, all primary not lattergroupincludeshabitualnonvoters whose differences fromprimary voterswould be atin tributable not to the primary process but to generalpatterns participation the United of as States. Second, thearticlearguesideologymustbe defined morethanextremism. Ideology can be construed a sophisticated as Onlyon these beliefsystem a psychological or identification. votersand latter twodefinitions a fewminor do differences arisebetweenpresidential primary generalelectionvoters.

Presidential primary critics that"becausethemoreideologically fear fervent percent theeligible the primaries, 19 of voters tothepollsduring go . who the nomination system . . fails produce to moderate nominees canwin general election govern and effectively intheWhite once House" (Walker, 1988, 65). Besidesthese p. critics, ideological the nature thepresidential of and primary electorate concerned has political scientists. Crotty Jackson presi(1985),Ladd (1978),Lengle(1981),and Polsby (1983) all proclaim has dentialprimary votersto be highly ideological. This position even by reached status political the of science convention wisdom, signified inin textbooks 1986,p. clusion introductory American government (Wilson, on these conclusions 160).Yet,two flaws plague much theresearch which of an was are based. First, ideological representativeness judgedonlyunder unwise choices weremadeinthe of extremism definitionideology. Second, of selection comparison groups. as is ideology An alternative defining to ideology extremismto define an on as a sophisticated political viewpoint; abstract philosophy howthepoet be litical world should structured 1964). (Campbell al., 1960;Converse, voter of Under definitionideological, this unrepresenpresidential primary than different ofconsequences those set tativeness wouldhavean entirely cause nature primary of voters would The feared thecritics. sophisticated by
the are *Analysespresented based on data from AmericanNationalElectionStudy,1980, and ContentCodingsofLevel ofPoliticalConceptualization, 1956-84, made availablethrough for theInter-University Consortium Politicaland Social Research. The authorwishesto thank on Neither for comments the manuscript. RichardBrodyand BernardGrofman theirhelpful for of theseindividuals theICPSR bear anyresponsibility analysesor interpretations nor presentedhere. American Journalof PoliticalScience, Vol. 33, No. 3, August 1989, Pp. 570-87 ? 1989 by the University Texas Press, P.O. Box 7819, Austin,TX 78713 of

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

571

withconsistent, them seek candidates to issuepositions. well-developed Ideologically sophisticated primary voters would makethepresidential prian for mary process ideal method screening presidential candidates. A second alternative definition describes ideology a psychological as or identification symbolic (Conoverand Feldman, 1981;Levitin Miller, and inthe 1979).As a psychological identification, modeofparty identification, ideology "as a perceptual acts screen filtering or device which requires little in in thewayof sophistication orderto influence choice." (Knight, vote in 1985, 833). Ideology, this p. sense,hasmoreto do with symbols as such socialcontrol, radical and capitalism, left, reformist thanit does with left and issues (Conover Feldman, 1981).Underthis definition, primary voters would attracted candidates be to these and evoking symbols notnecessarily to extremist candidates. Theideological ofpresidential representativeness primary voters needs all definitionsideology: of tobe judgedunder three extremism, sophisticahas identification. definition itsownunique Each tion,and psychological of But we consequences thetype candidate for chosen. before canspeakto we theseconsequences, haveto establish howideological nonrepresentativeness wouldbe uncovered. Previous research ideological on representativeness is often confused over what thebestgroup is with which compare to votpresidential primary and ers.Crotty Jackson non(1985)andRanney (1972)employed primary as Even at that, voters their comparison groups. someconfusion existed aboutpotential of in that usedtwo sources nonrepresentativeness Ranney statewide one the included samples, ofwhich only registered voters, other ofwhich included wholevoting population. the age Lengle(1981)usedall with votmembers thepolitical of party, a sample including registered only ers.The application a variety comparison of of makescross-study groups but useda none generalizations difficult, moreimportant, oftheseauthors that distortions causedbytheprimary comparison group wouldreveal process. are or Not all primary nonvoters political party adherents activecitiin include significant a elections. nonvoters zens,voting other Primary proof nonvoters portion habitual (Geer, 1985). In contrast, presidential priare a election voters mary voters mainly moreactivesubsetof general distortions causedbythepres(Norrander, 1986a).Ifwe wantto uncover of idential and primary system, notthosecausedbygeneral patterns parin the is election ticipation America, correct comparison group general voterswhofailed votein theprimaries.When comparisons made, to such are fewdistinctions between and arise voters nonvoters primary (Geer, 1985; Keeterand Zukin,1983;Kritzer, 1980;Norrander, 1986a). The search theideological for of representativenesspresidential prihas voters indeed is The mary important. representativeness question con-

572

Barbara Norrander

of nominees theevaluation and of for sequences theselection presidential of will eachofthethree types ideThisarticle explore theprimary process. a but discussion thedata. of ology sequentially, first, brief Data data for (NES) provide The 1980NationalElectionStudiessurveys Thesesurveys theonly are turnout. onesintheNES series studying primary and verified variables boththegeneral vote for election presthat contain can idential This information be used to producea relatively primaries. variable with one category containing dichotomous dependent error-free, in and conprimaries a second those who respondents voted thepresidential whovotedin thegeneral election but residents primary of states taining a of Besides failed voteintheprimaries.' to providinggoodmeasure thedemeasures theinthese the of timely surveys provide most pendent variable, of variables. wereaskedmany therelevant quesRespondents dependent in the season,either Aprilfortheminor tions ideology on during primary of For for panelor inJune themajorpanelrespondents. somedefinitions questions wereaskedonlyat thetimeof thegeneral ideology, however, Thistime maypresent for election. problems thesedeflag interpretation initions. as Ideology Extremism in definition ideology of previously, mostcommon rthe As mentioned ofpresidential priwith abouttherepresentativeness conjunction questions voters wouldbe exis Democratic primary mary voters one ofextremism. voters wouldbe extremely conliberal whileRepublican primary tremely that voters choosecandidates will who servative. leadstofears primary This in found general extreme themore for moderate voters aretooideologically be of voters would manifested extremismprimary elections. ideological The scaleor on issues. in their on a single ideological either positions that A variety arguments beengiven bolster have to claims presidenof the are extreme. Sometimes extremvoters more tialprimary ideologically of voters inferred thesuccess "extremist" is from ism presidential of primary and candidates suchas GeorgeMcGovern RonaldReagan(Ladd, 1978). lead that low rates Sometimes turnout in-primaries toconclusions those paraddressedin thisarticle, thus,is betweena nationalsampleofpri'The representativeness stateswho did not vote in the primaries. maryvotersand generalelectionvotersin primary in will of This analysis notindicatewhether primary votersare representative voters nonpri(1) of votersin one state,suchas New Hampshire,are representative primary states;(2) primary of votersare representative maryvotersin anotherstate, such as California;or (3) primary whichhave potentially themedianvoterin thegeneralelection.These comparisons, interesting versusinconclusionsabout the consequences of presidential sequencingand intraprimary analyses. terparty choices, are leftto future

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

573

must oriented ticipating be moreideologically (Walker, 1988).Sometimes delprimary extremismextrapolated evidence convention voter is from that extreme are the egates more ideologically (Ladd, 1978).Sometimes higher status educational and levelsofpresidential voters socioeconomic primary are used to suggest they that also are moreideologically extreme (Ladd, sometimes position issue be1978;Lengle,1981),andfinally, comparisons voters nonvoters usedtodemonstrate ideological and are the tween primary of primary voters nonrepresentativeness (Ranney,1972). Each assertion that voters ideologically extreme either are more is logpresidential primary or flawed. ically empirically that votThe success extremist of candidates notindicate primary does in extreme. Votechoice presidential ersaremore primaries is ideologically effort influenced a variety factors: by of campaign (Aldrich, 1980;Grush, and 1980;Parent, Jillson, Weber,1987); candidate qualities (Norrander, et effects (Bar1986b; Marshall, 1984;Williams al., 1976);andbandwagon are found have very to little effect tels,1985).Issuesand ideology often (Gopoian, 1982; Keeterand Zukin, 1983; Marshall,1984; Norrander, Thesuccess "ideological" of candidates most is due 1986b). likely tofactors nature theelectorate thecandidates. of or other thantheideological This In election, the phenomenon notuniqueto primaries. the1980general is on ideological RonaldReaganwaselected a considerable to extent an antiCarter vote. electorate The secondargument an ideologically for extreme primary flawed. and positions, is logically Convention delegates, their ideological sincetheformer a is havelittle relevance presidential to primary voters, in than latter. example, 1980, the For much more elitegroup 5,325 smaller, in but tookpart thetwomajor-party peodelegates conventions, 32 million in of ple voted theprimaries (Pomper, 1981).Giventhelargenumber prifit wouldmoreclosely themodelof thenonideological mary voters, they masspublic thantheideological delegates. in closescrutiny. The final three also Voters arguments failto survive are different from elections notnecessarily low turnout, low stimulus, or the in elections. for voters high or turnout, high stimulus, See, instance, evelections of idenceon thecomposition off-year congressional (Wolfinger, claims that and presiRosenstone, McIntosh, 1981).In a similar manner, becausethey are dential voters ideologically are primary unrepresentative in fall Primary respects, demographically unrepresentativeflat. voters, most election voters fail vote who to arenotdemographically dissimilar general to inthe and primaries (Norrander, 1986a;Keeter Zukin, 1983).Demographic levels prias educational of differences do exist, that such theslightly higher linked a conceptualization to of be mary voters, would moreappropriately rather thanextremism. final The belief as ideology a sophisticated system

574

Barbara Norrander

prebecausetheevidence falls based on issueextremism short argument rather genthan wasbasedon comparisons all primary to nonvoters sented voters. eralelection be The would on can extremism be tested twogrounds. first Ideological taken. positions scale.Thesecond would onissue be ona general ideological wouldhaveto takemoreextreme voters poUnderbothaspects, primary election in claims of to thangeneral voters thesamestates support sitions unrepresentativeness. ideological
Scale Extremism an Ideological Self-Placement on

in primaries.2

scale with NES the Extremism anideology on self-placement wastested on scale, -to asking respondents place themselves a seven-point question to liberal moderate extreme conservative. folding By extreme through from at scale with moderates scaleinhalf, four-point a extremism is created this liberals conservatives theother. and at Results preone end and extreme are in that voters no primary sented Table 1.A demonstrate presidential in voters do notvote who election extreme their ideology general than more

Extremism Issue Positions in

was answers six to Issue extremism measured taking respondents' by valueoffour, adding each issue folding atthecenter seven-point questions, anof and thesescorestogether, dividing thenumber issuequestions by in scale.3 shown Table As is issueextremism swered. result an average The and election voters exist voters general primary 1.B,nodifferences between for separately. on thisscale,either thetotalsampleor foreach party between scale genissueextremism mayhidedifferences The average if: scores hidedifferences voters (1) average and eralelection voters primary scale on or on individual issues, (2) thevalueoffour a seven-point is not case folding four at in for thetruemoderate position a population, which of of The a of produces biasedindicator extremism. possibility thefirst these each question can separately. by objections be solvedsimply examining
the to and (1985)usedthesame1980NES setofsurveys examine ideo2Crotty Jackson Difideological. weremore that and voters found suchvoters of extremism primary logical voters allnonto primary in and (1) occur Crotty Jackson compared ferences findings because instead election of askedat thetime thegeneral on and voters (2) useda question ideology of ofthoseaskedat thetime theprimaries. season.Theseincluded the askedduring primary werethose 3Issues usedintheanalysis panel), v1048 theminor for panel, for (v1060 themajor versus unemployment inflation reduce services present levelsofgovernment reduce continue or v1081), (v1093, spending defense (v1222 aid with and v1189), government tominorities relations Russia(v1201, v1114), (v1126, all for respondents).

REPRESENTATIVENESS OF PRESIDENTIAL PRIMARY VOTERS

575

TABLE 1 Ideological Extremism A. Extremism foldedideologicalplacementscale on Total Sample Republicans Democrats pVa GEV PV GEV PV GEV Moderate Slight ideology Moderateideology Extremeideology Nof Cases X2Significance 32% 39% 26% 3% 27% 41% 28% 4% 28% 37% 34% 1% 20% 36% 42% 3% 32% 42% 22% 5% 32% 44% 20% 5%

(265) (255) 1.77 .622

(110) (106) 3.57 .312

(134) (133) .19 .978

B. Issue extremism indicatedby averageplacementon sixseven-point as issue questionsfoldedat themiddle Total Sample Republicans Democrats pVa GEV PV GEV PV GEV Mean positionb Standarddev. Nof cases T-test Significance 2.55 2.52 2.58 2.53 2.52 2.54 .66 .64 (305) (307) .45 .656 .62 .66 (122) (123) .54 .593 .69 .63 (157) (167) - .26 .795

aPV = presidential primary voters; GEV = general election voters whodidnotvotein the primaries. is 1 bCoding such that = moderate, = extreme. 4

difference When thisis done, onlyone instanceof a statistically significant and in issue positions betweenprimary generalelectionvotersappeared. In view on the apvoterstook a more conservative 1980 Republicanprimary services thandid Republicanswho onlyvoted levelofgovernment propriate in thegeneralelection.The second objectionabout theaverage extremism issue scale, thatabout the truemoderatepositionfora population,can be the The standardized scale thatresultedgave exactly same issue extremism scale.4For all concerned,and forDemocratsand Reansweras theoriginal
issue scale version theaverage of extremism areas follows: 4Results the from standardized solved by standardizing each issue before folding it at the neutral position.

= t = forall voters, = .35, significance .728; forRepublicans,t = 1.06, significance .289; forDemocrats,t =-.77, significance=.442.

576

Barbara Norrander

votersare no more extremein theirissue or primary publicansseparately, ideologicalpositionsthanare generalelectionvoterswho do notvote in the primaries. primary votersare Some may stillnot be convincedthat presidential have been based on a nasince the findings ideologicallyrepresentative, method. tional sample and primariesare conducted on a state-by-state varywidelyand are closelyassociatedwith Turnoutratesin theseprimaries (Moran suchas thenatureoftherace and legal structures contextual factors, and Fenster, 1982; Norranderand Smith,1985). Perhaps primarieswith multicandidate states,earlyprimaries, lower turnout rates,closed primary have an electoratethatis moreideoor primaries, some otherconfiguration the was extreme. Such a contingency exploredbybreaking national logically date of the prisubsetsbased on 10 characteristics: sample intocontextual in mary,amountof campaigneffort a state as measuredby campaignexof number candidates,closenessofnationalrace, open or closed penditures, directelectionof delegates,presrepresentation, rules,use ofproportional length timepresof primary, withcongressional run identialprimary jointly rate. In none ofthese used bya state,and overallturnout identialprimaries electorateprove to be more ideologicallyexsubsamplesdid the primary tremein termsof placementon the ideological scale. In only two cases and representation combined (Democrats in statesnot usingproportional on primary the Democratsand Republicansin stateswithno congressional levweretheresignificantly different same date as thepresidential primary) and els of issue extremism, in one of these cases (Democrats in statesnot it representation), was generalelectionvoterswho apusingproportional peared to be more ideologicallyextreme.Given the large numberof comcould be expectedto ocrelationships parisonsmade, thesetwo significant do cur by chance.5These findings not suggestthatthe national sample is differences. coveringup state-level fear the smallerelectorateis ideologCriticsof presidential primaries is icallyextreme.This positedas the rootof the perceivedproblemsassociin nomination proated withthe dominanceof primaries the presidential cess. The evidence presented above clearly demonstratesthat primary thangeneralelectionvoters.Yet, extreme votersare notmoreideologically thisevidencedoes notspeak to all aspectsofideologicalrepresentativeness, definitions. and I shall now turnto alternative

coefficient at 5One out of 20 tables would produce a correlation significant the .05 level by chance alone.

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

577

IdeologicalSophistication Ideologycan be defined politicalsophistication: ability discuss as the to and think about politicsin an abstract manner(Campbell et al., 1960; Converse, 1964). Ideology of thissortleads people to discusspoliticsin terms It ofliberalism conservatism. also leads people to organizetheir and political viewsin a consistent manner,such thatthosewho are liberalon one issue are also liberalon otherissues. If primary votersare "core voters"(Campbell, 1966) who are bettereducatedand moreinvolvedin politics,themanner in whichtheywould be ideologicallynonrepresentative would be in ideologicalsophistication. Ideology as a sophisticated approach to politics has been tested by two criteria:spontaneousconceptualization and constraint. Ideological Sophistication Measured by SpontaneousConceptualization teststhenotionthatpeople witha more Spontaneousconceptualization discusspoliticsin termsof ideology. sophisticated viewpoint automatically in discussions about politicsare imitated survey Ordinary settings asking by what theylike or dislikeabout each politicalparty.Based on respondents are in responsesto theseopen-endedquestions,respondents classified one of fourcategories:ideologues, grouporiented,natureof the times,or no issue content.Considerablecontroversy existsover thevalidity, reliability, of of and comparability thetwo methods measuring levelsofconceptualization: Converse's (1964) method,based on thefullinterview protocols,and the Nie, Verba, and Petrocik's(1979) surrogate measure,based on coded responses(see, e.g., Smith,1980). Cassel (1984), however,demonstrates thatboth measures are valid indicatorsof politicalsophistication withas measures. Hagner, Pierce, and Knight muchreliability mostattitudinal as based on theorig(1986) providea measureofideologicalconceptualization inal protocolsforeach yearfrom1956 to 1984. Mergingthisdata withthe willreveal whether 1980 survey votersare more likelyto be ideoprimary logues than are general electionvoters. Table 2 presents rather half intriguing results.Nearly ofthe1980Republicanprimary voters wereideologuesversusone-third Republicangeneral of election voters. These lattervoters were most likely to be classifiedas nature-of-the-times voters.AmongDemocrats,less than20 percent both of wereideologues.The plurality all Democratswereclassified of setsofvoters as grouporiented.Apparently, thereare interand intraparty differences in levelsofconceptualization primary and These nationaldifparticipation. for ferences factors. persist thesubsamplesbased on thecontextual Repubvotersweremorelikelyto be ideologuesin six of the 10 types licanprimary of states.In one case, Democraticvotersappeared to be different, not but

578

Barbara Norrander

of in being more likelyto be ideologues. In stateswitha shorthistory privoterswere morelikelyto fallin the nature-of-themaryuse, Democratic whatis happenagain,does notdistort nationalpattern, timescategory.The ing in various typesof states. measure of about thesetting thelevel ofconceptualization Two factors of the need to be keptin mindwhenconsidering significance thesefindings. First,the measure comes fromquestionsasked at the timeof the general Second, theelectionwas 1980,whentheRepubelection,nottheprimaries. candidatein Ronald Reagan. The Reaconservative ran licanparty a clearly by ideologicalcomments themostactive mayhave stimulated gan candidacy Republicans. (See Nie, Verba, and Petrocik,1979, fora similarargument of about theeffects Goldwaterin 1964.) In otheryearsRepublicanprimary to maybe no morelikely be ideologuesthanRepublicangeneralelecvoters tion voters. of is by the If,however, classification notconfounded thetiming thesurof to mayhave something say about the composition the veys,the findings The Dempartiesand the portionof those partiesvotingin the primaries. ocraticpartyclearlyremainsa group-basedparty.The Republican party with one mayhave twoconstituencies: ideologicaland one dissatisfied Demin ocraticperformance the late 1970s. Such a divisionhas been noted by citizens movMiller(1987) to have existedin 1984,withnature-of-the-times maybe the citizens These nature-of-the-times ingintotheRepublicanparty. Additionvoters(Campbell, 1966) who do notvotein primaries. peripheral votersmaymovein and out ofpartiesas timeschange, ally,theseperipheral leavingthe less activecomponentof the favoredpartyto have a different component. ideological base thanthe more active primary-voting
TABLE 2 Levels ofIdeological Sophistication Total Sample pVa GEV Ideologues Group Natureoftimes No issue content N ofcases
X2

Republicans PV GEV 48% 18% 24% 11% 32% 16% 41% 11%

Democrats GEV PV 16% 46% 21% 17% 18% 49% 21% 12%

30% 34% 23% 14%

23% 33% 32% 13%

Significance

(292) (349) 7.63 .054

(114) (133) 9.81 .020

(160) (171) 1.44 .697

voters; GEV = generalelectionvoterswho did not vote in the aPV = presidential primary primaries.

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

579

Ideological Sophistication Constraint as A second testofideologyas politicalsophistication involvesconstraint. If ideologyconstitutes deductivesystemof logic, a liberal on one issue a shouldbe a liberalon all otherissues. Constraint measuredbythemagwas nitudeof correlation coefficients betweenthe six politicalissues. Table 3.A presents Pearson's r correlations betweenthe six issues and tests significant to indicatewhentwo correlations statistically are different. Only twoinstances votersare moreconstrained theirissuggest primary in sue positionsthan are generalelectionvoters.Republican primary voters have more consistent attitudes betweendesiredlevels of government services and opinionson guaranteedstandardof living(whichalso is reflected in the value forthe total population). Additionally, Democratic primary votersare moreconstrained their in opinionson aid to minorities levels and of defensespending.Yet general electionvotersappear to be more conin strained one case. Republicangeneralelectionvotershave moreconsistentopinionsbetweenlevelsof defensespending and relations withRussia. Finally,in two otherinstances, correlation the coefficients primary for voters are statistically different from coefficients generalelectionvoters the for but in a patternwhichdoes not suggesteithergroupis more constrained. AmongDemocrats,betweenopinionson solving inflation unemployment or withrelations withRussia, and opinionson levels of defensespendingand levels of government votersshow no relationship services,primary while Given thispeculiarpatgeneralelectionvotersshownegativerelationships. ternand theoffsetting cases ofmoreconstraint each group,theevidence by does not supportthe claim thatpresidential votersare more ideoprimary set logicalbypossessinga moreconsistent of beliefs.Again, thecontextual subsamplepresentsthe same picture. The use of correlation coefficients demonstrate to has constraint been criticized because thesize ofcorrelation coefficients dependson theamount of variationin attitudes the group studied(Barton and Parsons, 1977). of For instance,Republican primary votersmay possess a greatdeal of consensuson theissuesofrelations withRussia and defensespending. Suchconto sensuswould leave littlevariation be matchedacrossthetwo issues. Yet thisdoes not appear to be the case. Barton and Parsons devised a method formeasuring constraint based on averagestandarddeviations betweenrespondents'mean issue positionsin a policyarea and theirpositionson specificissues. Resultsof thistest,providedin Table 3.B, again demonstrate votersare not more constrained is primary (constraint indicatedby lower is values). Indeed, the mostdramaticdifference betweenRepublicangeneral electionvoterswho demonstrate constraint foreign on policyissuesand voterswho do not. These same patterns Republicanprimary reappear for the contextual-based subsets.

TABLE 3 Issue Constraints A. Constraint coefficients measuredbyPearson's r correlation Total Democrat Republican PV GEV PV GEV z z pVa GEV z M!Eb M/G M/S E/G E/S G/S D/R M/D M/R E/D E/R G/D S/D S/R
G/R

.14* .22** .35** .35** .20** .39** .09

.27** -1.58 .16** .72 .31** .44 .36** -.10 .27** -.76 .20** 2.46 .28** -2.32

Domestic Policy .26** .23* .23 .10 .08 .16 .23* .16* .54 .47 .34** .28** .24* .20* .28 .50** .24** 2.16 ForeignPolicy -.04 .35** -3.03

.02 .20** .35** .30** .07 .28** .11

.25** -1.82 .15* .44 .29** .56 .39** -.80 .27** -1.60 .11 1.44 .26** -1.30 2.09 .96 .88 2.14 2.39 .91 1.48
.50

AcrossPolicyAreas -.27 .07 .11 .26** .17** 1.11 -.04 .10 -1.09 .11* .08 .29 1.42 .06 -.08 1.47 .20* -.00 -.59 -.04 .05 .06 -.06 1.23 .27 -.17* -.20* -.03 -.19** 1.79
.07

.40** .17* .17* .05 -.06 -.17* .08 -.19* .05 -.23**
- .01

.13* .08 .18** .05

- .01

-.50 1.56

.87

-.07 .10

.08

.01

.52

.10 .07

-1.29 .22

.21** .10 .19* .01

- .07

methodc measuredbyBarton-Parsons B. Constraint Democrat Total Republican PV GEV PV GEV PV GEV Domesticpolicy N ofcases Foreignpolicy N ofcases .633 .654 (195) (221) .511 .459 (285) (279) .575 .624 (80) (90) .535 .392 (117) (118) .690 .677 (97) (118) .511 .516 (143) (144)

* *Statistically at significant the .01 level.

*Statistically atthe significant .05level.

aPV = presidential election voters whodidnotvotein the primary voters; GEV = general
primaries.
bM =

G aid to minorities; = solveemployment versus E inflation; = levelof government D R with of S standard living; = level defense of services; = guaranteed spending; = relations Russia. cBarton-Parson measure involves average the standard deviation between respondents' mean area in issue. values indicate positions a policy andrespondents' positions eachspecific on Small more constraint.

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

581

Lookingforideologicalunrepresentativeness terms ideologicalsoin of phistication ratherthan ideological extremism providessome evidence of nonrepresentativeness, notoverwhelming but evidencethatprimary voters are more ideological. In fact,sometimesgeneralelectionvotersappear to in be more ideologicallysophisticated termsof constraint, mostly but primaryvotersand general electionvoterslook the same. Ideologyas a Psychological Identification The third definition ideologyis as a psychological of identification. Unas der thisconceptualization, people develop identifications liberals,modThese identifications, muchlike party erates,or conservatives. functioning how individuals view the politicalworldand influence identification, affect votechoice,butonlyifvoterscan linkcandidateswiththeideologicallabels. This linking ideological labels to candidatesmay be verydifficult acof to in with complish theprimary setting. Ideologicallabels are intertwined party labels (Conover and Feldman, 1981) makingintraparty ideological distincin tionsdifficult perceive. Many of the candidatesrunning primaries to do not have well-known ideological positionsat the startof the race. Indeed, The learningof canmanyof the candidateseven lack name recognition. didates' ideological positionsduringthe primary season may be difficult, since it is not always rationalfor candidates to stressideology (Aldrich, voters have stronger 1980). Thus, even if primary ideological identities, in theseidentities of maybe irrelevant thattheideologicalpositions thecandidates are unknown. PsychologicalIdentification identification The first step in investigating ideologyas a psychological is to uncoverthe extentof ideological identification. Respondentsto the 1980 NES surveys were giventwo chancesto proclaiman ideologicalidentity.From these two chances, a three-category ideological usage variable who placed was created. The highestcategoryincludesthose respondents the includesthosewho had themselves thefirst on inquiry; second category the to be pushedto adopt an identification through use ofa secondquestion; and the finalcategory includesthosewho stillrefusedto place themselves. votersare notmore Table 4.A providestheresults.Once againprimary themselves when first ideological. They are not anymorelikelyto identify themselves nor are theymorelikelyto do so asked to classify ideologically for suba after second probe. This lack of difference persists thecontextual minorexceptions:Democraticprimary votersin highexsampleswithtwo themselves or penditure delegate electionstatesare more likelyto classify voters.Again, these probablyreprethanare Democraticgeneralelection sent chance occurrences.

582

Barbara Norrander TABLE 4 Identities Ideological

scale on oneself anideological to A. Ideological identity basedonwillingnessclassify TotalSample Republican Democrat PV GEV pVa GEV PV GEV First question Second question Never Nof Cases
x2

78% 16% 6%

74% 16% 10%

84% 12% 5%

78% 15% 7%

75% 17% 8%

72% 16% 12%

Significance

(340) (345) 3.67 .160

(131) (136) 1.70 .428

(179) (186) 2.08 .354

B. Ideological identities basedonfeelings toward ideological groupsb Democrats Republicans PV PV GEV GEV Liberal 42.2 40.8 60.3 58.2 Nof cases (102) (113) (124) (135) T-test .48 .84 Significance .634 .399 Conservative 73.4 68.6 55.5 59.4 Nof cases (105) (118) (125) (131) T-test -1.62 2.16 .107 .032 Significance aPV= presidential primary voters; GEV = general election voters didnot who vote the in
primaries.

bValues given average are thermometer ratings.

as Levitinand Miller (1979) used feeling thermometers theirmeasure of symbolic Conover and Feldman (1981) maintainthatatidentification. titudestowardliberalsand conservatives onlyprecursors psychologto are ical identities. a in Nevertheless, checkfordifferences thermometer ratings might desirable,ifonlyforan exhaustivelook at all possible measures. be Thermometer ratings,however,were asked only in the general election phase of the 1980 surveys and may be confoundedby changesin opinions over time. Table 4.B liststhe check fordifferences ratings liberalsand conin of servatives. The onlydifference forRepublicanprimary is votersto rateconservatives morewarmly thanRepublicangeneralelectionvoters.This may warmerfeelings generallyrepresent towardconservatives among primary voters,or strong Republicans(the mostlikelyprimary voters)mayhave increased theirevaluationsof conservatism duringthe fallcampaignto bal-

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

583

TABLE 5 Ideological Classification Candidates of Republicans pVa GEV Numberofcandidatesplacedb Nof cases T-test Significance 2.58 (125) 1.39 .167 2.33 (126) Democrats PV GEV 2.07 (165) .75 .453 1.98 (163)

aPV = presidential whodidnotvotein the primary voters; GEV = general election voters primaries. of bValues givenare themeannumber candidates scale. placedon a liberal-conservative could place a maximum fourcandidates of and Republicans (Reagan,Bush, Connally, three and Democrats, Anderson); (Carter, Kennedy, Brown).

ance theirwarmfeelings towardthe conservative Ronald Reagan. Again, the contextual subsamplesproducedthe same results withonlyone exception. Democraticprimary votersin low-expenditure statesratedconservativesmore coldlythan did generalelectionvoters. UsingIdeological Labels to ClassifyCandidates The finalaspectto thispsychological identity involvesthe ability use to thisidentity evaluate candidates.For thistesta simplecountwas made to ofthenumber candidateseach respondent of placed on theseven-point ideologyscale. Ifpresidential primary votersare moreideologicalbecause they are more likelyto evaluate candidateson ideology,we would expectthem to be able to classify morecandidates.Again thisdoes notappear to be the case. As shownin Table 5, presidential votersare no more likely primary to classify candidateson ideologicalscales thanare generalelectionvoters. The onlyexception thecontextual in subsamples thatDemocraticprimary is votersin delegate electionstatescould place more candidates. The searchforideologicalrepresentativeness based on a psychological definition ideologyproducesthe same resultsas thatbased on a sophisof tication definition: mixedbag. A little a evidenceexiststhatpresidential primaryvotersare more ideological,withRepublican primary votersfeeling morewarmly towardtheconservative label, butmostly primary votersmirrorgeneralelectionvotersin feelings towardand usage of the ideological labels. Conclusions Conventional wisdomheld thatpresidential voterswere more primary ideological. If by "more ideological" it is meant that primary voters are

584

Barbara Norrander

voters conventional wisdomis wrong.Primary moreideologically extreme, positionson issues nor on a genare neither morelikelyto adopt extremist primary votersare notcomthisconclusion, eral ideologyscale. In drawing portion nonvoters, sincethelatter includesa significant pared to all primary that of habitual nonvoters.Nonrepresentativeness would occur between and voters thesehabitualnonvoters primary wouldbe due to thegeneralnain system. When tureof participation the United States not to the primary system, primary caused by the primary lookingfornonrepresentativeness votersmustbe comparedto generalelectionvoterswho failto vote in the are primary voterssimply not moreideoprimaries. Withthiscomparison, logicallyextreme. of priThe questionof the ideological representativeness presidential question. maryvoters,however,is more complicatedthanthe extremism Two alterof usages, and measurements. Ideologyhas a variety definitions, beliefsystem as a or nativedefinitions describeideologyas a sophisticated identification. the Under thesetwo definitions, answerto the psychological Some eviquestion is less straightforward. ideological representativeness dence suggests votersare moreideological.Some evidencesuggests primary generalelectionvotersare moreideological. But mostof the evidencesugevidencesupporting gestsneither groupis more ideological. The strongest thecontention thatprimary votersare moreideologicalis that1980 Republabel and lican primary votersfeltmore warmlytowardthe conservative were more likelyto be ideologues thanRepublicangeneralelectionvoters These findings, however,mayhave been who did notvote in theprimaries. of affected the year and the timing the surveys. by of electoratearisebeConcernsovertherepresentativeness theprimary the cause the natureof the electoratecould influence selectionofpresidenvoters selectingextremist tial candidates. Fears about extremist primary candidatesunpalatableto themoremoderategeneralelectionvotersare unextreme.The efvotersjust are notmore ideologically supported.Primary of votersare less clear. fectsof the actual ideological differences primary IfRepublicanprimary towardtheconservative votersfeelmorewarmly label, theymay be more likelyto respondto candidateswho professsuch This does notmean thatprimary voterswould respondto a conidentities. thattheywould respondto thesymof servative platform issues,but rather the label. ConoverandFeldman(1981) strongbolssurrounding conservative and label definitions centeron symbols ly argue thatideological identities than statusquo, social control,and "left"groupsrather such as capitalism, that issues. Candidatescould winRepublicanprimaries evokingsymbols by ratherthan espousingactual conservative conservatism surround policies. If primary votersare morelikelyto be ideologues, such as the Repuba licansin 1980,theelectoralconsequencesmayor maynot suggest greater

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

585

candidates.Ifthelevelofconroleforissuesin attracting to support primary ceptualizationmeasure is highlysituational,such that the public merely by media,orpolitical mimics greater ofideologicalterms candidates, the use elites (Nie, Verba, and Petrocik,1979), thenthe ideologicalnatureof primaryvotersis notone connectedto issuesbutone connectedto the greater attention paid to thenationalpoliticaldebate. Iflevelsofconceptualization (Knight,1985),thenprimary represent actualsophisticated beliefstructures votersmay be more attracted candidateswho presentcoherent,rather to thanextreme, ofpolicies.Underanyofthesearguments, primary voters sets candidatesfordifferent reasonsthando generalelectionvoters, mayprefer butthetwogroupsmaybothfavor same candidate.After thenaturethe all, of-the-times ideologue Republicans both overwhelmingly and supported Ronald Reagan in the 1980 generalelection. primary critics thatan This articlebegan withthe fearsof presidential electoratewould choose poor candidates: ideologicallyunrepresentative candidateswho could notwingeneralelections,or iftheydid win,could not has govern.The premiseofthisargument been shownto be false.Presidential primary votersare not more ideologically extremeand, at most,differ If and identification. the current on prislightly ideological sophistication mustbe thecause. mary system does producepoor candidates,otherfactors of critics' Of course,theconclusion theprimary argument mayalso be false. in candidates conventions theirheydaynominated Afterall, nationalparty not win (i.e., WilliamJennings Bryan and Al Smith)and were who could candidate could not followedby electionswhere the winning presidential Grant and WarrenHarding). govern(i.e., Ulysses 17 submitted February1988 Manuscript received July1988 25 Final manuscript
REFERENCES nomination and Strategies choicesinpresidential John 1980. Beforetheconvention: H. Aldrich, of campaigns.Chicago: University Chicago Press. campaigns. in nominating Bartels,LarryM. 1985. Expectationsand preferences presidential AmericanPoliticalScience Review,79:804-15. Public Barton, Allen H., and R. Wayne Parsons. 1977. Measuringbeliefsystemstructure. 41:159-80. Opinion Quarterly, of Campbell,Angus. 1966. Surgeand decline: A study electoralchange. In Angus Campbell, PhilipE. Converse,WarrenE. Miller,and Donald E. Stokes,eds., Electionsand thepoliticalorder.New York: Wiley. Campbell, Angus, PhilipE. Converse,WarrenE. Miller,and Donald E. Stokes. 1960. The Americanvoter.New York: Wiley. indexofideoThe "level ofconceptualization" Cassel, Carol A. 1984. Issues in measurement: AmericanJournalof PoliticalScience,28:418-29. logical sophistication.

586

Barbara Norrander

and StanleyFeldman. 1981. The originsand meaningof liberal/ Conover, Pamela Johnston, AmericanJournalof PoliticalScience, 25:617-45. conservative self-identifications. in Converse, PhilipE. 1964. The natureof beliefsystems mass publics. In David E. Apter, New York: Free Press. ed., Ideologyand discontent. Washand primaries nominations. Crotty, William,and JohnS. JacksonIII. 1985. Presidential Press. ington,DC: CongressionalQuarterly electorates.Presentedat primary of Geer, JohnG. 1985. The representativeness presidential the annual meetingof the AmericanPoliticalScience Association,New Orleans. Amerprimaries. and Gopoian, David. 1982.Issue preferences candidatechoiceinpresidential ican Journalof PoliticalScience, 26:523-46. and regionality, prioroutcomeson Grush,JosephE. 1980. Impactof candidateexpenditures, Journal Personality Social Psychology, and of primaries. the1976Democraticpresidential 38:337-47. codingsof levelofpoHagner, Paul R., JohnC. Pierce, and Kathleen Knight.1986. Content 1956-1984. Ann Arbor, MI: CenterforPoliticalStudies. liticalconceptualization, choice: Thefailureof thenewpresidential Keeter, Scott, and CliffZukin. 1983. Uninformed New York: Praeger. system. nominating does matter. Kathleen. 1985. Ideologyin the 1980election:Ideological sophistication Knight, Journalof Politics,47:828-53. of Kritzer,Herbert M. 1980. The representativeness the 1972 presidentialprimaries.In WilliamCrotty, ed., The partysymbol.San Francisco: Freeman. gone? New York: Norton. Ladd, EverettCarll. 1978. Wherehave all thevoters Westport, CT: Greenwood. primaries. and Lengle, JamesI. 1981. Representation presidential elecof Levitin, Teresa E., and WarrenE. Miller. 1979. Ideologicalinterpretation presidential tions.AmericanPoliticalScience Review,73:751-71. and presidential voters.Social Sciprimary Marshall,Thomas R. 1984. Issues, personalities, ence Quarterly, 65:750-60. The Reagan era. In WarrenE. Miller re-examined: E. identification Miller,Warren 1987.Party DC: CenterforNationalPolthe and JohnR. Petrocik, eds., Where's party?Washington, icy. A in primaries: diachronic Moran, Jack,and Mark Fenster. 1982. Voter turnout presidential 10:453-76. analysis.AmericanPoliticsQuarterly, Americanvoter. Nie, NormanH., SidneyVerba, and JohnR. Petrocik.1979. The changing Press. Enlarged ed. Cambridge:Harvard University votersas a subsetof Presidential primary Barbara. 1986a. Selectiveparticipation: Norrander, 14:35-53. generalelectionvoters.AmericanPoliticsQuarterly, Journal Politics, of primaries. . 1986b. Correlatesof vote choice in the 1980 presidential 48:156-67. and candidatestrategy, turnBarbara,and GreggW. Smith.1985.Type ofcontest, Norrander, 13:28-50. out in presidential AmericanPoliticsQuarterly, primaries. and Parent,T. Wayne,CalvinC. Jillson, Ronald E. Weber. 1987. Votingoutcomesin the 1984 and caucuses. AmericanPoliticalScience Review,81:67-84. Democraticprimaries Press. New York: OxfordUniversity Polsby,Nelson W. 1983. Consequencesofpartyreform. contests.In Gerald M. Pomper,ed., The election Pomper,Gerald M. 1981. The nominating of 1980. Chatham,NJ: ChathamHouse. elections.American in primary Ranney,Austin.1972.Turnoutand representation presidential PoliticalScience Review,66:21-37. False measuresof ideologicalsoSmith,Eric R. A. N. 1980. The levels of conceptualization: AmericanPoliticalScience Review,74:685-96. phistication. 12:64-77. game. WilsonQuarterly, Walker,Jack. 1988. The primary

REPRESENTATIVENESS

OF PRESIDENTIAL

PRIMARY VOTERS

587

Williams, Daniel C., StephenJ. Weber, Gordon A. Haaland, Ronald H. Mueller,and Robert in E. Craig. 1976.Voterdecisionmaking a primary election:An evaluationofthreemodels of choice. AmericanJournalof PoliticalScience, 22:37-49. Wilson,JamesQ. 1986. Americangovernment. Lexington,MA: Heath. Wolfinger, RaymondE., StevenJ. Rosenstone,and RichardA. McIntosh. 1981. Presidential and congressional voterscompared. AmericanPoliticsQuarterly, 9:245-56.

Вам также может понравиться