INFORMATION TO USERS This reproduction was made from a copy o f a document sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this document, the quality o f the reproduction is heavily dependent upon the quality o f the material submitted. The following explanation o f techniques is provided to help clarify markings or notations which may appear on this reproduction. 1 .T h e sign or “ target” for pages apparently lacking from the document photographed is “ Missing Page(s)” . I f it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure complete continuity. 2. When an image on the film is obliterated with a round black mark, it is an indication o f either blurred copy because o f movement during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image o f the page can be found in the adjacent frame. I f copyrighted materials were deleted, a target note w ill appear listing the pages in the adjacent frame. 3. When a map, drawing or chart, etc., is part o f the material being photographed, a definite method o f “ sectioning” the material has been followed. It is customary to begin film ing at the upper le ft hand comer o f a large sheet and to continue from left to right in equal sections with small overlaps. I f necessary, sectioning is continued again-beginning below the first row and c o n t in u in g on until complete. 4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Customer Services Department. 5. Some pages in any document may have indistinct print. In all cases the best available copy has been film ed. University Micixxilms International 300 N. Zeeb Road Ann Arbor, Ml 48106 8400638 W a ll, D ia n e Eve BASES OF ELECTORAL C O M P E T IT IO N IN MICHIGAN APPELLATE COURT ELECTIONS, 1948-1982 Ph.D. M ichigan State University University Microfilms International 300 N. Zeeb Road, Ann Arbor, Ml 48106 Copyright 1983 by W all, D iane Eve All Rights Reserved 1983 BASES OF ELECTORAL COMPETITION IN MICHIGAN APPELLATE COURT ELECTIONS, 1948-1982 By Diane Eve Wall A DISSERTATION Submitted to Michigan State University in partial fu lfillm e n t of the requirements fo r the degree of DOCTOR OF PHILOSOPHY Department of P o litic a l Science 1983 ABSTRACT BASES OF ELECTORAL COMPETITION IN MICHIGAN APPELLATE COURT ELECTIONS, 1948-1982 By Diane Eve Wall This dissertation analyzes the re la tiv e influence of candidate attributes which may explain who wins and who loses in contested Michigan appellate court elections. the appellate lev e l. The electoral rules fo r the Michigan Supreme Court and Court of Appeals d i f f e r . this study. Two courts currently comprise Two research objectives guide The f i r s t is to determine the re la tiv e influence of bases of candidate competition which may explain the election out­ come fo r each of the Michigan appellate courts. This analysis refines the information already established about bases of candidate competition fo r Michigan Supreme Court elections and provides in fo r­ mation about those bases fo r Michigan Court of Appeals contested elec­ tions. The second objective is to compare the re la tiv e influence of the bases of candidate competition between two time periods and court levels. The two periods analyzed were fo r Supreme Court elections before and a fte r adoption of the 1963 Michigan Constitution. The other comparison is between the rank ordering of the re la tiv e influence Diane Eve Wall of candidate attributes in Supreme Court elections and the ranking fo r Court of Appeals contests. The data are aggregate voting s ta tis tic s and candidate bio­ graphical information fo r a ll Michigan Supreme Court elections from 1948 to 1982 and a ll contested Michigan Court of Appeals general elections from 1970 to 1982. The appellate court models are tested with Probit, an ordinal analog of regression. Four major conclusions from this analysis d i f f e r from e a r lie r Michigan studies. F ir s t , fiv e candidate attributes have a positive relationship to winning an appellate court election, the most impor­ tant of which is support from the candidate's geographical section. Only this a ttrib u te provides an advantage during the total period studied and fo r both courts. Second, incumbency and strength of the candidate's p o litic a l party are positively related to only Supreme Court election outcomes during the e a rlie r period. Third, ethnic appeal of the candidate's name and the candidate's unearned name fa m ilia r ity are positively related to winning contested Court of Appeals elections. Fourth, single estimation of one appellate court model of candidate attributes is inappropriate. C learly, the rules of the game and national trends have had an impact on Michigan's appellate court elections. Dedicated to God and my special friends ii ACKNOWLEDGMENTS I wish to acknowledge the assistance I had in the preparation of th is dissertation. The personnel and policies at Michigan State University from the time of my undergraduate training through gradu­ ate degrees promoted a scholarly and humane environment. The four members of the dissertation committee provided invaluable expertise. As chair of my program and dissertation committees, I am indebted to Harold Spaeth for his s k i l l f u l coordination. His expeditious chapter comments enabled me to complete this dissertation in a timely and e ffe c tiv e manner. Michael A ltfe ld provided methodological consulta­ tion and understanding fo r the deadline constraints. Charles Press supplied insight on Michigan p o litic s and attention to d e t a il. Cheryl Reedy successfully blended strong doses of encouragement with criticism during several long work sessions. Other professionals also f a c ilit a t e d the progress of this dissertation. H arriet Dhanak, Data Archive, emitted inexhaustible time, patience, and wisdom about data management and people. Roger Lane, Michigan Supreme Court Information Center, and the Michigan Law Library s t a ff generously aided my collection of data. I would lik e to thank a substantial number of special friends who encouraged me and a t the sam e time understood. Unfortunately, the contribution of a ll of these friends cannot be acknowledged here, but i t w ill never be forgotten. Without the insight and caring of Susanne Sommers, Linda Covey, and Joanne Hamachek, this dissertation would not have been attempted. Because, with God, Pastor Ken Karlzen knows no lim it s , he enabled my completion of this dissertation. graduate students provided invaluable sustained assistance. Benson has been so much more than a colleague. Two Gaye She gave extensively of herself in s ittin g down with me "to get to business" and at other times in lis ten in g . Margaret Gilkison shared an o ffic e , academic exchanges, and tim e-off from studies. I eternally thank Tony R itt e r , a very special frie n d , whose support and a b i l i t y to help me rejuvenate sustained me. Deep appreciation goes to my parents, Jean and George Bradley, and my s is te rs , Jeanne Planet, Penny Bradley, and Sheree Bradley; words cannot describe th e ir contribution. The acknowledgements would not be complete without mention of Abagail's sacrifices. Abby good naturedly accepted periods when she did not receive the attention she deserved and fostered my physi­ cal and mental health. iv TABLE OF CONTENTS Page LIST OF TABLES.......................................................................................... v ii Chapter I. II. INTRODUCTION AND THEORY ............................................................ 1 Objectives .............................................................................. Theory .......................................................................................... Judges as Policy Makers ................................................ Legitimacy of the Judge's Authority ........................ State Judicial Selection Methods .............................. Sources of electoral support .................................... Implications fo r Michigan ju d ic ia l elections . . Bases of candidate support in Michigan . . . . State Judicial Selection Research .............................. Comparative studies ...................................................... Elective studies ............................................................ Michigan's Judicial System ................................................ Appellate Courts .................................................................. Trial C o u r t s ........................................................................ Selection Issue in Michigan .......................................... Footnotes—Chapter I ............................................................ 1 3 3 4 5 5 7 9 12 12 16 24 25 27 28 33 METHODOLOGY.................................................................................... 38 Chapter Objectives .................................................................. Research Design ........................................................................ Data S e t .................................................................................... Research Hypotheses ............................................................ Hypothesis I ........................................................................ Model and o p e r a t i o n a l i z a t i o n .................................... Technique and c rite rio n ................................................ Hypothesis I I ........................................................................ Conceptualizations contrasted .................................... Operationalization ...................................................... M o d e l.................................................................................... Techniques and c r it e r ia ................................................ Hypothesis I I I .................................................................. Technique and c rite rio n ................................................ Hypothesis I V ........................................................................ Techniques and c r it e r ia ................................................ 38 38 41 44 44 45 46 46 47 52 58 59 62 63 63 65 v Chapter III. IV. Page Hypothesis V ........................................................................ Techniques and c rite rio n .......................................... Summary.................................................................................... Footnotes— Chapter I I ...................................................... 66 67 67 69 .................................... 75 Chapter Objectives ............................................................ Hannah Model ........................................................................ Updated Period .................................................................. Pre- and P o st-P eriod s...................................................... Hypothesized Model ............................................................ Tests of M o dels.................................................................. Relationship of Attributes .......................................... Summary.................................................................................... Footnotes— Chapter I I I ...................................................... 75 75 76 85 87 87 97 104 106 ANALYSIS OF SUPREME COURT MODELS .............................. 109 Chapter Objectives ............................................................ Hypothesized Model ............................................................ Positive and S ignificant Coefficients . . . . Ex Post F o r e c a s t i n g ...................................................... Relationship of Attributes ................................................ Summary.................................................................................... Footnotes— Chapter I V ...................................................... 109 110 Ill 114 117 120 122 CONCLUSIONS AND FUTURE RESEARCH .......................................... 124 Chapter Objectives ............................................................ Conclusions.............................................................................. Candidate Attributes ...................................................... Model P e rfo rm a n c e ............................................................ Model E s t i m a t i o n ............................................................ Future Research .................................................................. Additional D a t a .................................................................. Expanded Scope .................................................................. 124 125 125 127 131 133 134 135 A P P E N D IX ................................................................................................ 138 SOURCES CONSULTED 157 V. ANALYSIS OF COURT OF APPEALS MODELS .............................................................................. vi LIST OF TABLES Table Page 2.1 Data s e t ...................................................................................... 42 3.1 Tabulation of candidateadvantage in Supreme Court elections, 1948-1980................................................................... 77 3.2 Supreme Court advantages of winners, 1971-1980 3.3 Test of two Supreme Court models, 1948-1963 . . . 90 3.4 Test of two Supreme Court models, 1964-1980 . . . 91 3.5 Dummy technique Supreme Court model, 1948-1980 3.6 Rank order of Supreme Court candidate attrib u tes . 3.7 Forecasting Supreme Court e le ctio n , 1982 4.1 Test of two Court of Appeals models, 1970-1980 4.2 Forecasting Court of Appeals election, 1982 4.3 Rank order of appellate court candidate a ttrib u te s , 1970-1980 .................................................................................... 118 Voter drop-off between state partisan o ffic e and Supreme Court ...................................................................... 139 A.2 Variable's sources of information .......................................... 141 A.3 Strength of p o litic a l party in lower state partisan c o n t e s t s .................................................................................... 142 Supreme Court candidate's sectional support, 19481982 145 Court of Appeals candidate sectional support, 19701982 149 Supreme Court candidates, 151 A .l A.4 A.5 A.6 1948-1982 v ii . . . . . . . . 83 . 93 . 101 103 . . . . 112 115 Table Page A .7 Court of Appeals candidates, 1970-1982 A.8 Comparison of Supreme Court Models fo r v iii ........................... two periods 154 . 156 CHAPTER I INTRODUCTION AND THEORY Objectives Two courts currently comprise the appellate court level of Michigan's ju d ic ia l system. highest court of appeal. The Michigan Supreme Court is the sta te 's The seven Justices who serve eight-year terms on th is court are elected on a nonpartisan b a llo t although they are nominated by a partisan convention. The Court of Appeals of Michigan is organized into three d is tr ic ts fo r purposes of selection. Each d i s t r i c t has fiv e judges who are nominated by nonpartisan p e titio n and elected on nonpartisan ballots fo r six-year terms. A Court of Appeals primary is only conducted when the number of candidates exceeds double the number of seats. This dissertation is a study of one aspect of the selection process. My research is an analysis of the re la tiv e influence of the candidate a ttrib u tes which may explain who wins and who loses in a contested Michigan appellate court general or vacancy election. The objectives of my research are twofold. My f i r s t objective is to determine the r e la tiv e influence of the bases of candidate competition which may explain the election outcomes fo r each of the Michigan appellate courts. I hope th is study w ill add to the develop­ ment of a framework fo r the analysis of the other states which have ju d ic ia l elections. L ikely , modifications w ill be needed in this 1 2 model to adapt fo r each s ta te 's p a rticu lar h is to r ic a l, c u ltu r a l, and p o litic a l context. Also I wish to refine the information already known about the bases of candidate competition fo r Michigan Supreme Court elections and to provide information about those bases fo r Michigan Court of Appeals contested elections. At this time, none exists. My second objective is to compare the r e la tiv e influence of the bases of candidate competition between two time periods and court levels. The two time periods analyzed are fo r the Supreme Court elections before and a f t e r the adoption of the Michigan Constitution of 1963. This approach provides an opportunity to study the d i f f e r ­ ences resulting from changes in the formal rules. That is desirable because the Constitution of 1963 and the implementing le g is la tio n created sig n ifican t changes in the election procedures governing the Supreme Court contest. The other comparison is between the rank ordering of the re la tiv e influence of candidate a ttrib u tes which may explain who wins and loses in Supreme Court elections and the ranking fo r the Court of Appeals contests. This investigation w ill add to our understanding of the differences between the levels of state courts. These two research objectives address my general goal of r e f in ­ ing and expanding our knowledge about ju d ic ia l elections in Michigan. In the past, policy makers in Michigan have urged drastic changes in the state 's ju d ic ia l selection process without adequate information. I hope my research is a step toward f u l f i l l i n g the need fo r more 3 empirically based information to be available for policy makers. The following are my objectives fo r the rest of Chapter I . A scholar's theoretical and methodological orientations in te r ­ act in any research endeavor. However, theory informs the researcher and should be preeminent over methodology. Therefore, the next por­ tion of this chapter contains my theory; methodology is reserved fo r Chapter I I . A sketch of Michigan's ju d ic ia l system follows discussion of the theoretical frame fo r my study. Chapter I is concluded with a summary. Theory Judges as Policy Makers The ju d icia ry is one of three branches of government in the American p o litic a l system. I t is conceptualized as a p o litic a l i n s t i ­ tution which interacts with the other in s titu tio n s and interests in our p o litic a l system. This Systems Theory conceptualization is an essential foundation for my research because i t places the courts under conventional p o litic a l norms. The " p o litic a l jurisprudence" approach is well established among American p o litic a l s c ie n tis ts .* Most scholars envision the key component of p o litic s as c o n flic t—c o n flic t over the determination of p the allocation of society's values. The function of American courts as: (1) administrator of the law, (2) resolver of disputes between p a rtie s , and (3) policy maker places the ju d icia ry in the core of the 3 p o litic a l process. Court decisions advantage one party over another. 4 As Justice Cardozo(1921) noted, judging is making choices. is At issue how the choices are made and on what grounds. Courts, as p o litic a l in s titu tio n s , can be analyzed by the decision-making models which were developed fo r other p o litic a l i n s t i ­ tutions. Herbert Jacob and Kenneth Vines were frontrunners'in the p o litic a l process approach to state courts. They maintained that state courts have h is to r ic a lly been actors in the state p o litic a l system.^ Legitimacy of the Judge's Authority The next theoretical issue is the means by which ju d ic ia l power is legitim atized. Does legitimacy derive from the inherent authority of the Law, custom, or popular consent? Jack Ladinsky and Allan S ilver (1967) review the debate over ju d ic ia l legitimacy. They clearly establish that your view of the ju d ic ia l role determines your answer. Declaratory theorists view the judge as an impartial declarer of the known Law and find ju d ic ia l legitimacy in legal and tra d itio n a l sanc­ tions. Legal Realists view judges as policy makers and believe that ju d ic ia l authority flows from popular consent in a democracy. My research is based on the b e lie f that judges are policy makers. In a constitutional democracy judges should obtain th e ir authority to make p o litic a l decisions from the consent of the people. This adheres to conventional democratic theory. All p o litic a l officers in a democracy th e o re tic a lly gain th e ir authority from election, e ith e r d ire c tly or in d ire c tly . Therefore, the selection process fo r obtaining ju d ic ia l o ffic e is an important f ie ld fo r research by p o litic a l scien tists. 5 State Judicial Selection Methods The debate over the role of the ju d ic ia ry is p a r t ia lly respon­ sible fo r the variety of selection methods at the state le v e l. Cur- re n tly , there are fiv e main methods fo r selecting state judges. The alternatives range from gubernatorial appointment of judges to exclu­ sive control by the people through elections. In some appointive systems the governor has exclusive control over selecting the nominee. Some governors must share th e ir power with a commission which provides a l i s t of e lig ib le candidates. th is . Several states have modifications of In some states the appointed judge is required to run on a retention b a llo t. Almost three-fourths of the states provide fo r the election of some or a ll of th e ir judges. lature elects the judges. In a small number of states the le g is ­ Twenty-two states provide fo r the people to elect the justices of the highest appellate court. These states are divided almost equally between those conducting partisan ju d ic ia l elec­ tions and nonpartisan ones. Although p o litic a l parties are formally removed from the electoral process in nonpartisan elections, p a r t i­ san can s t i l l be present.® A few states even combine a partisan nomination procedure with a nonpartisan general election.^ In elec­ tiv e systems an explanation fo r who does and does not become a judge is grounded in the voter's decision-making process. Sources of electoral support. Since the role of elections is to leg itim atize ju d ic ia l policy making, i t is s ig n ifican t to id entify the re la tiv e influence of the bases of candidate support—those 6 a ttrib u tes which may explain who wins and who loses. O Decision-making theory, in the context of how a voter decides to cast his/her b a llo t, is linked to an id e n tific a tio n of what factors can advantage a candi­ date over the other candidates. P o litic a l scientists recognize three basic approaches to such voter decision-making. These approaches are now discussed in order to develop which bases of candidate support would be suggested by each approach. F ir s t , those researchers in the American Voter theoretical orientation believe that p o litic a l party id e n tific a tio n acts as a q perceptual screen. Therefore, these theorists would suggest the strength of the candidate's p o litic a l party in the election is a power­ ful base of support fo r a candidate. A dditionally, Campbell et a l . , found that when attitudes are weak, such as in issueless elections, the voter w ill rely more on personal characteristics of the candi­ date to determine the vote d e c is io n .^ Due to the American Bar Asso­ c ia tio n 's Code of Judicial Conduct discouraging substantive issue debates, most ju d ic ia l elections are issueless. 11 Therefore, the personal characteristics of the candidate or p o litic a l party a f f i l i a ­ tion can be important bases fo r candidate support. Second, there are variations in the theoretical orientation that voters are rational decision makers. 12 However, there is general agreement among these theorists on two points. F ir s t , the voter is s e lf-in te re s te d , attempting to act in a manner which is best fo r her/ himself. Second, information is processed to determine a decision. Many rational choice theorists stress the importance of issues. How­ ever, as has been noted, ju d ic ia l elections have special constraints 7 which result in campaigns lacking issue positions that are debated by the candidates. Thus, the voter has to turn to other kinds of informa­ tion which are accessible. Third, within Herbert Simon's (1969) cybernetic approach to decision making, a decision can be portrayed as a complex problem. In order to cope, the decision maker decomposes the problem into less complex parts which are solved and then reconstructed. In a vote decision, these less complex components could be the cues the voter envisions as relevant. Cues are simple bits of information which represent a greater wealth of detailed information. Therefore, within this theoretical frame the bases of support fo r a candidate might be the candidate's p o litic a l party a f f i l i a t i o n , i f known; indicators of achievement; or personal candidate characteristics. Although some tension exists between these theoretical approaches, adherence to any one does not preclude the p o ssib ility of a role fo r the bases of support suggested by the other theories. The three types of bases of candidate support suggested by the theoretical frameworks above are incorporated into my model of factors which pro­ vide a Michigan appellate court candidate an advantage over the other candidates who possess those attributes to a lesser degree. Implications for Michigan ju d icia l elections. The sources of electoral support I discussed in general are now developed in regard to Michigan's electoral system. Michigan is discussed. F ir s t, the ju d icia l electorate in Then, the specific bases of candidate support in Michigan are id e n tifie d . 8 Judicial voters in Michigan possess candidate preferences which vary in degree of inten sity and saliency. Those preferences are the underlying theoretical dependent variable in voting studies. Michigan's electoral scheme does not provide a preferential b a llo t in which the candidates in a contest can be rank ordered. Therefore, the ju d ic ia l voters must transform th e ir preferences into a dichotomous vote decision which is used in my study. The ju d ic ia l electorate is a special subset of the to tal electorate particip atin g in those elections. have explored the concept of voter subsets. P o litic a l scientists Joseph Schlesinger found that a t the state level d iffe re n t public offices have d iffe re n t elec­ to ra te s .13 The following are my four assumptions about the ju d ic ia l electorate in Michigan. F ir s t , the ju d ic ia l electorate w ill be influenced by the attrib u tes of the ju d ic ia l candidates since most ju d ic ia l elections lack the issue focus which ty p ifie s nonjudicial elections. 14 Second, the ju d ic ia l electorate is more interested in ju d ic ia l contests and aware of candidate a ttrib u tes than the whole set of voters. Richard Watson and Rondal Downing found the existence of a subset--"attentive publics" of the c o u r t.13 These are individuals active in le g a lly related fie ld s and those appearing regularly in court. In her Michigan study, Susan Hannah is in agreement with Watson and Downing that the uninterested voter quits voting before reaching the ju d ic ia l contests at the end of the b a l l o t . 13 Third, Michigan's ju d ic ia l electorate is more aware of the party a f f i l i a t i o n of ju d ic ia l candidates than the whole set of voters. 9 Frequently in Michigan contests at the top of the state b a llo t , e .g ., Governor and Secretary of State, focus on personalities. a ttra c ts additional voters. That Those voters attracted only by personali­ tie s have ceased voting before reaching the bottom of the partisan race contests, e . g ., Michigan State University Board of Trustees. Addi­ tio n a lly , p o litic a l stalwarts would be less deterred from voting by the multi-winner characteristic of most of the ju d ic ia l r a c e s .^ For the period from 1948 to 1980 the vote totals fo r ju d icia l contests do reveal only a s lig h t continued drop-off from the number of voters casting ballots fo r the Michigan State University Board of Trustees— a multi-winner partisan race. These votes appear in Table A .l in the Appendix. Fourth, not a ll of the ju d icia l electorate necessarily has such a pre-determined vote decision that they may not be influenced by the characteristics of the ju d icia l b a llo t. Normally, the only information present on the ju d icia l b a llo t in Michigan is the incum­ bency designation and the appearance of the candidate's printed name which may convey ethnic cues. Bases of candidate support in Michigan. The three general bases of support fo r a ju d ic ia l candidate discussed e a r lie r can be subdivided into more specific attributes in order to address Michigan's h is to r ic a l, c u ltu ra l, and p o litic a l environment. The American Voter theorists emphasize the significance of a candidate's p o litic a l party a ffilia tio n . Although Michigan's ju d ic ia l elections are nonpartisan, election laws provide fo r the nomination of Supreme Court candidates 10 by a p o litic a l party. Court of Appeals candidates have ty p ic a lly held previous public offices which are partisan, e .g ., county prosecutor. At the beginning point fo r my data set--1948, "spirited two party competition had come to Michigan." 18 Therefore, the strength of the appellate court candidate's p o litic a l party can provide a base fo r a ju d ic ia l candidate to win in Michigan. As stated e a r l i e r , the rational choice orientation stresses the importance of issue positions and information which indicates the quality of the candidate. In Michigan the formal and informal rules of ju d ic ia l campaigning mute an effe c tiv e issue dialogue among the candidates. 19 However, in Michigan elections only the ju d ic ia l candi­ dates are provided with an incumbency designation on the b a llo t. Also the people in this state tr a d itio n a lly have respected governmental service and Michigan has a substantial number of state and local posi­ tions. Therefore, the specific bases of candidate support in Michigan linked to this approach are such factors as the candidate's status as an incumbent and the length of public o ffic e service—appointive or ele ctive. Cybernetic theorists stress the role cues play in aiding the voter in deciding which candidates to support. gan may play a role as cues fo r ju d ic ia l voters. Four factors in Michi­ One factor is Michi­ gan h is to ric a lly has had a population containing a large and diverse pool of ethnic backgrounds. may be a cue. Therefore, a candidate's ethnic surname A second factor is Catholicism which had an early i n f l u ­ ence role in the development of the state. A candidate may be 11 associated as a Catholic from some of the organizations lis te d on the campaign lit e r a tu r e or from having a tr a d itio n a lly Catholic name. Thus, recognition of a candidate as a Catholic may be a cue fo r the voter. As is true in other states, a third factor is Michigan's p o litic a l culture which contains a group of family names prominent in Michigan p o litic s . A candidate's name which is the same or similar to an established name may be a possible cue. The fourth factor is sectionalism. Geographical and economic elements in the development of Michigan resulted in the existence of three d istin ctiv e sections. 20 These sections periodically come into c o n flic t in the p o litic a l arena over the allocation of scarce resources. One section is the D etroit Metropolitan Area in southeastern Michigan. I t is heavily industrialized and the largest urban center in the state. The greatest variety of Michgan's ethnic groups are located there. The second section is jo u r n a lis tic a lly referred to as "Outstate." contains the rest of the lower peninsula. It This section is predomi­ nately rural and a g ric u ltu ra lly oriented, although i t has several metropolitan areas. In this section the Hollanders have made a strong impact on the development of western Michigan. The third section is Michigan's Upper Peninsula. This land was added to Michigan in conjunction with the settlement of a dispute over Michigan's southern boundary. Folklore indicates that many people in the lower peninsula were outraged over this addition. The Upper Peninsula has a heritage as a mining and forestry center. It has an agricultural economic base as does Outstate Michigan. However, due to the Upper Peninsula's soil and climate it s agriculture is more 12 sim ilar to Wisconsin's than to the rest of Michigan's. The people of the Upper Peninsula have strong ethnic id e n tific a tio n s , especially those persons with a Scandanavian background. The media regularly refe r to the residence of each ju d ic ia l candidate in th e ir campaign coverage. 21 This reinforces the sectional id e n tific a tio n of ju d ic ia l candidates as a cue in Michigan. Any of the four factors discussed may be valuable to a Michigan voter as a cue to simplify his/her vote decision in ju d ic ia l contests. State Judicial Selection Research My review of the research on state ju d ic ia l selection is divided into two portions. In the f i r s t section I discuss the major studies which compare appointive and elective ju d ic ia l selection systmes. In the second section I focus on the lit e r a tu r e which concen­ trates on elective methods fo r ju d ic ia l selection. Comparative studies. The studies which compare state appoint­ ive and elective systems are an integral part of the continuing debate on the source of ju d ic ia l legitimacy. This lit e r a tu r e focuses on the characteristics of the judges as a basis fo r evaluating which selection process is best. profession. This approach is encouraged by many in the legal They view analysis of the characteristics of judges pro­ duced by a selection system as the appropriate c r ite rio n . Therefore, this orientation results in researchers analyzing data sets which encompass several states and are restricted to only the characteris­ tic s of the winners--the judges. 13 "The Effect of In s titu tio n a l Differences in the Recruitment Process" by Herbert Jacob is an early study which often is cited in other comparative research. Jacob reasserted the need fo r p o litic a l scientists to e x p lic itly study the impact of formal in stitu tio n s on behavior. He did so through assessing the e ffe c t of appointment and election procedures and o ffice level on the selection of judges. This was accomplished by examining the local t ie s , the p o litic a l experience, and the partisan a f f i l i a t i o n of judges in twelve states. Jacob concluded that the selection procedures and level of o ffic e do a ffe c t the ju d icia l selection outcome. 22 Michigan was not one of the states comprising Jacob's data set. However, Bradley Canon's replication of Jacob's research did include Michigan as one of the partisan election states. Canon in d i­ cated that the had been informed by Michigan scholars that Michigan Supreme Court elections actually were partisan. While Jacob drew his conclusions from analyzing state t r i a l judges, Canon's research was on state supreme court selection during the 1960's. The ju d icia l characteristics analyzed in this study were party a f f i l i a t i o n , re lig io n , career pattern, educational background, and localism. Canon concluded that in stitu tio n a l differences do not make as much of an impact on the variations in ju d icia l characteristics as Jacob had reported. For example, Canon agreed with Jacob that localism was emphasized in partisan and le g is la tiv e election systems; but Canon found that localism was re a lly a regional phenomena. 23 Since Canon found that regional variation accounts fo r some of the differences in 14 ju d icia l characteristics, a regional or state case study research design is warranted. Most of the studies of elective methods for ju d icia l selection have that approach and I w ill discuss them in the next section. There is one major comparative study with a data set restricted to an individual state. This is Watson and Downing's comprehensive study on Missouri's appointive and partisan election system. They systematically analyzed the selection process fo r justices to the C ircuit Court, Court of Appeals, and Supreme Court. In this state i t was possible to compare the characteristics of the judges of the Court of Appeals to those of the Supreme Court under both selection methods. The background characteristics they analyzed were: age, legal educa­ tio n , place of b irth , partisan a f f i l i a t i o n , prior ju d icia l experience, and prior p o litic a l experience. For the appellate courts, Watson and Downing concluded that the parochial character of the Missouri bench was not affected by the introduction of the Missouri Plan. However, they did find a d i f f e r ­ ence between the levels of appellate courts. Localism (the ju s tic e 's birth place is within the d i s t r ic t ) was less strong at the Supreme Court level than the Court of Appeals. Also Watson and Downing con­ cluded that the partisan composition of the appellate judges had not become more balanced. However, the major party's representation on the bench had not increased as much as i t would have under an elec­ tiv e system. Their analysis revealed that the judges possessing p o litic a l experience continued as a background characteristic. In teresting ly, with the change to the Missouri Plan the p o litic a l 15 experience of judges declined at the Court of Appeals level while i t increased as a characteristic a t the Supreme Court le v e l. ?4 Stuart Nagel's Comparing Elected and Appointed Judicial Sys­ tems was published soon a f t e r Canon's a r t ic le which had demonstrated the impact of regional differences on ju d ic ia l characteristics. The research design in th is book did not appear to control fo r regional differences. Nagel used a 1960 national sample of bipartisan Congress­ ional elections to compare to Supreme Court partisan elections. I t is unclear whether Michigan is included as a partisan election state. He concluded "that voters in ju d ic ia l elections consider party a f f i l i a ­ tion less than voters in general elections, although they seem more prone to consider ethnic a f f i l i a t i o n s . " 25 In regard to nonpartisan elections, Nagel's research did not contain any empirical evidence on the impact of ethnic a f f i l i a t i o n . However, he did suggest that espe­ c ia lly voters in nonpartisan elections would need to resort to ethnic surnames as a source of information about the ju d ic ia l candidates. 26 The research reviewed in this section was designed fo r com­ paring elected judges to appointed judges. The findings reported are in general agreement that localism, p o litic a l party a f f i l i a t i o n , p o litic a l experience, and ethnic appeal are the predominant character­ is tic s of the elected judges. However, these cross-state studies suffer from an inclusion of a variety of selection methods and the regional differences in p o litic a l culture lessens th e ir comparability. The next section contains studies which concentrate on the character­ is tic s of the judges produced by elective selection methods. these are studies of an individual state. Many of 16 Elective studies. F irs t I w ill review ju d ic ia l election stud­ ies which do not concentrate on Michigan. Then I w ill discuss the research specific to Michigan's ju d ic ia l elections. As with the studies comparing the systems, almost a ll of these researchers r e s t r ic t th e ir data to the characteristics of the winners. Two scholars conducted research on elections in more than one state. In a study on the aspirations of Southern attorney generals, Samuel Krislov found an important ch aracteristic of state Supreme Court judges was th e ir previous state and local public o ffic e expe­ rience. The high v i s i b i l i t y of the county prosecutor was the basis fo r Krislov's stressing the influence of the local level experience. 27 The other researcher also included in his study offices other than ju d icia l ones. James Eisenstein collected information on elections of state attorney generals, prosecuting attorneys, and judges to make some tentative generalizations. For the ju d ic ia l races he noted that the incumbent is rarely defeated. Eisenstein concluded that in partisan elections the p o litic a l party a f f i l i a t i o n of the ju d ic ia l candidate was the most important information. For nonpartisan races, Eisenstein noted that the voter has to rely on other cues such as incumbency, name recognition, and endorsement by groups. 28 Partisan elections fo r state ju d ic ia l offices seem to lack enough in terest fo r scholars (outside of the comparative orientation mentioned above) to conduct research on them. An exception is the exploratory investigation of Texas ju d ic ia l elections by Bancroft Henderson and T. C. S in c la ir. However, as they noted, these partisan 17 elections are conducted in a one-party state. Questionnaires and interviews of the 1962 federal judges in Texas and the state judges— appellate and t r i a l —provided the information Henderson and S in c la ir analyzed. Henderson and S in c la ir described the Texas process as domi­ nated by incumbents in mainly uncontested elections. More than one- h a lf of the appellate judges came to the court via appointment. The other characteristics they found the selection process rewarded were: high economic and social and religious status, legal tra d itio n in the fam ily, long residence by the judge and family in the d i s t r i c t , law degree from University of Texas Law school, p o litic a l party a c t iv i t y , public o ffic e experience, and an average age of 58.8 years. Henderson and S in c la ir id e n tifie d three characteristics as irre le v a n t. Sex was not considered a determining factor because the percentage of women judges was close to the percent of female lawyers. Also race and e th n ic ity were not considered as relevant characteristics because the minority ethnic and racial groups are not represented in the lawyer pool. This pool acts as a f i l t e r to attainment of a ju d ic ia l post. Henderson and S in c la ir suggested those characteristics revealed a process of selective "recruitment" which increases in exaggeration from the lower courts to the highest in Texas. Kennth Vines' 29 study of the open bipartisan elections in Louisiana produced conclusions sim ilar to Henderson and S in c la ir 's . The data set included any Louisiana state judge who served from 1945 to 1960. Vines found th at the longer a candidate's incumbency, the 18 less opposition the incumbent faced in the election which assured victory. The other characteristics favored in Louisiana included being early middle-aged, home born and raised, educated within the s tate , an established local resident of the election d i s t r i c t , and having a p o litic a l rather than private practice career. In his con­ clusion, Vines stated that "lik e legislators and many other elected o f f i c i a l s , state judges in Louisiana are selected according to the principle of localism. . . ." 30 Nonpartisan election of judges has received more attention by ju d ic ia l researchers. In an early study, Malcolm Moos investigated the e ffe c t of partisan endorsements of Supreme Court and D is t r ic t Court candidates in Minnesota's nonpartisan elections. Beginning in 1912 Minnesota removed party designations from the nomination and election process. At the Supreme Court le v e l, Moos found that party endorsement had a negligible e ffe c t on the election outcome. Actually, the best strategy fo r incumbents is to campaign as a team, irrespective of th e ir party a f f i l i a t i o n s . level of courts. Moos id e n tifie d a difference between the He noted that the partisan endorsements of d i s t r i c t court candidates have been more e ffe c tiv e because these races have more active campaigns. Robert Heiberg's study of the social backgrounds of Minnesota Supreme Court Justices confirmed Moos' conclusion. Heiberg found that ju d ic ia l incumbency has been impregnable in Minnesota elections since 1912. He posited that the incumbent's advantage was cemented by two election rule changes. Since 1949, candidates must id e n tify which 19 Supreme Court seat they are running fo r and incumbents are designated on the b a llo t. No incumbent has lo st a re-election bid since that date! Another notable ju d ic ia l characteristic analyzed by Heiberg was the justices' ethnic origins. Of the new justices from 1931 to 1968, 90 percent were either Scandanavian, Irish -C ath o lic , or German. The Irish-C atholic representation on the court has been increasing since 1931. Also previous public o ffic e experience was a prevalent characteristic with the state level experience greater than local. Name recognition also seemed to have an occasional role in Minnesota elections. Heiberg provided an example of an election where the other candidates attempted to reduce the name recognition advantage of one of the candidates who was running fo r his fa th e r's vacant se at.^ Research on a neighboring state also stressed the power of the incumbent in winning nonpartisan elections. In investigating Wisconsin's ju d ic ia l and d i s t r ic t attorney elections, Jacob found that the judges were insulated from the rest of the p o litic a l system. He noted that incumbents ty p ic a lly won regardless of the s h if t in the partisan a f f i l i a t i o n of the electorate. Jacob concluded that this was p a r t ia lly a result of the lack of competition in these elections. 33 Two other Midwestern states have been the focus of analysis by p o litic a l scientists. Michigan and Ohio have election rules fo r selection of the Supreme Court Justices which leads to these states 20 being labeled as "mixed." 34 In mixed states the nomination process is partisan but the election b a llo t is nonpartisan. The research reported in two studies comprise the substantial analysis of ju d ic ia l elections in Michigan. "An Evaluation of Judicial Elections in Michigan, 1948-1968" is the e a r lie r study, by Susan Hannah. She developed a m ulti-factored model of competition in Michigan Supreme Court elections from 1948 to 1970. Her data set included the challen­ gers as well as the winners wich was an improvement over most of the research discussed up to this point. The two scholars whom I dis­ cuss a f te r Hannah also included the challengers in th e ir data sets. However, a lim ita tio n of Hannah's data set is that her pool of candi­ dates was restricted to only the Democratic and Republican candidates, with one exception. The additional information on the excluded candi­ dates may provide a better understanding of Michigan's Supreme Court elections. The candidate characteristics which she hypothesized as l ik e ly personal advantages in a Supreme Court election are: the incum­ bency status of the candidate, the strength of the candidate's nomi­ nating p o litic a l party, the public o ffic e experience of the ca n d id a te state and lo c a l, the candidate's private practice reputation, sec­ tional support fo r the candidate, and the ethnic appeal of the candidate. 35 These characteristics and her methodology are exten­ sively discussed in my Chapter I I . Hannah concluded th at in Supreme Court elections the advan­ tage from being an incumbent was the strongest asset, but i t was rarely s u ffic ie n t to win. The other attributes advantaging a candi­ date were the strength of the candidate's nominating party and the 21 previous p o litic a l o ffice experience--state and local. Also, she noted that those candidates beating incumbents had strong local and ethnic appeal as well. In summary Hannah characterized Michigan's Supreme Court elections as "partisan elections in everything but name." 36 The candidate advantages in Supreme Court elections were not compared to Court of Appeals elections. The Court of Appeals elections were excluded from this analysis because of th e ir recent creation. In the other major study, From Ballot to Bench, Philip DuBois surveyed the research based on ju dicial elections held between the years 1948 to 1974. As an overview of voting in nonpartisan elec­ tions, DuBois suggested the following factors may be used by the voters in deciding th e ir vote choice. Voters may be guided by the candidate's p o litic a l party a f f i l i a t i o n when that information is accessible. DuBois indicated that the major guides fo r the voters are the candidate's incumbent status and fa m ilia r name. He asserted that name fa m ilia r ity is not to ta lly a function of incumbency. factor suggested was the candidate's residence. Another Additionally, cues drawn from the b a llo t may be used, such as the ethnic or religious appeal of the candidate's surname, the candidate's sex, a candidate's eye-catching nickname, or the position of the candidate on the b a llo t. DuBois believed that i t would be an impossible task to estimate the precise effects of those nonparty cues in nonpartisan elections. Therefore, he p a r t ia lly assessed the impact of the nonparty cues through examination of the disruption in the partisan division of the vote when the party cue was removed and replaced by the other factors. 37 22 DuBois' conclusions about which ju d icia l characteristics were in flu e n tia l in Michigan Supreme Court elections were basically the same as Hannah's. However, DuBois did not find strong evidence that Michigan's Supreme Court elections are partisan in every respect but name. He obtained a mean correlation ( r = .40) of the partisan division of the vote fo r Supreme Court Justice and Governor which was lower than fo r partisan states but higher than fo r nonpartisan states. Noting that some Michigan Supreme Court elections evoked partisan responses, DuBois concluded that Supreme Court elections in Michigan were determined largely independent of partisan p o litic s . His analysis revealed that when partisan lines were not f o l ­ lowed in Michigan, long-term incumbency or name f a m ilia r it y was present. All of the mixed states reported in From B allo t to Bench had incumbency and name f a m ili a r it y as important cues fo r the voters. Also, DuBois discovered that sometimes sectionalism, name confusion, and ethnic-religious cues influenced the outcome of the election in Michigan. As fo r the Irish -C ath olic cue, he believed survey data were needed to explain why i t was important. The other Midwestern state which fo r the selection of state Supreme Court Barber's paper at 38 has a mixed electoral process Justices in Ohio. Kathleen the 1982 Midwest P o litic a l Science Association Meeting builds on her e a r lie r studies of ju d ic ia l elections in Ohio. Her data on the Supreme Court and Court of Appeals were expanded to cover the years from 1960 to 1980. She found that the correlation between partisan vote fo r Governor and Supreme Court Justice had been 23 declining over those years. Barber characterized the voter in Ohio as following party cues when they were available and when they were not available the voter relie d on such cues as name f a m ili a r it y , e th n ic ity , or race. Barber suggested that holding public o ffic e could be a source fo r establishing name f a m ilia r it y in Ohio. Also, she maintained that even though incumbency did not guarantee re -e le c tio n , incumbents won more often they they lo s t. 39 How do the research findings fo r these two Midwestern states compare? They have a common denominator in presenting evidence fo r the influence of incumbency, name f a m i l i a r i t y , and eth n icity in advantaging a candidate to win a Supreme Court election. In contrast, in Michigan sectional tie s advantaged candidates but this is not mentioned as relevant in Ohio. Also, neither researcher studying Michigan reported a decline in the significance of partisan voting fo r Supreme Court Justices as Barber reported fo r Ohio. Hannah and DuBois did not seem to have th e ir analyses of ju d ic ia l competition directed toward studying temporal trends. The following are summary observations about the research on ju d ic ia l selection. F ir s t , i t is quite evident that the "factual vacuum" lamented by Jacob so many years ago has not been adequately fille d . 40 Second, most research has focused on the state Supreme Courts--a phenomena paralleled a t the federal le v e l. Third, research on ju d ic ia l elections only recently has progressed beyond the stage of inventorying the predominant characteristics of judges. Last, the f i r s t evaluation of Michigan's ju d ic ia l system by a p o litic a l 24 sc ie n tis t was completed ju s t ten years ago. Substantial changes in Michigan's system occurred near the end of that study which could not adequately be captured due to the long terms of o ffic e fo r the major ju d ic ia l seats and the r e la t iv e ly recent introduction of an in te r ­ mediate appellate court. In the rest of Chapter I , I outline Michi­ gan's ju d ic ia l system and the relevant changes brought by the adoption of a new state constitution. Michigan's Judicial System Michigan's courts are provided fo r in the state constitution. The state has had four constitutions, each refle ctin g the contemporary times and attitu des. The ju d ic ia l system established in the Constitu­ tion of 1908 was substantially altered by the current constitution, adopted in 1 9 6 3 . Thi s occurred at approximately the midpoint of the period I analyzed. The new constitution created an unitary ju dicial system comprised of a Supreme Court, a Court of Appeals, a C ircuit Court, a Probate Court, and other courts of limited ju ris d ic tio n that may be established by two-thirds vote of the state le g is la tu re . Also this constitution established these uniform qualifications fo r a ll judges in Michigan courts: they must be licensed to practice law in Michigan and under seventy years of age a t the time of election or appointment. I w ill sketch the main courts' organization, ju r is d ic ­ tio n , and selection process in order to provide the environmental context relevant to this study. 42 25 Appellate Courts The Michigan Supreme Court is the state's fin a l appellate court. The justices on th is court maintain th e ir o ffic e in th e ir resident c ity rather than in the c a p ita l. lated among the Justices via the mail. Draft opinions are circu­ Previous to the new constitu­ tio n , this court consisted of eight justices who were nominated in party convention. In the odd-numbered year Spring elections, candi­ dates ran at-large fo r staggered eight-year terms on a nonpartisan b a llo t. Incumbents were so designated on the b a llo t. The Governor had the power to appoint justices to f i l l vacancies until the next general election. The Michigan Constitution of 1963 and subsequent implementing election laws made the following changes at the Supreme Court le v e l. The number of justices was reduced to seven. in 1968. This was accomplished Elections fo r the justices are held in the Fall of even- numbered years. Supreme Court incumbents can f i l e an a f f id a v it of candidacy in place of a p o litic a l party nomination. Gubernatorial appointment fo r vacancies was abolished by the new constitution, but i t was reinstated in 1968. 43 The Michigan Court of Appeals was created by the Constitution of 1963 to reduce the burden on the Supreme Court. This was particu­ la r ly necessary because the new constitution provided the rig h t to appeal in criminal and c iv il cases. The Court of Appeals' ju ris d ictio n is provided fo r in the Rules of the Supreme Court. panels of three judges. Cases are heard by They are assigned by the Chief Justice of the 26 Court of Appeals who ensures that every judge s its equally with the other judges. I n i t i a l l y , the Court of Appeals consisted of nine justices who were elected in 1964 fo r varied terms of o ffic e in order to stagger the six-year term of o ffic e . The number of justices on the court may be increased by law with its size currently at eighteen. ships are f i l l e d by election. nominated by p e titio n . New judge­ Judges of the Court of Appeals are A primary election is required i f the number of candidates from the petitions becomes more than double the number of seats. Nonpartisan elections fo r the Court of Appeals are held in the Fall of even-numbered years. Justices are elected from three approximately equally populated d is tr ic ts drawn to follow county lin es. Since this court's creation in 1964, the d is tr ic ts have been reappor­ tioned only once--in 1972. Beginning with the 1972 e le c tio n , two coun­ tie s were shifted from the Second D is t r ic t to the F irs t D is t r ic t . The incumbents are designated on the b a llo t and may f i l e fo r re-election by a f f id a v it . Provision fo r vacancy appointment by the Governor is the same as fo r the Supreme Court. There are several outstanding differences between these two courts relevant to my candidate selection study. The Court of Appeals candidate has an electoral d i s t r ic t which is geographically smaller than the Supreme Court's and with approximately one th ird the number of potential voters. In Court of Appeals elections, the p o litic a l parties are formally removed from the process. Often candidates fo r the Court of Appeals face less competition in the election because 27 the requirement fo r party nomination of Supreme Court candidates has resulted in every race under this system being contested. The Court of Appeals candidate may have to face a primary election, but Supreme Court candidates never do. The whole existence of the Court of Appeals until 1983 has been under Republican administrations--Romney's and Mi 11iken' s--while fo r the period studied the Supreme Court has been under both party administrations (Democratic from 1949 to 1963 and Republican from 1963 to 1982). Trial Courts The C ircuit Court is Michigan's court of general original j u r ­ isdiction. Although C ircuit Court judges also hear appeals from lower courts and state administrative agencies, the C ircu it Court is not classified as an appellate court. Candidates are placed on the b a llo t via a petition fo r six-year terms. Incumbents may f i l e an a f f id a v it and are designated as incumbents on the b a llo t. The non­ partisan elections before 1963 were held in the Spring of odd-numbered years and are not held in the Fall of even-numbered years. Vacancies may be f i l l e d by Gubernatorial appointment un til the next election. The Circuits are designed along county lines and may be multi-judge and/or multi-county in composition. The number of c irc u its and judge­ ships may be increased by law with the number currently at fifty -tw o and six ty-th ree , respectively. The Probate Court has original ju ris d ic tio n in juvenile delin­ quents and dependents cases. The seventy-nine d is tr ic ts (105 judges) are organized along county lin es. However, eight counties are 28 consolidated into four d is tr ic ts . Q ualifications, election, f i l l i n g of vacancies, and term of o ffic e are the same as fo r the C irc u it Court. The D is t r ic t Court has ju ris d ic tio n in c iv il cases up to $10,000; a ll misdemeanors where less than one year punishment; arraign­ ment, setting and accepting b a il; preliminary examination in felony cases; evictions; garnishments; mortgage and land contract fo re ­ closures; and a small claims division fo r c iv il cases under $600. This court was created because the 1963 Constitution required the state leg is la tu re to create a court or courts to replace the offices of Justice of Peace and C irc u it Court Commissioner. The d is tr ic ts (cur­ rently 97) are determined by population and existing p o litic a l boundar­ ies. Q u alifications, e le ctio n , f i l l i n g of vacancies, and term of o ffic e are the same as fo r the C irc u it Court. Selection Issue in Michigan Except fo r the f i r s t twenty-two years of Michigan's statehood, its appellate ju d icia ry has been elected. However, this does not mean that ju d ic ia l selection in Michigan is a settled matter. A run­ ning dialogue continues in Michigan's law journals and as recently as two years ago a serious movement was underway to in s titu te an appointed Supreme Court. an issue? Why may the selection procedure fo r ju d ic ia l offices be Jacob (1964) discovered that the d iffe re n t selection methods resulted in variations in the characteristics of the judges. Also, partisan motives have been a strong basis fo r the changes in ju d ic ia l selection methods in the United States. This was documented by William S. Carpenter in his Judicial Tenure in the United States. 44 29 Several times during the period before 1948, proposed a lt e r a ­ tions of the selection process were on the b a llo t. In 1934, reformers placed a constitutional amendment on the b a llo t which would have provided fo r the nonpartisan election of judges; i t fa ile d . 45 Next, under the sponsorship of the Michigan State Bar, a form of the Missouri Plan was placed on the b a llo t in 1938. This fa ile d due to the poor use of propaganda and opposition by the Wayne County C irc u it ju d g e s .^ In the following year these judges and sympathetic Democrats who feared discrimination under the proposed Missouri Plan, placed on the b a llo t a proposal fo r the nonpartisan election of state judges. passed. This one 47 For the period included in my study, from 1948 to 1982, a lt e r a ­ tions in the electoral system were s t i l l being urged. Considerable energy was exerted on the question of ju d ic ia l selection in the Consti­ tutional Convention of 1961-1952. Albert Sturm noted that no set of issues gave more trouble in the Committee of the Whole debate than the ju dicia ry a r t i c l e . Generally, the Republican delegates supported a merit selection plan while the Democratic delegates backed nonpartisan AQ election. A poorly worded poll of the state bar membership revealed that they were divided over this issue. 49 The outcome was that the existing method prevailed fo r the Supreme Court and the Court of Appeals was required to have complete nonpartisan procedures. Another issue fo r contention was vacancy appointments. The abolition of Gubernatorial vacancy appointments to the court in the Constitution of 1953 had l i t t l e impact on the appellate courts because 30 of it s short l i f e . A constitutional amendment passed in 1968 restored the Governor's appointment power. A s ig n ifican t change brought about by the Constitution of 1963 was switching the ju d ic ia l elections to the F a ll. A second s ig n ific a n t change was the provision in the new constitution allowing incumbents to f i l e an a f f id a v it fo r re-e lection . Another attempt to place a merit selection plan on the b a llo t occurred in 1968. A co alitio n of associations which met in 1967 had agreed upon a l i s t of the deficiencies in the current process and a plan to correct those procedures. However, th e ir campaign to "take the courts out of p o litic s " f a i l e d . ^ More recently, in 1970 a ll four of the Supreme Court candidates publicly denounced the prevailing selection procedure. They a ll favored some form of merit plan.^1 A merit plan also was the recommendation of the Special Commission to Review A rtic le VI of the Michigan Constitution. BP Not a ll of the Supreme Court Justices nor a ll of those on the Special Commission rejected an elective process. Justices Paul Adams and George Edwards supported ju d ic ia l selection by election because they claimed that i t provides fo r participation by the people and ensures the justices w ill know the people and th e ir needs. 53 The Commission's dissenting report included the point that a referenda vote on judges, as suggested by the Commission, would not be as meaningful a contest as we have now because the Supreme Court candidates face opposition in the election. 54 Current support fo r an elective system can also be found in the on-going dialogue carried out in the legal periodicals. For a recent example of th is interchange, see "The Way I See I t " in the Michigan Bar Journal. 55 31 Two recent episodes reinforce the seriousness of the struggle over Michigan's selection procedure. In 1981, L t. Governor Brickley actively supported a reactivation of the movement to appoint Michigan's Supreme Court. candidate. 56 Brickley had experience, as a losing Court of Appeals At the time he sought the Republican party's gubernatorial nomination in 1981, his v i s i b i l i t y in the selection issue diminished. A fter a Democratic gubernatorial v ic to ry , ju s t before leaving o ffic e in 1982, Republican Governor Mil liken appointed Brickley to the Supreme Court. The next Mi H i ken appointment created the second episode. With the November, 1982, death of a re-elected Democratic Supreme Court incumbent, Mil liken appointed Court of Appeals Justice Dorothy Com­ stock Riley to both the unexpired term and the new term of the deceased Justice. Blanchard, the new Democratic Governor, challenged in the Supreme Court Mi 11fken' s power to f i l l the new term vacancy. The i n i t i a l vote by the Supreme Court s p lit along p o litic a l party lines with Justice Levin's deciding vote creating a 3-3 t i e which allowed her to remain on the Supreme Court. The Supreme Court and in particu­ la r Justice Levin drew heavy critic ism in the press. Soon afterward Levin switched his vote which resulted in the removal of Riley from the bench. Riley who had lost in the 1982 Supreme Court election now lo st in the Supreme Court! The whole series of actions has l e f t the dignity of the Supreme Court and the power of gubernatorial appointment in an uncertain state. 57 What is certain is that the selection process used to determine Michigan's appellate ju d icia ry is extremely important to various 32 interests in Michigan. The debate on which method should be adopted is based on the assumption that any method influences the selection out­ come. Unfortunately, there s t i l l exists a "factual vacuum" about the nature of that influence. My goal is to f i l l information supported by empirical evidence. some of that vacuum with I have developed a model of appellate court candidate a ttrib u tes to determine the r e la tiv e influence of those bases of competition. These candidate a ttrib u tes are grounded in the theoretical approaches of the American Voter, rational choice, and cybernetic theorists. As Watson and Downing sought to id e n tify the groups advantaged by the Missouri Plan and Hannah sought to id e n tify the groups advantaged in Michigan, my study is intended to ad ditionally compare th e ir power as reflected by the re la tiv e influence of the candidate a ttrib u te s . How the existing nonpartisan election procedures operate should be known before further changes are adopted. next chapter. My methodology fo r this model is outlined in the FOOTNOTES— CHAPTER I ^Glendon Schubert, "Academic Ideology and the Study of Adjudi­ cation," American P o litic a l Science Review (March 1967): 106-29. 2 This is one of the many definitions based on Systems Theory. See David Easton, The P o litic a l System (New York: Knopf, 1953); and A Framework fo r P o litic a l Analysis (Enqlewood C l i f f s , N.J.: PrenticeH a ll, 1965). q Harold J. Spaeth, Supreme Court Policy Making: Explanation and Prediction (San Francisco: W. H. Freeman and Company, 1979), pp. 1-8. ^Herbert Jacob and Kenneth Vines, "Studies in Judicial P o li­ t ic s ," Tulane Studies in P o litic a l Science, V I I I (New Orleans: Tulane University, 1962). Also see James Herndon, "The Role of the Judiciary in State P o litic a l Systems," in Judicial Behavior, ed.: Glendon Schubert (Chicago: Rand McNally, 1964), pp. 153-61; and Wallace S. Sayre and Herbert Kaufman, Governing New York City (New York: Russell Sage, 1960). c Book of the States, 1982 (Chicago: Council of State Govern­ ments, 260-261). ^Oliver P. Williams and Charles R. Adrian, "The Insulation of Local P o litic s Under the Nonpartisan B a llo t," American P o litic a l Science Review (1959), p. 1053; and Eugene Lee, The P o litic s of Nonpartisanship (Berkeley: University of C alifornia Press, 1960), pp. 98 and 117. ^The "mixed" states are Arizona, Michigan, and Ohio and are analyzed in P hilip DuBois, From B allot to Bench: Judicial Elections and the Quest fo r Accountability (Austin: University of Texas Press, 1980): O My study does not include other factors which are hypotheti­ ca lly sig n ific a n t in ju d ic ia l elections. The lim ita tion s of aggre­ gate voting data and secondary biographical material as well as lim ited funds precludes analysis of such factors as: the influence of bar p o lls, newspaper coverage, and campaign a c t iv i t ie s . The l a t t e r two factors in p articular provide an opportunity fo r fu rth er state ju d ic ia l research. 33 34 9Campbell, e t a 1 . , The American Voter (Chicago: Chicago Press, 1960), p. 131. University of 10I b i d . , p. 141. U Canons 7 ( B ) (1 ) (c ) , 7 (A )(2 ), and 7 (A )(4 ). 12 Early leaders in rational choice decision-making include Herbert A. Simon's early work such as Administrative Behavior, 3rd ed. (New York: The Free Press, 1976); and Anthony Downs, An Economic Theory of Democracy (New York: Harper and Row, 1957). 13 Joseph A. Schlesinger, "The Structure of Competition fo r Office in the American States," Behavioral Science 5 (July 1960): 198. 14 A study in Michigan found as partisan issues decreased, other dimensions formed the electorate divisions. See Norman C. Thomas, "The Electorate and State Constitutional Revision: An Analysis of Four Michigan Referenda," Midwest Journal of P o litic a l Science 12 (February 1968): 115-29. 15 Richard A. Watson and Rondal G. Downing, The P o litic s of The Bench and the Bar: Judicial Selection Under the Missouri Nonpartisan Court Plan (New York: John Wiley, 1969), pp. 334-335, 352. 16 Susan B. Hannah, "An Evaluation of Judicial Elections in Michigan, 1948-1968," (Ph.D. dissertation, Michigan State University, 1972), Chapter I I . 17Campbell, et a l . , argue that the form of the b a llo t has the greatest impact on p o litic a l behavior when the attitudes are weak. American, p. 283. 1Q Lansing: "“Carolyn S tieber, The P o litic s of Change in Michigan (East Michigan State University Press, 1970), p. 1. 19DuBois, B a llo t, pp. 65-66, 69. 20 The development of sectionalism in a state often has at its root economic facto rs. Therefore, the suburbs surrounding D etroit and D etroit are economically bound together to form a section--D etroit Metropolitan Area. They have more intertwined interests versus the rest of the state than they have differences between them. For a scholarly history of Michigan, see W illis Frederick Dunbar, Michigan: A History of the Wolverine State (Grand Rapids: William B. E. Erdmans Publishing Co., 1965); and Carolyn Stieber, P o lit ic s , fo r recent r iv a l r ie s . 21 I t is not unknown fo r a rtic le s on the court elections to elaborate on sectional r iv a l r ie s . See D etroit Free Press, 5 February 1961, p. 1. 35 22 Herbert Jacob, "The Effect of In s titu tio n a l Differences in the Recruitment Process: The Case of State Judges," Journal of Public Law (1964): 116. 23 Bradley C. Canon, "The Impact of Formal Selection Processes on the Characteristics of Judges--Reconsidered," Law and Society Review 6 (May 1972): 582-588. 24 Watson and Downing, Bench and Bar, pp. 206, 217-219. 25 Stuart Nagel, Comparing Elected and Appointed Judicial Sys­ tems (Beverly H ills : Sage Professional Papers in American P o litic s , 1973), p. 37. 26I b i d . , p. 23. 27 Samuel Krislov, "Constituency vs. Constitutionalism: The Desegregation Issue and Tensions and Aspirations of Southern Attorneys General," Midwest Journal of P o litic a l Science 3 (1959): 88. 28 James Eisenstein, P o litic s and the Legal Process (New York: Harper and Row, 1973), p. 31. 29 Bancroft C. Henderson and T. C. S in c la ir , Judicial Selection in Texas: An Exploratory Study, University of Houston Studies in Social Science (Houston: Public A ffa irs Research Center, 1965), pp. 19-21, 51-68. 30 Kenneth N. Vines, "The Selection of Judges in Louisiana," in Studies, pp. 104-113, 118. 31 Malcolm Moos, "Judicial Elections and Partisan Endorsements of Judicial Candidates in Minnesota," American P o litic a l Science Review 35 (February 1941): 71-73. 3? Robert A. Heiberg, "Social Backgrounds of the Minnesota Supreme Court Justices: 1858-1968," Minnesota Law Review 53 (1969): 903-908, 914-931. 33 Herbert Jacob, "Judicial Insulation--Elections, Direct P articipation, and Public Attention to the Courts in Wisconsin," Wisconsin Law Review (1966): 818. ^DuBois, B a llo t, p. 73. 35Hannah, "Evaluation," p. 172. 35I b i d . , pp. 181-182. 3^DuBois, B a llo t, pp. 72-81. 36 38I b i d . , pp. 81-88, 132. 39 Kathleen L. Barber, "Nonpartisan Ballots and Voter Confusion in Judicial Elections," paper prepared fo r delivery a t the 1982 Annual Meeting of the Midwest P o litic a l Science Association, Milwaukee, Wisconsin, pp. 28-32. ^H e rb e rt Jacob, Justice in America (Boston: 1965), p. 207. L i t t l e Brown, 41 Albert L. Sturm and Margaret Whitaker, Implementing a New Constitution: The Michigan Experience, Michigan Governmental Studies No. 50 (Ann Arbor: University of Michigan Press, 1968), p. 138. 42Michigan State Constitution (1908), A rtic le V I I ; Michigan State Constitution (1963), A rtic le VI; and Laws Relating to Elections, compiled by the Michigan Secretary of State, Chapters XVIII and XIX. 43 Amendment proposed by H.J.R. F, R atified August 6, 1968, P.A. 1968, 711. 44 William S. Carpenter, Judicial Tenure in the United States (New Haven: Yale University Press, 1918), p. 73. 45 Daniel McHargue, Direct Government in Michigan, Michigan Constitutional Convention Studies, No. 17, Lansing, 1961, p. 38. 46 George E. Brand, "Michigan State Bar's Work fo r Judicial Appointment," Journal of the American Judicature Society 22 (February 1939): 197-202. 47 McHargue, D ir e c t, p. 39. * r> ^ O f f i c i a l Record of the Michigan Constitutional Convention, 1961-1962, V. I , pp. 1256, 1313-1342, 1355-1372, 1596-1604; and Albert L. Sturm, Constitution-Making in Michigan, 1961-1962, Michigan Governmental Studies No. 43 (Ann Arbor: University of Michigan Press, 1963), p. 201. 4Q Michigan State Bar Journal 41 (1962): 12-16. 38Hannah, "Evaluation," p. 35. 51Ka1amazoo Gazette, 14 October 1970, p. A-10. 52H.C.R. 22, 1971 Session, pp. 6-7. 53I b i d . , pp. 32-35. 54I b i d . , p. 37. 37 1981): ^"The Way I See I t , " The Michigan Bar Journal 80 (October 761-762; and (December 1981): 963-965. 56 Address to the Michigan Press Association by Lt. Governor James Brickley, January 30, 1981. *^As well as regular television coverage of this episode, the following is indicative of the press coverage: Lansing State Journal, 4 January 1983; 5 January 1983; 6 January 1983; 12 February 1983; D etroit Free Press, 13 February 1983; 12 February 1983; 15 February 1983. Even an a r t ic le on the "Riley Ouster" by an United Press International w rite r appeared in the Michigan State University news­ paper, The State News, 1 March 1983, p. 5. CHAPTER I I METHODOLOGY Chapter Objectives With the theoretical framework established in Chapter I , the methodology is indicated in this chapter. design is described. Next, the data set I n i t i a l l y , the research is id e n tifie d . Then dis­ cussed are the fiv e research hypotheses, techniques to te st them, and the c r i t e r i a . Chapter I I is concluded with a summary. Research Design Approximately three-fourths of the states elect some of th e ir ju d icia l o ffic e rs . Refining our knowledge by identifying the re la tiv e relationships between the bases of competition fo r these judgeships is long overdue. As a beginning, this is a case study of Michigan's appellate court elections. There is l i t t l e doubt that changes in the rules of the game may a ffe c t behavior.* Adoption of the 1963 Michigan Constitution and the resulting election laws, which were described in Chapter I , provide a prime opportunity fo r investigation of changes in the influence of the bases of candidate competition in Supreme Court elections. The formal selection rules fo r the recently created Court of Appeals are d iffe re n t than fo r the Supreme Court. This also provides a special opportunity to investigate the differences in the re la tiv e influence of the candidate a ttrib u tes which may explain the 38 39 election outcome in each of these courts. Therefore, Michigan's ju d ic ia l electoral scheme has several q u alitie s which warrant its separate analysis. The research focus is to determine the r e la tiv e influence of selected candidate a ttrib u tes which may explain the election outcome fo r Michigan's appellate court races. The a ttrib u tes analyzed are derived from three basic approaches to decision making. Aggregate voting s ta tis tic s and candidate biographical information from the years 1948 to 1982 comprise the data set. 2 These data are analyzed with Probit, an ordinal analog of regression, to answer the following questions about Michigan's appellate court elections. What are the r e la tiv e influences of the bases of competition fo r Michigan Supreme Court elections before and a f te r the Michigan Constitution of 1963? Does the Michigan Court of Appeals have the same rank ordering of these bases as the Supreme Court? I began the analysis of Supreme Court elections by updating Susan Hannah's model of the bases of candidate competition in Michigan Supreme Court races fo r the years 1948 to 1970. With the addition of the elections fo r the years from 1971 to 1980 (elections in 1981 and 1982 were reserved fo r predictive purposes), I evaluated her model and used i t as a standard fo r my model. Part of the analysis included testing Hannah's model with the data divided into two periods--before and a f te r the adoption of the 1963 Constitution. posttest quasi-experimental research design. This is a pretest- The changes resulting from this Constitution may have s ig n ific a n tly affected the bases of competition in Supreme Court elections. 40 Next, I determined the r e la tiv e influence of the set of candi­ date a ttrib u tes which may explain who wins and who loses in a con­ tested Supreme Court election. This model contains an important a t t r i ­ bute omitted by Hannah and a refinement of her operationalization of the bases of candidate competition. The coefficients produced by Probit represent the r e la tiv e influences of the bases of candidate support fo r the two periods: from 1948 to 1963 and from 1964 to 1980. The analysis included a comparison of my model fo r the period from 1948 to 1963 to Hannah's model fo r the same period. Of greater interest is the comparison of these two periods to determine changes in the re la tiv e influence of the candidate attributes which may explain who wins or loses a contested Supreme Court election. Chapter I I I presents the results of the Supreme Court election analysis. The other appellate court analyzed is the Michigan Court of Appeals. As f a r as I am aware, no model on the bases of competition fo r Michigan Court of Appeals elections exists. Again, I used Probit to determine the r e la tiv e influence of the bases of competition which may provide a candidate fo r a Court of Appeals contested race an advantage over the other candidates not possessing those a ttrib u tes or possessing them to a lesser degree. Even though the Court of Appeals began in 1964, only the elections from 1970 to 1980 were u t iliz e d to determine the re la tiv e influence of the candidate a ttrib u te s . The reason fo r exclusion of the elections from 1964 to 1969 is indicated when the data set is discussed. reserved fo r predictive purposes. The elections in 1981 and 1982 were 41 There are differences in the formal selection processes fo r the two appellate courts. Therefore, the analysis also included a comparison of the rank order of the re la tiv e influence of candidate attributes which may explain who wins and loses in Court of Appeals elections and the ranking fo r the Supreme Court contests. Chapter IV contains the findings from the Court of Appeals elections analysis. The preceding discussion of the research design provides a framework fo r elaboration of the methodology in the rest of this chapter. Data Set The data set contains a ll Michigan Supreme Court elections fo r the period from 1948 to 1982 and a ll contested Michigan Court of Appeals general elections from 1970 to 1982. See Table 2.1 fo r the dates of these ju d ic ia l elections and the number of candidates. 1948 was chosen as the starting date fo r the Supreme Court in order to include in the data set the en tire period analyzed by Hannah. The l a t t e r starting date fo r the Court of Appeals elections was necessary fo r the following reason. Because this is a recently created appellate court, i t is not u n til 1970 that an incumbent runs in a contested elec­ tio n. Since incumbency is one of the candidate attributes under investigation, I wished to avoid possible bias due to the inclusion of the e a r lie r elections. Another concern about possible bias resulted in the decision to omit unopposed elections. The prerequisite fo r inclusion of an election was that each ju d ic ia l o ffic e had to have more than one s e r i­ ous candidate on the b a llo t. I adopted V. 0. Key's c rite rio n that a 42 TABLE 2 . 1 . — Data set Office Year Number of Candidates Judicial Office Supreme Court 49,51,52,53,55,56,57, 59,60,61,62,63 51 66,68,70,72,74,76,78, 80,82 41* 33 Court of Appeals F irs t D is t r ic t 74,76,82 Second D is tr ic t 70,74,76,78,80 Third D is t r ic t 70,74,78,80 Nonjudicial O ffic e ** State Board of Agriculture 49,51,53,55,57,59 MSU Board of Trustees 61,63,66,68,70,74,78,80,82 State Treasurer 52 Auditor General 56,60,62 C f a f a D na v+A «■ »wu ow L^WUI U \J t Education 72 Wayne State Univer­ s ity Board of Governors 76 *42 actual candidates; one was excluded as a nonserious candidate. **Used to operationalize strength of the candidate's p o litic a l party. 43 serious candidate had to receive at least 5 percent of the vote. 3 Inclusion of unopposed elections could bias the results with the attributes of candidates who were not in a competitive situation. This particular data set was chosen fo r several reasons. F ir s t, appellate courts are the more pervasive policy-making courts. Therefore, there is greater voter attention on state appellate court elections than on the lower state court elections.^ Second, informa­ tion is more readily available fo r these more prestigious courts. The a v a ila b ilit y of information is often a greater problem a t the state level than at the federal le v e l. because i t includes the recent past. Third, this period was chosen Hopefully, analysis of this data set w ill provide the best indication of how Michigan's electoral process presently operates. Fourth, at approximately midpoint in this period a new Michigan 5 Constitution was adopted. I t contained and led to provisions s i g n i f i ­ cant to the competition fo r ju d icia l seats. Of particular interest was the creation of the Court of Appeals, change from Spring to Fall elec­ tions fo r ju dicial seats, and b a llo t admission fo r incumbents via filin g an a f fid a v it of candidacy. these changes to be examined. The time span of my data set allows Last, the time span is long enough to provide fo r elections with incumbents running fo r re-election and short enough to exclude the pre-1948 period in which information in many instances was not recorded or has been destroyed. Unfortunately, irre trie v a b le information was lost in a Michigan State Library build­ ing f i r e in the 19501s. 44 Research Hypotheses Five research hypotheses guide the empirical analysis of Michigan appellate court elections. hypotheses is introduced. In this section, each of the Within the discussion of the second hypo­ thesis is a comparison of Hannah's and my conceptualizations of the bases of candidate competition. contrasted. Their operationalization is also For each hypothesis, the analytical techniques and c r it e r ia used to test i t are id e n tifie d . Hypothesis I My research on Michigan Judicial candidate a ttrib u tes evolved from in terest piqued by Hannah's preliminary study of Michigan ju d i­ cial elections. I t contained a model of partisan and nonpartisan com­ p e titiv e bases fo r Supreme Court candidates.^ The partisan base which would advantage a candidate was the strength of the state p o litic a l party nominating that candidate. The nonpartisan bases were a candi­ date's incumbency, p o litic a l background of the candidate, the candi­ date's private practice experience, sectional support fo r the candidate, and ethnic appeal of the candidate. The data Hannah used were a ll Supreme Court elections from 1948 to 1970. An important question is: With the inclusion of the elections since 1970, does her model explain Supreme Court election results with the same a b i l i t y as fo r the 1948 to 1970 period? There­ fo re , the f i r s t hypothesis is: Hypothesis I : The bases of competition fo r Michigan Supreme Court elections as id e n tifie d by Hannah fo r the period 1948 to 1970 are s t i l l bases of competition fo r the period from 1971 to 1980. 45 Model and operatio nalization. The following are the bases of competition in the Hannah model and th e ir o p eratio nalization:7 1. Incumbent. A candidate was scored po sitively i f at the time of election the candidate was a Supreme Court Justice--appointed or elected. 2. Party. A candidate was scored positively i f the nominating p o litic a l party won the state o ffic e having the most votes. 3. State Office. A positive score was recorded fo r candi­ dates who held or previously had held state elective or appointive o ffic e . 4. Local O ffice. The candidate who held or previously had held local appointive or e le ctive o ffic e was scored p o sitively . State judgeships other than Michigan Supreme Court were scored as local o f f i ces. 5. Private. A candidate with a lif t im e private prac­ tic e and an "A" Martindale-Hubbel Law Directory rating was scored p o sitively. 6. Section. A candidate from D etroit or the Upper Peninsula was scored p o sitively . 7. Ethnic. A candidate was scored po sitively i f the candidate was Roman Catholic, or Black, or had a name easily id e n tifie d with an ethnic group. Hannah tested her model with the requirement that fo r each election the winning Supreme Court candidates were to have higher 46 scores than th e ir opponents. The s t a tis tic Hannah used to evaluate her model was the percent of winners predicted correctly. A score fo r each candidate was determined by adding points awarded fo r the presence of each of the bases of competition. In assigning points fo r each of these bases, Hannah awarded each fa c to r, in the order lis te d above, an equally decreasing value. Thus, a candidate scored positively fo r Incumbent received seven points and so forth down to one point fo r the presence of Ethnic. Technique and c r it e r io n . I tested the f i r s t hypothesis by updating the Hannah model with her operationalization. cated her scoring procedure and s t a t is t ic . Also, I dupli­ The c rite rio n was that the updated model must have a percent predicted correctly no less than fiv e percentage points of the original ta b le 's . Therefore, at least 81 percent predicted correctly by the updated model was required. Hypothesis I I The original Hannah model predicted ju d ic ia l winners from 1948 to 1970 remarkably w ell. However, I believe that a very important candidate a t tr ib u t e , name f a m ili a r it y , was not included in her model Q and other attributes can be more d is tin c tly conceptualized. More advanced analytical techniques have become available since Hannah's study. These techniques f a c i l i t a t e the addition of a time dimension to the meaning of three of the variables in the Hannah model. Another variable in her model is separated into three bases of candidate competition in order to provide further conceptual refinement. fo re , building upon the preliminary study by Hannah (1972) and There­ 47 P hilip DuBois' (1980) la t e r research leads to a more sophisticated model of the bases of competition fo r candidates in Michigan ju d ic ia l g elections. Thus, the second hypothesis, which is the foundation of my research, is: Hypothesis I I : There is a positive relationship between winning an appellate court election in Michigan and the incumbency status of the candidate, strength of the candidate's p o litic a l party, state and local p o litic a l o ffic e experience of the candidate, support from the candidate's geographical section, unearned fa m ilia r it y of the candidate's name, Ir is h or other ethnic appeal of the candidate's name, and support from the candidate being recognized as a Catholic. Discussion of Hypothesis I I continues with a comparison of Hannah's and my conceptualizations of the bases of competition. Next, my variable operationalizations are presented and contrasted with Hannah's. Last discussed are my model and the techniques used to test i t , including the c r i t e r i a . Conceptualizations contrasted. Two of the candidate a t t r i ­ butes in my model were conceptualized as in Hannah's. They are the strength of the candidate's p o litic a l party and support from a candi­ date's geographical section. competition in Hannah's model. I reconceptualized four of the bases of These candidate attributes are incumbency, state public o ffic e experience, local public o ffic e experience, and ethnic appeal. A comparison of those differences follows. Hannah conceptualized a candidate's incumbency as an advantage in ju d ic ia l elections because that incumbency is designated on the b a llo t fo r a ll Michigan ju d ic ia l o ffic e s . Other than the appearance 48 of the candidate's name and what association they may evoke the only information available from the b a llo t is whether the candidate cur­ rently holds that o f f i c e . 1^ Hannah concluded that since the public seems to desire s t a b i l i t y in the ju d ic ia l arena, the incumbent has an advantage over the challengers who are on the b a llo t without this designation. An a lte rn a tiv e conceptualization of incumbency is suggested in the voting lit e r a t u r e . Although these scholars do not agree on the magnitude of incumbency's impact on election outcomes, they agree that the length of time a candidate has been an incumbent is an important research consideration.11 Therefore, the notion of incum­ bency is broadened to include the frequency and nature of the candi­ date's b a llo t appearances. The judge who has held o ffic e fo r more terms and whose name has appeared on the b a llo t more often with the incumbency designation would have an advantage over the other candi­ dates. In addition to the extraordinary advantage present at the polls in the form of the b a llo t designation, the incumbent has the benefit of wide and numerous contacts with potential voters through the performance of his/her ju d ic ia l o ffic e . To give this base of competition a f u l l e r dimension, the other p o s s ib ility fo r a candidate's b a llo t appearances was also incorporated into my conceptualization of incumbency. Remember, one p o s s ib ility is that the candidate, as an incumbent, is thought of by the voters as a previous winner who is the current officeholder. Another p o s s ib ility fo r a candidate is repeated appearances on the b a llo t without the 49 incumbency designation. A fte r an in i t i a l candidacy, the repeat candi­ date may become disadvantage due to a growing image as "a lo s e r." I do not believe the i n i t i a l b a llo t appearance portrays a candidate as a success or fa ilu r e ; i t was viewed as neutral fo r a ll candidates, except fo r those candidates who were designated incumbents in th e ir f i r s t race. They have the advantage of th e ir incumbency. My conceptualization of ju d ic ia l incumbency is a m odification of the conceptualization found in the voting lite r a tu r e due to the influence of the rules of the game fo r Michigan ju d ic ia l elections. I posit a positive lin e a r relationship between incumbency and the elec­ tion outcome while some scholars have suggested a nonlinear re la tio n ­ ship in which the influence of a candidate's incumbency begins to decline fo r the long time incumbent. Even i f nonjudicial elections are nonlinear as ju s t described, the type of election I modeled would contain ju s t the e a r lie r gradually ascending portion of th at curve which can be represented lin e a r ly . There are three reasons why a Michigan appellate court incum­ bent can have very few re -e lectio n bids. F ir s t, admission to the Michigan State Bar is a q u a lific a tio n fo r ju d ic ia l o ffic e . This along with the higher status of appellate court o ffic e results in an over­ whelming number of candidates who are a t least middle-aged. Second, the term of o ffic e is long which reduces the frequency of re-electio n bids. Third, a candidate who is seventy years of age or older is in e lig ib le fo r ju d ic ia l o ffic e in Michigan. career in comparison to other o ffic e s . This shortens a ju d ic ia l Therefore, neither Hannah's 50 nor my conceptualization of th is base of competition is c u rv ilin e a r. My conceptualization of incumbency has a broader dimension because I included the frequency and nature of the candidate's b a llo t appear­ ances.1^ The second and th ird candidate a ttrib u te s --s ta te public o ffic e experience and local public o ffic e experience—are contrasted together because the differences in conceptualization between Hannah and me are the same fo r both. Hannah's conceptualization of the advantage derived from holding public o ffic e was re s tric te d to whether the can­ didate had ever held a public o ffic e , irrespective of how long that o ffic e had been held. My conceptualization of the advantage derived from holding public o ffic e was extended to include the length of that service. I t would seem th at the longer a candidate's officeholding experience, the greater the advantage. Longer periods in o ffic e may provide more opportunities fo r the candidate to establish her/his v i s i b i l i t y and a communications network; both may be more f u lly estab­ lished over a longer period of time. The fourth candidate a ttrib u te which Hannah and I conceptual­ ize d iffe r e n tly is ethnic appeal. Hannah grouped together Catholics, Blacks, and those having a name ea sily associated with an ethnic group to comprise one base of competition. d is tin c t bases of competition. receive from: I reconceptualized i t into three They are the support a candidate may (1) having an Iris h name, (2) having a name easily associated with another major ethnic group in Michigan, and (3) being recognized as a Catholic. Also, I would have included a separate base 51 of competition fo r being a Black, but there are too few Black candi­ dates in my data to s t a t is t ic a lly ju s tif y such a category. This ra c ia l candidate a ttrib u te should not be incorporated with e ith e r of the other two a ttrib u te s , as Hannah d id, because the basis fo r the support a Black would receive is d iffe re n t than the support from being recognized as a Catholic or having an ea s ily iden­ t if ia b le ethnic name. Recognition as a Catholic is a d is tin c t candi­ date a ttrib u te fo r the following reasons. Michigan h isto rie s note the early introduction and influence of Catholicism in Michigan. cism is one of Michigan's largest relig ious groups. C ath oli­ Also, th is re lig io n has a highly organized and unified nature, weekly a tte n ­ dance as a relig ious te n e t, and the presence of diverse social a c t iv i­ tie s . These ch aracteristics of Catholicism provide a Catholic candi­ date read ily availab le organizational resources and communication channels. F in a lly , the special nature of Catholicism results in higher voter p a rtic ip a tio n by Catholics. 13 As a group, Blacks in Michigan do not have these ch a ra c te ris tic s , especially the higher vot­ ing ra te , which may provide a source of candidate support. Being a Black should not be incorporated with the candidate a ttrib u te of ethnic appeal. One reason ethnic appeal is conceptualized as a separate candidate a ttrib u te is because a candidate with an easily id e n tifia b le ethnic name can gain support a t the polls from the appearance of his/her name on the b a llo t. Blacks do not have this advantage because, in many instances, they are not ea sily id e n tifie d as Blacks from the appearance of th e ir name on the b a llo t. 52 Instead of only two categories fo r the ethnic appeal of a candidate's name, i t would be ideal to have a separate one fo r each major ethnic group in M ic h ig an .^ could be established. Then the influence of each group Unfortunately, th is was not possible because there were too few cases fo r a ll of the ethnic groups except the Ir is h . H is to ric a lly , the Iris h in America have been p o litic a lly a c t iv e .^ Also, the Kennedy mystique in the 1960's may have contributed to popularization of Iris h candidacies. For example, 25 percent of the Michigan Supreme Court candidates between 1964 and 1980 had Iris h names. Since th is ethnic group would seem to have a d iffe re n t e ffe c t than the other prevalent ethnic groups, I established i t as an in d i­ vidual base o f competition. O p eratio n alizatio n . The op erationalization of the nine candi­ date a ttrib u te s which comprise the model d iffe rs in some respects from those in the Hannah model. These differences are noted as I present the operatio nalization of each of the candidate a ttrib u te s which may explain who wins or loses in a Michigan ju d ic ia l contested e le ctio n . In many instances I was able to re ta in a greater amount of information through use of a higher level of measurement and/or more precise conceptualization. Greater v a lid ity would be obtained from reducing the loss of information about the candidates. Sources of information fo r each candidate a ttrib u te are lis te d in Table A .2 in the Appendix. I w ill operationalize the dependent variable f i r s t , then proceed with the independent variables. 53 The dependent variable is the appellate court election outcome. Each candidate's election re s u lt is measured by collapsing the theor­ e tic a l in terval level of measurement into an em pirically available le v e l--a dichotomy. The m ulti-winner nature of these ju d ic ia l ele c­ tions complicates the use of any other level of measurement. Winning candidates received a "1" while losing candidates were scored a "0" fo r a ll contested term and vacancy elections. Now I operationalize the nine independent variables in my research. The f i r s t candidate a ttrib u te is the advantage derived from appearing on the b a llo t as an incumbent. I operationalized th is variab le as the number of consecutive elections in which a can­ didate was designated on the b a llo t as an incumbent minus the number of electio n s, a fte r the i n i t i a l one, in which the candidate ran w ith­ out b enefit of the incumbency lis tin g fo r th at o ffic e . Therefore, the i n i t i a l b a llo t appearance of a candidate, except fo r appointed incum­ bents, was coded a "0." The second b a llo t appearance of a success­ fu l candidate was coded a "1." The second b a llo t appearance of an unsuccessful candidate was coded a " -1 ." I used is higher than Hannah's. The level of measurement Therefore, more information about a candidate was incorporated in th is v a ria b le . The second candidate a ttrib u te is the strength of a candidate's p o litic a l party. I measured th is variable as the percent of the vote fo r the candidate's party fo r the partisan state o ffic e which consis­ te n tly received the amount of votes closest to the ju d ic ia l o ffic e 's . As mentioned in Chapter I , use of th is lower partisan race more accurately re fle c ts the strength of the p o litic a l p a rtie s . IS Most of 54 these races are m ulti-winner. Therefore, I determined each p o litic a l p arty's strength by adding the votes fo r that party's candidates and dividing th at sum by the to ta l votes cast in that ele ctio n . See Table A.3 in the Appendix fo r my manipulation of the vote s ta tis tic s . During the 1948 to 1963 period when most of the ju d ic ia l elections were held in the Spring, the partisan state o ffic e was the Board of Agriculture (now t it le d Trustees of Michigan State U n iv ersity ). For the four November vacancy electio n s, the Auditor General race had the closest vote to ta l. From 1964 to 1982 the Supreme Court and Court of Appeals elections were held in the F a ll. The Michigan State Univer­ s ity Trustees race consistently had the closest vote to ta l and there­ fore I used i t to determine the strength of the p o litic a l parties fo r these general and vacancy elections. For a Court of Appeals candidate to q u a lify fo r a percentage score, the candidate must have been associated with a p o litic a l party to the extent th at there was a party endorsement or the candidate had previously run under th at party label without a switch in party allegience being declared. Unfortunately, the strength of the candi­ date's p o litic a l party could not be determined fo r the Court of Appeals elections. During the period under study, there were instances of c o n flictin g candidate endorsements by the county p o litic a l party organizations.1^ The basic differences between my operationalization and Hannah's are th at I used a lower level of state o ffic e and my level of measurement is higher— in te r v a l. I captured the degree of party 55 support through the use of percentages. Therefore, the gradual changes over the years in p o litic a l party strength in Michigan can be seen more f u lly . My fin e r gradations than Hannah's " l'V O " scor­ ing add to the va ria tio n of th is v a ria b le . The th ird candidate a ttrib u te is a candidate's state public o ffic e experience. I operationalized the advantage obtained from a candidate having state p o litic a l o ffic e experience as the number of years the candidate held a substantial state e le c tiv e or appointive o ffic e in Michigan. Offices q u a lifie d i f they were f u ll time and eith e r referred to in p rin t during the campaign, lis te d by the candi­ date in e ith e r the Michigan State Bar Journal Roster or M artindaleHubbel Law D irectory, or recorded in the Michigan Manual. The records I used only allowed me to round the length of public o ffic e experience to tenths of a year as a meaningful measurement. My opera­ tio n a liz a tio n of th is variable and the next one d iffe re d from Hannah's due to the differences in our conceptualizations. My higher level of measurement than Hannah's should increase the variatio n contained by these variables. The fourth candidate a ttrib u te is a candidate's local public o ffic e experience. The measurement procedure used fo r the advantage a candidate has from the years of holding local public o ffic e was basically the same as fo r the state level experience. I added the Directory of the Michigan Municipal League to the above indicators of a substantial o ffic e a t the local le v e l. For Supreme Court candi­ dates, I recorded holding a Court of Appeals judgeship as a local o ffic e . 56 The f i f t h candidate a ttr ib u te , support from the candidate's geographical section, is measured as a percentage. The vote in the candidate's section which was cast fo r th at candidate was divided by the to ta l vote in that section. For m ulti-winner contests, the "total vote" was approximated by dividing the recorded vote by the number of seats in the contest. I id e n tifie d a candidate's section by the loca­ tion reported in my biographical sources. Most often th is was where the candidate was currently working. For the Supreme Court, I summed the county vote to ta ls which comprise the candidate's section in order to determine each candidate's percentage. In the Appendix, Table A .4 indicates the division of the counties fo r the three Supreme Court sections and the aggregated votes I used to determine the percentages. For the Court of Appeals, Table A .5 id e n tifie s the counties in each D is tric t and each candidate's home county vote percentage. Hannah's and my operationalizations d iffe re d in two respects. F ir s t , we used d iffe re n t levels of measurement. 54 to 55 The comments on pages in which I contrasted Hannah's and my operationalizations of the strength of the candidate's p o litic a l party also apply here. Second, Hannah included only two geographical sections—D e tro it and the Upper Peninsula. measurement procedure. Outstate. I e x p lic itly included a ll three sections in the The highest Supreme Court voter turnout occurs However, support from th is section is tempered in re la tio n to the others by it s expansive geographical size . The D e tro it Metro­ politan area regularly has fewer votes cast fo r Supreme Court candi­ dates but th is area also has more issues arising which reinforce 57 sectional id e n tific a tio n . The Upper Peninsula has the fewest votes cast in Supreme Court contests. This is compensated fo r by stronger sectional unity with candidates from this area having fewer competi­ tors from th e ir own section than do candidates in the other two sec­ tio ns. Thus, i t is easier to r a lly around a single "fa v o rite son." The sixth candidate a ttrib u te is name fa m ilia r ity . I measured the advantage a candidate has from possessing a name which was made popular by others as a dichotomy. A candidate was scored a "1" i f he/she had a name which was the same as or sim ila r to a name popular in Michigan p o litic s or a ju d ic ia l incumbent's. I determined which candidate names were fa m ilia r by a scrutiny of the Michigan election documents, legal sources, and Michigan histo ries indicated in Table A .2 in the Appendix. A ll other candidates were scored as a "0," including candidates with fa m ilia r names due to th e ir having estab­ lished its popularity. To remain meaningful and d is tin c t from the other bases of competition, this variable must be re s tric te d to an advantage obtained by a candidate who has not her/him self earned the voters' recognition. This variable was not included in Hannah's model and th erefo re, th is operationalization is not compared to hers. Now, I operationalize the seventh and eighth candidate a t t r i butes--the advantage from having an Iris h name or having a name ea sily associated with a major ethnic group in Michigan. For a ll of the candidates, including married females, I focused on the name as i t appeared on the b a llo t. both of these a ttrib u te s . The measurement procedure was the same fo r I scored a candidate possessing a name easily 58 id e n tifie d with an ethnic group prevalent in Michigan as a "1." I coded a ll the rest of the candidates as a "0." For the Court of Appeals electio n s, the ethnic group prevalent in Michigan a d d itio n a lly had to be sizeable in that D is t r ic t. An ethnic group had to be a t lea st four-tenths of a percent of the popu­ la tio n in Michigan to be considered prevalent. For an ethnic group to be considered sizeable in the D is t r ic t, i t had to be id e n tifie d in a cu ltu ral history as in flu e n tia l. 18 The level of measurement I used fo r my seventh through ninth candidate a ttrib u te s was the same as Hannah's fo r her Ethnic. Last, I operationalize my ninth candidate a ttrib u te --th e advantage an appellate court candidate may have from being recognized as a Catholic. I scored a candidate who is recognized as a Catholic as a "1"; a ll other candidates I coded as a "0." I determined the presence of th at recognition from candidate actions, e .g ., an obvi­ ous reference to her/his Catholicism in campaign lite r a tu r e ; or from the candidate's having a tr a d itio n a lly Catholic name. A candidate was not coded as a Catholic because of his/her name i f the candidate's non-Catholicism was widely known. For example, non-Catholic John W. F itzg e rald , son of a former Michigan Governor and a well-known p o litic ia n in his own r ig h t, was scored a "0." Model. In research, correct specification of the model is a basic assumption. The second hypothesis establishes a model in which the dependent variable is a lin e a r function of the independent variab les. I assume th at function is lin e a r because the approach to 59 studying ju d ic ia l elections I use is in its infancy. A d d itio nally, I assume th at the value of none of the candidate a ttrib u te s in te r ­ acts with the value of any of the others. Until fu rth er understand­ ing can be obtained, a simpler lin e a r equation without in teractive terms is preferred. A crucial difference between my model and Hannah's is that I did not predetermine the re la tiv e influence of the candidate a t t r i ­ butes which may explain who wins and loses in a contested Michigan appellate court election. She assigned a value to each of ther bases of competition in order to id e n tify the most successful combinations of advantages fo r winning. I did not do so because there is not yet enough research to know th at any base is , fo r instance, exactly twice as in flu e n tia l as another. The newer techniques, such as P robit, enable avoidance of predetermined weightings of the independent v a ri­ ables. Techniques and c r it e r i a . The following are the techniques and c r ite r ia used to te s t the second hypothesis. I tested th is hypo­ thesis with P ro bit, an ordinal analog of regression. For my research question, Probit or Logit are the most appropriate s ta tis tic a l techniques. IQ Probit was designed fo r research inquiries in which the underlying theoretical dependent variable is interval but was measured o rd in a lly . The p ro b ab ility of winning cannot be measured--just whether the candidate won or lo s t. Probit provides maximum likelih ood e s ti­ mates (MLE) which indicate the re la tiv e influence of each of the independent variables. 60 Least squares regression is currently being used to analyze ju d ic ia l data and has the advantage of an extensive groundwork already la id to f a c ilit a t e its use with p o litic a l science issues. Specialized techniques have been developed to correct fo r vio la tio n of the regres­ sion assumptions. However, lea st squares regression cannot adequately f i t the data when the dependent variable is measured as a dichotomy— such as ju d ic ia l case votes or the win/loss outcome of ju d ic ia l elec­ tio ns. Frequently, the re s u lt is th at a model tested with least squares regression is rejected due to a low goodness of f i t measure when the goodness of f i t s t a tis tic from the more appropriate Probit would suggest th at the model not be rejected. The c r ite r ia fo r the Probit te s t of the model follow s. 2 goodness of f i t measure, estimated R , must be a t le a s t .60. 20 The Addi­ tio n a lly , the summary s t a t is t ic , Spearman's r , should be s ig n ific a n t with alpha a t the conventional value of .05. Also included is lambda^, a reduction in erro r s t a t is t ic , in order to take into account the size of the marginals. Lambdah should be a t lea st .50. This is an a rb itra ry decision in which a reduction in erro r by 50 percent is a reasonable expectation. In regard to the maximum likelih o o d e s t i­ mates fo r each model, the c r ite r ia were th at they are positive and s t a t is t ic a lly s ig n ific a n t. For each c o e ffic ie n t, th is was determined by a one-tailed t - t e s t of significance with alpha set at .05. The s t a t is t ic , -2 Times Log Likelihood R atio, is used to te st the overall significance of the independent variables. -2 Times Log Likelihood Ratio is analogous to the F-Test in regression. Chi-square fo r -2 Times Log Likelihood Ratio must be s ig n ific a n t at alpha equal to .05. 61 Another technique used to evaluate both appellate court models was ex post forecasting. 21 d ic tiv e a b ilit y of a model. This type of forecasting can te s t the pre­ In order to ex post forecast the most recent data must be withheld from the data which is used to estimate the model c o e ffic ie n ts . The data excluded were the appellate candi­ dates running in the 1981 and 1982 elections. I calculated a score fo r each of those candidates by entering my withheld data into the estimated equation. Then, I predicted whether a candidate won or lo s t by comparing that candidate's forecasted score to the scores in my original Probit run. Last, I determined the percent of candidates predicted co rrectly and lambda^. The models fo r the two appellate courts should be able to predict the outcome of those withheld elec­ tions with a t lea st 60 percent accuracy and at lea st 50 percent reduction in erro r fo r the lambdab. For the Supreme Court model, I also used a dummy variable technique to te s t my b e lie f th at there are two d is tin c t periods fo r Supreme Court elections between 1948 and 1980. For each independent variable and the constant term, I added to the equation a binary variable which captures the change in the c o e ffic ie n t from the f i r s t period to the second. This technique is not used fo r the Court of Appeals elections to establish two periods because a ll of the Court of Appeals elections occur in the second time period. A t - t e s t of significance (alpha equal to .05) can be performed on these binary variables to substantiate that the co efficie n ts representing the re la tiv e influence of the candidate a ttrib u te s are d iffe re n t in the two time periods. 22 62 In regard to the Supreme Court models which were estimated fo r the two time periods, another basis used fo r evaluating them is that they should predict b e tte r than the Hannah model when i t also is divided into the same time periods. The original Hannah model cor­ re c tly predicted 86 percent of the candidates from 1948 to 1970. The percent predicted co rrectly by the Hannah model fo r the 1948 to 1963 period and the 1964 to 1980 period may be d iffe re n t than the percent reported in the Hannah study. Thus, the percentages calculated fo r each time period in the Hannah model are the more f it t in g standards fo r evaluating my Supreme Court models. Therefore, the percent pre­ dicted correctly obtained from the two Probit runs must be higher than the respective Hannah percentages. Hypothesis I I I In the remaining hypothesis, I investigate the differences between the models in the influence of the candidate a ttrib u te s which may explain the election outcome in an appellate court contest. The Supreme Court elections are compared fo r the two time periods. Then, the Supreme Court and Court of Appeals contests are compared fo r the same time period. The th ird hypothesis begins th is examination with the e a r lie r period Supreme Court races. Which are the strongest bases of competition which may pro­ vide support fo r a Supreme Court candidate's electo ral success? The ju d ic ia l election lite r a tu r e reviewed in Chapter I repeatedly in d i­ cated th at during the 1948 to 1963 period three candidate a ttrib u te s had a strong influence on who would win. The incumbency status of the 63 candidate overwhelmingly was id e n tifie d in th at lite r a tu r e as the prime fa cto r leading to v ic to ry . The other two candidate a ttrib u te s were the candidate's p o litic a l party a f f ilia t io n and state p o litic a l o ffic e experience of the candidate. According to Hannah's study, the order in which I ju s t presented these three candidate a ttrib u te s indicates the order of the magnitude of th e ir influence. 23 Therefore, the th ird hypothesis is: Hypothesis I I I : For Supreme Court elections during the period from 1948 to 1963, the incumbency status of the candidate, strength of the candidate's p o litic a l party, and state p o litic a l o ffic e experience of the candidate are the candidate a ttrib u te s which have the strongest re la tiv e influence, respectively. Technique and c r ite r io n . I standardized the value of the maximum likelih ood estimate co efficie n ts in order to be able to compare the re la tiv e influence of the candidate a ttrib u te s . 24 This was necessary because the maximum likelih ood estimate co efficie n ts fo r the independent variables were estimated using d iffe re n t scales of measure­ ment. I deemed the r e la tiv e influence of a candidate a ttrib u te as greater than another i f the value of it s standardized c o e ffic ie n t was la rg e r. The c rite rio n was th at the larg e r o f the two standardized co efficien ts must be a t le a s t 10 percent la rg e r. 25 Hypothesis IV How do the re la tiv e influences of the candidate a ttrib u te s which may explain who wins and loses in the more recent Supreme Court elections compare to the e a r lie r period? The fourth hypothesis answered th is question by re fle c tin g the changes in the rules of the game brought into e ffe c t by the Michigan Constitution of 1963 and 64 national trends. 26 I believe the influence of two candidate a t t r i ­ butes were negatively affected . They are the candidate's incumbency status and strength of the p o litic a l party nominating the candidate. The following are my reasons fo r focusing on these two a t t r i ­ butes. F ir s t, n a tio n ally and w ithin Michigan i t is becoming more f r e ­ quent and easier to create a p o litic a l party in order to place a special candidate on the b a llo t. I t would seem th a t th is would have an impact on both of the candidate a ttrib u te s ju s t mentioned, espe­ c ia lly incumbency. The creation of a new party enables a b a llo t position fo r a candidate who may be able to e ffe c tiv e ly challenge the ju d ic ia l incumbents who regu larly capture th e ir p arty's nomination. Remember, in Michigan a nonincumbent candidate must be nominated by a p o litic a l party in order to run fo r a Supreme Court seat. Second, the la s t several national elections have revealed that in a strongly challenged election incumbency is not the automatic tic k e t i t once seemed to be. This recent "Throw-Out-The-In's" psy­ chology of some o f the voters may have been extended to ju d ic ia l elec­ tions and has an impact on the Michigan voters' desire fo r s t a b ility in the ju d ic ia ry . This especially may be a p o s s ib ility as more voters in Michigan come to acknowledge that the courts are in the p o litic a l arena. Th ird, the influence of the strength of the candidate's p o l i t i ­ cal party may be lessened in relationship to the other candidate a ttrib u te s due to the provision since 1963 which permits a ju d ic ia l incumbent to f i l e an a ffid a v it of candidacy. A judge may even have established h is/h e r incumbent status by the viable a lte rn a tiv e of 65 creating a p o litic a l party in which she/he was the sole candidate. Therefore, a candidate who has other strong bases of support may e ffe c tiv e ly by-pass the established p o litic a l p a rties. 27 A fourth reason is that the degree of p o litic a l party ident if ic a t io n in Michigan is decreasing; as i t is n a tio n ally . 28 This may have altered the previous balance of party supporters and non­ id e n tifie rs comprising the Michigan Judicial electo rate. Therefore, the fourth hypothesis is: Hypothesis IV : For the Supreme Court elections from 1964 to 1980, the incumbency status of a candidate and strength of the p o litic a l party nominating the candidate have less of an influence than during the 1948 to 1963 period. Techniques and c r it e r i a . Techniques are available to inves­ tig a te two aspects of the decreased influence of the two candidate a ttrib u tes specified in th is hypothesis. I determined the existence of a decrease in the influence of the a ttrib u te s and also whether that decrease resulted in a change in the relationship of the influence of a ll of the a ttrib u te s . F ir s t, the amount and direction of the change from the f i r s t period to the second is indicated by the c o e ffi­ cients fo r the binary variables created in the dummy variable tech­ nique. I used a one-tailed t - t e s t (alpha set a t .05) to determine the s ta tis tic a l significance of the decrease in the re la tiv e influence of the two candidate a ttrib u te s . Second, a change in the re la tio n ­ ship of the influence of the candidate attrib u te s to each other is indicated by a difference in the rank ordering of the standardized co efficients fo r the two periods. The c rite rio n was that both of the 66 hypothesized candidate a ttrib u te s had to have a lower rank in the second time period than in the f i r s t . Hypothesis V In th is la s t hypothesis, I explore differences between the re la tiv e influence of Court of Appeals candidate a ttrib u te s and Supreme Court candidate a ttrib u te s . The nomination process fo r the Court of Appeals candidates does not provide a formal role fo r Michi­ gan's p o litic a l p a rties. Therefore, the Court of Appeals elections are more lik e nonpartisan elections than are the Michigan Supreme Court elections. 29 Another difference is that the Michigan Court of Appeals elections have less media coverage than the Supreme Court elections. Thus, less information about the Court of Appeals con­ tests is availab le to the ju d ic ia l voters. Three studies provide guidance on how the re la tiv e influence of the Court of Appeals candidate a ttrib u te s may d if f e r from the Supreme C ourt's. In a nonjudicial study, Gerald Pomper found that nonpartisan elections make ethnic factors the ch ief basis of p o litic a l division while partisan elections incorporate ethnic factors into more general cleavages. In a recent a r t ic l e , Hannah also indicated that ethnic names, as well as sex, were cues used by ju d ic ia l voters in elections which lacked information. Dubois' research found name id e n tific a tio n and incumbency as strong cues in nonpartisan states. Therefore, the f i f t h hypothesis is: 30 67 Hypothesis V: In contested Court of Appeals elections from 1970 to 1980, the ethnic appeal of a candidate's name and a candidate's unearned name fa m ilia r ity have greater re la tiv e influence than in Supreme Court contests fo r the same period. Techniques and c r ite r io n . I standardized the maximum lik e ­ lihood estimate co efficie n ts fo r the re la tiv e influence of the candi­ date a ttrib u te s in order to compare the Court of Appeals to the Supreme Court. Standardization enabled a rank ordering of the a t t r i ­ butes in both models to be meaningful. pared. These rank orders were com­ The c rite rio n was th at fo r an a ttrib u te to display a greater influence in one model it s rank order had to be higher than in the other model. Summary Chapter I I presented the methodology fo r an empirical analysis of several unanswered questions about Michigan's appellate Court elections. What is the r e la tiv e influence of each of the bases of candidate competition which may explain the election outcome fo r each of the Michigan appellate courts? What are the differences in the re la tiv e influence of the bases of competition fo r Supreme Court elections fo r the two time periods—before and a fte r the adoption of the Michigan Constitution of 1963? What are the differences in the re la tiv e influence of the Court of Appeals and Supreme Court bases of competition fo r the same period? F ir s t, I discussed the research design used to answer these questions—a case study. Second, the data set was described. It contains a ll Michigan Supreme Court elections fo r the period from 68 1948 to 1982 and a ll contested Michigan Court of Appeals general elections from 1970 to 1982. Third, the fiv e research hypotheses, techniques to te s t them, and c r ite r ia were elaborated. The major technique used was P ro bit, an ordinal analog of regression. It provided estimates fo r the re la tiv e influence of the candidate a t t r i ­ butes and s ta tis tic s fo r evaluation of the models. For the tests of significance, alpha was set a t the conventional value of .05. Results of the analysis of Michigan Supreme Court and Court of Appeals elec­ tions follow in Chapters I I and IV , respectively. FOOTNOTES— CHAPTER I I V. 0. Key, J r ., American State P o litic s : An Introduction (New York: Alfred A. Knopf, 1956), Chapter 7; and Eugene Lee, The P o litic s of Nonpartisanship (Berkeley: University of C alifo rn ia Press, 1960), p. 6. 2 The use of aggregate voting data was introduced by Rice and perfected by V. 0. Key, J r. Ranney cautioned that the use of aggre­ gate data cannot provide predictions about individual voting behavior. Stuart Rice, Quantitative Methods in P o litic s (New York: Alfred A. Knopf, 1928); V. 0. Key, J r . , A Primer of S ta tis tic s fo r P o litic a l Scientists (New York: Thomas Y. Crowell, 1966); and Austin Ranney, "The U t ilit y and Limitations of Aggregate Data in the Study of Electoral Behavior," Essays on the Behavioral Study of P o litic s , ed.: Austin Ranney (Urbanal University of Illin o is Press, 1962), pp. 91102. 3 V. 0. Key, J r ., Southern P o litic s (New York: Vintage Books, 1949), p. 421. Because most of the ju d ic ia l elections were m u ltiwinner, the vote cast fo r an o ffic e was approximated by dividing the to tal vote by the number of seats. 4 A basic tenet in the behavioral ju d ic ia l lite ra tu r e is that appellate courts make p o licy, although there is disagreement as to the extent. For a lucid treatment of the policy making of a national appellate court, see Harold J. Spaeth, Supreme Court Policy Making: Explanation and Prediction (San Francisco: W. H. Freeman and Company, 1979). At the state le v e l, see Herbert Jacob and Kenneth Vines, "Studies in Judicial P o litic s ," Tulane Studies in P o litic a l Science 8 (New Orleans: Tulane U niversity, 1962). For lite ra tu r e specific to Michigan, see Glendon Schubert, "The Packing of the Michigan Supreme Court," Quantitative Analysis of Judicial Behavior (Glencoe: The Free Press, 1959), pp. 129-41; and Sidney Ulmer, "The P o litc ia l Party Variable in the Michigan Supreme Court," Journal of Public Law 11 (1962): 353-62. An additional study which also has mean voter turn-out rates in Michigan (see Figure 2.1) is Susan B. Hannah, "An Evaluation of Judicial Elections in Michigan, 1948-1968" (Ph.D. d issertatio n , Michigan State U niversity, 1972). ^For th is and the remaining references in Chapter I I to con­ s titu tio n a l provisions and elections laws, see Michigan State Constitu­ tion (1963), A rtic le 6; and Laws Relating to Elections, compiled by the Michigan Secretary of S tate, Chapters 18 and 19. 69 70 Hannah, "Evaluation." This exploratory model was only one component of her analysis which was the f i r s t comprehensive evaluation of Michigan ju d ic ia l elections. Hannah did not update the Supreme Court model nor extent i t to the Court of Appeals elections. Instead, her recent research has concentrated on other Michigan ju d ic ia l elec­ tion issues, such as the degree of competition and voter p a rtic ip a ­ tio n . See, "Competition in Michigan's Judicial Elections: Democratic Ideals vs. Judicial R e a litie s ," Wayne Law Review 24 (July 1978): 12671306; and "Voting in Local Judicial Elections: The Case of the Faithful E lectorate," paper delivered a t the 1982 Annual Meeting of the Midwest P o litic a l Science Association, Milwaukee, Wisconsin. ^The following are the sources fo r each Factor: Factor I: Biographical m aterial in the Michigan Manual Factor I I : State of Michigan, O ffic ia l Canvass of Votes fo r the election concerned. Factor I I I : Michigan Manual, Martindale-Hubbell Legal (s ic .) Directory Factor IV: Same as Factor I I I Factor V: Michigan State Bar Journal Roster, Michigan Manual, Martindale-Hubbell Legal Directory Factor VI: Same as Factor V Factor V II: Same as Factor V. Quoted from Footnote 6 in Hannah, "Evaluation," p. 214. O °See my discussion of the studies by Eisenstein, Heiberg, DuBois, and Barber fo r th e ir recognition of the role played by name fa m ilia r ity in Chapter I , pp. 13-20. g See my Footnote 8 in Chapter I . Several other lik e ly v a r i­ ables were also not included in my model. Often the b a llo t position of a candidate's name is a consideration in studying electo ral success. In Michigan the names are alphabetically rotated by pre­ c in c t. James Chapman of the Elections O ffic e , Secretary of S tate, indicated in a November 3, 1981, phone conversation th at th is has been in e ffe c t fo r a t least the la s t th ir t y years. My model also does not include a variable fo r gender or Blacks because there were too few cases in my data set. This was also true in regard to the role of appointment. 10When candidate names are s im ila r, the occupations fo r those candidates are indicated on the b a llo t. This exception occurred in the 1959 Supreme Court contest with Kenneth Cole and Maurice Cole as candidates. For a discussion of b a llo t designations which includes the appropriate legal c ita tio n s fo r Michigan, see Maurice Kelman, "B allot Designations: Their Nature, Function and C o n s titu tio n a lity ," 12 Wayne State Law Review (Summer 1966): 777. 71 Richard G. Niemi and Herbert F. Weisberg, Controversies in American Voting Behavior (San Francisco: W. H. Freeman, 1976), p. 243. For an update on the lite ra tu r e on incumbency, see Herbert Asher, "Voting Behavior Research in the 1980s: An Examination of Some Old and New Problem Areas," paper delivered a t the 1982 Annual Meeting of The American P o litic a l Science Association, Denver Hilton H otel, September 2-5, 1982, pp. 43-50. For incumbency at the state le v e l, see Malcolm E. Jewell and David M. Olson, American State P o litic a l Parties and Elections (Homewood, I I . : Dorsey Press, 1978), pp. 237244. 12 The two conceptualizations of incumbency provide an oppor­ tu n ity to te s t two models. The naive model used in this te s t w ill conceptualize incumbency as in flu e n tia l only in its immediate state. I t was coded as a "1" i f the candidate was the incumbent and "0" otherwise. A te s t of the difference in the -2 Times Log Likelihood Ratio was performed on both models. The model with the higher sig n ific a n t (alpha equal to .05) chi-square is preferred. 13 See Table A .2. In nonpartisan elections religious iden­ t i t y and religious groups can perform some of the m obilization and cue-giving functions performed by p o litic a l p a rtie s . However, since the religious a f f ilia t io n of candidates is less frequently known to the whole ele cto ra te, the bulk of the impact is more lik e ly to come from the organizational resources and communication channels religious groups provide. See W illis D. Hawley, Nonpartisan Elections and the Case fo r Party P o litic s (New York: John Wiley and Sons, 1973), pp. 95-96. For the greater p articip atio n by Catholics, see Robert Lane, P o litic a l L ife (Glencoe, I I . : Free Press, 1959), p. 236, 244-247. ^Hawley cites studies of nonjudicial elections which in d i­ cate th at the ethnic id e n tity of p o litic a l candidates is an important resource among persons of the same ethnic background. See footnotes 24 and 25 in Hawley, Nonpartisan, p. 103. In 1940, Michigan was f i f t h in the country fo r w hite, foreign born ele cto ra te, comprising 16 percent of Michigan's electo rate. See James K, Pollock and Samuel Eldersveld, Michigan P o litic s in T ran sitio n , University of Michigan Government Studies #10 (Ann Arbor: University of Michigan Press, 1942), pp. 5960 fo r the counties with a high ethnic percentage. ^Robert Lane, P o litic a l L if e , p. 236. In an August 23, 1982, conversation with Archie Fraser, State Public Adm inistrator, re tire d , he indicated that the Iris h name popularity in Michigan started in the early 1930s. 16 In partisan elections, the popular vote fo r lesser offices is more party determined than fo r the top of the ballot--Presidency. P h ilip Converse et a l . , " S ta b ility and Change in 1960: A Reinstating Election," American P o litic a l Science Review 55 (June 1961): 273-280. 72 In 1970 Robert Danhof was a candidate in the Second D is tric t Court of Appeals and he received both the Ingham County Democratic Party and Ingham County Republican Party endorsement. As another example, in the 1978 Second D is tric t Court of Appeals primary and Macomb County Democratic Party endorsed Walter Cynar while the Genessee County Democratic Party endorsed Luke Quinn. Information confirmed by Paul Mooradian, Former Assistant to the Chair, Michigan Democratic Party. 18 See Table A .2 fo r the sources I used to determine the eth­ n ic ity of a candidate's name. The following are the percents fo r the countries and the appellate candidates easily id e n tifie d with each ethnic groups: Poland 255,467 3.3% Germany 234,183 2.9% U.K.--Scot 179,826 2.3% Ita ly 120,363 1.5% Netherlands 85,797 1.1% USSR 77,441 1.0% Hungary 46,811 .6% Finland 45,671 .6% Sweden 44,991 .6% Austria 43,675 .6% Czech 38,475 .5% Ireland 34,527 .4% Yugoslavia 30,736 .4% Walter Cynar, Roman Gribbs, and Meyer Warshawsky Emerson R. Boyles, George E. Bushnell, Ross Campbell, Maurice F. Cole, Neil E. Reid, Gary McDonald, Louis McGregor, Barbara Mackenzie, Edward M. Sharpe Robert Danhof, John R. Dethmers, John D. Voelker Zolton Ferency and Stephen J. Roth John Swainson Thomas Brennan, Vincent Brennan, Michael Cavanagh, E. T. F itz ­ gerald, John W. F itzg erald , Michael Hegarty, T. G. Kavanagh, T. M. Kavanagh, Michael K elly, Harry K elly , Richard Maher, James McLaughlin, Joseph Moynihan, John Murray, Michael O'Hara, Dorothy C. R ile y , James L. Ryan, Theodore Ryan, Joseph A. S ullivan , Joseph B. S ullivan , Joseph Swallow, and Daniel Walsh 73 Population s ta tis tic s obtained from Anthony V. Rizzo, Atlas of Michi­ gan's Foreign Born Population (Dearborn, M i.: Free World Press, 1968]"! 19 I used the Probit program brought to Michigan State Uni­ ve rs ity by John A ldrich. Logit was unavailable. 20 2 2 P ro bit's estimated R is not as meaningful as the R from le a s t squares regression because the residual sum of squares is set a t the number of cases. Therefore, I avoided use of tests of sig ­ nificance which are based on the r2. 21 Charles W. Ostrom, J r . , Time Series Analysis: Regression Techniques (Beverly H ills : Sage Publications, 1978), pp. 58-59. 22 Robert Pindyck and Daniel L. Rubinfeld, Econometric Models and Economic Forecasts (New York: McGraw-Hill, 1976), pp. 77-84. 23Hannah, "Evaluation," pp. 204-5. 24 ai The formula B* = fL ” is from Richard D. McKelvey and y William Zavoina, "A S ta tis tic a l Model fo r the Analysis of Ordinal Level Dependent V ariables," Journal of Mathematical Sociology 4 (1975): 115. 25 Probit does not provide the information necessary to per­ form tests of significance fo r differences of co efficien ts as described in Jan Kmenta, Elements of Econometrics (New York: Macmillan, 1971), p. 372. 26 The o rig inal change to nonpartisan Spring elections in Michigan was an attempt to remove these o ffices from national and state partisan influence. O liver P. Williams and Charles R. Adrian, Four C itie s (P hiladelph ia, Pa.: U niversity of Pennsylvania Press, 1963), pp. 33-90. Believing th at th is resulted in a malleable small voter turnout, framers of the 1963 Constitution of Michigan switched the nonpartisan ju d ic ia l elections back to the F a ll. See A lbert Lee Sturm, Constitution Making in Michigan, 1961-62 (Ann Arbor: In s titu te of Public Adm inistration, U niversity of Michigan, 1963). 27 For example, in 1976 T. G. Kavanagh was a successful incum­ bent candidate who had f ile d an a f fid a v it. According to a November 3, 1976, D etro it News a r t ic le , the Democrats refused to nominate Kavanagh because of his refusal to campaign fo r the other Democratic Supreme Court candidates. Another interesting example is CharlesL. Levin's candidacies. A p o litic a l party, Non-Partisan Judiciary Party, was established ju s t fo r his 1976 successful Supreme Court candidacy. The party most lik e ly to nominate Levin was already committed. Then, in 1980 Levin file d an a f fid a v it of candidacy and won. 74 28 Data and le t t e r from Robert M. Teeter, President, Market Opinion Research on February 24, 1983. 29 Using Adrian's typology fo r nonpartisan electio n s, Michigan Supreme Court elections would be c la s s ifie d as Type I and Michigan Court of Appeals elections as Type I I I . Since the same election rules apply fo r the Court of Appeals and C irc u it Court, Hannah's research may be used as support fo r this c la s s ific a tio n of Michigan's appellate courts. Charles Adrian, "A Typology fo r Nonpartisan Elections," Western P o litic a l Quarterly 12 (1959): 449-58; and Hannah, "Evaluation," pp. 182, 210-212. 30 Gerald Pomper, "Ethnic and Group Voting in Nonpartisan Municipal Elections," Public Opinion Quarterly (Spring 1966): 96; Hannah, "Voting," p. 4; and DuBois, B a llo t, p. 89. CHAPTER I I I ANALYSIS OF SUPREME COURT MODELS Chapter Objectives The research objectives addressed in Chapter I I I are twofold. The f i r s t objective is to determine the re la tiv e influence of the bases of candidate competition which may explain the election out­ come fo r the Michigan Supreme Court. The second objective is to com­ pare the re la tiv e influence of the bases of Michigan Supreme Court candidate competition between two time periods. Michigan Supreme Court data fo r elections from 1948 to 1982 are used to evaluate the f i r s t four research hypotheses discussed in Chapter I I . in the Appendix. See Table A .6 Therefore, Chapter I I I is divided in to : f i r s t , pre­ senting the results from testing the updated Hannah model and second, reporting the results from testing the hypothesized models. This chapter concludes with a summary of the findings fo r the highest appellate court in Michigan— the Supreme Court. Hannah Model Susan B. Hannah's prelim inary study of Michigan ju d ic ia l elec­ tions from 1948 to 1970 contained a model of partisan and nonpartisan competitive bases fo r Supreme Court candidates. Does her model explain Supreme Court election results fo r the period since 1970 with the same a b ilit y as fo r the 1948 to 1970 period? 75 The findings from a 76 te s t of the f i r s t research hypothesis provide an answer to this question. Hypothesis I : The bases of competition fo r Michigan Supreme Court elections as id e n tifie d by Hannah fo r the period 1948 to 1970 are s t i l l bases of competition fo r the period from 1971 to 1980. Analysis of the results from updating the Hannah model is divided into two sections. Discussed f i r s t is the a b ilit y of the updated model to co rrectly predict the Supreme Court winners candidate's to ta l score. from the Discussed next is the a b ilit y of Hannah's model to co rrec tly predict Supreme Court winners when the whole time period is divided into two periods--before and a fte r the adoption of the Constitution of 1963. Updated Period Hannah's model was tested with the elections from 1971 to 1980. Her op eratio nalization of the bases of competition, scoring procedure, and s ta tis tic were duplicated. A to ta l score was deter­ mined fo r each candidate by adding the points fo r each p o sitively scored a ttrib u te . Candidates with the highest to ta l scores in each election were predicted to the score column. win and th is was indicated with a star in The actual winners were designated with a star in the candidate-party column. Table 3.1 presents the Hannah model fo r Michigan Supreme Court elections from 1948 to 1980. The model's a b ilit y to co rrectly pre­ d ic t Supreme Court winners fo r the whole period, from 1948 to 1980, is 85 percent. This s t a tis tic is very close to the 86 percent pre­ dicted co rrec tly fo r the o rig in al period--from 1948 to 1970. The Table 3 . 1 . --T a b u la tio n o f candid ate advantage in Supreme Court e le c tio n s , 1943-1980 Candidate-Party Incumbent n j. Party State o ffic e Local O ffice n • ^ P n vate Section Ethnic Score Hannah's, 1948-1970 April 1948 Bushnell (D )* Sharpe (D}*1 Moore (R )l Rigney (R )1 + + - + H- - + + + + + + - - 12* 7 13 13* A pril 1951 Boyles (R )*1 Reid (R )* Ryan (DH Lee (D )l + + - + + - + + — + + + + - + - 25* 17* 5 6 Nov. 1952 Adams (D )*1 King (R )l + - + - + - + - + - + - 25* 2 April 1953 Dethmers (R )* K elly (R )*1 Adams (D )1 Smith (D) + + - + + - + + + + + + + - + + - + - + - 22* 21* 19 5 April 1955 Black (D )*1 Carr (R )* Roth (D) Brake (R )l + + - + + + + + + + + + + + + - + - 19* 22* 10 13 - Table 3 . 1 . — Continued Candidate-Party Incumbent Nov. 1956 O'Hara (R )1 T. Smith (D )*1 Edwards (D )* Simpson (R) April 1957 T. Smith (D )*1 T. M. Kavanagh O'Hara (R )l Childs (R )l Voelker (D )* Moynihan (R) . (D )* Party O ffice Private Section - - - 4 4 4 - - + 4 - + + - - 4 4 - 4 - 4 4 4 4 - 4 4 - - - - - - - - - - + 4 _ 4 - - - 4 - 4 + - A pril 1959 Edwards (D )* Voelker (D )* Baldwin (R) K. Cole1 , M. Cole (R )1 + 4 + - 4 Nov. 1960 i Souris (D )* Breakley (R) + - 4 + + A pril 1961 Dethmers (R )* Kelly (R )*l McLaughlin (D )1 Boehm (D) o ffic e Ethnic 4 3 - - 22* 19* 4 - 4 21* 16* 3 2 4 4 4 _ 4 4 - 4 4 - - 4 _ 4 4 - 4 — 4 4 - - 4 4 4 - 4 - - - - - - - 4 4 4 4 - - - 4 4 4 - - 4 4 - 4 - - - 4 4 4 - 4 - - 4 - Score 20* 7 20* 21* 3 0 7 21* 4 16* 22* 16 12 Table 3 . 1 . --C o n tin u ed Candidate-Party Nov. 1962 . O'Hara (R )*1 Adams (D)* 0. Smith (D )* McGregor (R )l April 1963 P. Adams (D )* Black (D )* l Holbrook (R )1 R. Smith (R) Nov. 1966 T. M. Kavanagh (D )* T. Brennan ( R )*I 0. Smith (D )l Warshawsky (R) Nov. 1968 T. G. Kavanagh (D )* O'Hara (R) Nov. 1970 Williams (D )*l Swainson (D )*l Dethmers (R )1 Piggins (R )l Incumbent Party + + - + + - + + - + - o ffic e O ffice + + + + + + + + + + + + + + + + + + + + + + + + + + Ethnic + + + + Score 9 18* 13* 11 17* 25* 4 4 + 17* 13 17* 16 + + 13* 10 + + + + + + + - Section + + + Private + + 7 12 22 * 15* Table 3 . 1 . — Continued Candidate-Party Incumbent Party State O ffice O ffice Private Section Ethnic Score R eplication, 1971-19802 Nov. 1972 Coleman (R )* Levin ( - ) * Gilmore (D) Evans (D) Thorburn (R) Nov. 1974 J. Fitzgerald (R )* T. M. Kavanagh (D )* M ille r (R) Moody, J r. (D) Nov. 1976 T. G. Kavanagh ( - ) * Gribbs (D) Swallow (R) Nov. 1976 Vac Ryan (R )* Kaufman (D) Nov. 1976 Vac Moody, J r. (D )* Lindemer (R) - + - - - - - - + - - - - + - + + + - - + - + - - - - + - - - + - - - - + - - - + + + + + + - - + + + + - - - + - + + 10* 10 11* 7 10 - - + - 18* 17* 10 6 + + + + + - + + + 14* 13 5 - + + + + + + 14* 13 + + + - + + + + + + + + - + 12 19* Table 3 . 1 . --C o n tin u ed Candidate-Party Incumbent Party + + + Nov. 1978 Williams (D )* Ryan (R )* McDonald (D) G ilb e rt (R) Nov. 1980 Coleman (R )* Levin ( - ) * Burch (D) Hegarty (R) S ullivan (D) State O ffice + - - - + + + + - - - - - + - - - + + + - - - Local O ffi ce + + + + + P rivate Section - Ethnic + + - - - + + - + - - + - - - - + - - - + + + + + - + + Score 14* 20* 5 12 17* 17* 7 14 12 Percentage of Winners Predicted Correctly Replication Pre-Period 1948-1970 1971-1980 1948-1963 % Predicted by Party 68 45 % Predicted by Score 86 82 Hannah's Post-Period Total 1964-1980 1948-1980 74 44 62 91 75 85 ♦Denotes winner of electio n when under candidate-party column and predicted winner when under score column. ICorrection of Hannah data. 2 For the 1971-1980 period, i t was necessary to include candidates who were not nominated by a major p o litic a l party which was one of Hannah's re s tric tio n s . 82 percent predicted co rrec tly fo r ju s t the updated period, from 1971 to 1980, is 82 percent. The c rite rio n was that the value of th is s t a tis tic must be a t le a s t 81 percent. This c rite rio n was s a tis fie d . the f i r s t hypothesis was confirmed. Therefore, The bases of competition fo r Michigan Supreme Court elections as id e n tifie d in the Hannah model are s t i l l bases of competition fo r the period from 1971 to 1980. Further analysis of the expanded data revealed interesting d i f ­ ferences from the e a r lie r study. F ir s t, a s lig h tly d iffe re n t conclu­ sion may be drawn fo r the two periods through investigation of a subset--the Supreme Court winners. Hannah organized the Supreme Court candidate winners according to th e ir to ta l score—from the highest points to lowest. The winners of the 1971 to 1980 Supreme Court elections were so organized. Table 3.2 presents the 1971 to 1980 winners and the percentages fo r both periods. Hannah's analysis of the percentage of winners possessing various combinations of the candidate a ttrib u te s led to the following conclusion. Incumbency was the strongest candidate advantage, but i t was ra re ly s u ffic ie n t to provide a v ic to ry . The other a ttrib u te s providing a candidate an advantage were the strength of the p o litic a l party which nominated the candidate and previous p o litic a l o ffic e experience— state or lo c a l. The three candidates who defeated incum­ bents had strong local and ethnic appeal as w e ll. A comparison of the percentages fo r the combinations of a ttrib u te s fo r the two time periods supports the conclusion that the relationships fo r the d iffe re n t periods are s im ila r, except th at in Table 3 . 2 . --Supreme Court advantages o f w in n ers, 1971-1980 Winner J. J. T. M. C. T. J. G. B. M. C. State O ffice Incumbent Party + + + + + + + + - + + + - + - - - - + - Ryan (R-78) Fitzgerald (R-74) M. Kavanagh (D-74) Coleman (R-80) Levin (--8 0 ) G. Kavanagh ( - -76) Ryan (R-76V) Williams (D-78) Moody, J r. (D-76V)* Coleman (R-72) Levin ( - -72) Combination of A ttrib u tes Incumbent, Party, State Incumbent, Party, Local Incumbent and Party Incumbent and State Incumbent and Local P arty, State or Local Incumbent and Section Incumbent ♦Defeated an incumbent. - + + - Local O ffice P rivate + + + + + + + + - - - + + + + Hannah's, 1948-1970 26% 38 23 23 8 20 31 77 - Section + - + + + + + + Ethnic + + + + + + - + Score 20 18 17 17 17 14 14 14 12 10 10 R eplication, 1971-1980 0% 27 27 18 27 18 45 63 84 the period from 1971 to 1980 the advantage from a candidate's local public o ffic e experience was more in flu e n tia l than a candidate's state public o ffic e experience. Also in that period, the only candidate who defeated an incumbent did so with an advantage from the strength of the nominating p o litic a l party, local o ffic e experience, and sec­ tional support, but not ethnic appeal. The advantage derived from the combination of incumbency and sectional support was not reported by Hannah. The percentage of the winners possessing th is combination fo r the o rig inal period and updated period is 31 percent and 45 percent, respectively. For both periods, the percentage of winners having th is combination of bases of competition is higher than fo r the combina­ tions reported above. Second, an important change in the influence of another of the candidate a ttrib u te s exists between the orig in al period, from 1948 to 1970, and the updated period, from 1971 to 1980. In Hannah's explora­ tion of the bases of competition, she noted th at a high percentage of the winners could be co rrectly predicted solely on the basis of th e ir having been nominated by the p o litic a l party which won the state o ffic e with the greatest voter p a rtic ip a tio n in th at e le c tio n . Table 3.2 presents the calculations fo r both time periods. While the percent predicted co rrectly from the to ta l score fo r each candidate is very s im ila r between those two periods, a consider­ able drop--23 percentage points--occurs when the winners are predicted from only the candidate's ratin g on the p o litic a l party fa c to r. C le a rly , the influence of the strength of the candidate's p o litic a l 85 party has not remained stable. For the period since 1970, both of the Supreme Court candidates who file d an a ffid a v it of candidacy were winners. The new constitution provided incumbents th is a lte rn a tiv e . Also, one of the two candidates who created a p o litic a l party to q u a lify fo r the b a llo t during th is period was a winner. These results warranted fu rth e r analysis of Hannah's model with the data divided into two periods to r e fle c t the change in Michigan's constitutions. Pre- and Post-Periods Table 3.2 also provides the values fo r the percent predicted co rrectly by the whole model and by only u t iliz in g the p o litic a l party variable with the data s p lit into the pre- and post-1963 Con­ s titu tio n periods. Reorganization of the data is inform ative. The complete Hannah model has a greater a b ilit y to co rrectly predict the winners during the 1948 to 1963 period than the 1964 to 1980 p e rio d 91 percent as compared to 69 percent. Also, note that the pre­ period's percentage is higher than the percentage fo r the o rig inal time period. the change in the model's This supports the b e lie f th at a b ilit y to predict winners is associated with the adoption of the new co nstitu tion . Further support fo r th is conclusion is obtained from focusing on the success rate fo r the incumbents.1 In the 1948 to 1963 period, only two incumbents lo s t th e ir re-e lectio n bid (10 percent of incum­ bents running), while in the la te r period, four incumbents lo s t (31 percent of incumbents running). This represents a 100 percent increase in the number of incumbents who lo s t. Also, in the e a r lie r 86 period only 50 percent of the election outcomes of the incumbents were in co rre ctly predicted. However, 75 percent were mispredicted in the period a fte r 1963. S im ilar results occurred with the analysis above was repeated fo r candidate scores based only on the p o litic a l party fa c to r. The 1948 to 1963 period has a higher percent predicted co rrectly than fo r the original time period. With the periods s p lit into the years from 1948 to 1963 and from 1964 to 1980, the percent predicted co rrectly plunged from 74 percent to 44 percent. A ll of the candidates before 1964 were nominated by a p o litic a l party. Since 1964, four candidates e ith e r ran without a party nomination or created a party fo r th e ir candidacy. Empirical evidence provided confirmation of research Hypothe­ sis I- - t h e bases of competition id e n tifie d by Hannah are s t i l l bases of competition. However, additional manipulation of the data on candidate bases of competition fo r Supreme Court elections during the period from 1948 to 1980 c le a rly indicated the necessity fo r creating two d is tin c t time periods. The electoral changes associated with the adoption of the 1963 Michigan Constitution suggest that 1964 is an appropriate dividing point. Also, th is analysis of the Hannah model led to the conclusion th at an investigation of the bases of candidate com petition--especially incumbency, strength of the p o litic a l party which nominated the candidate, and support from the candidate's sectio n --w ith a newer an alytical technique would be p ro fita b le . 87 Hypothesized Model Next presented is an expanded and reconceptualized model of the candidate attrib u te s which may explain who wins and who loses in a contested Michigan Supreme Court e le ctio n . The following section discusses the results from a set of s ta tis tic a l tests of two models which were conducted in order to te s t Research Hypothesis I I . Hypothesis I I : There is a positive relation ship between winning an appellate court election in Michigan and the incumbency status of the candidate, strength of the candidate's p o litic a l party, state and local p o litic a l o ffic e experience of the candidate, support from the candidate's geographical section, unearned fa m ilia r ity of the candidate's name, Iris h or other ethnic appeal of the candidate's name, and support from the candidate being recognized as a Catholic. Tests of Models The f i r s t question resolved by the -2 Times Log Likelihood Ratio Test of Differences is which conceptualization of incumbency more adequately explains the election outcome. 2 In one conceptualiza­ tio n , incumbency is in flu e n tia l only in its immediate s ta te . The other conceptualization o f incumbency incorporates the influence derived from the length o f the incumbency or image developed from a series of b a llo t appearances. Hannah had remarked th at two variables— incumbency and strength o f p o litic a l party—enabled a loose categoriza­ tio n of the Supreme Court election outcomes. Therefore as a s im p lifi­ cation, only those two variables were used in the -2 Times Log Like­ lihood Ratio Test of Differences. p o litic a l party varia b le . The constrained model contained the Each of the unconstrained models contained one of the conceptualizations of incumbency and the party v a ria b le . 88 The c rite rio n was that the model with the higher s ig n ific a n t chi-square fo r the difference in the Log Likelihood Ratio provides greater explanatory power than the other model. The naive model in which incumbency was conceptualized as in flu e n tia l only in its imme­ diate state has a su b stantially higher chi-square—approximately twentysix to fourteen. accepted. 3 Based on the c r ite rio n , the naive model was Also, a conventional rule in model building is th at when a ll else is equal, the more parsimonious model is to be preferred. There­ fo re , in the rest of th is research, incumbency was conceptualized as the immediate status of the candidate and operationalized as a dichotomy— scored as a one when present and as a zero otherwise. For the period analyzed, the simple incumbent designation b e tte r explains appellate court election outcomes even though both conceptualizations of incumbency s ig n ific a n tly added explanatory power to the model. I t is possible that the impact of the incumbency designation on the b a llo t overshadows the influence of the candi­ date's length of incumbency or image developed from a series of b a llo t appearances. However, i t also may be possible th at my conclusion about which conceptualization is best applies only fo r appellate court elec­ tions from 1948 to 1980. That period contained a few unusual elections which suggests that the more complex conceptualization of incumbency should not be to ta lly rejected fo r future study.^ A second question is whether certain of the candidate a t t r i ­ butes in the hypothesized model (unconstrained model) add s ig n ific a n tly to the explanatory power of th at model. A constrained model was developed as a means to conduct the -2 Time Log Likelihood Ratio Test of Differences described above. Selection o f the candidate a t t r i ­ butes fo r the constrained model was guided by the preliminary in v e s ti­ gation of the Hannah model and the media's perpetual focus on three candidate ch aracteristics: incumbency status, strength of p o litic a l party, and sectional support. Therefore, the constrained model posited th a t the incumbency status of the candidate, strength of the candi­ date's p o litic a l party and the support from the candidate's section are p o s itiv e ly related to winning a Supreme Court e le ctio n . Tables 3.3 and 3.4 present the Probit estimates of the models and the results of th is te s t conducted fo r two time periods— before and a fte r the adoption of the Michigan Constitution of 1963, respectively. The re la tiv e influence of the candidate a ttrib u te s is addressed in the discussion of Hypotheses I I I and IV. The hypothesized model (unconstrained) in both of the periods s a tis fie d the goodness of f i t c r it e r ia . The -2 Times Log Likelihood Ratio fo r both of the unconstrained models is s t a tis tic a lly s ig n if i­ cant. The constrained models fo r both of the periods also s a tis fie d these c r it e r ia . I t may be concluded th at both the unconstrained and constrained models adequately explain Supreme Court election outcomes during the periods under study. needed. Thus, additional information is The -2 Times Log Likelihood Ratio Test of Differences can provide the required information on whether the additional candidate a ttrib u te s in the unconstrained model add s ig n ific a n tly to the explana­ tory power of the model. Table 3 . 3 . — T est o f two Supreme C ourt m odels, 1948-1963 Constrained Model Unconstrained Model A ttrib u tes MLE Incumbency Party Sectional State Local Ethnic Catholic Constant Est. R2 -2 x LLR PPC rs 2.251 .386 .135 -.0 68 -.061 -.3 17 -.3 14 SE 1.004 .301 .078 .162 .069 1.212 1.142 MLE/SE MLE MLE* SE MLE/SE 2.244** 1.280 2.144 .303 .118 2.104 9.293 4.800 .816 .178 .050 2.622** 1.699** 2.368** 1.745** -.421 -.876 -.2 62 -.275 -26.014 .985 5 2 .3 5 3 ^ 88.235 .762** .739 xb -2 x LLR(d iffe re n c e ) ~ 1.262 -21.811 .979*** 51.091** 90.196 .803** .783 (Chi square, 4 degrees of freedom) N = 51 ♦Indicates s ig n ific a n t standardized MLE ♦♦Indicates significance a t .05 level ? ? ? (C r itic a l values, Z = 1.64, r g = .305, x 7 = 1 4 .1 , x 4 = 9 .4 9 , x 3 = 7.81) O ♦♦♦For comparison: lin e a r regression R = .629 Farrar-Glauber R2 = .586 as estimate of degree of m u ltic o llin e a rity . I do not consider th is a serious level in which the model should not be used. Table 3 . 4 . --T e s t o f two Supreme Court models, 1964-1980 Unconstrained Model Constrained Model A ttrib u tes MLE Incumbency Party Sectional State Local Iris h Other Ethnic Catholic Name Constant Est. R2 -2 x LLR Percent Predicted Correctly -.7 3 8 -.0 42 .117 .096 .100 1.409 -.6 5 2 -.3 8 3 -.3 1 0 SE .869 .027 .041 .122 .097 1.626 1.291 1.498 1.298 -5.085 .832 33.645** 82.927 .646** .562 Ab - 2 x LLR,(d iffe re n ce ) = 3.4 (Chi square, 6 degrees of freedom) MLE/SE MLE MLE* SE MLE/SE - .850 -1.561 2.865* .786 1.027 .867 -.5 05 -.2 56 -.239 -.437 -.032 .103 0 0 4.430 .645 .020 .032 - .678 -1.632 3.276** -3.611 .7 7 4 *** 30.168** 90.244 .795** .750 N = 41 ♦Indicates s ig n ific a n t standardized MLE ♦♦Indicates significance a t .05 level 2 2 2 (C r itic a l values, Z = 1.64, r s = .305, x g = 16.92, x 6 = 12.59, x ^ 3 ?»82 ***F o r comparison: lin e a r regression R^ = .229 Farrar-Glauber R2 = .085 as estimate of degree of m u ltic o llin e a r ity . I do not consider th is a serious level in which the model should not be used. 92 For both periods, -2 Times Log Likelihood R atl0( cj-jfference) is not s t a tis tic a lly s ig n ific a n t with alpha set a t .05. This is a cle ar signal that the additional candidate variables— length of the candidate's state and local public o ffic e experience, Iris h or other ethnic appeal of the candidate's name, and the candidate being recognized as a Catholic— do not add to the a b ilit y of the model to explain who wins and who loses Supreme Court elections from 1948 to 1980. Additional guidance is available from ex post forecasting. In the ex post fo recast, both models in co rrectly predicted a victory fo r Dorothy Comstock R iley in the 1982 e le ctio n . However, the con­ strained model was closer to co rrectly predicting her election outcome than the hypothesized (unconstrained) model. results w ill be discussed la te r . The ex post forecasting Therefore, with Ockham's Razor as the conventional wisdom, the parsimonious three variable model is used in the dummy technique discussed next. The th ird question is whether there exists a s ig n ific a n t d i f ­ ference before and a fte r adoption of the 1963 Constitution in the r e la tiv e influence of one or more of the candidate a ttrib u te s which may provide an electo ral advantage in Supreme Court elections. A dummy variable technique is availab le in which binary variables are added to a model in order to id e n tify the possible change in the re la tiv e influence of the candidate a ttrib u te s from the f i r s t time period to the second. The Probit estimates of the model and s ta tis tic s present in Table 3.5 provide empirical evidence to support my theoretical position 93 Table 3 . 5 . --Dummy technique Supreme Court model, 1948-1980 A ttributes MLE SE MLE/SE Incumbency 2.144 .818 2.622** -2.576 1.042 -2 .4 7 2 ** .303 .178 1.699** - .334 .179 -1 .8 6 3 ** .118 .050 2.368** - .014 .059 -21.811 10.699 -2 .0 3 9 ** 18.071 10.771 1.678** Incumbency x D* Party Party x D Sectional Sectional x D Constant D Est. R2 .965 -2 x LLR 81.463** Percent Predicted Correctly 90.217 r .801** s N = 9? *n = (* w*1en e^ectlon occurs a fte r 1963 <-0 otherwise **In d icates s ig n ific a n t at .05 level ( c r it ic a l values, Z = 1.64, r g = .305, x2 = 14.1) - .230 94 that the two periods under study are d iffe re n t. The binary variables fo r the strength of the candidate's p o litic a l party and the incumbency status of the candidate are s t a tis tic a lly s ig n ific a n t. Also, the binary variable fo r the constant is s t a tis tic a lly s ig n ific a n t. These results le g itim a tiz e separate estimation of the two time periods. The maximum likelih ood estimate co efficie n ts from the dummy variable tech­ nique estimation and from separate estimation fo r the two periods 5 should be e s sen tia lly equivalent. They are. Separate estimation of the two periods is required fo r th is study because i t provides the information necessary in order to rank order the re la tiv e influence of the Supreme Court candidate a ttrib u te s . Now, the la s t c r ite r ia fo r the te s t of Hypothesis I I may be addressed. The req u isite theoretical condition was th at the hypo­ thesized candidate a ttrib u te s p o sitively influence the pro b ab ility of a candidate's winning an appellate court ele ctio n . The c r ite r ia were that in a one-tailed t - t e s t each c o e ffic ie n t be positive and s ig n if i­ cant a t alpha equal to .05= In Table 3=3 and 3 .4 , the positive and s ig n ific a n t co efficie n ts fo r the re la tiv e influence of the candidate a ttrib u te s are id e n tifie d fo r each period. In the e a r lie r period, winning a Supreme Court election is p o sitively related to the incumbency status of the candidate, strength of the candidate's p o litic a l party, and support from the candidate's £ geographical section. Although Hannah concluded th at state and local public o ffic e experience were a ttrib u te s which advantaged a candidate, the length of such experience was not in flu e n tia l in the Supreme Court 95 elections from 1948 to 1963. Also, lacking s ig n ific a n t influence were the ethnic appeal o f a candidate's name and support from the candidate being recognized as a C atholic. A lack of s ig n ific a n t cases during th is period prevented estimation of the advantage derived from the unearned fa m ilia r ity of a candidate's name. The results are d iffe re n t fo r the period from 1964 to 1980. The only candidate a ttrib u te which continued to be p o sitively related to winning a Supreme Court election was the support from a candidate's geographical section. relevant changes: The new constitution contained two Supreme Court elections were moved from the Spring to Fall and provision fo r incumbents to f i l e an a f fid a v it of candidacy. Since 1963, the strength o f a candidate's nominating p o litic a l party and incumbency status of the candidate were not in flu e n tia l in deter­ mining the election outcome. The length of a candidate's state or local o ffic e experience, Iris h or other'ethnic appeal of the candi­ date's name, unearned fa m ilia r ity of a candidate's name, and being recognized as a Catholic also lacked s ig n ific a n t influence during the second period. In regard to Supreme Court electio n s, testing Hypothesis I I with data from two d is tin c t time periods produced mixed support fo r th is hypothesis. The positive relationship hypothesized fo r most of the bases of competition and winning an appellate court election was not substantiated. At the Michigan Supreme Court le v e l, the a ttrib u te s which did not provide an advantage fo r a candidate were the length of elected or appointed public o ffic e experience a t the state or local 96 le v e l, ethnic appeal of the candidate's name, and the candidate being recognized as a C atholic. I t is doubtful th at the unearned fa m ilia r ity of a candidate's name provides any advantage to a candi­ date. Two candidate a ttrib u te s had provided an electo ral advantage in the Supreme Court elections from the years 1948 to 1963. These were the strength of the p o litic a l party which nominated the candidate and the candidate's incumbency status. However, th e ir r e la tiv e influence decreased to such an extent th at in the second period they no longer provided a candidate an advantage. The only candidate a ttrib u te which supplied an electo ral advantage during the whole period studied was support from the candidate's geographical section. Therefore, I conclude the second research hypothesis was con­ firm ed, with two caveats. F ir s t, some of the a ttrib u te s did not pro­ vide a candidate an additional advantage to defeat the other Michigan Supreme Court contestants. Even though the hypothesized positive relation ship was not supported fo r a ll of the candidate a ttrib u te s , none of them exhibited a s ig n ific a n tly negative re la tio n s h ip . Second, although a ll the rest of the c r it e r ia were s a tis fie d by the hypothe­ sized Supreme Court model, a more parsimonious model was preferred because i t b e tte r s a tis fie d those c r it e r ia . Next, the results of the tests of Hypotheses I I I and IV explore the relation ship of the Supreme Court candidate a ttrib u te s . 97 R elatio n sh ip o f A ttrib u te s The focus of th is section is comparison of the re la tiv e in flu ­ ence of the candidate attrib u te s within each period and between the two time periods. The ju d ic ia l election lite r a tu r e reviewed in Chapter I consistently indicated th at during the 1948 to 1963 period, three candidate a ttrib u te s had a strong influence on who won and who lo s t. A candidate's incumbency status overwhelmingly was id e n tifie d as the prime fa cto r leading to an appellate court v ic to ry. The can­ didate's p o litic a l party a f f ilia t io n and state p o litic a l o ffic e expe­ rience were the other two a ttrib u te s . According to Hannah's study, the order in which these three a ttrib u te s were presented indicates the order of the magnitude of th e ir influence.^ Corroboration of Hannah's research finding may be provided in the conclusion drawn from testing Hypothesis I I I . Hypothesis I I I : For Supreme Court elections during the period from 1948 to 1963, the incumbency status of the candidate, strength of the candidate's p o litic a l party, and state p o litic a l o ffic e experience of the candidate are the candidate a ttrib u tes which have the strongest re la tiv e influence, respectively. The standardized maximum likelih ood estimates indicate the re la tiv e influence of the candidate a ttrib u te s in a form which allows comparisons. The c rite rio n fo r one standardized maximum likelihood estimate being larger than another was that the larger of the two standardized maximum likelihood estimates must be a t least 10 percent larg e r. Inspection of Table 3.3 reveals th at the relationship of the re la tiv e influence of the a ttrib u te s was not as expected. 98 F ir s t, although the candidate's incumbency status did provide a positive influence, i t was not the most powerful fa cto r as hypothe­ sized. The most in flu e n tia l of the candidate a ttrib u te s was the strength of the p o litic a l party which nominated the candidate. Its influence is more than four times greater than the influence of incum­ bency and obviously s a tis fie d the c rite rio n fo r one c o e ffic ie n t to be larger than another. Second, the length of a candidate's state public o ffic e experience—elected or appointed--did not even add s ig n ific a n tly to the explanatory power of the model. This a ttr ib u te , as measured in this study, can not be considered the second most powerful fa cto r to pro­ vide an advantage fo r a candidate in a Michigan Supreme Court election, A candidate a ttrib u te which was not included in Hypothesis I I I was one of the three variables having the strongest r e la tiv e influence. The support from a candidate's geographical section provided more than two times greater an advantage fo r a candidate to win than the candidate's incumbency status. Therefore, results from the te s t of research Hypothesis I I I with data from Supreme Court elections during the period from 1948 to 1963 c le a rly led to the conclusion that th is hypothesis was not con­ firmed. The advantage derived from being an incumbent was th ird strongest--not f i r s t . The strength of the candidate's nominating p o litic a l party was the most in flu e n tia l, rather than second. As measured, the length of a candidate's state public o ffic e experience had no influence over the election outcome. 99 Thus Hannah's conclusion about the strength of the bases of competition was not corroborated by the findings in th is study. Her conclusion was predicated on the magnitude of incumbency being seven points, party as six points, state o ffic e as fiv e points, and section as two points. Using data to estimate the co efficien ts fo r these variables revealed th at the re la tiv e influence of these candidate a ttrib u te s was not of equally decreasing u n its. The rounded standard­ ized maximum likelih ood estimates fo r these a ttrib u te s are two, nine, zero, and f iv e , respectively. The large differences between these values is obvious. Does th is difference between the models account fo r the two incorrect predictions in the Hannah model? Yes. Hannah's model erred in predicting Charles Rigney as winner of the Supreme Court seat rather than predicting a voctory fo r Edward M. Sharpe in 1949, while the Probit model showed Sharpe had a higher p ro b ab ility fo r a victory than a loss. Both models in correctly predicted a loss fo r Michael O'Hara in 1962. O'Hara defeated the recently appointed Supreme Court Justice Paul L. Adams. The p o litic a l party which nominated Adams was s lig h tly stronger a t the polls than O'Hara's. sectional support was much greater than Adams'. However, O'Hara's Because the Hannah model established such a low value fo r sectional support, th is model would have predicted a victo ry fo r Adams over O'Hara even i f Adams' incumbency had not been included in his score. Both of the errors in the Hannah model were on predicting loses fo r candidates who had greater sectional support than the incorrectly predicted winners. 100 Does the importance of the support from a candidate's section continue in the more recent period? This question w ill be investigated in the following discussion o f the te s t of the fourth hypothesis. How do the r e la tiv e influences of the candidate a ttrib u te s which may explain who wins and loses in Supreme Court elections during the period from 1964 to 1980 compare to the e a r lie r period? Hypothe­ sis IV addressed th is question by re fle c tin g the changes in the rules of the game brought into e ffe c t by the Michigan Constitution of 1963 and national trends as described in Chapter I I . I believe th at the re la tiv e influence of the candidate's incumbency status and strength of the p o litic a l party which nominated the candidate were negatively affected . Information on the a ttrib u te s of the serious Supreme Court candidates running fo r o ffic e from 1964 to 1980 were used to te s t Hypothesis IV . Hypothesis IV : For the Supreme Court elections from 1964 to 1980, the incumbency status of a candidate and strength of the p o litic a l party nominating the candi­ date have less of an influence than during the 1948 to 1963 period. Dramatic changes in the re la tiv e influence of the candidate a ttrib u te s are displayed in Table 3 .4 . decrease were analyzed. Two aspects of the hypothesized F ir s t, the two a ttrib u te s hypothesized to provide a candidate less of an advantage than e a r lie r have such a decreased influence th a t they are s t a tis tic a lly in s ig n ific a n t in th is period. The one-tailed t - t e s t of the co efficie n ts of these binary variables in the dummy technique equation substantiated a sig n ific a n t reduction in the influence of the candidate's incumbency status and strength of the candidate's nominating p o litic a l party. 101 Second, fo r the change in the relationship of the influence of the two candidate a ttrib u te s , the c rite rio n was that the rank of both of the a ttrib u te s had to be lower than in the e a r lie r period. rank orders are indicated in Table 3 .6 . The Obviously, th is standard was s a tis fie d . Table 3 .6 .—Rank order of Supreme Court candidate attrib u te s 1948-1963 A ttributes MLE* 1964-1980 Rank A ttributes MLE* Party 9.293 1 Sectional 4.430 Sectional 4.800 2 Party 0 Incumbency 2.104 3 Incumbency 0 State 0 State 0 Local 0 Local 0 Ethnic 0 Name 0 Catholic 0 Ethnic 0 Catholic 0 Rank 1 ♦Indicates s ig n ific a n t standardized MLE. In the period from 1948 to 1963, the strength of the p o litic a l party which nominated the candidate was ranked in f i r s t place and the incumbency status of a candidate was ranked as th ird in the rank order­ ing of the candidate a ttrib u te s which had a positive influence on the election outcome. In the more recent period, these two candidate a ttrib u te s were not even ranked as one of the variables providing an electoral advantage to a Supreme Court candidate. The only s ig n ific a n t 102 base of competition, sectional support, was ranked in f i r s t place. In te re s tin g ly , the standardized maximum likelihood estimate fo r the support from a candidate's geographical section remained stable between the two periods. I t would seem that the foundation fo r a candidate's competitive position was h is/h e r a b ilit y to e ffe c tiv e ly communicate he/she is the "local boy." An evaluation of the model suggested in Hypothesis IV can be conducted via ex post forecasting. The Supreme Court candidates in 1981 and 1982 were used to predict who won and who lo s t in those elections. The established standards were th at fo r the set of fo re ­ casted candidates the percent predicted correctly s ta tis tic must be at least .60 and lambda^ be a t lea st .50. Inspection of Table 3.7 id e n tifie s the only incorrect prediction was the forecasted victory fo r Dorothy Comstock R iley. The percent predicted co rrectly was .80. This is well above the c rite rio n set fo r the model. Also, lambda^ was .50 which is the reduction in error required of the model. There­ fo re , I conclude that th is model more than adequately predicts an election outcome which was excluded from the data used to estimate the re la tiv e influence of the Supreme Court candidate a ttrib u te s . The c rite rio n fo r both o f the analyzed aspects of a decrease in the re la tiv e influence of candidate a ttrib u tes was s a tis fie d . The decrease in the re la tiv e influence of a candidate's status as an incumbent and strength of the p o litic a l party which nominated the candidate was so drastic that both of these a ttrib u tes cannot be con­ sidered as providing a competitive advantage in a Supreme Court Table 3 . 7 . — F o recasting Supreme Court e le c t io n , 1982 Candidates A ttrib utes Party Sectional MLE M. Cavanagh B. Moody, J r. R. Campbel1 - .032 56. 56. 41. 41. 2. .103 53. 79. 16. 48. 10. 0 1 0 0 0 Incumbency - .437 Constant -3.611 Sum .059 P(Z = 1) .53 2.308 D. Riley -3.280 .99 .026 .00 P. Avery -2.642 .53 .00 Prediction 1 1 0 1 0 Outcome 1 1 0 0 0 Correct Predictions _ Total Predictions = .50 pn 104 election since 1963. Also, the values fo r the re la tiv e influence of the candidate a ttrib u te s , as suggested in th is hypothesis, enabled prediction of a future election which was beyond expectation. There­ fo re , Hypothesis IV was deemed confirmed. Does the drastic drop in influence of two of the candidate a ttrib u te s account fo r which Supreme Court winners since 1963 were in co rrectly predicted in the Hannah model? Yes, the fiv e Supreme Court winners which the Hannah model scored as losers had p ro b a b ili­ tie s indicating a victory in the model estimated by P ro bit. Thomas E. Brennan in 1966, G. Mennen Williams and John Swainson in 1970, Charles Levin in 1972, and B la ir Moody, J r . , in 1976 a ll had estab­ lished stronger support in th e ir geographical section than had the opponents they defeated. Summary Chapter I I I contained the findings from tests of the f i r s t four research hypotheses. The models of the candidate a ttrib u te s which may explain who wins and who loses Michigan Supreme Court elec­ tions were tested with data obtained from the years 1948 to 1980. The te s t results fo r elections a t the Supreme Court level were mixed. The f i r s t hypothesis was confirmed. The bases of competition fo r Supreme Court elections as id e n tifie d by Hannah are s t i l l bases of competition fo r the period from 1971 to 1980. However, fu rth e r analy­ sis of the elections from 1948 to 1980 led to a d iffe re n t model posited in Hypothesis I I . Extensive tests of Hypothesis I I suggest that caveats should be attached to the conclusion th at th is hypothesis was 105 also confirmed. Several o f the candidate a ttrib u te s , as measured, did not provide a base of support fo r a candidate to win. Therefore, I concluded th at a more parsimonious model than the one posited in the second hypothesis should be preferred. Empirical evidence pertaining to the re la tiv e strength of the three candidate a ttrib u te s , indicated by Hannah, which provide the strongest advantage fo r a candidate to win did not lend adequate support fo r confirmation of Hypothesis I I I . The incumbency status of the candidate, strength o f the candidate's p o litic a l p arty, and length of the candidate's state public o ffic e experience were not the strong­ est bases of competition, respectively, fo r candidates running in Supreme Court elections from 1948 to 1963. State public o ffic e expe­ rience of a candidate, as operationalized in th is study, had no in flu ­ ence on the election outcome. However, the fourth hypothesis was con­ firmed when i t was tested with data on Supreme Court elections from 1964 to 1980. The r e la tiv e influence of the candidate's status as an incumbent and strength of the p o litic a l party which nominated the candidate not only decreased, as hypothesized, they no longer provided a candidate an advantage in competition fo r a Supreme Court seat. The th ird and fourth hypotheses enabled a comparison of two time periods fo r one appellate court le v e l. The next chapter presents the results of a te s t of Hypothesis I I with data from another appellate court—the Michigan Court o f Appeals. Also, the re la tiv e influence of the candidate a ttrib u te s is compared between the two appellate courts fo r the same time period. FOOTNOTES— CHAPTER I I I In analyzing the success rate fo r Supreme Court incumbents, possible bias due to the candidate being an appointee rather than an elected incumbent must be considered. The number of cases is sm all, but I conclude th at being an appointed incumbent does not strongly a ffe c t the success rate fo r Supreme Court incumbents during the period studied. Also, both Democratic and Republican governors appointed Supreme Court Justices during the time period studied and party control of the governorship did not have a substantial influence on the success rate fo r Supreme Court incumbents. Comparing Appointed and Elected Supreme Court Incumbents Elected Incumbents F irs t Re-election Appointed Incumbents F irs t Election 1948-1963 Harry Kelly Eugene Black 1961 1963 won won Clark Adams Talbot Smith George Edwards John Voelker Theodore Souris Paul Adams Otis Smith i D D D D D D D 1952 1956 1956 1957 1960 1962 1962 won won won won won lo s t won 0% of elected lo s t 14% of appointed lo s t (a Democrat) 100% of f i r s t election as incumbent losers were appointed 1964-1980 T. M. Kavanagh M. D. O'Hara T. G. Kavanagh G. M. Williams M. S. Coleman Charles Levin 1966 1968 1976 1978 1980 1980 won lo s t won won won won John W. Fitzgerald James L. Ryan Lawrence Lindemer R R R 1974 1976 1976 won won lo s t 17% of elected lo s t (a Republican) 33% of appointed lo s t (a Republican) 50% of f i r s t election as incumbent losers were appointed *D = Appointed by Democratic governor 2 R = Appointed by Republican governor. 106 107 2 -2 Times Log Likelihood Ratio Test of Differences provides information on whether additional variables in a model add s ig n ific a n tly to the explanatory power of the model. This is determined by comparing the Log Likelihood Function of two models. Both models have a common set of variables and the unconstrained model contains additional v a ri­ ables. The difference in the Log Likelihood Function between the constrained and unconstrained models measure the difference in goodness of f i t . When th is difference is m ultiplied by -2 i t is a chi-square s ta tis tic under the hypothesis that constraints (the model without the additional variables) do not s ig n ific a n tly change the goodness of f i t . I f the chi-square is sig n ific a n t a t the alpha level established, then i t may be concluded that the additional variables s ig n ific a n tly add to the explanatory power. This te st may be conducted twice with two d i f ­ feren t unconstrained models; in th is case, two d iffe re n t conceptualiza­ tions of incumbency. Between the two unconstrained models, the model with the higher chi-square calculated in the -2 Times Log Likelihood Ratio Test of Differences may be concluded to have the additional variable with the stronger explanatory power. See John H. A ldrich, "Electoral Choice in 1972: A Test of Some Theorems of The Spatial Model of Electoral Competition," Journal of Mathematical Sociology 5 (1977): 234. 3 A ll Probit estimations were achieved before ten ite ra tio n s . -2 Times the difference in the Log Likelihood Functions (the Log Likelihood Function of the constrained model minus the Log Likelihood Function of the unconstrained model) is called the -2 Times Log Like­ lihood Ratio and is chi-square d istrib u ted . Naive Incumbency 2 X - - 2[LLFcons£ra .jne£j - LLF|jncon5£pa.jnecj] = - 2 [ - 29.3566 - (-16.0429)] x \ = 26.6274 p < .001 Hypothesized Incumbency X - _2t LLFCOnstrained ~ LLFunconstrained^ = -2 [-29.3566 - (-22.1645)] x \= 14.3842 p < .001 108 4 For instance, in 1970 a long-time incumbent, John Dethmers, was defeated when two former Governors of Michigan won in th at elec­ tio n . Also, a Supreme Court candidate who unsuccessfully ran fo r o ffic e several times f in a lly won in 1962. Michael D. O'Hara won dur­ ing the aura of the Kennedy years. In te re s tin g ly , O'Hara lo s t his re-e lectio n bid in 1968 when his only challenger was T. G. Kavanagh. ^Robert Pindyck and Daniel L. Rubinfeld, Econometric Models and Economic Forecasts (New York: McGraw-Hill, 1976), pp. 81-82. The party variable is not s ig n ific a n t in the unconstrained model but is in the constrained model. Of possible concern is the phenomenon that in the constrained model th is variable is acting as a proxy fo r the omitted variables. I do not believe th is to be the case, especially since under the same pair of models in the second period the party variab le remained in s ig n ific a n t. ^Susan B. Hannah, "An Evaluation of Judicial Elections in Michigan, 1948-1968" (Ph.D. d is s e rta tio n , Michigan State U n iversity, 1972), pp. 204-205. CHAPTER IV ANALYSIS OF COURT OF APPEALS MODELS Chapter Objectives The research objectives addressed in Chapter IV pertain to issues which, to the best of my knowledge, have not been analyzed by p o litic a l s c ie n tis ts . The f i r s t objective is to determine the re la tiv e influence of the bases of candidate competition which may explain the election outcome fo r the Michigan Court of Appeals. The second objec­ tiv e is to compare the re la tiv e influence of the bases of candidate competition between the two appellate courts--Michigan Supreme Court and Michigan Court of Appeals. The Court of Appeals is included in th is study because the two appellate courts d if f e r in ways th at are relevant to candidate selection. The Court of Appeals candidate has an electo ral d is t r ic t which is geographically smaller than the Supreme Court's and with approximately one-third the number of potential voters. In Court of Appeals electio n s, p o litic a l parties are form ally removed from the process. Often candidates fo r the Court of Appeals face less compe­ t it io n in the election because the requirement fo r party nomination of Supreme Court candidates has resulted in every Supreme Court race under th is system being contested. The Court of Appeals candidate may have to face a primary e le c tio n , but Supreme Court candidates never do. 109 110 The whole existence of the Court of Appeals through 1982 has been under Republican administrations—Romney's and M illik e n 1 s--w hile fo r the period studied the Supreme Court has been under both party admin­ istra tio n s (Democratic from 1949 to 1963 and Republican from 1963 to 1982). Data on Michigan Court of Appeals contested elections from 1970 to 1982 are used to evaluate Hypotheses I I and V. in the Appendix. See Table A.7 Hypothesis V also required use of Michigan Supreme Court data fo r elections from 1970 to 1980 in order to make appro­ p riate comparisons between the appellate courts. Therefore, Chapter IV f i r s t presents the results from testing the Michigan Court of Appeals model and second, resports the results from comparing the re la tiv e influence of the candidate a ttrib u te s . This chapter concludes with a summary of the findings fo r the Court of Appeals of Michigan. Hypothesized Model The hypothesized model of the re la tiv e influence of candidate attrib u te s which may explain who wins and who loses in appellate court elections was tested with Michigan Court of Appeals contested elections. This provided additional information from another level appellate court in order to more adequately te s t research Hypothesis II. Hypothesis I I : There is a positive relationship between winning an appellate court election in Michigan and the incumbency status of the candidate, strength of the can­ didate's p o litic a l party, state and local p o litic a l o ffic e experience of the candidate, support from the candidate's geographical section, unearned fa m ilia r ity of the candidate's name, Iris h or other ethnic appeal of the candidate's name, and support from the candidate being recognized as a Catholic. Ill In th is section, the hypothesized model's goodness of f i t is reported. Next, the significance of the co efficien ts which indicate the re la ­ tiv e influence of the candidate a ttrib u te s is discussed. a te s t of two models. presented. This includes Then, the forecasting a b ilit y of the model is Table 4.1 contains the estimation of two Court of Appeals models and th e ir s t a tis tic s .* The goodness of f i t of the hypothesized model is evaluated fir s t. The c r ite r ia fo r three goodness of f i t s ta tis tic s fo llow . 2 Estimated R must be a t lea st .60. Spearman's r , another summary s t a t is t ic , must be s ig n ific a n t at alpha equal to .05. measures reduction in e rro r, must be a t le a s t .50. Lambda^, which The goodness of f i t s ta tis tic s , reported in Table 4 .1 , fo r the hypothesized model handily surpassed the minimum requirements. Spearman's r is .82 and lambda^ is .80. The estimated R is .95. Also, the percent predicted co rrectly is 91 percent which outperforms the Supreme Court model fo r both periods. Positive and S ig n ifica n t C oefficients The req u isite theoretical condition was th at the hypothesized candidate a ttrib u te s p o s itiv e ly influence the p ro b ab ility of a candi­ date's winning an appellate court e le ctio n . The c r ite r ia were th at in a one-tailed t - t e s t each c o e ffic ie n t be positive and s ig n ific a n t at alpha equal to .05. The three co efficie n ts which s a tis fie d that requirement are indicated in Table 4 .1 . The support from a candi­ date's geographical section, ethnic appeal of a candidate's name, and unearned fa m ilia r ity of a candidate's name were p o s itiv e ly related to Table 4 . 1 . --T e s t o f two Court o f Appeals m odels, 1970-1980 Unconstrained Model Constrained Model A ttrib utes MLE MLE^ MLE/SE MLE SE MLE/SE 5.127 .034 38.039 .018 .135 1 .8 6 5 ^ Incumbency Sectional Local Ethnic Catholic Name 7.411 .079 .084 4.086 -2.814 3.455 Constant Est. R2 -2 x LLR Percent Predicted Correctly -9.748 .9 4 8 ^ 3 3 .7 4 8 ^ -2.688 .876 2 2 .0 2 0 ^ 90.909 .8 1 7 ^ .800 78.788 .5 8 9 ^ .533 Ah . 2.653 0 . 4.057 0 . 3.090 SE 0 35.086 .038 .064 2.191 2.060 1.790 .2 1 1 2 .0 7 5 ^ 1.323 1 .8 6 5 ^ -1.366 1 .9 3 1 ^ * “ -"(d iffe re n c e ) = n - 728” (Chi square, 4 degrees of freedom) N = 33 ♦Indicates s ig n ific a n t standardized MLE ♦♦Indicates significance a t .05 level 2 2 ? (C r itic a l values, Z - 1.64, r s = .305, X 5 = 12.59, X 4 = 9 .4 9 , X 2 = 5.99) ♦♦♦For comparison: lin e a r regression = .665 Farrar-Glauber R2 = .482 as the estimate of degree of m u ltic o llin e a rity . I do not consider th is a serious level in which the model should not be used. 113 winning a contested Court of Appeals election during the years from 1970 to 1980. However, three other candidate attributes--incumbency status of the candidate, length of the candidate's local public o ffic e experience, and a candidate's being recognized as a C ath o lic-lacked s ig n ific a n t influence in determining the election outcome. finding fo r incumbency is not to ta lly surprising. elections, the incumbent often is not challenged. This In Court of Appeals The incumbents in th is data set are the ones who, fo r whatever reasons, were chal­ lenged in th e ir re-election bid. The opposite results fo r one of the candidate attrib u tes is suggestive. In From B allo t to Bench, P h ilip DuBois referred to the 3 Irish -C ath o lic winning phenomenon in Michigan ju d ic ia l elections. For the Court of Appeals elections from 1970 to 1980, being recognized as a Catholic was not an advantage, while the ethnic appeal of a can­ didate's name was. Unfortunately, too few cases prohibited a separate category fo r the Iris h names, but my results suggest that the appeal of a candidate's Iris h name provided the advantage—not the candi­ date's also being recognized as a Catholic. Survey data from voters who p articip ate in Michigan ju d ic ia l elections would fu rth e r enlighten Michigan's Irish -C ath o lic phenomenon. Two candidate attrib u te s were not estimated in th is model. A lack of s u ffic ie n t cases during the period investigated prevented estimation of the advantage derived from the length of a candidate's state public o ffic e experience. Also, the advantage provided a 114 candidate from h is/h e r p o litic a l party a f f ilia t io n was not estimated because of data lim ita tio n s . There were discussed in Chapter I I . Additional information about the significance of the c o e ffi­ cients is availa b le from the -2 Times Log Likelihood Ratio Test of Differences. This te s t indicates whether certain of the candidate a ttrib u te s in the hypothesized model (unconstrained model) add sig­ n ific a n tly to the explanatory power of th at model. The constrained model used fo r the Court of Appeals elections is sim ila r to the one developed fo r th is te s t a t the Supreme Court level in Chapter I I I . The p o litic a l party variable was omitted from the Court of Appeals constrained model. The -2 Times Log Likelihood R atl 0 ( c|.jfference) ls s t a t is t i4 c a lly s ig n ific a n t with alpha set a t .05. The c rite rio n was s a tis ­ fie d . Therefore, th is re s u lt supports the conclusion th at the ethnic appeal of the candidate's name and unearned fa m ilia r ity of the can­ didate's name add to the explanatory power of the hypothesized model. Ex Post Forecasting Another source of information fo r evaluation of the hypothe­ sized model is availab le in the form of ex post forecasting. The Court of Appeals elections from 1981 to 1982 were excluded from the data used to estimate th is model in order to be used in the ex post forecast. When the 1982 election was held (a fte r th is research design was firm ly s e t), only one Court of Appeals seat was contested. were only two candidates. There This small data set makes in terp reta tio n of the summary goodness of f i t s ta tis tic s d i f f i c u l t . 115 Inspection of Table 4.2 id e n tifie s the only incorrect predic­ tio n was the forecasted loss fo r Roman S. Gribbs. Both of the Table 4 .2 .— Forecasting Court of Appeals e le c tio n , 1982 Candidates A ttributes MLE Roman S. Gribbs Incumbency Sectional Local Ethnic Catholic Name 7.411 .079 .084 4.086 -2.814 3.455 Constant -9.748 0 0 55. 45. 18 20 1 0 1 0 0 0 -2.443 Sum Geraldine B. Ford -4.675 .07 P(Z=1) .0 0 Prediction 0 0 Outcome 1 0 candidates in th is race were forecasted to lose. Gribbs did have a s lig h t forecasted p ro b ab ility of winning, while Geraldine B. Ford did not. model. Gribbs won the e le ctio n . This does supply some support to the Therefore, the ex post forecast results are not c le a r. Testing Hypothesis I I with data obtained from Court o f Appeals elections from 1970 to 1980 preduced mixed support fo r th is hypothe­ sis . The model's goodness of f i t provided exceptional support. Also, three candidate a ttrib u te s , as hypothesized, provided an electoral advantage in Court of Appeals contested elections from the years 1970 116 to 1980. These were the support from the candidate's geographical section, appeal of the candidate's ethnic name, and unearned fa m ilia r ­ it y of the candidate's name. However, the positive relationship hypothesized fo r the rest of the bases of competition and winning an appellate court election was not substantiated. At the Court of Appeals le v e l, the a t t r i ­ butes which did not provide an advantage fo r a candidate were the incumbency status of the candidate, length of the candidate's local public o ffic e experience—appointed or elected, and the candidate being recognized as a Catholic. I t is doubtful th at the length of a candidate's state public o ffic e experience provides any advantage to a candidate. Electoral laws fo r the two appellate courts d if f e r over the role fo r p o litic a l parties in the respective con­ te sts . Therefore, whether the strength of the candidate's p o litic a l party provides or does not provide an advantage to a Court of Appeals candidate is s t i l l an open question. Also, the ex post forecast performance of the model does not c le a rly support the second hypothesis. In the ex post fo recast, the actual winner of the 1982 Court of Appeals contest was not predicted to win. However, th is candidate did have a small forecasted proba­ b i l i t y o f winning. Therefore, I conclude Hypothesis I I was confirmed, with a caveat. Some of the a ttrib u te s in the model did not provide a candidate an additional advantage to win over the other Court of Appeals contestants who possessed those attrib u te s to a lesser extent. 117 Even though the hypothesized positive relationship was not supported fo r a ll of the candidate a ttrib u te s , none of them exhibited a sig­ n ific a n tly negative relatio n sh ip . In the next section, the results of the te s t of Hypothesis V explore the relationship of the Court of Appeals candidate a ttrib u te s . Relationship of A ttributes This section compares the re la tiv e influence of the candi­ date a ttrib u te s which may explain who wins and who loses a contested Court of Appeals election with the re la tiv e influence of Supreme Court candidate a ttrib u te s . Also, compared is the relationship of the re la tiv e influence of the Court of Appeals candidate a ttrib u te s . Michigan's Court of Appeals contests are more lik e nonpartisan elections than Michigan's Supreme Court races. elections also have less media coverage. The Court of Appeals Thus, three studies which were discussed in Chapter I I provide guidance on how the re la tiv e influence of the Court of Appeals candidate a ttrib u te s may d if f e r from the Supreme C ourt's. Based on these studies, 1 believe th at ethnic fa cto rs , such as the ethnic appeal of a candidate's name, and a candidate's unearned name fa m ilia r ity are more in flu e n tia l a ttrib u te s in Court of Appeals elections than Supreme Court races. Data from Supreme Court and Court of Appeals contested elections from 1970 to 1980 tested Hypothesis V . 5 Hypothesis V: In contested Court of Appeals elections from 1970 to 1980, the ethnic appeal of a candidate's name and a candidate's unearned name fa m ilia r ity have greater re la tiv e influence than in Supreme Court contests fo r the same period. 118 Rank ordering of the standardized maximum likelih ood estimate co efficie n ts enabled comparison of the re la tiv e influence of the Court of Appeals candidate a ttrib u te s to the Supreme C ourt's. The c rite rio n was that fo r an a ttrib u te to display a greater influence in one model its rank order had to be higher than in the other model. The rank orderings fo r both models are reproduced in Table 4 .3 . Table 4 .3 .--Rank order of appellate court candidate a ttrib u te s , 19701980 Supreme Court Court of Appeals A ttributes MLE* Rank MLE* A ttributes Ethnic 4.057 1 Sectional 4.788 Name 3.090 2 Name 0 . Sectional 2.653 3 Ethnic 0 . Incumbency 0 . Incumbency 0 . Local 0 . Party 0 . Catholic 0 . State 0 . Local 0 . Catholic 0 . Rank 1 ♦Indicates s ig n ific a n t standardized MLE. In the contested Court of Appeals elections from 1970 to 1980, the ethnic appeal of the candidate's name has a f i r s t place rank. The unearned fa m ilia r ity of the candidate's name has a second place ranking. Neither of those two candidate a ttrib u te s even q u a lifie d fo r ranking in the Supreme Court model fo r the same period. the c rite rio n was s a tis fie d . D e fin ite ly , Therefore, Hypothesis V was confirmed. 119 The ethnic appeal of the candidate's name and a candidate's unearned name fa m ilia r ity have a greater re la tiv e influence in contested Court of Appeals elections than in Supreme Court races fo r the same period. S urprisingly, in Court of Appeals elections from 1970 to 1980, the re la tiv e influence of the ethnic appeal of the candidate's name and unearned fa m ilia r ity of the candidate's name was even greater than the re la tiv e influence of support from the candidate's geographi­ cal section. Remember, in the Supreme Court contests support from the candidate's section was viewed as the foundation of the candi­ date's electoral advantage. Only the strength of the candidate's p o litic a l party was higher ranked in one of the Supreme Court models. Also, in the i n i t i a l Court of Appeals race one of the candidates even e x p lic itly campaigned as the "local boy" who would give th e ir county its deserved representation on the court.^ My theoretical expectation was th at these two candidate a t t r i ­ butes would have greater influence in the Court of Appeals contests than the Supreme Court elections because p o litic a l parties do not have a formal ro le a t the Court of Appeals le v e l. With a reduced party cue, other cues would take on greater importance. Perhaps the surprising degree of strength of the advantage provided by the ethnic appeal of the candidate's name and the candidate's unearned name fa m ilia r ity is due to an even greater reduction in the role fo r the strength of the candidate's p o litic a l party a f f il ia t io n than an ticipated. Informa­ tion from a survey of the electorate who ac tu a lly vote in Court of Appeals elections would be tremendously helpful in explaining th is re s u lt. 120 Summary Chapter IV contained the results from tests of the second and f i f t h research hypotheses. The data set included contested Court of Appeals elections from 1970 to 1982 and Supreme Court elections from 1970 to 1980. Test results fo r elections a t the Court of Appeals level were mixed. Hypothesis I I was confirmed, with a caveat. Some of the can­ didate a ttrib u te s , as measured, did not have the hypothesized posi­ tiv e relationship to winning an appellate court ele ctio n . They are the incumbency status of the candidate, length of the candidate's local public o ffic e experience, and recognition of the candidate as a Catholic. The results of the te s t of the model via ex post forecasting were not as straightward as desired. However, the set of candidates was very small. Hypothesis V was confirmed. In contested Court of Appeals elec­ tio n s, the re la tiv e influence of the ethnic appeal of the candidate's name and a candidate's unearned name fa m ilia r ity was greater than th e ir re la tiv e influence in Supreme Court elections fo r the same period. The re la tiv e strength of these two Court of Appeals candi­ date a ttrib u te s was even greater than anticipated. Tests of Hypothesis I I and V incorporated information from both of the appellate courts. Chapter V contains my conclusions which w ill draw together the separate findings in Chapters I I I and IV on the re la tiv e influence of Supreme Court and Court of Appeals candidate a ttrib u te s . Also, Chapter V discusses suggestions fo r 121 fu rth e r research on the re la tiv e influence of candidate a ttrib u te s which may explain who wins and who loses appellate court elections. FOOTNOTES— CHAPTER IV Both estimates of the Court of Appeals models fa ile d to con­ verge w ithin ten ite ra tio n s . That means that a local maximum fo r the estimates was not achieved. I do not consider th is a serious problem, however, due to the fa c t th at the fa ilu r e to achieve con­ vergence did not appear to be due to serious data problems. 2 Of the seats having an incumbent in the race, 64 percent were unchallenged contests. Of the nine incumbents who were challenged, four were recent appointees—about 44 percent. Taking th at into consideration, only fiv e incumbents who were not recent appointees had competitors. This is 20 percent of a ll of the seats having an incumbent in the race. Court of Appeals Incumbent Races 1970-1980 Year F irs t D is tr ic t Second D is tr ic t Third D is tr ic t 1970 J. H. G ill is R. Danhof1 , 2 L. McGregor R. B. Burns1 1972 T. J. Lesinski T. C. Quinn D. E. Holbrook 1974 V. J. Brennan ? G. Bashara, Jr. S. Bronson* R. Danhof1 T. M. Burns 0 G. A lle n , J r. 1976 J. H. G ill is D. C. R iley 1 W. R. Beasley1 , 2 R. B. Burns * 2 1978 D. C. Riley W. Cynar1 , 2 1980 G. Bashara, J r. V. J. Brennan R. Danhof1, S. Bronson T. M. Burns G. A lle n , J r. Opposed incumbent 2 Appointee q P h ilip DuBois, From B allo t to Bench: Judicial Elections and the Quest fo r Accountably!ty (Austin: University of Texas Press, 1980), pp. 81-88, 132". 122 123 -2 Times the difference in the Log Likelihood Functions (the Log Likelihood Function of the constrained model minus the Log Like­ lihood Function of the unconstrained model) is called the -2 Times Log Likelihood Ratio and is chi-square d istrib u ted . -2 x LLR - -2CLLFconstrained - ( LLfruncons tra in e d ^ = -2 [-11.7274 - (-5 .8 6 3 3 )] X 2 = 11.7282 (c r it ic a l value p p < .02 = 9 .4 9 , a = .05) 5 The estimations of the Supreme Court model fo r the two d i f ­ fe re n t time periods--1964 to 1980 and 1970 to 1980--are very s im ila r. Therefore, the conclusions drawn here about the differences between the two courts also appropriately apply to the Supreme Court elec­ tions from 1964 to 1980. See Table A. 8 in the Appendix. C In the November 1, 1964, p. 23, issue of the Grand Rapids Press, Robert B. Burns advertised himself as "Kent County's Only Candidate" and pointed out "Kent County should have one out of three judges on the Court of Appeals." CHAPTER V CONCLUSIONS AND FUTURE RESEARCH Chapter Objectives In the past, policy makers in Michigan have urged drastic changes in the s ta te 's ju d ic ia l selection process without adequate information. Therefore, the general goal of th is research is to refin e and expand our knowledge about ju d ic ia l elections in Michigan. This study.builds on the prelim inary findings by Susan B. Hannah. The theoretical framework fo r my study included viewing judges as policy makers whose au tho rity is leg itim ized through e le ctio n . Three approaches to decision-making th eo ry--American Voter, rational choice, and cybernetic--aid the id e n tific a tio n of what factors may provide a candidate an advantage over the other candidates when the voter casts his/h er b a llo t. Therefore, the f i r s t research objective was to determine the re la tiv e influence of the bases of candidate competition which may explain the election outcome fo r each of the Michigan appellate courts. The second research objective was to compare the re la tiv e influence of the bases of candidate competition between two time periods and court le v e ls . Five research hypotheses were tested with data from Michigan Supreme Court elections fo r the years from 1948 to 1982 and Michigan Court of Appeals elections fo r the period from 1970 to 1982. 124 125 As the la s t chapter in th is d issertatio n , two objectives structure Chapter V. The two previous chapters separately presented the results of the tests of the hypotheses fo r the two appellate courts in Michigan. Therefore, the f i r s t objective is to compare these findings in order to develop conclusions about Michigan's appellate court elections. Research is a cumulative endeavor. Thus, the second objective is to id e n tify avenues fo r continued research on the re la tiv e influence of candidate attrib u tes which may explain who wins or loses a major ju d ic ia l election. Conclusions This section compares the findings from my separate analysis of the two appellate courts in Michigan. this synthesis. Three questions organize F ir s t, which candidate attrib u tes provide an advan­ tage in Michigan appellate court elections? models perform? Second, how well do the Third, would estimation of one model fo r appellate court elections from 1948 to 1982 create an appropriate model? Candidate A ttributes Which candidate attrib u tes provide an appellate court candi­ date an electoral advantage? Hypothesis 1 was confirmed. The bases of competition fo r Supreme Court elections as id e n tifie d by Hannah fo r the period from 1948 to 1970 are s t i l l bases of competition fo r the period from 1971 to 1980. However, fu rth er analysis revealed that a more sophisticated model estimated with a newer technique was better able to determine the re la tiv e strength of the candidate attrib u tes 126 which may explain who wins and who loses an appellate court elec­ tio n . For both appellate courts, Hypothesis I I , which specified the new model, was confirmed, with caveats. They are: fo r both courts, not a ll of the candidate attrib u tes are p o sitively related to the election outcome and fo r the Supreme Court le v e l, a more parsimonious model is preferred. The only candidate a ttrib u te which remained in flu e n tia l in elections fo r both courts and during both time periods was the support from the candidate's geographical sec­ tio n . Four candidate a ttrib u te s were in flu e n tia l only in one court level fo r one time period. The incumbency status of the candidate and strength of the candidate's p o litic a l party were p o sitively related to only Supreme Court election outcomes fo r the period from 1948 to 1963. The relationship of these two a ttrib u te s ' re la tiv e influence, as specified in Hypothesis I I I , was not confirmed. ever, Hypothesis IV was confirmed. How­ The re la tiv e influence of the Supreme Court candidate's incumbency status and strength of the p o litic a l party nominating the candidate decreased to such an extent th at in the period from 1964 to 1980 they no longer provided an elec­ to ral advantage. The other two candidate a ttrib u tes which had a positive in flu ­ ence in only one court fo r one period were the ethnic appeal of the candidate's name and unearned fa m ilia r ity of the candidate's name. In Hypothesis V, they were hypothesized to be more in flu e n tia l in 127 Court of Appeals elections than Supreme Court elections fo r the same period. This hypothesis was confirmed. In fa c t, these two a t t r i ­ butes only provided an electoral advantage in Court of Appeals ele c­ tions during the years from 1970 to 1980. Therefore, fiv e candidate a ttrib u te s had a positve relationship to winning an appellate court ele ctio n . The key to an appellate court victo ry was support from the candidate's geographical section. Only th is a ttrib u te consistently provided an advantage during the to ta l period and fo r both courts. The status of the candidate as an incumbent and strength of the p o litic a l party nominating the candi­ date were p o sitively related to only Supreme Court election outcomes during the years from 1948-1963. The ethnic appeal of the candidate's name and the candidate's unearned name fa m ilia r ity established a p o sitive relation ship to winning contested Court of Appeals elec­ tions during the years from 1970 to 1980. Model Performance The models fo r both levels of appellate courts performed remarkably w e ll. The c r ite r ia fo r the -2 Times Log Likelihood 2 r a tio , estimated R , percent predicted c o rre c tly , Spearman's r , and lambda^ were s a tis fie d by the Supreme Court models and Court of Appeals model. The percent predicted co rrectly are e s sen tia lly the same fo r both appellate court models. Also, the lambda^ values fo r the models are very sim ila r. The nature of the models' incorrect prediction is another basis fo r comparison. Incorrect predictions from the estimation procedure 128 and ex post forecasting are used to make th is comparison. B asically, the nature of a ll of the models' incorrect prediction is the in a b ility of Probit to accommodate the constraint of the number of seats to be f i l l e d in each ele ctio n . For some electio n s, too few winners were predicted and the actual winner, who was in co rrectly predicted as a lo ser, had the highest score among the predicted losers. cases the d ire c t opposite occurred. In other Too many winners were predicted with the actual loser having the lowest score among the predicted winners. Two examples to illu s tr a t e the f i r s t situ atio n are selected. One example is from an estimates Supreme Court model. from a forecasted Court of Appeals electio n . The other is In 1972, Mary Coleman surprised some people when she became the f i r s t female to be elected to Michigan's Supreme Court. The Supreme Court model had predicted a loss fo r her, but only one winner was predicted fo r the two seats in th is election. She did have the second highest score among the nine candidates in th is race. As an "Outstate" candidate, Coleman had the highest percentage of sectional support and probably won because the six D etro it Metropolitan candidates s p lit the support from th e ir geographical section. In the ex post forecast of the 1982 Court of Appeals e le c tio n , no winners were predicted fo r the contested seat. The actual winner, Roman S. Gribbs, did have a higher score than his opponent, Geraldine B. Ford. The example fo r the other s itu a tio n --to o many winners predicted--comes from the ex post forecast of the 1982 Supreme Court 129 ele ctio n . win. In this e le c tio n , Dorothy Comstock Riley was predicted to She lo s t. However, the model predicted three winners fo r the two seats in th is contest and Riley did have the th ird highest score. Also, the vote to ta ls fo r second and th ird place, as recorded in Michigan's O ffic ia l Canvass of Votes, were very close—Michael Cavanagh (1,092,507) and Dorothy C. Riley (1,079,437). Two prediction errors fo r Supreme Court elections during the period from 1948 to 1963 had other dynamics operating. Supreme Court candidate Michael O'Hara. predicted in a two-seat race. Both involved In 1957, three winners were O'Hara, the lo s e r, had the second highest score while Thomas M. Kavanagh, a winner, had the th ird highest score. Both were scored p o sitively fo r the ethnic appeal of th e ir name and unearned name fa m ilia r ity . The model could not discern th at "Kavanagh" was and is a more p o litic a lly powerful name than "O'Hara." In 1962, O'Hara was predicted to lose to an appointed incum­ bent, Paul Adams. O'Hara's values on the other variables were as good as or better than Adams', except fo r a 1 percent difference between the two candidates' strength of p o litic a l party. Credit as an incumbent resulted in a higher score fo r th is recent appointee than O'Hara's. However, no winner would have been predicted i f Adams' incumbency had not been scored. In th at case, O'Hara would have had the higher score, although not predicted as a winner. From the 1962 e le c tio n , not distinguishing between a recently appointed and elected incumbent appeared to be a weakness of the model. 130 However, in the 1956 Supreme Court contest the recently appointed incumbent, Talbot Smith, was co rrectly predicted in his race against O'Hara. During the period from 1948 to 1963, fiv e additional appointed incumbents were co rrectly predicted as winners in th e ir f i r s t race as incumbents. Two of the three appointed Supreme Court incumbents in the 1964 to 1980 period were also predicted correctly in th e ir f i r s t race as an incumbent. On the whole, the appointed incumbent phenomenon does not misguide the model's predictions. Supreme Court elections from 1964 to 1980 also had other dynamics operating in three incorrect predictions. F ir s t , in 1968 O'Hara was predicted to win but lo s t to his challenger, Thomas G. Kavanagh, who was also predicted as a winner fo r the single Supreme Court seat. Another more powerful "Kavanagh" name outperformed an "O'Hara" name. Second, in 1974 too many winners were predicted and even in order of the scores of the predicted winners was in correct. The actual lo ser, B la ir Moody, J r . , had the second highest score in a two-seat race. Moody's sectional vote was higher than one of the other winners. However, Moody's geographical section was D etro it Metropolitan Area which ty p ic a lly did not have the highest voter pa rticip atio n in Supreme Court elections. did not predict any winners. not have the highest score. Th ird, in 1976 the model James Ryan, the actual winner, did He had the same percentage of sectional support as his opponent, Charles Kaufman. Ryan's victo ry is th at he is a "jo in e r." are included in his biography. A possible explanation fo r A long l i s t of associations 131 Therefore, my conclusion about the comparative performance of the models is th at they performed equally w e ll. The a b ilit y of the models to explain who wins and who loses appellate court elections exceeds expectations. However, a ll of the models have the same weakness--an in a b ility to account fo r the constraint of the number of seats availab le in the elections. For a few elections, too many winners were predicted; fo r other elections, too few winners were predicted. I f the prediction results were adjusted by using the candidate's ranking in those elections, the models' predictive a b ilit y would approach perfection. However, recalculation of the prediction results is unnecessary considering the models' outstanding performance and inappropriate fo r the technique used to te s t the models. Model Estimation Would a single estimation of one appellate court model of the candidate a ttrib u te s which may explain who wins and who loses an appellate court election during the years from 1948 to 1982 be an appropriate model? reasons. No, i t would not be appropriate fo r the following The rules of the game and national trends which were described in Chapter 1 have had an impact on Michigan's appellate court elections. F ir s t, in Michigan the set of candidate a ttrib u te s which provide an advantage are d iffe re n t fo r the two appellate courts. Also, the standardized c o e ffic ie n t fo r the re la tiv e influence of the a ttrib u te which both courts have in common--sectional support--is d iffe re n t fo r each court. The value fo r the Supreme Court is 132 approximately twice the value fo r the Court of Appeals. Therefore, these two courts have great enough differences that they should have the re la tiv e influence of th e ir candidate attrib u tes separately determined. Second, there is a need fo r special organization of the time periods used to establish the re la tiv e influence of the candidate a ttrib u te s . At the Michigan Constitution Convention, some of the delegates intended to change the basis fo r election to the Supreme Court. S ign ificant changes in the re la tiv e influence of some Supreme Court candidate a ttrib u tes occurred. Thus, the re la tiv e influence of candidate attrib u tes fo r Supreme Court elections should not have one estimated value fo r the whole period, fo r the year 1948 to 1982. The appropriate dividing point is 1964. 1964 is also the year in which the f i r s t elections fo r Justices of the Court of Appeals were held. However, the re la tiv e influence of candidate a ttrib u tes which may explain who wins and who loses Court of Appeals elections should not be determined by using data fo r elections from 1964 to the present. I t is essential that the early elections in which no incumbents ran fo r re -e lectio n be excluded. The election in 1970 is the appropriate i n it i a l election fo r the data set. Therefore, three d is tin c t sets of periods exist fo r appellate court elections and should be used in determining the re la tiv e influence of candidate a ttrib u te s . The Supreme Court has two periods--from 1948 to 1963 and 1964 to the present. The Court of Appeals period—from 1970 to the present— is a subset of the recent Supreme Court period. 133 This section of Chapter V contained three major conclusions from my research on Michigan appellate court elections. F ir s t, the support from the candidate's geographical section, incumbency status of the candidate, strength of the candidate's nominating p o litic a l party, ethnic appeal of the candidate's name, and the candidate's unearned name fa m ilia r ity are the fiv e candidate a ttrib u te s which had a positive relation ship to winning an appellate court e le c tio n . They key to an appellate court victory was the candidate's support from her/h is geographical section. Second, the models of candidate a ttrib u te s which may explain who wins and who loses an appellate court election performed e x tra o rd in a rily w e ll. However, a ll the models had the same weakness--an in a b ility to adjust fo r the number of seats to be f i l l e d in an e le ctio n . Third, a single estimation of one appellate court model of the re la tiv e influence of the a ttrib u te s which provide a candidate an electoral advantage would not be an appropriate model. The electo ral phenomena were d iffe re n t fo r the two appellate courts and estimation using d iffe re n t time periods is requ i red. Future Research Conventionally, i t is the resp o n s ib ility of the scholar to conclude a research project with suggestions fo r future research. suggestions are directed to: My f i r s t , fu rth e r research with additional data and second, expanding the scope of the research. 134 Additional Data I suggest future research be conducted with two kinds of addi­ tio nal data—more elections and survey information. F ir s t, incorpora­ tio n of future elections into the data set w ill enhance investigation of the re la tiv e influence of candidate a ttrib u te s fo r both appellate courts. The Supreme Court and Court of Appeals should be compared again when the Court of Appeals is more established and a longer time period than ten years is a v a ila b le . Re-estimation of the re la tiv e influence of Supreme Court candidate attrib u te s may provide new insights because in the future the Court of Appeals may act as a career step fo r Supreme Court candidates. Also, an interesting question fo r the Court of Appeals is whether there are s ig n ific a n t differences in the re la tiv e influence of candidate a ttrib u te s between the three Court of Appeals d is tr ic ts . I suspect s ig n ific a n t differences in the re la tiv e influence of the candidate a ttrib u te s e x is t because o f the greater number of contested elections in the Second D is tr ic t. An attempt to answer th is question was made by using the dummy variable technique. prevented meaningful re su lts. Data lim ita tio n s With additional electio n s, estimation of the changes between the three d is tr ic ts should be possible. Second, I suggest continued pursuit of th is research topic with the addition of survey data. I t is essential that the electorate surveyed be actual appellate court voters. in atio n is not implemented. from survey data. Too often th is discrim­ At le a s t three benefits would be gained The assumptions on the composition and knowledge 135 base of the appellate court electo rate could be tested and hopefully corroborated. Greater insight on Michigan's Iris h -C a th o lic candidate popularity might be obtained. Perhaps a survey could help c la r ify whether the media led Supreme Court voters to focus on three candi­ date attributes--incum bency, party, and section—or the media's cover­ age of these is in response to the voters. Expanded Scope The scope of th is research topic may be expanded to analyze other Michigan ju d ic ia l elections or include other state ju d ic ia l elections. F ir s t , th is study of the bases of competition fo r appel­ la te court elections could p ro fita b ly be expanded to become an inves­ tig a tio n of Michigan's ju d ic ia l elections. Inclusion of C irc u it Court elections would provide fo r two comparisons. The re la tiv e influence of candidate a ttrib u te s could be compared between the d i f ­ fe re n t types of courts—appellate and t r i a l . Also, a comparison could be made between the Supreme Court and the two courts with the same electoral rules— tne Court of Appeals and C irc u it Court. Another in teresting issue is the possible differences in the re la tiv e influence of Court of Appeals candidate a ttrib u te s fo r primary candidates and general election candidates. In primary elections the number of competitors fo r a seat is greater and the ju d ic ia l electorate probably d iffe re n t than in general elections. Second, an expanded study could compare the re la tiv e in flu ­ ence of candidate a ttrib u te s in Michigan's appellate court elections to those in another s ta te . Are the candidate a ttrib u te s and th e ir 136 re la tiv e influence sim ila r in another "mixed" state? I f Michigan is compared to Ohio, the comparison would be informative about appellate court elections within the same geographical region of the United States. I f Arizona were selected as the other "mixed" s ta te , then appellate court elections would be contrasted between regions. An ambitious project would incorporate a ll three states. What are the differences in the r e la tiv e influence of appellate court a ttrib u te s which may explain who wins and who loses in a "mixed" state election and a partisan election state? The differences in the formal roles fo r the p o litic a l parties should produce in te r ­ esting resu lts. Again, the partisan ju d ic ia l election state selected could be one within Michigan's region or outside of that region. A more sophisticated research design might include both--states within and outside of Michigan's region. mentation by state government. Our federal system fosters experi­ With substantial differences between each sta te 's ju d ic ia l electo ral process, opportunities fo r fu rth er research abound. My intention was fo r th is study to reduce the "factual vacuum" in the area of state ju d ic ia l selection. This research provides scholars and policy makers em pirically based evidence on the re la ­ tiv e influence of candidate attrib u te s which may explain who wins and who loses contested appellate court elections. At the Supreme Court le v e l, greater refinement of our knowledge was achieved. Information was provided a t the Court of Appeals le v e l, where none existed. I hope th is study w ill add to the development of a 137 frame work fo r future analysis of the other states which have ju d ic ia l elections. Also, th is research provides a comparison of the re la tiv e influence of the bases of candidate competition between two time periods and court le v e ls . Evidence was provided to substantiate that changes in the rules of the game do a ffe c t behavior. D iff e r ­ ences between the two appellate courts were revealed. Hannah pioneered analysis by p o litic a l s c ien tists of ju d ic ia l elections in Michigan with her extensive evaluation of Michigan ju d ic ia l elections from 1948 to 1968. th at information. This study refined and expanded Development of newer an aly tica l techniques w ill enable even greater refinement and accumulation of knowledge. APPENDIX 138 Table A . I . - - V o t e r d ro p -o ff between s ta te p a rtis a n o f f ic e and Supreme Court Election State Partisan O ffice Supreme Court A pril 1949 State Board of A griculture 1,687,180 Regular 1,353-444 A pril 1951 State Board o f A griculture 1,383,061 Regular 1,133,670 2,762,362 Vacancy 1,941,768 Nov. 1952 (A uditor General)* State Treasurer 2 2,766,954 A pril 1953 State Board of A griculture 1,608,409 Regular 1,348,983 April 1955 State Board of A griculture 2,057,453 Regular 1,593,384 Nov. 1956 Auditor General 2,968,909 Vacancy 2,003,438 1,964,299 April 1957 State Board of A griculture 2,174,931 Regular Vacancy 1,647,039 801,705 April 1959 State Board of A griculture 2,386,146 Regular 1,630,739 Nov. 1960 Auditor General 3,179,722 Vacancy 1,966,904 AprfI 1961 Trustees of Mich. State Univ.'* 2,434,014 Regular 1,822,418 Nov. 1962 Trustees of MSU 2,638,903 Vacancy 1,846,009 1,760,339 April 1963 Trustees of Mich.. State Univ. 3,181,626 Regular 2,480,695 Nov. 1966 Trustees of Mich.. State Univ. 4,455,529 Regular 3,131,439 Nov. 1968 Trustees of Mich. State Univ. 5,757,532 Regular^ 2,098,134 Nov. 1970 Trustees of Mich. State Univ. 4,689,065 Regular 3,945,389 Nov. 1972 State Board of Education 5,984,618 Regular 4,749,517 Regular 3,593,754 Nov. 1974 2 (Trustees of Mich. State U n iv .) 5,886,128 Trustees of Mich. State Univ. 4,425,457 Table A . I . — Continued Election Nov. 1976 State Partisan O ffice Governors of Wayne State Univ. (Trustees of Mich. State U n iv.) 2 5,995,132 Supreme Court Regular 4 2,638,153 6,014,778 Vacancy 2,484,417 Vacancy 2,331,169 Nov. 1978 Trustees of Mich, State Univ. 4,575,498 Regular 4,261,360 Nov. 1980 Trustees o f Mich. State Univ. 6,350,508 Regular 5,243,952 *No m ilti-w in n e r partisan state o ffic e on b a llo t. p These contests were required in order to have the necessary p o litic a l p arties represented. O Name change from State Board of A griculture was e ffe c tiv e May 6 , 1959. 4 Only one regular term seat to be f i l l e d due to reduction in size of the Supreme Court. 141 Table A .2 . - - V a r ia b le ' sources o f in fo rm atio n Candidate Outcome State of Michigan, O ffic ia l Canvass of Votes, Michigan Manual Incumbency Same as Candidate Outcome Strength of Candi­ date's P o litic a l Party Elections Division of Secretary of State of Michigan; State of Michigan, O ffic ia l Canvass of Votes; Lansing State Journal; D e tro it News; and D e tro it Free Press State P o litic a l O ffice Experience Michigan Manual, Martindale-Hubbell Law Directo ry , Michigan State Bar Journal Roster, D e tro it News, D e tro it Free Press, and Lansing State Journal Local P o litic a l O ffice Experience Same as above and Directory of Municipal O f f i­ c ia ls , Grand Rapids Press, and Marquette's The Mining Journal Sectional Support Michian Manual, O ffic ia l Canvass of Votes, Michigan Bar Association Journal Roster, D e tro it Free Press, D etro it News, and Lansing State Journal Name F a m ilia rity Michigan Manual, Michigan State Bar Journal; Dunbar's Michigan; A History of the Wolve r­ ine S ta te ; Martindale-Hubbel Law Directory biographical section; Michigan Reports! Michigan Court of AppeaTs Reports; Michigan Biographies, 1-2; Mho's Who in Michigan; Bench and Bar of Michigan, 1938; Michigan Through the Centuries: Family and Personal H isto ry, 3-4; Mho's Who Among Black America; Michigan Chronicle; E. Krasicky, Deputy Attorney General; John Stenger, Michigan Court of Appeals Research S ta ff; Archie Fraser, State Public Adm inistrator, re tire d ; and the newspapers lis te d above Iris h Ethnic Appeal Same as Name F a m ilia rity plus Rizzo's Atlas of Michigan's Foreign Born Population, Michigan: A Guide to the Wolverine S ta te , Smith's New Directory of American Family Names, Pollock and Eldersveld's Michigan P o litic s in Transi­ tio n , and G ra ff's HTie People o f Michigan: A History and Selected Bibliography of the Races and N a tio n a litie s Who Settled Our State Other Ethnic Appeal Same as Ir is h Ethnic Appeal Catholicism Same as Iris h Ethnic Appeal Table A .3 . --S tre n g th o f p o l i t i c a l p a rty in low er s ta te p a rtis a n contests Year April 1949 April 1951 Nov. 1952 April 1953 O ffice Party's Percentage of Vote --------------------------------------------------------------------------------------------------------------------------1 2 3 4 5 6 7 8 9 10 11 12 13 State Board of A griculture 43.60 54.69 .99 .72 State Board of A griculture 38.01 60.17 .54 .96 .31 State Treasurer 47.12 (Auditor General) 46.77 52.42 .13 .28 .05 52.90 .28 .05 State Board of Agriculture 37.97 60.95 .74 .33 State Board of A griculture 50.20 49.05 .53 .2 2 Nov. 1956 Auditor Gen. 50.42 49.34 .25 A pril 1957 State Board of Agriculture 51.69 47.38 .94 State Board of A griculture 49.36 49.53 .53 .41 Nov. 1960 Auditor Gen. 52.03 47.71 .05 .08 .06 A pril 1961 Trustees of Mich. S. Univ. 50.65 48.30 .29 .32 .27 Nov. 1962 Auditor Gen. 50.32 49.46 .2 2 A pril 1963 Trustees of MSU 49.28 50.05 .67 A pril 1955 April 1959 Table A .3 . --C o n tin u ed P arty's Percentage of Vote Year Nov. 1966 Nov. 1968 Nov. 1970 Nov. 1972 Nov. 1974 Nov. 1976 Nov. 1978 O ffi ce 3 4 5 1 2 Trustees of MSU 46.38 53.24 Trustees of MSU 52.32 46.97 Trustees of MSU 54.02 44.50 State Board of Education 45.64 50.86 (Trustees of MSU) 47.89 50.86 Trustees of MSU 53.30 43.15 Governors of Wayne State U niversity 52.06 45.32 .43 (Trustees of MSU) 52.44 45.50 .97 Trustees of MSU 51.85 48.14 6 7 8 9 10 11 12 .38 .28 .92 2 .0 2 1.28 .23 .2 0 .57 .27 .71 .50 .72 .52 1.38 .28 .41 .55 .28 .28 .79 .31 Table A .3 . --C o n tin u e d Party's Percentage of Vote Year O ffice 1 Nov. 1980 Nov. 1982 2 3 4 5 6 7 Trustees of MSU 47.69 48.70 1.36 2.24 Trustees of MSU 55.65 41.31 1.07 1.97 ^Democratic o Republican 3 American Independent Party 4 Conservative 5 Human Rights ^Independent American ^Libertarian O New P o litic s g Progressive ^ P r o h ib itio n is t ^ S o c ia lis t Labor 12 S o c ia lis t Worker ^ U .S . Labor Party 8 9 10 11 12 13 Table A .4 . --Supreme Court c a n d id a te 's s e c tio n a l su p p o rt, -1982 Year S ection V o te /S ta te Vote April 1949 April 1949 April 1951 November 1952 April 1953 April 1955 April 1955 November 1956 November 1956 Candidate S e c tio n * % Support2 E.M. G.E. C.W. A.E. M.A. Sharpe* Bushnell** Rigney Moore Eden Outstate D e tro it Metropolitan Outstate Detroi t Metropoli tan D e tro it Metropolitan 212727 178101 90117 79700 36989 374064 252193 374064 252193 252193 E.R. N.E. M.A. J.H. T.P. C.L. Boyles** Reid* 1 Eden Lee Ryan Simmons Outstate Outstate D e tro it Metropolitan D e tro it Metropolitan Outstate Detroi t Metropolitan 201923 220801 20493 63788 75931 26058 326764 326764 201855 201855 326764 201855 C.J. Adams** M.A. Eden C.H. King Outstate D e tro it Metropolitan D e tro it Metropolitan 588255 77908 209113 1090892 760963 760963 54 J.R. Dethmers** H.F. K elly* C.J. Adams* T. Smith Outstate D e tro it Metropolitan Outstate Outstate 215918 148123 198579 75377 373994 254354 373994 373994 58 58 53 L.W. E.F. S .J. D.H. Outstate Outstate Outstate Outstate 299123 217367 153963 264763 467611 467611 467611 467611 64 47 33 57 Outstate Upper Peninsula D e tro it Metropolitan Outstate 552000 53394 655149 384595 1133368 90485 799811 1084873 49 59 82 36 Carr** Black* Roth Brake T. Smith* 1 M.D. O'Hara G. Edwards** J. Simpson 57 71 24 32 15 62 6 8 5 16 23 7 10 28 20 Table A .4 . — Continued Year Candidate Section^ S ection V o te /S ta te Vote % Support2 April 1957 T. Smith* 1 T.M. Kavanagh* R.E. Childs M.D. O'Hara Outstate Outstate D e tro it Metropolitan Upper Peninsula 249088 254657 71332 34035 471052 471052 301640 50829 53 54 24 67 April 1957 J.D. Voelker** J.A. Moynihan Upper Peninsula Detroi t Metropoli tan 34194 131966 49704 305774 69 43 April 1959 J.D. Voelker** G. Edwards*I K.W. Cole M.F. Cole W.H. Baldwin Upper Peninsula D e tro it Metropolitan Outstate D e tro it Metropolitan D e tro it Metropolitan 34059 240719 131163 66364 15748 44595 317522 453254 317522 317522 76 76 29 21 5 November 1960 T. Souris* 1 J.R. Breakey, Jr. D e tro it Metropolitan Outstate 449603 617361 671791 1211182 67 50 April 1961 J.R. H.F. J.H . E.C. Dethmers** K e lly *! McLaughlin Boehm Outstate D e tro it Metropolitan Outstate D e tro it Metropolitan 392481 185624 195388 104486 593816 269298 593816 269298 M.D. P.L. O.M. L.D. O'Hara* Adams!. Smith* McGregor Upper Peninsula Upper Peninsula Outstate Outstate 44551 36809 521861 564888 81360 81360 1760339 1760339 55 45 48 52 P.L. E.F. R.G. D.E. Adams* I Black*J Smith Holbrook Upper Peninsula Outstate Outstate Outstate 35910 455084 376937 362139 57713 786145 786145 786145 62 58 48 46 November 1962 November 1962 A pril 1963 66 69 33 39 Table A .4 . — Continued Year S e c tio n ! Candidate S ection V o te /S ta te Vote % Support* November 1966 T.M. Kavanagh*! T.E. Brennan* O.M. Smith1 M. Warshawsky Outstate D e tro it Metropolitan Outstate Outstate 712679 267144 506348 308027 1024998 472134 1024998 1024998 70 57 49 30 November 1968 T.G. Kavanagh* M.D. O'Hara! D e tro it Metropolitan Upper Peninsula 326152 57868 603787 91871 54 63 November 1970 G.M. W illiam s* J. Swainson* J.R. Dethmers I E.S. Piggins D e tro it Metropolitan D e tro it Metropolitan Outstate D e tro it Metropolitan 375926 437536 600744 293640 637963 637963 1264272 637963 59 67 48 46 November 1972 M.S. Coleman* C.L. Levin* V .J. Brennan Z. Ferency H.W. Gilmore W.J. Beer R.L. Evans W.A. Ortman J.S . Thorburn Outstate D e tro it Metropolitan D e tro it Metropolitan D e tro it Metropolitan D e tro it Metropolitan Outstate D e tro it Metropolitan D e tro it Metropolitan Outstate 703190 25729 206909 133237 143033 168217 160722 40566 431407 1641005 650386 650386 650386 650386 1641005 650386 650386 1641005 43 40 32 20 22 10 25 Outstate Outstate Outstate D e tro it Metropolitan 693273 903135 319354 292826 1258244 1258244 1258244 468046 55 72 25 63 D e tro it Metropolitan D e tro it Metropolitan Outstate D e tro it Metropolitan 307226 245443 176590 27467 638031 638031 1902861 638031 48 38 9 4 November 1974 J.W. Fitzg erald * T.M. Kavanagh*! A.C. M ille r B. Moody, Jr. November 1976 T.G. R.S. J.P . J.E . Kavanagh* Gribbs Swallow Wells 1 1 6 26 Table A .4 . --C o n tin u ed Year Section! Candidate November 1976 November 1976 J.L . Ryan*I C. Kaufman B. Moody, J r .* L.B. Lindemerl Z. Ferency Section Vote/State Vote D e tro it Metropolitan D e tro it Metropolitan D e tro it Metropolitan Outstate D e tro it Metropolitan 283963 282911 307373 720667 124612 566877 566877 595878 1795235 595878 50 50 52 40 D e tro it Metropolitan D e tro it Metropolitan Outstate D e tro it Metropolitan 394355 305370 479440 146891 488817 488817 1563220 488817 81 62 31 30 70 70 23 November 1978 G.M. W illiams * J .L . Ryan*I G.R. McDonald A. G ilb e rt November 1980 M.S. Coleman** C.L. Levin** E. Burch M. Hegarty J.B. Sullivan Outstate D e tro it Metropolitan D e tro it Metropolitan D e tro it Metropolitan D e tro it Metropolitan 1344832 425496 140641 73485 197700 1923275 603973 603973 603973 603973 B. Moody, J r . * 1 M.F. Cavanagh* D.C. R iley R. Campbell P.W. Avery D e tro it Metropolitan Outstate D e tro it Metropolitan Outstate Outstate 382844 793952 232028 244755 155952 483824 1504645 483824 1504645 1504645 November 1982 Section: 1 * Incumbent 21 12 33 79 53 48 16 10 D e tro it Metropolitan Area: Wayne County Outstate: A ll counties not included in the other two sections Upper Peninsula: A lger, Baraga, Chippewa, D e lta , Dickinson, Gogebic, Houghton, Iro n , Keweenaw, Luce, Mackinac, Marquette, Menominee, Ontonagon, and Schoolcraft counties. 2The vote per seat was approximated by dividing by the number of contested seats. in someinstances, the percentages do not sum exactly to 1 0 0 percent. *Winner % Support2 Therefore, 149 Table A .5 . — Court o f Appeals can didate s e c tio n a l supp ort, 1970-1982 Year Candidate County Votes % Support D is tric t 1970 1974 R.J. Danhof Ingham Second 33865 55677 61 D.R. Freeman Genesee Second 71241 99116 72 R.B. Burns Kent Third 67582 86655 78 J.E. Carr Branch Third 4185 7169 58 R.M. Maher Wayne F irs t 276055 450328 61 N.J. Kaufman Wayne F irs t 259356 450328 58 J.A. Sullivan Wayne F irs t 176221 450328 39 H. Heading Wayne F irs t 189023 450328 42 R.J. Danhof Ingham Second 45575 60433 75 S.J. Bronson Oakland Second 128724 154067 84 Genesee 71765 OCCUIIU SI254 M.J. Kelly Oakland Second 83185 161350 52 M.F. Cavanagh Ingham Second 47407 62041 76 G.E. Montgomery Macomb Second 47002 120859 39 B.L. Kaufman Oakland Second 89696 161350 56 D.F. Walsh Otsego Third 2903 2630 110 D.E. Holbrook, Jr. Clare Third 3555 4182 85 D.R. Freeman 88 150 Table A .5 . --C o n tin u ed County Year Candidate Votes % Support D is tr ic t 1974 1976 1978 1980 1982 R. Richardson Saginaw Third 24956 38366 65 J.T . Letts Kent Third 77167 91068 85 Wayne F irs t 314397 529925 59 R.L. Evans Wayne F irs t 215527 529925 41 W.R. Beasley Oakland Second 149223 237298 63 E.T. Fitzgerald Shiawassee Second 12520 19432 64 W.P. Cynar Macomb Second 112029 163078 69 E.T. Fitzgerald Shiawassee Second 14019 19165 73 B.B. MacKenzie Emmet Th i rd 5720 7218 79 K.L. Block Kent Third 55483 113354 49 R.J. Danhof Ingham Second 63124 79610 79 S.J. Bronson Oakland Second 196770 239872 82 J .J . Murray Macomb Second 79138 149122 53 G.S. A lle n , Jr. Kalamazoo Third 48098 55050 87 T.M. Burns Saginaw Third 47363 55616 85 P.S. Hirzel Calhoun Third 15083 31270 48 R.S. Gribbs Wayne F irs t 287965 527650 55 G.B. Ford Wayne F irs t 239654 527650 45 D.C. Riley Table A.6.--Supreme Court candidates, 1948-1982 C and idate 102491* 102491 102491 102491 102491 102511 102511 102511 102511 102511 102511 104522 104522 104522 102531 102531 102531 102531 102551 102551 102551 102551 104562 104562 104563 104563 102571 102571 102571 102571 102572 102572 102591 102591 102591 102591 E.M. Sharpe G .E. Bushnell C.W. Rigney A .E . Moore M.A. Eden E .R . Boyles N .E . Reid M.A. Eden J .H . Lee T .P . Ryan C .L . Siimions M.A. Eden C .H . King C .J . Adams J .R . Dethmers C .J . Adams T . Smith H .F . K e lly L.W. C a rr E .F . B lack S .J . Roth D .H . Brake T . Smith M.D. O 'H ara G. Edwards J . Simpson T . Smith T .M . Kavanagh R. C h ild s M .D. O 'H ara J .D . V o e lk e r J .A . Moynihan J .D . V o e lk e r G. Edwards K.W. Cole M .F. Cole Outcome Incumbency P a rty 1 1 0 0 0 1 1 0 0 0 0 0 0 1 1 0 0 1 1 1 0 0 1 0 1 0 1 1 0 0 1 0 1 1 0 0 2 1 0 0 0 2 1 -1 0 0 0 -2 0 1 2 2 0 0 3 0 0 0 1 0 1 0 2 0 0 -1 1 0 2 2 3 0 44. 44. 55. 55. 1. 60. 60. 1. 38. 38. 1. 0. 52. 47. 61. 38. 38. 61. 49. 50. 50. 49. 50. 49. 50. 49. 52. 52. 47. 47. 52. 47. 49. 49. 1. 50 . S ta te 0. 0. 0. 0. 0. 8. 0. 0. 0. 5. 0. 0. 0. 4. 3 .7 4. 0. 8. 9. 2. 2. 12. 3. 0. 0. 0. 3. 3. 0. 0. 0. 0. 0. 0. 0. 0. Local 4. 7. 8 .3 1 7 .3 0. 11. 5 0 .3 0. 4 0 .3 0. 2. 0. 0. 8. 11. 8. 0. 8. 2 4 .8 1. 8 .3 13. 0. 0. 15. 36. 0. 11. 7 .3 0. 20. 4 3 .3 20. 15. 3. 2 9 .3 Name 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 0 0 0 0 S e c tio n a l O ther E th n ic Cathol ic Ir is h Dichotomous Incumbency 57. 71. 24. 32. 15. 62 . 68. 5. 16. 23. 7. 10. 28. 54. 58. 53. 20. 58. 64. 47 . 33 . 57. 49. 59. 82. 36. 53. 54. 24. 67. 69. 43 . 76. 76. 29. 21. 1 1 0 0 0 1 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 1 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 1 0 0 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 1 0 1 0 1 0 0 0 0 1 1 0 0 0 1 1 0 0 0 0 0 0 1 1 1 0 0 1 0 0 0 1 0 1 0 1 0 0 0 1 0 1 1 0 0 Table A.6.—Continued C and idate 102591 104602 104602 102611 102611 102611 102611 104622 104622 104623 104623 102631 102631 102631 102631 104661 104661 104661 104661 104681 104681 104701 104701 104701 104701 104721 104721 104721 104721 103721 104721 103721 104721 104721 104741 104741 W.H. Baldwin T. S o u ris J .R . B reakey, J r . H .F . K e lly J .H . M cLaughlin E.C . Boehm J .R . Dethmers M.D. O 'H ara P .L . Adams O.M. Smith L .D . McGregor P .L . Adams E .F . Black R.G. Smith D .E . Holbrook T.M . Kavanagh T .E . Brennan 0 . Smith M. Warshawsky T .G . Kavanagh M.D. O 'Hara G.M. W illia m s J . Swainson J . Dethmers E. P ig g in s M .S. Coleman C .L . Levin V .J . Brennen Z. Ferency H. G ilm ore W. Beer R .L . Evans W. Ortman J. Thorburn J.W . F itz g e r a ld T.M . Kavanagh Outcome Incumbency P a rty 0 1 0 1 0 0 1 1 0 1 0 1 1 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 0 0 0 0 0 0 1 1 0 1 0 1 0 0 3 -2 1 1 0 0 1 0 0 1 0 2 0 0 -1 0 0 4 0 0 0 0 0 0 0 0 0 0 1 2 50. 52. 48. 48. 51. 51. 48. 49. 5 0. 5 0. 49. 49. 49. 5 0. 50. 46. 5 3. 46. 53. 52. 47. 54. 54. 45. 45. 51. 0. 0. 1. 46. 2. 46. 0. 51. 43. 53. S ta te 0. 5. 0. 8. 5. 0. 3 .7 0. 6. 4 .3 0. 6. 2. 0. 0. 3. 0. 4 .3 3. 0. 0. 16. 4. 3 .7 0. 0. 0. 3. 12. 4 .5 0. 0. 0. 0. 0. 3. Local 0. 1. 1 4 .7 8. 0. 2 6 .3 11. 0. 6. 4. 3 .7 6. 1. 9 .8 2 2 .3 11. 5. 4. 13. 4. 0. 0. 9 .5 11. 15. 15. 6. 7. 0. 18. 14. 6. 4. 15. 16. 11. N 0 0 0 0 0 0 0 1 1 0 0 1 0 1 0 1 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 S E C tW 5. 67. 50. 69. 33. 39. 66. 55. 45. 48. 52. 62. 58. 48. 46. 70. 57. 49. 3 0. 54. 63. 59. 6 7. 48. 46. 43. 40. 3 2. 2 0. 22. 10. 2 5. 6. 26. 5 5. 73. ° ^ c catholic 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1 0 0 1 0 1 0 0 0 0 0 1 1 1 0 1 1 0 0 0 0 0 0 1 0 0 0 0 1 0 0 1 ■H.h 0 0 0 1 1 0 0 1 0 0 0 0 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1 1 0 1 0 1 0 0 1 0 1 1 0 0 1 0 0 1 0 1 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0 0 1 1 err ro Table A.6.—Continued C and idate 104741 104741 104761 104761 104761 104761 104762 104762 104763 104763 104763 104781 104781 104781 104781 104801 104801 104801 104801 104801 104821 104821 104821 104821 104821 A .C . M i l l e r B. Moody, J r . T .G . Kavanagh R .S . G ribbs J .P . Swallow J .E . W ells J . L . Ryan C. Kaufman L. Lindemer Z. Ferency B. Moody, J r . G.M. W illia m s J . L . Ryan G .R. McDonald A. G ilb e r t M .S. Coleman C .L . Levin E. Burch M. H egarty J .B . S u lliv a n M .F . Cavanagh B. Moody, J r . R. Campbell D .C . R ile y P.W. Avery ♦ I d e n t i f i e d Code: Outcome Incumbency P a rty 0 0 1 0 0 0 1 0 0 0 1 1 1 0 0 1 1 0 0 0 1 1 0 0 0 0 0 1 0 0 0 1 0 1 -1 -1 1 2 0 0 1 1 0 0 0 43. 53. 0. 52. 45. 1. 45. 52. 45. 0. 52. 52. 48. 52. 48. 49. 0. 48. 49. 48. 56. 56. 41. 41. 2. S ta te 0. 0. 0. 0. 0. 0. 0. 0. 5 .8 12. 0. 16. 0. 0. 0. 0. 0. 1. 4. 2. Local 16. 9. 5. 12. 14. 0. 13. 18. 3. 0. 11. 0. 13. 7. 12. 15. 6. 0. 0. 9. Name 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 S e c tio n a l 25. 63. 48. 38. 9. 4. 50. 50. 40. 21. 52. 81. 62. 31. 30. 70. 70. 23 . 12. 33. 53. 79. 16. 48. 10. O ther E th n ic C a th o lic Ir is h 0 0 0 1 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 1 1 0 0 0 0 1 0 0 0 1 0 1 0 1 0 0 0 0 0 1 0 0 0 0 0 1 1 Dichotomous Incumbency 0 0 1 0 0 0 1 0 1 0 0 1 1 0 0 1 1 0 0 0 0 1 0 0 0 F i r s t and second d i g i t s on l e f t — le v e l 10 * Supreme C o u rt T h ird d i g i t — month 1 = F eb ru ary ; 2 = A p r i l , 3 = Septem ber; 4 = November Fourth and F i f t h D i g i t s — Year S ix th d i g i t — Type E le c tio n . 1 = r e g u la r ; 2 = vacancy; 3 = a d d it io n a l vacancy Table A .7 . --C o u rt o f Appeals c a n d id a te s , 1970-1982 Candidate 224702* R.J. Danhof Outcome S ectio n al Name 1 61. 0 1 0 6 . 1 . 1 E thnic C a th o lic S ta te Local Incumbency D.R. Freeman 0 72. 0 0 0 0 . 14. 0 234701 R.D. Burns 1 78. 0 0 0 0 . 6 . 1 234701 J.E . Carr 0 58. 1 0 0 0 . 4. 0 214741 R.M. Maher 1 61. 1 1 1 0 . 14. 0 214741 N.J. Kaufman 1 1 0 0 0 . 29. 0 214741 J.A. Sul1ivan 0 39. 0 1 0 2 . 23.7 0 214741 H. Heading 0 42. 0 0 0 0 . 7. 0 224741 R.J. Danhof 1 75. 0 1 0 6 . 1 . 1 224741 S.J. Bronson 1 84. 0 0 0 0 . 4. 1 224741 D.R. Freeman 0 8 8 . 0 0 0 0 . 18. 0 224741 M.J. Kelly 1 52. 1 1 1 0 . 0 . 0 224741 M.F. Cavanagh 1 76. 1 1 1 0 . 5. 0 224741 G.E. Montgomery 0 39. 0 0 0 0 . 3. 0 224741 B.L. Kaufman 0 56. 0 0 0 0 . 0 . 0 234741 D.F. Walsh 1 . 0 1 1 5. 7. 0 234741 D.E. Holbrook, J r. 1 85. 1 0 0 0 . 13. 0 234741 R. Richardson 0 65. 0 0 0 0 . 15. 0 234741 J .T . Letts 0 85. 0 0 0 0 . 19. 0 214762 D.C. R iley 1 59. 0 1 1 0 . 13.3 1 CO in 224702 1 1 0 Table A .7 . --C o n tin u ed Sectional Name Ethnic Catholic State Local Incumbency 214762 R.L. Evans 0 41. 0 0 0 0. 1 0 . 0 224761 W.R. Beasley 1 63. 0 0 0 0. 26. 1 224761 E.T. Fitzgerald 0 64. 1 1 1 0. 1. 0 224781 W.P. Cynar 1 69. 0 1 1 0. 224781 E.T. Fitzgerald 0 73. 1 1 1 0 . 1. 0 234781 B.B. MacKenzie 1 79. 0 1 0 0 . 0. 0 234781 K.L. Block 0 49. 0 0 0 0. 0. 0 224801 R.J. Danhof 1 79. 0 1 0 6 1. 1 224801 S.J. Bronson 1 82. 0 0 0 0. 4. 1 224801 J .J . Murray 0 53. 0 1 1 0. 15. 0 234801 6 1 • Outcome 00 Candidate 0 0 0 234801 T.M. Burns 1 85. 1 0 234801 P.S. H irzel 0 214821 R.S. Gribbs 1 214821 G.B. Ford 0 45. * Id e n tific a tio n Code: 1 1 2 . 1 1 7. 1 0 . 1 CO • . . . 0 0 0 0. 55. 0 1 1 0. 2 0 . 0 0 0 0 0. 18. 0 j* . S. A lle n , J r. 1 2 1 1 0. 0 F irs t and second d ig its on l e f t - - l e v e l 21 = f i r s t court appeals; 22 = second court appeals; 23 = th ird court of appeals Third d ig it — month 1 = February; 2 = A p ril; 3 = September; 4 = November Fourth and F ifth d ig its — year Sixth d i g i t - - type election 1 = regular; 2 = vacancy; 3 = additional vacancy 156 Table A .8 . --Comparison o f Supreme Court models fo r two periods 1970-:1980 1964-1980 A ttributes MLE SE MLE/SE MLE SE MLE/SE Incumbency -.738 .869 - .850 .637 1.078 .591 Party -.042 .027 -1.561 -.007 .049 -.140 Sectional .117 .041 2.865** .109 .047 2.295** State .096 .1 2 2 .786 .089 .164 .540 Local .1 0 0 .097 1.027 .075 .142 .531 1.626 .867 -1.117 1.275 -.876 Iris h 1.409 Other Ethnic -.652 1.291 -.505 Catholic -.383 1.498 -.256 1.097 1.538 .713 Name -.310 1.298 -.239 1.810 2.142 .845 -5.085 -6.310 .832 .894 -2 x LLR 33.645** 30.562** Percent Predicted Correctly 82.927 88.571 Constant Est. R2 rs K .646** .755** .562 .692 N = 41 **Indicates significance a t .05 le v e l. (C ritic a l values, Z = 1.64, r g = .305, x*g= 16.92 X2S= 15.51). SOURCES CONSULTED 157 SOURCES CONSULTED Cited Sources Address to the Michigan Press Association by L t. Governor James B rickley, January 30, 1981. Adrian, Charles R. "A Typology fo r Nonpartisan Elections." P o litic a l Quarterly 12 (1959): 449-58. Western A ldrich, John H. "Electoral Choice in 1972: A Test of Some Theorems of the Spatial Model of Electoral Competition." Journal of Mathematical Sociology 5 (1977): 215-237. Asher, Herbert. ""Voting Behavior Research in the 1980s: An Examina­ tio n of Some Old and New Problem Areas." Paper presented a t the 1982 Annual Meeting of the American P o litic a l Society Association, Denver, September 2 -5 , 1982. Barber, Kathleen. "Nonpartisan Ballots and Voter Confusion in Judicial Elections." Paper presented a t the Midwest P o litic a l Science Meeting, Milwaukee, 1982. ________ . "Selection of Ohio Appellate Judges: A Case Study in In v is ib le P o litic s ." In P o litic a l Behavior and Public Issues in Ohio, pp. 175-230. Edited by John J. Gargan and James G. Coke. Kent, Ohio: Kent State University Press, 1972. Book of the States, 1982-83. Governments, 1982. Vol. 24. Braden, John A. "The Way I See I t . " ber 1981): 963-965. Lexington's Council of State Michigan Bar Journal 80 (Decem­ Brand, George E. "Michigan State Bar's Work fo r Judicial Appointment." Journal of the American Judicature Society 22 (February 1939): 197-202. Campbell, Angus; Converse, P h ilip E.; M ille r , Warren E .; and Stokes, Donald E. The American V oter. Chicago: University of Chicago Press, 1960. Canon, Bradley C. "The Impact of Formal Selection Processes on the Characteristics o f Judges--Reconsidered." Law and Society 6 (1972): 579-593. Cardozo, Benjamin N. The Nature of the Judicial Process. Yale University Press, 1921. 158 New Haven: 159 Carpenter, W illiam S. Judicial Tenure in the United S tates. Haven: Yale U niversity Press, 1918. Case, Herbert S ., ed. Who's Who in Michigan. New 1936. Chapman, James. Elections O ffic e , Secretary of State of Michigan. In terview , 3 November 1981. Converse, P h ilip E.; Campbell, Angus; M ille r , Warren E.; and Stokes, Donald E. " S ta b ility and Change in 1960: A Reinstating E lection." American P o litic a l Science Review 55 (June 1961): 269-280. ; D etro it Free Press, January 1949 to March 1983. ( D e tro it Daily Press August-September 1964, substitute paper during s tr ik e ). D etro it News, January 1949 to January 1983. Donohue, Dennis. State Board of Law Examiners, State of Michigan. Interview , 21 July 1982. Downs, Anthony. An Economic Theory of Democracy. and Row, 1957. New York: Harper Dubois, P h ilip L. From B allo t to Bench: Judicial Elections and the Quest fo r A ccountability. Austin, Texas: University of Texas Press, 1930. Dunbar, W illis F. Michigan: A History of the Wolverine S ta te . Grand Rapids, M i.: W illiam B. E. Erdmans, 1965. ________• Michigan Through the Centuries: Family and Personal H isto ry, 3-4. New York: Lewi H istorical Publishing Co., i Ar r J L 3 0 0 . Easton, David. The P o litic a l System. New York: ________ . A Framework fo r P o litic a l Analysis. N.J7! P re n tic e -H a ll, 1965. Knopf, 1953. Englewood C lif f s , Eisenstein, James. P o litic s and the Legal Process. and Row, 1973. New York: Harper Fraser, Archie. Retired State of Michigan Public Administrator. Interview , August 1982. G raff, George. The People of Michigan: A History and Selected Bibliography of the Races and N a tio n a litie s Who Settled Our S tate. Lansing: Michigan Department of Education, 1970. Grand Rapids Press, September 1970 to November 1980. 160 Hannah, Susan B. "An Evaluation of Judicial Elections in Michigan, 1948-1968." Ph.D. d is s e rta tio n , Michigan State U n iversity, 1972. ________ . "Competition in Michigan's Judicial Elections: Democratic Ideals vs. Judicial R e a litie s ." Wayne Law Review 24 (July 1978): 1267-1306. ________ . "Voting in Local Judicial Elections: The Case of the Faithful E lecto rate." Paper delivered a t the 1982 Annual Meeting of the Midwest P o litic a l Science Association, Milwaukee, Wisconsin, 1982. Hawley, W illis D. P o litic s . Nonpartisan Elections and the Case fo r Party New York: John W iley, 1973. Heiberg, Robert A. "Social Backgrounds of the Minnesota Supreme Court Justices, 1858-1968." Minnesota Law Review 53: 901-937. Henderson, Bancroft C ., and S in c la ir, T. C. Judicial Selection in Texas: An Exploratory Study. University of Houston Studies in Social Science. Houston: Public A ffa irs Research Center, 1965. Herndon, James. "The Role of the Judiciary in State P o litic a l Sys­ tems." In Judicial Behavior. Edited by Glendon Schubert. Chicago: Rand McNally, 1964. Jacob, Herbert. "The E ffe ct of In s titu tio n a l Differences in the Recruitment Process: The Case of State Judges." Journal of Public Law (1964): 104-119. ________ . "Judicial In sulation— Elections, D irect P a rtic ip a tio n , and Public A ttention to the Courts in Wisconsin." Wisconsin Law Review (1966): 801-819. ________ . Justice in America. Boston: L i t t l e , Brown, 1965. Jew ell, Malcolm, and Olson, David M. American State P o litic a l Parties and E lections. Homewood, I I . : Dorsey Press, 1978. Kalamazoo Gazette, 14 October 1970, A-10. Kelman, Maurice. "B allot Designations: Their Nature, Function and C o n s titu tio n a lity ." Wayne Law Review 12: 756-779. Key, V. 0 . , J r. American State P o litic s : Alfred Knopf, 1956. An Introduction. New York: 161 Key, V. 0 . , J r. A Primer of S ta tis tic s fo r P o litic a l S c ie n tis ts . New York: Thomas Y. Crowell, 1966. ________• Southern P o litic s . Kmenta, Jan. New York: Elements of Econometrics. Vintage Books, 1949. New York: Macmillan, 1971. Krasicky, Eugene. Deputy Attorney General of Michigan. 3 September 1982. Interview , K rislov, Samuel. "Constituency vs. Constitutionalism : The Desegre­ gation Issue and Tensions and Aspirations of Southern Attorneys General." Midwest Journal of P o litic a l Science 3 (1959): 75-92. Ladinsky, Jack, and S ilv e r, A llan . "Popular Democracy and Judicial Independence: Electorate and E lite Reactions to Two Wisconsin Supreme Court Elections." Wisconsin Law Review (1967): 128-69. Lane, Robert. P o litic a l L i f e . Glencoe, I I . : Free Press, 1959. Lansing State Journal, January 1949 to March 1983. Laws Relating to E lections. Compiled by The Michigan Secretary of S tate, Chapters 18 and 19. Lee, Eugene. The P o litic s of Nonpartisanship. of C a lifo rn ia Press, 1960. Martindale-Hubbell Law D irecto ry. Inc. New York: Berkeley: University Martindale-Hubbel1, McHargue, Daniel. D irect Government in Michigan. Michigan Consti­ tutional Convention Studies #17. Prepared fo r MI. Con Con Preparatory Commission. McKelvey, Richard, and Zavoina, W illiam . "A S ta tis tic a l Model fo r the Analysis of Ordinal Level Dependent V ariables." Journal of Mathematical Sociology 4 (1975): 103-120. Michigan: A Guide to the Wolverine S ta te . W riters' Program of the Work Project Administration in the State of Michigan. New York: Oxford University Press, 1941. Michigan Biographies, 1-2. 1924. Lansing: Michigan H isto rical Commission, Michigan Chronicle, January 1966 to November 1980. 162 Michigan Court of Appeals Reports. Co-operative Publishing. Michigan Court Rulers, 1980. Company, 1980. Rochester, New York: St. Paul, Minn.: Michigan Reports. Rochester, New York: Publishier. Lawyers' West Publishing Lawyer's Co-operative Michigan State C onstitution, 1908. The Michigan State C onstitution, 1963. The Michigan Manual. Published Biannually by the Secretary of State, Lansing, Michigan. The Mining Journal. 1980. Marquette, Michigan. September 1970 to November Michigan Municipal League. Directory of Michigan Municipal O ff ic ia ls . Ann Arbor: Michigan Municipal League. Mooradian, Paul. Former Assistant to the Chair, Michigan Democratic Party. Interview , November 1982. Moos, Malcolm. "Judicial Elections and Partisan Endorsements of Judicial Candidates in Minnesota." American P o litic a l Science Review 35 (February 19411: 65-75. Nagel, Stuart S. Comparing Elected and Appointed Judicial Systems. Sage Professional Paper in American P o litic s , 04-001. Beverly H ills : Sage P ublications,1973. Nichols, Rudy J. "The Way I See I t . " (October 1981): 761-762. Michiqan Bar Journal 80 Niemi, Richard, and Weisberg, Herbert F ., ed. Controversies in American Voting Behavior. San Francisco: W. H. Freeman, 1976. O ffic ia l Canvass of Votes, State of Michigan. Published a fte r each regular state election by the Secretary of S tate's O ffic e , Lansing, Michigan. Ostrom, Charles W., Jr. Time Series Analysis: Regression Techniques. Beverly H ills : Sage Publications, 1978. Pindyck, Robert, and Rubinfeld, Daniel L. Econometric Models and Economic Forecasts. New York: McGraw-Hill, 1976. 163 Pollock, James K ., and Eldersveld, Samuel. Michigan P o litic s in T ran sitio n . U niversity of Michigan Government Studies #10. Ann Arbor: U niversity of Michigan Press, 1942. Pomper, Gerald. "Ethnic and Group Voting in Nonpartisan Municipal Elections." Public Opinion Quarterly (Spring 1966): 79-97. Ranney, Austin. "Parties in State P o litic s ." In P o litic s in the American S tates. Edited by Herbert Jacob and Kenneth N. Vines. Boston: L i t t l e , Brown and Co., 1965. ________ . " U t ilit y and Lim itations of Aggregate Date in the Study of Electoral Behavior." In Essays on the Behavioral Study of P o litic s , pp. 91-102. Edited by Austin Ranney. Urbana: University of Illin o is Press, 1982. Rice, S tuart. Q uantitative Methods in P o litic s . Knopf, 1928. New York: Alfred Rizzo, Anthony V. Atlas of Michigan's Foreign Born Population. born, M i.: Free World Press, 1968. Ryan, Theodore. Dear­ Interview , 4 August 1982. Sayre, Wallace S ., and Kaufman, Herbert. New York: Russell Sage, 1960. Governing New York C ity . Schlesinger, Joseph.- "The Structure of Competition fo r O ffice in the American States." Behavioral Science 5 (July 1960): 197210 . Schubert, Glendon. "Academic Ideology and the Study of Adjudication." American P o litic a l Science Review (March 1967): 106-29. ________ . "The Packing of the Michigan Supreme Court." In Quantita­ tiv e Analysis of Judicial Behavior. Glencoe: The Free Press, 1959. Simon, Herbert A. Adm inistrative Behavior, 3rd ed. Press, 1976. _. The Sciences of the A r t i f i c i a l . M .I.T . Press, 1969. New York: Cambridge, Mass.: Smith, Elsdon C. New Directory of American Family Names. Harper and Row, 1956. Free The New York: Spaeth, Harold J. Supreme Court Policy Making: Explanation and Pre­ d ic tio n . San Francisco: W. H. Freeman, 1979. 164 Special Commission to Review A rtic le VI of the Michigan Constitution. The State News, 1 March 1983. Stenger, John. Michigan Court of Appeals Research S ta ff. 7 September 1982. S tieb er, Carolyn. The P o litic s of Change in Michigan. Michigan State U niversity Press, 1970. Interview , East Lansing: Strum, A lbert. Constitution-Making in Michigan, 1961-1962. Michigan Governmental Studies No. 43. Ann Arbor: University of Michigan Press, 1963. ________, and Whitaker, Margaret. Implementing a New Constitution: The Michigan Experience. The University of Michigan Governmental Studies, #50. Ann Arbor: In s titu te of Public Adm inistration, the University of Michigan, 1968. Supreme Court of Michigan: Annual Report and Judicial S ta tis tic s . Published annually by the O ffice of the Court Administrator. Lansing, Michigan. Taylor, C. W. J r ., ed. Bench and Bar of Michigan, 1938. Francisco: Knight-Counihan. San Teeter, Robert M ., President. Market Opinion Research, D e tro it, Michigan. L e tte r, 24 February 1983. Thomas, Norman. C. "The Electorate and State Constitutional Revisions: An Analysis of Four Michigan Referenda." Mid­ west Journal of P o litic a l Science 12 (February 1968): 115-29. Ulmer, Sidney. "The P o litic a l Party Variable in the Michigan Supreme Court." Journal of Public Law 11 (1962): 353-362. Vines, Kenneth N. "The Selection of Judges in Louisiana." In Studies in Judicial P o litic s V I I I . Edited by Kenneth N. Vines and Herbert Jacob. Tulane U niversity, New Orleans, La.: Tulane Studies in P o litic a l Science, 1962. Watson, Richard A ., and Downing, Rondal G. The P o litic s of the Bench and the Bar: Judicial Selections under the Missouri Nonpartisan Court Plan. New York: John W iley, 1969. Who's Who Among Black America, 3rd ed. America Inc. Northbrook, I I . : Who's Who Among Black Am. Wolk Krouse, 1980. W illiam s, O liver P ., and Adrian, Charles R. "The Insulation of Local P o litic s Under the Nonpartisan B a llo t." American P o litic a l Science Association 53 (1959): 1052-1063. 165 W illiams, O liv er, and Adrian, Charles R. Four C itie s : A Study in Comparative Policy-Making. Philadelphia: University of Pennsylvania Press, 1963. General References Adamany, David, and DuBois, P h ilip . "Electing State Judges." Wisconsin Law Review (1976): 731-779. A ldrich, John H ., and Cnudde, Charles F. "Probing the Bounds of Conventional Wisdom: A Comparison of Regression, P ro bit, and Discriminant Analysis." American Journal of P o litic a l Science 19 (August 1975): 571-608. Baum, Lawrence. "Judicial P o litic s in the 1980s: One View of the F ield ." Paper delivered a t the 1982 Annual Meeting American P o litic a l Science Association, Denver, September 2 -5 , 1982. Berle, Adolf A ., Jr. "Elected Judges--or Appointed?" In The Courts A Reader in the Judicial Process, pp. 97-103. Edited by Robert Scigliano. Boston: L i t t l e , Brown and Co., 1962. Eden, Morton A. L etter 19 August 1982. Eubank, Robert B ., and Gow, David John. "The Pro-incumbent Bias in the 1978 and 1980 National Election Studies." American Journal of P o litic a l Science 27 (1983): 122-139. F u lle r, George N ., ed. H istoric Michigan: Land of the Great Lakes. V. 1-3. National H istorical Association, In c ., N.D. G ilb e rt, Charles E. "Some Aspects of Nonpartisan Elections in Large C itie s ." Midwest Journal of P o litic a l Science 6 (1962): 34562. Haynes, Evan. "Judicial Selection and the Democratic S p ir it." In The Courts: A Reader in the Judicial Process, pp. 57-69. Edited by Rovert Scigliano. Boston: L i t t le Brown and Company, 1962. Jacob, Herbert, and Vines, Kenneth N ., eds. P o litic s in the American States: A Comparative Analysis. Boston: L i t t l e , Brown and Co., 1965. Johnson, Charles A .; Shaefer, Roger C .; and McKnight, R. Neal. "The Salience of Judicial Candidates and Elections." Social Science Quarterly 59 (September 1978): 371-378. 166 K rehbiel, K eith, and W right, John R. "Incumbency E ffe ct in Congres­ sional Elections: A Test of Two Explanations." American Journal of P o litic a l Science 27 (1983): 140-157. McKnight, R. Neal; Schaefer, Roger; and Johnson, Charles A. "Choos­ ing Judges: Do the Voters Know What They're Doing?" Judicature 62 (August 1978): 94-99. M ille r , Arthur H ., and Wattenberg, Martin P. "Measuring Party Id e n tific a tio n : Independent or No Partisan Preference?" American Journal of P o litic a l Science 27 (1983): 106-121. O ffic ia l Record of the Michigan Constitutional Convention, 1961-62. Lansing, Michigan, 1962. P arenti, Michael. "Ethnic P o litic s and the Persistence of Ethnic Id e n tific a tio n ." American P o litic a l Science Review 61 (1967): 717-726. Quinn, Timothy C ., and B ea ttie , Stanley E. "A New Method of Judicial S election." Michigan State Bar Journal 32 (May 1953): 30-31, 39-44. Rae, D. The P o litic a l Consequences of Electoral Laws. Yale U niversity Press, 1967. New Haven: Sheldon, Charles H. "Influencing the Selection of Judges: The Variety and Effectiveness of State Bar A c tiv itie s ." Western P o litic a l Quarterly 30 (1977): 397-400. Solomon, Freda. "E ffect of Judicial Selection on Judge C haracteristics and A ttitu d e s ." Paper presented a t the Annual Meeting of the Midwest P o litic a l Science Association, Chicago, I l l i n o i s , April 24-26, 1980. Stokes, Donald, and M ille r , Warren E. "Party Government and the Saliency of Congress." Public Opinion Quarterly 26 (Winter 1962): 531-546. Uppal, Jag C. "Approaches to the Selection of Judges." State Govern­ ment 47 (1974): 46-49. Walker, Jack L. "B allot Forms and Voter Fatique: An Analysis of the O ffice Block and Party Column B a llo t." Midwest Journal of P o litic a l Science 10 (1966): 448-463. Watson, Richard A.; Downing, Rondal G.; and Spiegel, Frederick C. "Bar P o litic s , Judicial Selection and Representation of Social In te re s ts ." American P o litic a l Science Review 61 (1967): 54-71.