The Politics of Supreme Court Nominations: A Theory of Institutional Constraints and Choices Author(s): Bryon J. Moraski and Charles R. Shipan Reviewed work(s): Source: American Journal of Political Science, Vol. 43, No. 4 (Oct., 1999), pp. 1069-1095 Published by: Midwest Political Science Association Stable URL: http://www.jstor.org/stable/2991818 . Accessed: 14/05/2012 13:04 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 [email protected].

Midwest Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Political Science.

http://www.jstor.org

ThePoliticsofSupremeCourtNominations: A TheoryofInstitutional Constraints and Choices BryonJ.Moraski, UniversityofIowa Charles R. Shipan, UniversityofIowa can attempt touse hispower Whena vacancyoccurson theSupremeCourt,thepresident in ordertobringtheCourtin linewithhisownpolicypreferofnomination strategically to do so. First,he facestwoconstraints whenattempting ences.However,thepresident Courtmedian. justicesandtheexisting bythepresenceofcontinuing maybe constrained bytheSenate,whichmustapprovehis nominee.In this Second,he maybe constrained underwhicha president thatexaminestheconditions paperwe developandtesta theory can,anddo, in his choiceof a nominee.Ourresultsshowthatpresidents is constrained behavestrategically withrespectto SupremeCourtnominations. 1. INTRODUCTION

In recentdecades,vacancieson theSupremeCourthaveappearedapeverytwoyears.BecausetheCourtplayssucha centralrole proximately in politicsandpolicymaking, presidents place a greatdeal ofimportance on fillingthesevacancieswithnomineeswhowillproducea Courtthat whilewe on theiragendas.Surprisingly, however, looksmorefavorably havelearneda greatdeal inrecentyearsaboutSenatevotingon Supreme Courtnominations, analysishasbeendoneon presiverylittlesystematic of thejudiciaryin dents'choicesof nominees.Giventheinvolvement of the and thecentrality area of policymaking nearlyeveryimportant ofthechecksandbalancesin our nomination processto an understanding tounderstand theprocessbywhichpresiitis imperative politicalsystem, dentsselectnominees. Our Courtnominees. Herewe analyzepresidential choicesofSupreme SuthattheSenatealmostalwaysapproves starting pointis theobservation onlyfournominees Sincetheturnofthecentury, premeCourtnominees. havebeenapproved-a havebeenrejectedbytheSenate,whilefifty-five PoliticalScience at the1996 meetingof theSouthern Earlierversionsofthispaperwerepresented Associationandthe1997 meetingoftheAmericanPoliticalScienceAssociation.We wouldliketo thankChrisAchen,MattGabel,Michael Giles,JohnHuber,Eric Lawrence,Ed Schwartz,Jerry andBeckyMortonforhelpful Sorokin,KatieTenpas,andespeciallyKathyBawn,CaryCovington, We also wouldliketo thankTimGroseclose,GregAdams,and George comments anddiscussions. us withideologyscores. Krauseforproviding AmericanJournalofPoliticalScience,Vol.43, No. 4, October1999,Pp. 1069-1095(1999 bythe MidwestPoliticalScienceAssociation

1070

BryonJ.Moraskiand CharlesR. Shipan

successrateofgreater than93 percent.1 Giventhistremendous rateofsuccess,we believetheimportant toexamineis howthepresident question decideswhatkindofnominee he shouldchoose. Ourprimary is in determining howthepoliticalcontext interest influIn particular, encesnominations. we lookto see whether theexistenceof other political thepreferences ofthoseother institutions, andthe institutions, constitutionally prescribed sequenceoftheprocessaffect thepresident's choice.Wealso explorewhether otherfactors, suchas presidential popularityandyearsremaining in office, affect nominations. Ourgoal is todeterminewhichofthesefactors influence thepresident's choice. 2. THESUPREME COURTANDTHESEPARATION OFPOWERS Theselection ofa Supreme Courtjusticebeginswithan opening onthe Courtduetoeither thedeathorretirement ofa justice.Whena vacancyoca candidate, curs,thepresident nominates andthecandidate is confirmed or a vote of rejected by majority theSenate.Iftheposition ofchiefjusticebecomesvacant, thepresident botha sitting mayeither nominate justiceforthe anda newassociatejustice,orhemaynominate a newchiefjustice position fromoutsidetheCourt.2 eventhough thepresident Surprisingly, playsa majorrolein thisproof cess,mostrecentsystematic analyses SupremeCourtnominations have focusedmainly ontheSenate'sconfirmation vote.3 Wehavelearnedthatthe Senatevoteis morelikelytobe conflictual whenthenominee is lessqualifiedor is ideologically outofstepwiththeSenate(Cameron, Cover,and Segal 1990b).We havealso seenthatthetiming ofthenomination (Segal 1987)andthepreferences ofconstituencies andinterest groups(Overbyet al. 1992;Segal,Cameron, andCover1992;CaldeiraandWright 1998)influencetheSenate'svote.Additional evidencesuggests thattheideologyofthe departing justicealso maymatter (Ruckman 1993). 'The fournomineesrejectedbytheSenatewereJohnJ.Parkerin 1930,ClementHaynsworth, Jr.,in 1969,G. HarroldCarswellin 1970,andRobertH. Borkin 1987.Despitebeingrejected, these nomineesall receiveda gooddeal ofsupport whentheSenatevoted,ranging fromBork's42 percent toParker's49 percent. In additiontotheserejections, twoothernomineeswerenotconfirmed bythe hisnomination Senate:Johnson withdrew ofAbe Fortasforthepositionofchiefjusticebeforevotingtookplace,andtheSenatetookno actionon Johnson's nomination ofHomerThornberry. 2WilliamRehnquist, forexample,was an associatejusticewhennominated andconfirmed for thepositionofchiefjustice.His predecessor, Warren Burger, was noton theCourtwhennominated andconfirmed. 3A wealthof descriptive materialon howpresidents choosefromamongpotentialnominees canbe foundina number ofstudies.Forrecentexamples,see Abraham(1991),Maltese(1995), and Silverstein(1994). For systematic evaluationsof presidentialnominations see Massaro (1978; 1990), Cameron,Cover,and Segal (1990a), Watsonand Stookey(1995), and Nemacheckand Wahlbeck(1998).

SUPREME COURT NOMINATIONS

1071

Thesestudiesoftheconfirmation votetreatthechoiceofa nominee as exogenous.Putsomewhat differently, although manyofthesestudiesasindeciding sumethatmembers oftheSenatebehaverationally howtovote, forthepresident. theydo notmakethesameassumption Instead, thestrategic mannerin whicha president mightchoosea nomineeis ignored. Yet mostnominees receivea largemajority ofvotes,whichsuggests theneedfor a careful analysisofthenominationstage.Lopsidedconfirmation votesare almostcertainly dueinparttothecareful consideration givenbypresidents tothechoiceofnominees, consideration thecongivenwithan eyetoward firmation vote. How mightthepresident behavestrategically? GiventheCourt'skey rolein setting publicpolicy,thepresident willwanta Courtthatshareshis ideology andthuswillnominate someonewhowillbringtheCourtcloserto hispreferences. Atthesametime,however, thepresident is constrained by institutional features ofthenomination process.First,theSenatehas the or denythepresident's powerto confirm choice,whichmayforcehimto takethepreferences oftheSenateintoaccountwhennominating someone fora seaton theCourt.Second,depending on theconfiguration ofpreferencesofcontinuing thepresident justicesandhisownideology, maynotbe Inableto shift theCourtso thatitcompletely shareshisownpreferences. hisidealpoint. distance toward stead,hemaybe abletomoveitonlya short 3. THE NOMINATIONGAME

nominee The requirement thattheSenatemustapprovea president's on thenomination placesa certainstructure game.In thefirststage,a vanominates someonetofill cancyoccurs;in thesecondstage,thepresident thatvacancy;and in thethirdstage,theSenatevoteson thepresident's nominee.4 Because thepresident knowsthattheSenatewill voteon his oftheSennominee, whenhe makeshischoicehewilltakethepreferences ate intoaccount.Giventheconstitutionally prescribed sequenceof this on theassumption thatactorsarerationalandforwardgame,andrelying looking, we canmodelthenomination gameusingtheequilibrium concept ofsubgame perfection.5 4Forreasonsdiscussedlaterin thispaper,we omittheJudiciary Committee fromouranalysis. 5Analysesusingthesameequilibrium conceptto examinethepresident's appointment power includeNokkenand Sala (forthcoming), Hammondand Hill (1993), and Snyderand Weingast (1994). Like SnyderandWeingast,we modeltheappointment processas occurring alonga single Nokkenand Sala forthcoming) dimension. Althoughsomerecentanalyses(mostnotably, indicate thata seconddimension forSupremeCourtconfirmation mightbe empirically relevant voting, preof the vious studieshave demonstrated thata singledimensionprovidesa usefulapproximation morecomplicated multi-dimensional realityof theappointment process(e.g.,Segal,Cameron,and Cover1992;Cameron,Cover,andSegal 1990b).In addition, single-dimensional analysesaremore

1072

BryonJ.Moraskiand CharlesR. Shipan

If theSenaterejectsthenominee, thegameends.In otherwords,we treatthisas a single-period game.Severalarguments providejustification forfocusing ona single-period game.To beginwith,as ouranalysisdemonofinteresting andtestable that strates below,therearea number implications in ordereventually derivefroma single-period model.6Furthermore, to tounmodelthenomination processas a multiperiod game,itis usefulfirst derstand thesingle-period game. Mostimportantly, a single-period gameis a veryplausiblerepresentaofreasons,thepresident will tionofthenomination process.Fora number notwantthegametocontinue. Thepresident's publicapproval andstanding withtheSenatemaysuffer fromsucha rejection. Thus,thereis a lossofpoliticalcapitalassociated withputting forth a nominee whois rejected. Morehas to expendthiscapitalin a losingcause,whereasit over,thepresident be spentmoreprofitably onother might otherwise policyissues.Inaddition, thereare time-related costs.If histermin officeis ending,thepresident notgetanother chancetonominate someone. theSenate might Alternatively, might changeandbecomemoreinhospitable tohispreferred nominees (e.g., thepresident's partyoftenloses Senateseatsat themidterm election).Fiwilllosetimein whichhe couldhavea Courtthatpronally,thepresident moreinlinewithhispreferences. Becauseofthesecosts,he ducesoutcomes hasa strong tonominate incentive someonewhowillbe approved. Wemakethefollowing inthegame.First, all players have assumptions perfect andcomplete information aboutthepreferences ofotheractorsand thesequenceof thegame.Second,playershaveEuclideanpreferences. Third,theactionsofplayersaredriven bytheirpreferences overpolicy,an onbelow.Fourth, we elaborate assumption recognizing thatthesize ofthe forwelloveronehundred Courthasremained at ninemembers years,we ordiestheCourtwillhaveeightmembers. assumethatafter a justiceretires we assumethatiftheSenatedoesnotapprovea nominee, Fifth, theseatremainsopen,andtheCourtfunctions witheightmembers. Finally, inthespatialmodelspresented belowwe also assume,without lossofgenerality, that thepresident's idealpointis totheleftoftheCourtmedian. Thegamebeginswhena seatopensontheCourt.7 Priortothecreation ofthevacancy, themedianofthenine-member Courtis equaltotheposition tractable, theoretically and empirically. We agreewithKrehbiel's(1996) assessment that"multidimensionalinstitutional theoriesaremorelikelyto yieldexamplesthangeneralpropositions, andas such it is difficult to discernwhetheror how thegeneralproperties of multidimensional choice withininter-branch institutional settings differ fromtheunidimensional results"(1996,34). 6Asingle-period approachalso facilitates empiricaltesting bynarrowing thepotential equilibriumpredictions (Morton1999).Allowinga gametobe repeatedoftenproducesmultiple equilibria, whichcouldprevent themodelfromyieldingclearandtestableimplications. 7Wetreattheexistenceof a vacancyon theCourtexogenously. In a morecompletemodel,vacanciescouldbe endogenous, as someopeningsoccurdue to strategic considerations (Squire1988; Hagle 1993).

SUPREME COURT NOMINATIONS

1073

Figure1. SupremeCourtwitha Vacancy | J

| J2

| J3

l

l

l

l

l

l

J4

J

J5

J6

J7

J8

Ideology

ofthefifth justice.Oncethevacancyoccurs,however, thischanges.In Figure1,eachoftheeightremaining justices(J1through J8)is locatedalonga scale accordingto his or herideology.Becausetherearenowonlyeight members on theCourt,no singlejusticeoccupiesthemedianposition.Inovertheinterval stead,thepolicyoutcomewillbe a lottery J5],andwe 1J4, ofthisinterval, J= (J4+ J5)/2, tobe considered therefore allowthemidpoint theCourt'smedianafter a vacancyoccurs. Oncea seatopens,thepresident hasanopportunity tonominate a potentialnewjustice.Weassumethatthepresident, whoseidealpointwe denote byP,is motivated bypolicyconcerns. Thus,hewantstomovethemedianof theCourtas close as possibleto hisownidealpoint.In otherwords,the president wantsto minimize thedistance betweenP andJ*,whereJ*is definedas thelocationofthenewmedianifthepresident's nomineeis approvedbytheSenate.8 Consider first whatthepresident woulddo intheabsenceoftherequirementthattheSenateconfirm his nominee.In sucha case, ofcourse,the stillcannotsimply"choose"a newmedianfortheCourt.Rather, president he is constrained bythepresenceofeightsitting justices.Becauseofthese continuing justices,thenewmedianwillremainin theinterval J5],re1J4, ofthenominee. ifthe oftheideological Morespecifically, gardless position choosesa nomineeN suchthatN < J4,thenthenewmedian,J*, president whichtheSupremeCourtreachesdecisions; 8Wedo notexplicitly modeltheprocessthrough instead,we relyon themedianvoteron theCourtto providea proxyforitsdecision-making outcomes.Somewhatsurprisingly, verylittleattention has beenpaid to theaccuracyof usingthemetheCourt'spolicypositions. We believethattherelevanceofthemedianvoter dianvotertorepresent itsuse. to Courtdecisionmakingdeservesfurther attention, buthereuse threearguments to support we relyon Black's medianvotertheorem, whichholdsthatin a votingbody Firstandmostsimply, whosemembers canbe arrayedalonga linewithsingle-peaked theoutcomewillbe the preferences, medianvoter.Second,although somestudieshavedemonstrated thatthechiefjusticecan actstrategicallyin assigning opinions(MaltzmanandWahlbeck1996),itremainstruethatboththechiefjusticeandtheopinionauthormusttakecaretosatisfy themedianmember oftheCourt,forwithout the ofthatmember, support theycannotsustaina majority. Thus,changesinthemedianlead monotoniclose observers callyto changesin theCourt'soutput.Finally,andmostimpressionistically, recognizetheimportance ofthemedianmember oftheCourt.WhenNationalJournalrecently published a listof themostimportant politicalactorsin Washington, SandraDay O'Connor andAnthony Kennedy-usuallyat or nearthemedianon theCourt-made thelist,butJusticesRehnquistand Scalia didnot.See "TheWashington 100,"NationalJournal, June14, 1997.

1074

J.Moraski andCharles R.Shipan Bryon

willbe equaltoJ4.Similarly, ifN > J5thenJ*= J5;andifN is locatedinthe = then J* interval [J4, J5], N. We maintain thatthepresident's choiceis motivated the bytheeffect willhaveontheCourt'smedian.Thepresident nominee willchoosea nomineewho,ifapproved, willbringtheCourt'smedianclosertohisownideal point.He candothisbychoosing a nominee whoseidealpointis closetohis own(i.e.,N = P). Thus,whatis keyfrom ourperspective is thatthepresident can achieve both of these goals-minimizing both IP - J*1and IP - NI

simultaneously. Consider Figure1,keepinginmindthatforthemoment we arestilligtheroleplayedbytheSenate.Ifthepresident noring is locatedtotheleftof J4,hewillwanttomovethemediantoJ4andcando so bychoosing a nomineesuchthatN = P. Similarly, ifP > J5,anynomineesuchthatN > J5will movethemediantoJ5;andthepresident againwillchooseN = P. Finally, whenthepresident is locatedintheinterval J5],choosinga nomineeat 1J4, hisownidealpointwillcausethenewmediantobe locatedathisidealpoint (i.e., N = P will lead to J*= P).

In otherwords,regardless ofhislocationrelative totheexisting Court thepresident canmovethemediantoward hisidealpointbychoosmedian, ingN - P.9In somecaseshe is constrained bythepresenceofthecontinuingjustices,inthesensethathe canonlymovethemediana portion ofthe distanceto hisownidealpoint.Butin all cases,bychoosingN = P he obtainsthebestpossiblenewmedian. Whathappens whenweintroduce theSenateintothemodel?Obviously, thepresident nowmusttakeintoaccountthepreferences oftheSenatebeforedeciding ona nominee. Hisgoalwillbe tochoosethenominee whowill

producethebestnewmedianand whoalso willbe approvedbytheSenate.10

Whether theSenateconstrains thepresident, however, dependson the ofinstitutional As themodelwe developin the configuration preferences. nextsectiondemonstrates, therearethreedistinct regimes, andwhichvariablesaffect theposition ofthenominee dependson whichregimeexists.In thefirst tworegimes discussed below,theSenateandpresident agreeonthe direction of changeforthemedian(althoughtheymaydisagreeon the amount).In thethirdregime,theydisagreeabouteventhedirection of change. 9ChoosingN = P therefore is a weaklydominantstrategy forthepresident.The president could,ofcourse,chooseothervaluesofN. However,bychoosingN = P he notonlymovesthemediantowardhisidealpoint,he also increasesthelikelihoodthatfuture medianswillbe locatedclose tohisidealpoint. 10WeassumethattheSenatealso is motivated primarily bypolicyconcernsandis mostconcernedwithhowthenewnomineewill,ifconfirmed, affect themedianon theCourt.We base this on thedemonstrably assumption strongrelationship betweensenators'policypreferences andtheir voteson SupremeCourtnominees(see, e.g.,Cameron,Cover,andSegal 1990b;Ruckman1993).

SUPREMECOURTNOMINATIONS

1075

3.1 Regime1: Unconstrained Presidential Power First, consider thecase whereS < P < J,as illustrated inFigure2a. The president knowsthatifhenominates someonewhoshareshisideology(i.e., N = P) andthatpersonis confirmed, thenthenewmedianwillbe equaltoJ4. In addition, he knowsthattheSenateprefers anynewmedianJ*suchthat J*< J.Sincea nominee J*= J4,thepresident locatedatP produces knowsthat theSenate,whichprefers sucha nominee. Ofcourse,the J4toJ,willapprove Senatemight a nominee prefer locatedclosertoS thantheonethepresident is willingtoputforward. Sinceitprefers theSenate'shand J4toJ,however, is forced: byrejecting thepresident's itwoulddecreaseitsownutilnominee, ofhavingthefirst ity.Thepresident, byvirtue theSenate move,canpresent withanoptionthat, whilenotperfect, is oneitprefers tothestatusquo. in thisregimethepresident Moregenerally, is unconstrained bythe Senate;bychoosinga nomineeathisidealpointhe also movesthemedian closerto theSenate'sidealpoint.Putsomewhat differently, anynominee likeswillcreatea newCourtmedianthatfallswithin thatthepresident the Senate'swinset.Thisholdstruewhenever S < P < Jandalso usuallywhen

Figure2. Presidential Nominating Regimes 2a: Unconstrained President (Regime1)

S

P

J

J4

2b: Semi-Constrained President (Regime2)

P

I

~~~~I~~

1

J4

Is

1

S

2c: FullyConstrained President (Regime3)

P

J

S

1

J

1076

BryonJ.Moraskiand CharlesR. Shipan

P < S < J,withtheexceptionoftheconditiondiscussedin thenextsection.11 In effect,when the presidentand the Senate are on the same side of the Courtmedian,thepresidentis almostalways unconstrained by the Senate position,and N will be a functionof P,butit will notbe a functionof S or J.

3.2 Regime2: Semi-Constrained President If theSenate is locatedcloserto themedianof theSupremeCourtthan is thepresident(i.e., IJ- SI < IJ- PI), thepresidentmay nothave as much poweras he does in Regime 1. More specifically, ifP < S < J,S > (J4+ J)/2, and P < 2S - J,thepresidentwill no longerbe able to presentthe Senate witha take-it-or-leave-it offerand be confident thattheSenatewill takeit.12 To see this,considerFigure2b. If thepresidentchooses a nomineesuchthat N = P, thenthenew medianwouldbe J*= J4.The Senate,however,prefers Jto J*and therefore wouldrejectthenominee.In fact,theSenateprefersany to of point theright Is, whichis definedas Is = 2S - Jand is theSenate'sindifference pointwithrespectto theCourtmedian,to J. As opposed to Regime 1, wherethepresidentis constrainedby neither the Senate's ideal pointnorthemedianof the Court,herethepresidentis constrainedby a combinationof thetwo.Whatnomineecan thepresident choose in orderto obtainthebestpossiblenew median?Workingbackward, we can see thatthe Senate will approveany nomineethatyieldsa median thatis closerto S thanis J.The presidentthenknowsthatthebesthe can do is to choose a nomineewho will producea new medianat Is. He can do so by choosinga nomineesuchthatN = Is. Hence,in thisregime,thepresident has to nominatesomeonewhose positionis determined by thecombination of theideal pointsof theSenate and theCourt.He is semi-constrained; that is, he can choose a nomineewho will bringtheCourt'smediancloserto his own ideal point,buthe cannotdo so by choosingN = P, and he cannotmove themedianall theway to J4.

3.3 Regime3: FullyConstrained President Finally,thepresidentand theSenatemightbe locatedon oppositesides of theCourt'smedian,as depictedin Figure2c. In thisregime,thepresident and theSenatedisagreecompletelyabouttheideologicaldirectiontheCourt shouldtake.Indeed,anyattempt by thepresidentto movethemediancloser to his ideal pointwill be rebuffed by theSenate; and theSenate will never I"When S < P < J,P is alwaysin theSenate'swin set,and thepresident can alwayschoose N = P. '21notherwords,iftheSenateis closertoJthantoJ*andthepresident is farther fromS thanS is fromJ,thenthepresident is semi-constrained.

SUPREME COURT NOMINATIONS

1077

see a presidential nomineewho is likelyto movetheCourt'smediantoward S. There is a standoff,and the only outcome thatwould please both the presidentand the Senate would be a nomineelocated at J. Regardlessof whether thedistancebetweenJand S is smallor large,as longas P and S are on oppositesides of Jthepresidentwill befullyconstrainedin his abilityto movetheCourtmediancloserto his ideal point. EquilibriumOutcomes The precedingdiscussiondemonstrates thatwhile P, J,and S each are important determinants ofthelocationofN, theirimportance is a function of thespecificconfiguration ofpreferences. More explicitly, themodeldemonstratesthatthereare threedistincttheoretical regimes:one in whichN is a functionof P, one in whichN is a functionof Is (whichin turnis a function of S and J),and finally, one in whichN is a functionof J. Figure3 depictstheequilibriumoutcomesas a functionof the Senate median.HoldingP and Jconstant,we see thatwhenS < P, theoutcomewill be equal to P, regardlessof thelocationof S. Similarly,whenS > J,theoutcome willbe J,regardlessofthelocationof S. OnlywhenS is locatedto the rightof (J4+ J)/2(i.e., themidpointof theinterval[J4,J]) does thelocation oftheSenatemedianhavea directinfluenceon theoutcome.Similarfigures couldbe drawnin whichP is allowedto varywhileS and Jare heldconstant or whereJis allowedto varywhileP and S are heldconstant.In each figure we would see thatwithinone regimethevariableof interestwouldhave an effecton theoutcome,whilein theothertworegimesit wouldnot.

Figure3. Nominee'sIdeologyas a FunctionoftheSenate'sLocation Nominee's Position J_

I

/1

3 3~~~~~~~~~egime

PI, Regime1 P

Reg.2 LocationofS J4J

1078

BryonJ.Moraskiand CharlesR. Shipan

This theoreticalfinding-thatdifferent variablesare influential in differentregimes-guides our empiricalapproach.In particular,the theory of theempiricalmodel: pointsto thefollowingspecification N = Do+ PI*DI* P + P2* D2* IS + P3*D3* J+ E,

(1)

whereDI D2, and D3 are dummyvariablesindicatingthenatureof theregime.'3As thetheoretical modeland theempiricalspecification in Equation 1 makeclear,thenominee'spositionis affected by eitherthepresident'sideology or the Senate's indifference pointor the Court's median. It is not, however,a functionof all of thesevariablesat thesame time.Whethereach variablemattersdependson thelocationof S relativeto J and P. In effect, inwheredifferent then,thisadoptstheapproachof switchingregressions, 14 conditions. dependentvariablesmatterunderdifferent

4. MEASURING

INFLUENCES ON THE NOMINATION

To examinethechoice of SupremeCourtnominees,we considerall justices nominatedto the SupremeCourtbetween 1949 and 1994 (i.e., from Tom Clark throughStephenBreyer).This sample is relativelysmall,with onlytwenty-eight observations,and it is limitedin time.1 However,since theseweretheonlynomineesforwhomwe couldobtaindataforall thevariables,thisproblemcould notbe avoided.'6 Our dependentvariable,whichfollowsfromour theoreticalmodel,is theideologicalpositionof thenominee.To operationalizethisvariablewe use the ideology scores developedforSupremeCourtnomineesby Segal and Cover(1989). Segal and Coverderivedthesescores,whichare meantto be predictiveof a justice's futurebehavior,froma contentanalysisof preconfirmation editorialsaboutthecivillibertiesand civilrightstendenciesof thenominees.Like ADA scores,thesepredictedscoresrangefromzero to 13Moreformally, D, = 1 if S < P < N or if P < S < N, S < (J4+ J)/2,and P < Is, otherwise 1if P