Q1 . . 57$ : . , . . .5 .fimmmmutsfi... $731.39... ._.V..,.V. ._..V-. 1...- - . : .5... 171 . .y. .. .... 2...... .... i . ma... gig m. . 3.. has-.. .. V . 3." 0.0.... Mkfiwughttzzfihri I» s». . 11 \1H 1... . \ .... (2.5.35. .1... . I . . . .54....fiazkrfifivsk! . $14.“... . . a . Rm? .mfiflxfimvhzi .. .3... . -.. \ 7V. 4.7.0§Jh7s53. u: f \\ V at \ I.) ... . .. ..\ . . .079? .sftflliumwi. , .. 1. _. 3.. m2... . 391......»“13241. 131$}. . .3. 2 \I 7. 57:7...1‘7.‘ shad? razzbsksfiinifilhpizg 1.\. 12“”... {$1} :1 .71.»... 187277.917: . . Ertfiwkkl z... hum... .3 _ V . . lwnfiulmélx afiggili 1:727...an . .fiifizrflhfiv .. 71;}7.7¢1..7E.Hm¥.twh.anuxn.bah“?.wd...1.t..t....L :77. .m. 1 . gt“ ~74M7xiw7‘77. . . .7755! 44747.7... 13!. is. 1. \ 1 Tuih‘uhtfill s.»%77{n777 31 . txfigti LIL... 87.7.. 97!.) Hi . . V 3:41...‘ 122.... 13%....3... hmmuflx ; .. , ............. . ha a... s 1 .17 71.2117 $132.2}xfixgz. .. x L . 31.11)... ‘flfik . 77311313211: W11 1.1.1.11. I . . a 735%.)...“1. c)... 14.33%? 59$ 3 i. “2711. . n n... fléfinnflmflfl . 1m 1‘ 71 \ . i 4.. 1 112.7.Eta11 3 . - . 2 .. . 5... We .5»... .. \ x» \. t. "6.. . t . )1! .7 7.1%! 711111 V . it.” .7171 a ?\ \.&. .7....71.0fl.§‘. . 11‘...» . . . .{h-hvvflvq . 1 .375) . r .l . 17.‘ 1311 ‘7 7) 771701...» . - V. .2 - 3. .mfiximmww... 7...... 1%?»31... . . . .1 V . .1 7.1.1 ”@1717 it . .1311111A1a V 71 1.1.1 I... \\ x131.un.1u......$. 11151.“.71 $1 - 1. 311....-71137». 113.931.31U3»... 11113.11 $6.1. .1 . . IV\37\.§, _ 1. .177) . . a .1. .2127... 1.1%.... .3... 7.7.5.... 43%..7733. . 2...... -_ .&§§%713$¥7 -23.»)... 1.7 .15... 7.174.. ». V .. VI 1 . . . ‘ . . . .Ql . . , . . V \. . 7»... ..\.a. 1:. his» 71 \ , . . . \ it{%}£§1\\ f... 1.71 . 1 x . ) 71!... .71.. \\...u.\(...t.\\\hv\mmx iiififigghhriil‘ué; 7194.2. ..}7§.i3‘1\51&3$}f§h.10fi7§ 5. .. 7.11.1“. Inflow 1 §?§§§ . .71.“...i Rn” )51531.u¢1ial‘\. 1, . 11935273371 alflafluufluun‘ 3...... A. . . . 31117373113317.7171 E17177... (VLPM... 17.14:...51 .. 1.1.“..113 V .1. 1.33% .. . \ H9415; .. 1171.. . u .’ a... a! I 7: 7.377777377». .v\. 171:7- .27.:iiizflfqtt.‘ . . 171.3% .33....vxv73717i297177l9. .. y a... 0 177779 E r?“ . . r t 1. .i. 7?% . ..1 .l 7711.. . \s‘llp}: 3353)... . u aghai 71).. E77117. ”(uh-77... .177. .. . 511%.. {ohm . - . gagig7iflfilfig .9021; . s. . V .v... . . .n... i3h7..sx....«$n.z§<1.. 15...}...1 194.113.. 3.1...iwaaiifln... §.§1{bvflanu.i2cfi3.§1 1. . . Ic‘ 171...“...w-{tm ‘v‘lxlafiflfi. 111.01% 11‘ i::% . , 71.72%... lug): 1111;431:1271. .31}... 2.1. iiiflisslcs:§i}aéi& . . . . 57.1 . 11 I . V '1 Iixiyklirw. .1 1% g»; zfl§2171277§31m1§ . . . \77417771uhw‘flfinitLi7717i‘HM-r 77gl‘1‘» . ‘Ig‘iz‘ii 2‘“\ ‘nvA'llilnl‘x‘ (Kill 2 K 4: . 3.2.}... . shakfidi... Lg..3.rzs...um...mm . 7 1) {7% 1 . N x . 7. 1.11.1 . . . 5 1‘. .77 flirtilélshurur 71.77.. . ii: .30..“ . 1 1717.133 . .\ . .77.... 7.. . . 177.177 \ {.{thfluflfliii‘fl‘lYi 7‘07 . In...“ 1 . . . 725137711... 111?}: 11.1.». 137‘. .11!) ifixfiixégglixW97ififio3§ 1‘72... $.51. .71....) 7‘ 1.: 7.1(37. 4.36%... . ...MH.§$. g. 01. 1.1. i. am: Influx-NV.“ 764151.117). :. . . , . v». a); .11 .2751‘lhe .1 1Efi~§sll %§ALLK.. .. r .. . ‘4‘, ‘ Q. (I . . . 13%. ”a: .. 111% \ . . . . 7757737. 4v. 7 Etlél...t 4177. V . 77413777317 . . V . aging: 171.117.7354). in}; 017.371.31.711. 317711 7 1 1)) 7‘4 1‘ \ z r . . .éli7zisiilhasllzaefiaawfizaisglan 7? . 31773 M . 113%“:13‘7 17‘ . \ ‘i‘iflwfigz‘778911112. .777». . E11111 . . I . I5. . u . 7777'. 447.777! «lazy-$71.57*“. . . . 751.417....»7..\7\«l7.01h I- .I .1 .. 7.. . .. $4.91.. V H.....u.s.zx..v§.i . . V . 7727.1371su... . il.‘ . a . l Ell-7.. . .EI-yxi’bt I: I‘ll-3311747 311177177 iia‘twflttai .- I 1 . 707117.... : .5. . 1712......»- g....li.1ni.“.\1~o7 . . «ammul‘ . . z . 71717151127717“ 10779111477\ 1771. .1. $.31. . 877:..(1. 37.. ... .a . \ u. .- .1 . f 1:71.17!!- . .V t t\. 17.3.17: .31.... {.17. , . .7 . i. W777 . . ..=..._. ....n. 1777177... .7191! ...h i. $41.7 [I E74717 $7777.74: 1 I 7777.. .V!L777.7v7¥777t:7 l . rt... . 777$} v 595771.747}? .. . 77.71771... (1.4171 .717 O 1 1577571777. . 5.3.773}... . .. v ‘7: 7.1..r..47...r\.21(.l>)u5 49727.7(; 1.15.7! 1771777.. u. nil! 3....- . . (7.77.7? 17721777777797; . ((77:77...) 7777173711.»... x7.7l.nsllli)r?9$7.7t7». 7! 79 "'77... » Invghzfixi. "HM-LN" I7 717.77.77.77! . 74577771077. {1:71. 7. a- . I v1.1 v? . 0'75. .7. ..»7(.v\(\7 ~ 7377. L .(II .. at. E37711...) ‘iigzlxl .77. 44.77.7716?! . _ \ 1.7477111177115771; V . .. 17977.) #7179775. 1.3241 .777. L111! .. .. 3 I777 .73.! .1 7... t v t . {1.777.717 777777917737. («3777.7 a 4. 7|. .5... I .17 757.77....5i 7!...71‘177 n r EgiéitLE. 77.757.57.777... \. .7177?!) (.777il..7lx.i.lrf!..7lv77 . .797.§7§E\77 {‘77 I ~ 7 77.5.7 - 715137177777 f3‘hi71l77 . I: k. V. WS$§V~777Y7717§777§§7I . .\ .... 9 O ..|...l.l..617p77..771r|\ . .X .. £1777 177.17.1171571; 17.797.77.777! El. .75775...o770617.177|1l0f\u7117afl 777.307....- ..‘. . L..17.fl.\n\.77.k..~£t.7l..7‘ 777A 1.... J. (77.5.7877. 751777-7‘7177... Lit-7% : I . :. 757.777.1777}!!! .‘LALKELI . . . .5754 . 15.17.7115 71771717175171 7......»7.i71\... L 7.7.74. {90.77.7771 A 5.. 7.3.7. 7:777: liz...§7.7lv~§ 27‘ . 72.77-7171 7717.. £17112. 4. .77». . igii77‘t » L thttxiitgg '1.»ng Ligts. . 4175.15.77 . L llk».7i7i.gflum.v»7777h7.vu LI . . 74.77.737.74! \L i417 . .4.» f {7717. 2.7.. .7?! .5. x ( . L 7 .772 .... . 77¢. 71»! LN» f L.‘ . . .\. . . s I .» [Iafiltz‘ . 717 gift .4? 5.7.7.4. (V! .l I 3 777‘. IE. .DLL. 77777.... 7)..»077 ,7.\»v.a\l.7iu .LIL 3.747 1.7.. 177177.07. ~§§ :v: (N 711.171.770.377!!! “an“: :1. . 17717717743777... t7. 7:41.719 .7. i. ((1.7: 5.7!..1... II 57.5.... ‘77....7I5 ... 71:77.51! . .. r .. .x. a. .1 77277.1..77.\ 75.771771744779745... r > 7.4.1.1. E... ii 1%.; u.....:m...i.c(ur.r..£§.. 71t..ley..unn..L.s§\... $717577: 7...»... . .. ainvyiolilt .7177: .11 7... s .1\Hw...¢v.[ 5.?47155. Pt:~%§iti 77.77%: \x 17¢ 777.177.77i7f1. 77771777377»... .53k.‘217&77v».9ni71l {kt-17L iiikl‘lkl . {iii-7. ......... 717177.}. a 75.7%...557571‘1‘. . . . .1. 7 . 5.771.275. V LL - .7 1.511.771.7377. YES)!» Lust . I?! 7.77.7.7 .757???be . -V. X 1.77.73.77... .. 215...-.. v5.5. .. . .Sltgscilkgtpli 7:77.? .1227. . 7.57....1..:\.7.7.L57\....i7 . {.11. A... .11....V77ikkl. 17777777.: » 1.177». 7.1.4.97}... I11 .77.... I k r1»l.757~777.l75~tv7.£\§7.itt.tt s\ L x {.7755 571.777.77.757 £5.17 7t.?7!l..L»777t777>7»1-\|77»~Z. x .77. 7. \v. .I77...7»A7..\417a\..\. s! 7:» \. 15C» . 557.373- ~§.wfl.fv71sku..77..7. . wag Uniiln‘ .7727??? 171.. a . N t... .. 7‘. .x.‘ V A 1-. ‘riifli‘ké .. A. L) 771:. . .7!!\..77..~Lb\..57€7~5 $7.7. 1 < 77?...10. 771.71.711.51! {trust I .I A; 577.774.171.777... i275727 . . Lil “inhkALrUA77T.,..7'7.L.7I..Vvt~...0»l§r7....|u7filblfilun‘fincxi.gdnflfitlhfiifi~l7»7n.7§77. .\.\\ . v».n»5.71{..1»?k7177.~! “WM?” 7.. .7» K »( . . (77.77)» 77.. 1.777». Ezég‘éi .. . .6717} .l . E.‘fllitrhg\ . x 73‘ Kati...” flint... -5 3! -1.L.h.u§t.27.t%7z¥uvl§7§vx§st. 7.1!...{5ltc 777?75:!¢Qub1.n¥£2 . L..(._....).t....\7?. .1...» ..$7.r(..t¥32§7 . .. (47.17.77.777- .& impriuuvansxk‘t.i7~ik7$i~z 7174.1..- \. L L4L7fvicfnvvati77s xié‘niilfithi tan!» a .2 .1 . [Lt-71:1,}...rflanh; . «.L I»......P!7~.72£..1L.§ 7.797717171711-iuilé. .7»: . ‘Lt 710.577.. I7..\\I(.»L\...N.\77~.7 .174 \1 :77 . T: trl:\.f77177.l.7£7.\v!777\rll\.fi 775.77 R. 4.41767 .477» 4417.. . 7‘77. 3714.37. .7737?» . x L b 577‘ 117117... 8.77.1770» 77:77:17.; 717.117.... » 777.77. 1.7!. l~\v(777...bl~khtcltkn-i. .- 7.le 7777.77711rIE71FEKL7177177{£47.75l\\..\7.1§\7a§71\\l71‘77\lt$7.77.“..vti LLt‘ktL‘i-trxtikLnlt I» . .l . . :i. .115.“ 717.777... . . .LLJUP _....I.LJL.\£L.t...r..41.\1..1>5\7771:.7.7x$L\.\.C7.\..wV7.‘7\...n-.‘777.i7..17..”fisu. an.“ .41....7A7L172Lt... INF. “Husk“... .L... . L . u 0! IE it; {E‘Niit V1.1 I .. .. {7{§~\ N7|l I. . t7 iti‘7rfuv7iil.177l§ £59137?! .1 77.. 79-41.). .Ih\r7§7114|7177771”¥.7.|i .\77..I\7\tl{. .T. A; 7.171177... (71.77.. £717.7\\$¥717‘|7.5\77« I. 4.7 .1 (77.27.37.794...» “7 x I}\ .v. \ . . . + $72.11. - .: i77§li§§hfiw§§fi7 1215..-.1......,....-i,tt....t.vb Ln... 2...... . . 3 L . x... x. A E:F777i7~7~.71797.§7. $974.7 1.7. . .7... K it 7. a. . .\.L\ r. 7.77. 77.1.”...77....\.§M L... t .Uu... law“... .. 4.2 . A I I». 4. x 1 IL (L... . |£1..... 7 L VI In. L I h» A . . K I\ . \L. .i. . .. .. (u . i . 771777.76. ‘7 . L V. .t 77 T7» .3777 I. 7.7... .. . s. . LL 1. .71.... 71 .71.?» 7....47 .9. »\ 7|. . L I »..»7v...lLL 757.7477... .7 .» t .itiitihnlstvftuikiiiinv. u17.h.th2t.-....-£ s. V.. l....t..!:.....lztixitkzo...l.z 2.2.1..- .553... . . ...L\ x.1.7».J!1fl.&v§filki~kfi§fiuit7{i§7157117257}? . . a . . . .I . ...\. W111. .7137.Lo.0n‘ll... a .70.. . 7907.. 331.2. nu."- .........i7. . ..7....i7\k.k.\t.. 7.1119 E77771.) ,5 In, . . . . 5.3.51“ .J... x..." E571 a. ._ P555... . . rt... .L....s...v\...»h.nwf(.flrl»{.a.7linut...7. ,. ..Ir .\ .,...»h7~..L$L..n...7I7|: .7t.777797v7!~ .9}~77.7LL:7~...7»}\ 4 ‘ n f. . .. .. . . ....:~.71.7vii§4 . . 3. .; ~ . 1:... I . I .C ”i {ii This is to certify that the thesis entitled TURNOVER IN AMERICAN STATE HOUSES: 1952-1978 presented by GARY LANE CRAWLEY has been accepted towards fulfilhnent of the requirements for Ph.D. d . Political Sciences egree 1n Major professor Date 5-17-82 0-7639 RETURNING MATERIALS: )V1ESI_J Place in book drop to LIBRARIES remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. \. 'r 6 //7Y5/ TURNOVER IN AMERICAN STATE HOUSES: 1952-1978 By Gary Lane Crawley A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1982 ABSTRACT TURNOVER IN AMERICAN STATE HOUSES: 1952-1978 By Gary Lane Crawley This study analyzes variations in turnover in American state houses. Recent studies demonstrate that turnover is declining. However, it is not clear which factors are most important and why. In this dissertation a theory of turnover is developed and tested. TWO conditions affecting the level of turnover in a legislature are examined. First, the willingness of incumbents to seek reelection. Second, the ability of incumbents to win. The willingness to seek reelection is posited to be a function of the attractiveness of the institutional environment, electoral environment, and the availability of opportunities for higher office. The ability of incumbents to win reelection is viewed as a function of competition in the district and candidates' ability to raise money. Turnover in forty-seven state houses for the period 1952-1978 was regressed on variables from the institutional and electoral environment, and the Gary Lane Crawley opportunity structure. In addition, district level data for Indiana and Michigan for 1952-1978 was analyzed focus- ing on voluntary versus involuntary withdrawals from office. Finally, Michigan state house elections in 1978 were examined to assess the impact of campaign expendi- tures and party competition on the vote. The data show that "professionalism" in the leg- islature has the greatest impact on turnover relative to other variables included in the study. Both reapportion— ment and district level competition also have a signifi- cant impact on turnover. Several variables of theoretical significance, such as, the distribution of influence within the house, the importance of the legislature within government and the importance of state government, were not statistically related to turnover. This was most likely a result of poor measurement. Both campaign expenditures and party competition were factors in the ability of incumbents to win reelec- tion in Michigan in 1978. Increasing the attractiveness of the institutional environment results in reduced turnover in state legisla- tures and perhaps incumbent advantages as well. ACKNOWLEDGMENTS Only those who have undertaken a project of this magnitude can appreciate the depth of feeling that goes into thanking people for their support. Without that sup- port this manuscript would not exist. Of that I am certain. Of course, a special thank you goes to the members of my dissertation committee. Professors David W. Rohde and Charles Press were patient, encouraging and, above all, insightful. They often helped me in subtle ways I'm quite sure even they don't realize. To Professor John Aldrich I am especially indebted. He took the time to critique and offer suggestions with regard to the numerous rough drafts of this work from the beginning to end. To my committee chairman, Professor Frank Pinner, I owe everlasting gratitude. He pulled and prodded me through what was truly an ordeal. He has taught me what political science is and what a political scientist should be. In my view, Dr. Pinner exemplifies scholarship of the highest order. There are of course a number of friends and rela- tives who should be acknowledged but must remain anonymous. However, I cannot fail to express my appreciate to Liz and ii iii George Hanks. They stuck by me when times were hard, were always understanding and supportive and, when the push came near the end they helped with the million and one things that had to be done. They are both true friends. I have saved my most important acknowledgments to last. My children, Adam, Luke and Karen have suffered through all the long hours, dark moods, worry, etc. with- out complaint. That contribution will never be forgotten. Finally, to my wife Conni, I cannot find the words to express my full appreciation. So I will only say thank you. TABLE OF CONTENTS LI ST OF TwLE S O O O O O O O O O O O O O O O O 0 LIST OF FIGURES O O O O O O O O O O O O O O O O O mTRODUC TI ON 0 O O O O O O O O O O O O O O O O 0 CHAPTER 1 A.REVIEW OF THE LITERATURE . . . . . . . . Studies of State Legislative Turnover. Studies Reporting High Turnover. . . Turnover as a Function of Benefits . . More Complex Explanations of’l‘urnover............ Electoral Success of Congressional Incumbents Party as a Voter Cue . . . Incumbency as a Voter Cue . . . . . . Simmary................ A THEORY OF TURNOVER IN STATE L EGI SLAMS O O C C O O O O O O O O O O O The Decision to Run for Reelection The Institutional Environment . The Electoral Environment . . . The Opportunity Structure . . . O O O O O O O O O O O 0 Election Outcomes for State Legislators. Summary................ TURNOVER IN FORTY—SEVEN STATE HOUSES:1952-780000000000000 menataoooooooo The Dependent Variable: over in the Heuse . . . Independent variables: The Institutional Environment . . . . . Turnp iv Page vi ix 10 11 11 12 15 20 26 3h #2 1&2 52 60 68 78 82 83 8h 91 CHAPTER Professionalism . . . . . . . . The Distribution of Influence within the Heuse . . . . . . The Importance of State Government Independent Variables: The Electoral Environment Party Competition . . Reapportionment . . . Summary . . . . . . . . O O O Q O O O O O 0 O O O O O O O O O O O O O 0 Analysis 0 O 0 O 0 O O O O O O O O O 0 Conclusions . . . . . . . . . . . . . 1: TURNOVER IN THE STATE HOUSE IN INDIANA ANDMICHIGAN: 1952-78 . . . . . . . . . me Data 0 O O O O O O O O O O O O The Dependent Variable: Incumbents' Electoral Decisions . . . . . . . The Independent Variables . . . . . Ana-17313 o o o o o s. o o o o o o The Electoral Environment . . . Opportunities for Higher Office . . canoluSion-Booooooooooooo 5 THE EFFECTS OF CAMPAIGN SPENDING IN THE 1978 ELECTION FOR STATE REPRE— SENTATIVE IN MICHIGAN . . . . . . . . . The Data . . . . . . . . . . . The Dependent Variable: Winner' 5 Vote . o . . . . . . . The Independent variables . . . . . Analysj-s o o o o o o o o o o o o o o 0 Conclusions . . . . . . . . . . . . . 6 CONCLUSIONS AND IMPLICATIONS o o 0.0 o o O O O 0 APPENDIX A, DATA FOR FORTYESEVEN STATES: 1952-78. BIBI’ImRAPHY O O O O O O O O O C O O O O O O O Page 92 100 103 107 107 111 117 119 137 143 lhh lhs 151 153 161 172 175 180 181 182 182 186 199 202 210 240 LIST OF TABLES Table Page 3-1 TURNOVER IN FORTY-SEVEN STATE HOUSES BYYEAR: 1952-1978 0 o o o o o o o o o o 86 3-2 TURNOVER IN FORTY-SEVEN STATE HOUSES FOR SELECTED TIME PERIODS . . . . . . . . 91 3-3 VARIABLES INCLUDED IN THE PROFES- SIONALISM INDEX FOR FORTY-SEVEN STATES . . . . . . . . . . . . 9h 3-h DISTRIBUTION OF INFLUENCE WITHIN THE HOUSE FOR FORTY-SEVEN STATES: 1952-1978 0 o o o o o o o o o o o 102 3-5 PER CAPITA TAX REVENUES FOR FORTY- SEVEN STATES: 1952-1978 . . . . , . . . , 106 3-6 INDEX OF PARTY COMPETITION FOR FORTY- SEVEN STATES: 1952-1978 . . . . . . . . . 110 3-7 REAPPORTIONMENT IN FORTY-SEVEN STATE LEGISLATURES: 1952-1978 . . . . . . . . . 112 3-8 STATE REPRESENTATIVE'S OPPORTUNITIES FOR HIGHER OFFICE IN FORTY- SEVEN STATES: 1952-1978 . . . . . . . . . 116 3-9 DETERMINANTS OF TURNOVER IN FORTT- 3-10 DETERMINANTS OF TURNOVER IN FORTY- SEVEN STATE HOUSES FOR THE PERIODS: 1952-62, 196h-70, AND 1972-78 . . . . . . 128 3-11 THE MEAN AND STANDARD DEVIATION OF THE VARIABLES IN THE MODEL CONTROL- LING FOR REGION AND THE PERIODS: 1952-62, 196h-70,AND 1972-78 . . . . . . . 130 vi Table 3-12 3-13 u-a h-h l4--6 4-7 h-B u-9 vii DETERMINANTS OF'TURNOVER IN THIRTY- SIX NON-SOUTHERN STATE HOUSES FOR THE PERIODS: 1952-62, 19614-70, AND 1972-78 0 o o o o o o o INTER-ITEM CORRELATION COEFFICIENTS FOR ENVIRONMENTAL VARIABLES AND TURNOVER CONTROLLING FOR REGION AND THE PERIODS: 1952-62, 196““‘70, AND 1972-78 0 o o o o o o o ELECTORAL DECISIONS OF INDIANA AND MICHIGAN REPRESENTATIVES, IN PERCENTAGES OF THE TOTAL HOUSE MEMBERSHIP, 1952-1978 . . . . . . . ELECTORAL DECISIONS OF REPRESENTA- TIVES IN INDIANA AND MICHIGAN FOR THE PERIODS 1952-1962, 196N-1970, m1972-197sooooooooooo PROFESSIONALISM INDEX SCORE FOR THE INDIANA AND MICHIGAN LEGISLATURES, 1952-7800000000000000 THE LIKELIHOOD OF REELECTION To THE HOUSE OF REPRESENTATIVES IN INDIANA AND MICHIGAN, 1952-1978 . . . . . . ELECTORAL DECISIONS OF INCUMBENTS BY STATE AND COMPETITIVENESS OF THE DISTRICT, 1952.1978 0 o o o o o o o ELECTORAL DECISIONS OF INDIANA AND MICHIGAN REPRESENTATIVES BY REAPPOR- TIONMENT, IN PERCENTAGES OF TOTAL HOUSE MEMBERSHIP, 1952-1978 . . . . ELECTORAL DECISIONS OF INDIANA REPRE- SENTATIVES BY PARTY, 1952-1978 . , . ELECTORAL DECISIONS OF MICHIGAN REPRE- SENTATIVES BY PARTY, 1952-1978 . . . OPPORTUNITIES TO RUN FOR HIGHER OFFICE FOR INDIANA.AND MICHIGAN REPRESENTATIVES . . . . . . . . . . Page 132 136 1A7 150 155 159 163 166 169 170 171+ viii Table Page 5-1 AVERAGE VOTE, PARTY STRENGTH AND CAMPAIGN EXPENDITURES FOR CANDIDATES RUNNING FOR THE MICHIGAN HOUSE OF REPRESENTATIVES, 1978 , . . . . . . . . 185 5-2 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS . . . . . . . . . 188 5-3 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS CON- TROLLING FOR UHETHER AN INCUMBENT“WAS RUNNING . . . . . . . . . . 190 5-14 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS USING Two STAGE LEAST SQUARES . . . . . . . . . 196 5-5 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS USING TWO STAGE LEAST SQUARES: OMITTING ESTIMATED INCUMBENT EXPENDITURES . . o o . 198 LIST OF FIGURES Figure Page 1. WHY LEGISLATORS SEEK REELECTION . . . . . . . 66 2 .A MODEL OF STATE LEGISLATIVE TURNOVER . . . . 76 :3 TURNOVER FOR FORTY-SEVEN STATE HOUSES FOR SELECTED YEARS . . . . . . . . . . . . 89 INTRODUCTION Assessing state government performance has been a preoccupation among political reformers for over a hundred years. It is an ongoing enterprise among legislative and policy scholars. Depending on the nature of their ques— tions, conclusions about performance have been as varied as the state governments themselves. However, at least one theme reappears with great regularity: the need for insti- tutional reform of many state legislatures.1 The call for legislative reform embraces a wide range of activities not the least important of Which are those intended to attract and retain qualified people. These reforms are aimed at reducing membership turnover and thus providing an eXper~ ienced legislative body capable of performing what has become a rather complex legislative task. Advocates of reform generally propose a more professionalized legislature. Reform measures include raising legislative salaries, providing adequate staff and research services, increasing the level of compensation for expenses incurred while engaged in government business, and making legislative work a full-time, as Opposed to a part-time, Job. While a reduction in membership turnover (that is, the percentage of a legislature composed of new 2 members) results in a more experienced legislative body, it is unclear what factors relate to this reduction. The purpose of this study is to develop and test a theory of turnover in state legislatures. A theory of turnover is important to an understanding of how well our democracy functions. Democracy must provide capable repre- sentatives who are at the same time responsive to legiti~ mate citizen demands. Legislators falling short on either criterion should be ousted from office. Yet, state legis- latures have traditionally received low grades in this regard. Indeed, the high level of turnover in state legis- latures has been associated with ”volunteerism" and "amateurism." Prewitt (1970) discusses the danger of volunteer- ism. He found a high incidence of volunteerism among office holders in 82 California cities. Four factors lead- ing to volunteerism were (1) reliance on appointment to office, (2) small electorates, (3) high electoral advantage by the incumbent, and (h) a high rate of voluntary retire- ment from office (1970, p. 10). The last three factors are likely to be present for state legislators. Prewitt (1970, p. 10) suggests volunteerism: can serve to undermine an already weakened election system. Although the volunteer in office, espec- ially if relatively indifferent to staying there, may be a devoted public servant as he defines the role, he is unlikely to be constantly sensitive to voter preferences. His political thinking has been formed by a series of experiences which min- imize for him the importance of mass electorates. 3 VOlunteerism thus reflects a breakdown in representative democracy. Responses to constituent demands occur at the whim of legislators who are unconcerned about reelection. What is often ignored by students of state legisla- tive turnover is the possibility that as state legislatures increase the incentives designed to attract and hold quali- fied people, they also create a situation whereby incum- bents are rarely defeated. Since the rewards are greater, incumbents become more serious about holding office. Longer sessions and increases in franking privileges result in higher visibility and greater name recognition. In addition, if the importance of the legislature within gov— ernment is enhanced, incumbents become convenient arbitra- tors for constituent--bureaucrat conflicts. As a result of these and related rectors, while turnover may decline, it may do so because of what scholars of congressional elections would call the increase in "safe seats." This situation is hardly conducive to insuring incumbents are responsive because of the electoral process. The diffi- culty is we do not know to what extent incumbents leave state legislatures voluntarily or why. We know even less about why incumbents might have an electoral advantage. A study of turnover not only provides insight into volunteerism, but also addresses the question of legisla- tive capability. Scholars cite the need for more profes- sionalism in state legislatures. For example, Price (1975, p. 21) contends that as salaries increase, sessions lengthen, staffing expands and office space is provided, members will desire committee stability and then committee seniority. This occurs only after membership stability is achieved. He suggests (1975, p. h) this process is an improvement over ”amateur" legislative bodies where the presiding officers have high concentrations of power and ”little capacity to deal with a sophisticated bureaucracy." Yet, there is little theoretical discussion in the literature on state politics concerning the effects of professionalism on turnover. It is generally assumed that salaries decrease turnover but authors fail to specify what effects longer sessions have. Further, they fail to consider the possibility that as legislatures become more professional they may also attract more members desiring to move on to higher office. Turnover may rise as a result of professionalism. The theory developed in this work addresses these concerns. The theory is tested for a number of election years to assess its general applicability. The thesis advanced in this study is that turnover levels result from two partially independent sources. The first of these is the incumbent legislator's decision to seek reelection. Salary increases, improved working conditions, and the growing importance of state government policy-making should have an impact upon this decision. A second source is a legislator's ability to win reelection. While incentives aimed at luring incumbents to seek reelection may also help them win, there are addi- tional factors to be considered. Among them, the ability of incumbent candidates to raise the funds necessary to conduct an effective campaign is probably the most significant. To examine these relationships, three sets of data are utilized. First, aggregate data are presented based on 47 states for the period 1952 through 1978. Measures of the level of professionalism, party competition, bureaucratic growth, the incidence of re-districting, and the availability of opportunities for higher office, are used for all elections in the period 1952-78. In addi- tion, averages on each variable over several time periods will be used to smooth out the variations in turnover in particular states caused by unique, short term factors. Since it is difficult to determine whether turnover resulted from voluntary or involuntary withdrawals over an extended period for a large number of states, a second set of data is presented focusing on the percentage of state representatives who sought reelection in Indiana and Michigan during the 1952 through 1978 period. Changes ever time in the percentage of incumbents running for reelection are compared to corresponding shifts in the level of pro- fessionalization, party competition, and the incidence of redistricting in each state. These comparisons are then used to examine legislators' willingness to run for reelection. Finally, a third data set on state representatives in Michigan for the period 1952-1978 will focus on the ability of incumbents to win reelection. The degree of party competition in the district and candidates' expendi- tures will be used to predict election outcomes. A comment should be made on what this study will not address. It will not address a consideration of turn- over as it affects what might be called "effective" legislative performance. To my knowledge, we simply have no evidence which illustrates the relationship between turnover, legislative experience, and legislative effec- tiveness. More to the point, we do not have adequate measures of effectiveness.2 Does effectiveness involve a definitive solution to problems? How do we decide which particular problems should be solved, or even make a determination as to what solved them and how and whether, indeed, they were successfully dealt with? Does effective- ness imply efficiency? If so, efficiency with respect to what: political, social, or economic criteria? Should effectiveness also include upholding underlying Democratic principles of the political system? During the 1950's and several periods from 1960 to 197k, average turnover in the states hovered around #0 percent.3 It is unlikely that all, or nearly all, of those legislators leaving office were doing so because their constituents were dissatisfied. More recently, average turnover has generally declined. Why? It is due to increased incentives or a greater likelihood of reelection? Chapter 1 describes previous research on turnover in legislatures. It is divided into two parts. The first reviews studies which focus on turnover. This part stresses explanations of incumbents' voluntary withdrawals from office. The second reviews literature that pertains to incumbents' defeats, much of which deals with Congress. Chapter 2 presents a theory of turnover in state legislatures. Several independentvariables are considered which might explain turnover and the impact of these vari- ables is discussed. Chapter 3 presents an analysis of turnover in the houses of 47 states from 1952 through 1978. Turnover is hypothesized to be a function of the attractiveness of the institutional environment, the level of uncertainty in the electoral environment and the availability of opportun- ities for higher office to house members. The relative impact of these variables and changes over time are examined using multivariate regression analysis. Chapter A analyzes turnover in the legislatures of Indiana and Michigan from 1952 through 1978. The proportions of incumbents leaving electoral politics, running for higher office and losing elections are pre- sented. These data are controlled by professionalism, reapportionment, competition in the district and the availability of higher Office. In Chapter 5, the effects of campaign spending on the vote in Michigan house elections in 1978 are estimated using two-stage least squares analysis. The effects of Party competition in the district, seniority, and primary election opposition on the vote are also examined. In Chapter 6 conclusions are presented. In addi- tion the implications of this study are explored. FOOTNOTES 1For a detailed account of suggested reforms see the Citizen's Conference on State Legislatures, State Legislatures: An Evaluation of their Effectiveness (New York: Praeger Publishers, 1971 2Government performance has been studied often but, in each case measures of effectiveness are subject to the charge of value bias implied by the questions I raise here. This is not to say such studies shouldn't be undertaken, only that objective definitions of effectiveness are hard to come by. Examples of this approach include Alan Rosenthal, Le islative Performance in the States (New York: The Free Press, 1974;; Thomas A. Flinn, "An Evaluation of Legislative Performance: The State Legislature in Ohio," in John J. Gorgan and James G. Code (eds.), Political gghaviorggnd Public Issues in Ohio (Kent State, Ohio: Kent State University Press, 1972;, p. 153: Ronald D. Hedlund and Keith E. Hamm, "Institutional Development and Legisla- tive Effectiveness: Rules Changes in the Wisconsin Assembly," in Abdo I. Baaklin and James J. Heaphey (eds.), Comparative Legislative Reforms and Innovation§_(Albany New Yerk: State University of New York at Albany, 1977 , P0 1710 3This is based on data gathered from volumes of The Council of State Governments, The Book of the States (Lexington, Kentucky). CHAPTER 1 A REVIEW OF THE LITERATURE Studies of turnover in American state legislatures are for the most part descriptive. Theoretical explana- tions which emerge are Often untested and fail to consider the possibility that suggested independent variables have a differential impact depending on the extent in which turn- over is due to voluntary retirement, running for higher office or electoral defeat. Since students Of turnover in state legislatures have largely ignored questions address- ing the ability of incumbents to win reelection, this review is divided into two sections. The first discusses work devoted specifically to turnover in state legisla- tures. The authors of these studies tend to assume implicity that turnover results from the incumbents' voluntary withdrawal from office. The second section is a review of the literature aimed at explaining the electoral success of U.S. Congressional incumbents. From this, a theoretical framework will be obtained which applies to state legislators as well. 10 11 Studies of State Legislative Turnover While turnover in the states has not prompted a great deal of study, several themes do emerge. First, studies note that turnover is relatively high at the state level and suggest that this is an indication of the ama- teur standing of most state legislatures. Second, when explanations of high turnover are offered, they invariably focus on the benefits of office, or rather, the lack of benefits, as primary determinants. Finally, recent studies note that turnover appears to be declining and somewhat more complex explanations emerge. Studies Reporting High Turnover As early as 1938, Hyneman, examining data for the period 1925-1935, found an average of 35.4 percent fresh- men in ten state legislatures studied.1 He suggested that several terms were necessary to attain the experience required to be an effective legislator and he viewed a high level of turnover as detrimental to Overall legislative performance. In a l95h report by the APSA Committee on American Legislatures the existence of high turnover was substantiated by the finding that over half of all state legislators are new at each session.2 Jewell and Patterson (1966), in one of the first studies to recognize the importance of examining turnover crver time, also examine the proportion of state legislators serving their first term. They compare the 1925-35 figures 11 Studies of State Legislative Turnover While turnover in the states has not prompted a great deal of study, several themes do emerge. First, studies note that turnover is relatively high at the state level and suggest that this is an indication of the ama- teur standing of most state legislatures. Second, when explanations of high turnover are offered, they invariably focus on the benefits of office, or rather, the lack of benefits, as primary determinants. Finally, recent studies note that turnover appears to be declining and somewhat more complex explanations emerge. Studies Reporting High Turnover As early as 1938, Hyneman, examining data for the period 1925-1935, found an average of 35.4 percent fresh— men in ten state legislatures studied.1 He suggested that several terms were necessary to attain the experience required to be an effective legislator and he viewed a high level of turnover as detrimental to overall legislative performance. In a 195h report by the APSA Committee on American Legislatures the existence of high turnover was substantiated by the finding that over half of all state legislators are new at each session.2 Jewell and Patterson (1966), in one of the first studies to recognize the importance of examining turnover over time, also examine the proportion of state legislators serving their first term. They compare the 1925-35 figures 12 reported by Hyneman and the 1950 levels as reported by the APSA Committee on American Legislatures in 1959, with data collected on 15 states at a later date (the year reported for each state varies from 1957 in California to l96h in Kentucky).3 Noting that the state legislature is "an amateur political body in a variety of ways, and one of the indicators of its amateurism is its high turnover rate," the authors observe that turnover was decreasing.“ Their only explanation is that a relatively greater political stability has occurred in the postwar years. Turnover as a Function of Benefits Barber noted, "The rate of turnover among state legislators is very high. More than half of the approxi- 5 In mately 7,800 members are replaced every other year." addition, he cited data from the Hyneman study and his own work (for the 1946-58 period) demonstrating that turnover exhibited no clear relationship to various measures of party competition and demographic characteristics.6 He concluded that reducing turnover requires making the job more attractive and thus altering the attitudes of legislative incumbents. The view that a major factor resulting in high turnover is the lack of benefits provided legislators is discussed by Wahlke, et. al., (1962, pp. 122-129). The authors report that, of 220 legislators interviewed who indicated that they did not expect to run for reelection, 13 51 percent cited reasons which could be related to the benefits of the office. Interestingly, of the 252 legis- lators who did expect to run again, 91 percent included their ”involvement" in the legislative job of "public 7 service" as reasons which led them to seek reelection. Neither "involvement", nor "public service", are neces- sarily related to the benefits Offered to a legislator in the sense accepted by political scientists. Barber (1965, p. 7) illustrates the inability of most state legislatures to provide an attractive setting in the following terms: The typical legislator must rent rooms for his wife and himself at the state capital for the duration of the session. These may be in one of the two or three principal hotels, or in a motor court. The average pay of a legislator, if he is frugal, will usually barely cover the extra expenses of his sojourn in the capital. If the legislator is at all convivial, he cannot live in a hotel for five months for less than $1,000 (a typical salarY). This appallingly low salary rate accounts in large measure for the quality and character of our state legislators. How many people can afford to drop their regular employment for three to five months every other year? Barber also points to the increased work load faced by state legislators in recent years and the added complex- ity of legislation as further disincentives of legislative life. Thus, turnover in state legislatures was high (at least through the early 60's) and theoretical justification for this fact centered on the unattractiveness of the state legislative office. Nevertheless, empirical evidence was lacking. 1h Lockard also reported that turnover was very rapid among state legislatures. Looking at all fifty states, he found an average of 34 percent new members during the 1963 session.8 Turnover varied from a high of 67 percent in the Kentucky house to a low of 18 percent in the New York house. Lockard argued that being an effective legislator requires a certain level of competence for the legislative task which comes in part from serving a few terms. In an attempt to explain turnover levels empirically, he examined the extent of party competition, size of legislative mem- bership and legislative salary levels. His findings indicated these variables had little effect on turnover rates. A difficulty with his analysis is that it focused on only one point in time and potentially understates the importance of increased legislative salaries. In 1963, for example, the highest salary for any state legislator was $10,000 per year in New York.9 Many salary levels fell into the $2,000 to $5,000 range. In the two decades since Lockard's study, state legislative salaries have changed thereby begging a reconsideration of his findings. Furthermore, it is not immediately apparent why the size of the legislature should make a difference in turn- over levels, unless one argues that the smaller number of legislators, the more prestige the office carries. This line of reasoning seems tenuous at best. Finally, with 15 regard to party competition, Lockard fails to note that he is examining the ability of incumbents to win reelection without first establishing that they are active candi- dates.10 Perhaps the lack of a relationship between party competition and turnover in the legislature results from the unwillingness of incumbents to seek reelection regard- less of party competition. More Complex Explanations of Turnover Ray (1974), observing that legislative turnover is but one element Of the more general phenomenon of member- ship stability, further substantiates that turnover is decreasing over time. He points to the need to examine stability using several indicators: the proportion of first term members in the state legislature, the proportion of incumbents seeking reelection, and the average prior service among the membership. Examining data for the period 1893-1969 from Michigan, Connecticut and Wisconsin, Ray finds that all three measures, in all three states, show an increase in membership stability of the legisla— ture.11 For example, in Michigan the percentage of first-term legislators declined steadily from an average of 68 percent for the 1893-1913 period to an average of 23 percent for 1953-1969.12 Recognizing that membership stability is determined by factors affecting an incumbent's willingness to run, as well as an incumbent's electoral success, Ray suggests a -;_ __:,~' ‘— 16 number of indicators which might explain membership stability: salary, length of legislative session, psycho- logical gratifications, and the business or professional benefits which legislative service may yield. These fac- tors will presumably improve the legislator's benefit— cost ratio and thus motivate him to run. Degree of party competition in the legislator's district, the extent of primary election opposition, and variations between presi- dential and off-year election voting behavior might on the other hand, contribute to an incumbent's success at the polls. Rosenthal (1976, p. 609) cites the lack of com- parative analysis and explanatory power as two problems unsolved by prior studies of turnover in the state legisla- tures. Relying on turnover data for all fifty state legislatures during the period 1963 through 1971, he examines the relationships between turnover and three categories of variables tapping different elements of the 13 state legislative political environment. Using step- wise multiple regression, Rosenthal finds that the number of elections and the number of reapportionments are especially important in accounting for the variation in 124 turnover. The Rosenthal study is important in at least two respects. First, the use of turnover data for all fifty states covering several elections is an improvement over 17 earlier efforts which either examined all states for a one year period or a few states for an extended period. Second, this study is the first attempt at exploring the combined effects of several independent variables. How- ever, two difficulties emerge which point to the need for further research. The first is that the 1963 through 1971 period is limited with respect to the number of elections included and the nature of those elections. Certainly, the mid-60's can be classified as somewhat unique with respect to recent American political history. The Vietnam War, campus unrest, and racial turmoil are but a few examples of the unsettled political environment of this period. In addi- tion, the years 1963 through 1971 include the most dramatic and far reaching reapportionment decisions in our entire history.15 Since the sixties may be atypical, it is desirable to look at a longer time series. Second, the Rosenthal study (and most others as well) fails to develop a theoretical framework which could serve as the basis of understanding turnover and subse- quently lead to more refined empirical testing. Rosenthal indicates the criteria used to determine which variables should be used in the final analysis reported involved an examination of the correlations between those variables and turnover, and intercorrelations among the variables themselves. This procedure could eliminate intervening 18 variables of theoretical value and certainly does little to justify the choice of the variables included in the first place.16 Rosenthal acknowledges, in the latter part of his article, that some variables affect the incumbents' elec- toral fortunes and others the incumbents' willingness to run: yet, there is no systematic attempt to explain just how or under what conditions important relationships are expected to emerge. The fault lies not with Rosenthal alone, however. Few authors have attempted to establish, a riori, what effect the two likely causes of turnover (i.e., that incumbents must be willing to run for reelec- tion Egg able to win) have on subsequent analyses.17 Recent research indicates that change has occurred in the kind of turnover traditionally found in state legis- latures. Data for the period 1897-1967 show a gradual decline in the proportion of voluntary withdrawals from the lower chamber in eight states (Ray, 1976, pp. h30-h31). On the other hand, no clear patterns were evident for the proportion of incumbents defeated during the same period. Calvert (1979) also examines the proportion of incumbents withdrawing from office and the proportion losing in a reelection bid. Data on 29 state legislatures for the period 1966-1976 show that voluntary withdrawals from office are declining while the proportion of members defeated is relatively constant. Calvert conducted a —,..- n - _=l;L—.._';_.' ' "“‘—1—-——t-m , ' Z.‘ - nM‘. 11;. {”_._ l9 multivariate regression analysis treating voluntary withe drawal from office as the dependent variable and examined the effects of nine independent variables.18 In general, the strongest predictor in most years was the number of opportunities available for higher office. Unfortunately, neither Ray nor Calvert, present a further breakdown of voluntary withdrawals into those who leave politics and those who seek higher office. In addi- tion, data from both studies show that the pr0portion of incumbents who are defeated varies among states and within states over time: yet, no attempt is made to explain why this is the case. Because of the limited availability of theoretical and empirical material on state legislative electoral outcomes, it is necessary to turn to the litera- ture on U.S. Congressional elections to examine theoretical premises which may apply to the state level as well. Electoral Success of Congpessional Incumbents A.review Of the literature on congressional elec- tions uncovers several interesting characteristics of incumbency. Since World War II, turnover within the U.S. Congress has been relatively low (around 10 percent) and at the same time, incumbent success at reelection has been consistently high (at times around 95 percent). Conse- quently, studies have increasingly focused on the need to explain.why incumbents can be so successful. 20 There are two lines of research on congressional turnover. The first type of study which makes up the earlier period seeks the causes of turnover in party com- petition and in one aspect of party conflict - reappor- tionment. More recent efforts refute the importance of party and reapportionment, and stress advantages peculiar to incumbency status on election outcomes. The task at hand is to examine these varying perspectives and later consider their relevance to state legislative election outcomes. Party as a Voter Cue Miller and Stokes (1962) contributed one of the earliest exploratory works dealing with the question of why congressmen are elected or reelected. The aim of their work was to examine the validity of the party responsibil- ity model of democracy. The model embodies the notion that "the people" will be represented through a party. That is, the parties will sponsor policy by means of a platform, candidates will be chosen by virtue of the posi- 19 In testing this proposition tions taken by the party. with 1958 data, Miller and Stokes found that people did indeed vote for their congressman on the basis of party loyalty. Eighty-four percent of the votes for a candidate by party matched the voters party. Also, when examining the salience of congressional candidates, the authors report that 39 percent of the respondents knew something 21 about the incumbent, while only 20 percent knew something about his opponent. Furthermore, only seven percent of the responses involved issue content as a basis for a congres- sional vote.20 People voted for candidates on the basis of party, but that vote involved little information beyond party label with regard to issues or candidates. Jones (1966) reexamined the question of issue voting in congressional elections. He also presented data demonstrating high levels of success for incumbents in get- ting reelected. He found that in the elections from 1954 through 1960, 91 percent (1958) to 96.5 percent (1956) of the incumbents seeking reelection were successful.21 He found a decided lack of issue voting in congressional elections and argued that this was a result of candidates using the campaign as a forum in which to present them- selves as capable representatives, not issue—oriented individuals. He supported this position by noting that the electoral structure in this country inhibits issue voting. First, as election dates are fixed at periodic intervals, those dates do not necessarily coincide with times when important issues are on the public agenda. Second, since elections are held every two years, their importance is de-emphasized. Finally, most ballots are long, thereby imposing heavy information costs on voters and prompting them to use other cues such as party and name recognition. 22 Thus, Jones found support for the importance of party as did Miller and Stokes. Examining elections in which.no incumbent was running, he found that the incum- bent's party's candidate won a.majority of those elections, from 67.h percent (1958) to 85.3 percent (1954).22 How- ever, it is not at all clear that party loyalty has continued to play a dominant role in congressional elec- tions since that time. In fact, the data presented by Jones provide some indication that party loyalty has been growing less effective as an explanatory variable, because the percent of successes by the incumbent's party's candi- date is decreasing over time.23 Erikson (1971) offers additional evidence for the importance of the party in winning a congressional elec- tion. He conducts a regression analysis in an attempt to demonstrate that incumbency represents an advantage. By comparing the vote change of first term incumbents running for reelection with the vote in their initial election, he finds that incumbency gains the congressman an additional one to two percent of the vote (over and above party, ).2h etc. He concludes by stating: . . the major explanation appears to be simply that most districts are safe for one party and that the rare challenger who is able to defeat an incum- bent tends to be a strggg enough.candidate to win reelection.on his own. This line of reasoning is further supported by Tufts (1973), who looks at the possible effects of reapportionment on 23 congressional elections. some (1973, p. 553) argues that reapportionment has made an independent contribution to the success of incumbent congressmen. He then explains: Control of districting by one political party can mean many additional congressional seats in the larger states or the difference between a majority and minority seats in a state legislature--often remarkably independent of voter preferences. The case for party loyalty as a primary factor in deter- mining the outcome of congressional elections can thus be stated in even stronger terms: not only do voters elect congressmen because of their party affiliation, but con- gressmen acting through the party organization can affect the electoral balance of party voting in a district through redistricting. Unfortunately, additional evidence does not support his contention. Ferejohn (1975), in commenting on Tufte's article, demands evidence indicating that incumbents can control redistricting. He argues that the line-drawing explanation does not hold and presents evidence demonstrating that there was a decline in competitive seats in non-redistricted states as well as redistricted ones (as noted by Tufts).26 Thus, Tufte's redistricting argument is dealt a severe blow. Furthermore, Erikson (1972), reports that gerry- mandering, especially by Republicans, had an effect in the North up to 196k, but from 1966 onward the effect seems to have diminished with an average of h7.h percent of the 2h districts in the North being more Democratic than the mean.27 He goes on to argue that the Republican gerryman- der declined primarily due to incumbency. Reproducing the regression analysis used in the 1971 study, Erikson includes the elections of '66, '68, and '70. He finds that incumbency advantage is worth an average of five percent of the vote as compared to the two percent in prior elec- tions.28 He argues this coincides with the erosion of party identification in the mid- to late-Sixties as noted by Burnham (1970). The erosion of party identification is reported elsewhere. Abramson (1976) demonstrates that there has been a steady decline in party identification since World War II. By 1970, among white voters, only 27 percent were strong party identifiers.29 Prior to 1961+, 75 percent identified with a party in some fashion, by 197k only 60 percent did.30 Nie, Verba, and Petrocik (1976), find an increase over time in the proportion of party identifiers voting for the can- didate of the other party in Presidential, Congressional, and state and local elections. In addition, Jewell and Olson (1978) examine the possible causes for divided govern- ment (i.e., control of the governor's office by one party, while the other party held a majority of the seats in the legislature) in ten states. They report that the major explanationis split ticket voting. Thus, there is evidence that party identification may be becoming less useful as an explanation of voting in congressional elections. 25 Two difficulties are apparent at this point. First, the evidence available does not directly examine the impact of party voting at the state legislative level. It may be the case that split ticket voting at the state level is a function of the higher visibility of some offices vis—a—vis others. The vote cast for governor may be based on candi- date characteristics rather than on the party represented by each nominee. Because of the higher visibility of the office, voters are more aware of the candidates and party becomes less important as a voting cue. However, state legislative offices are less visible, and it may be the case that voters, lacking information about the candidates, fall back on the party as the basis of their vote. If this is plausible, then party loyalties are still of major importance in electing state legislators. Unfortunately, no data have been used to test this proposition. In later chapters the relationship between the degree of party com- petition at the state legislative district level and voting behavior will be considered. A second difficulty which arises as a result of the apparent decline in party voting at the congressional level is that it does not tell us much about other factors which might have an effect on the vote. It is not neces- sary to argue that party loyalties have no effect on the vote whatsoever. Its importance as an explanatory vari- able has simply diminished, and that is the key point. Yet, congressional incumbents are still winning reelection, 26 and the number of competitive elections is declining. The question is: if incumbents are winning reelection and the basis for success is not issue voting or party voting, then what accounts for it? Incumbency as a Voter Cue David Mayhew (1973) examines the phenomenon of "partisan swing" as a central focus of his paper. Partisan swing involves both an aggregate gain in votes for one party and the translation of those votes into a comparable gain in seats. In order for a change in the number of seats held by a party to be significant, it is necessary for a large number of congressional districts to be mar- ginal (i.e., closely contested--usually h5-55 percent of the vote for each party). Mayhew examines congressional elections from 1956 to 1972 and finds that the number of incumbents running in districts which were previously mar- ginal has been out in half.31 His point is that incumbents not only do win, but their victory is related to the safety of their districts. Mayhew suggests five explanations for the decline in the number of marginal districts. These explanations also provide insight into incumbency success. The first explanation concerns redistricting. Using Erikson's line of reasoning, he points out that there is little relation- ship between malapportionment and electoral benefit. In conjunction with this argument he notes that if districting 27 had an effect, it should show up with regard to Presidential voting in the district, which it does not.32 Mayhew's next three explanations all relate to actions the incumbent can take to ensure reelection. His second explanation is that incumbents can advertise them- selves better than their opponents. He supports this by citing increases in mail to the congressional office from constituents and Gallup Poll results indicating better name recognition of candidates from 1966 to 1970. Third, Mayhew posits that congressmen can get more mileage out of federal programs because such programs are on the increase. This explanation was not tested, but it will be dealt with later. Fourth, he presents an explanation which notes that members are simply more skilled with regard to position- taking than in the past. This skill has come with increased use of polling techniques and the development of more sophistication in using them. Fifth, Mayhew observes that the decrease in marginal districts could be explained by a shift in voter behavior unrelated to incumbent behavior. That is, voters have become dissatisfied with party as a one and are now opting for incumbency in its place. These explanations are intuitively interesting but, by and large, most have not been tested adequately. It is to Mayhew's credit that with at least three of the explana- tions offered he goes beyond earlier studies in that he deals with incumbency itself and its advantages. '..— 'wv‘ w“ '_‘u-w.. :- file‘s—Fr ~ . -‘mvw _ 28 Ferejohn (1975) examines the three explanations advocated by Maynew which deal with increased name recog- nition.for incumbents. Tb test these propositions, Ferejohn uses regression analysis to see if changes in candidate saliency (i.e., name recognition) had any effect on the vote when incumbency and party identification were fixed. His findings indicate that candidate saliency did have some effect. But, he also found that incumbency in 1964, 1965, and 1970, had a significant effect when controlling for candidate saliency.33 In addition, he was able to demon- strate that within party identification levels, change in voting behavior had occurred. Ferejohn, citing Tufte, con- cludes that the scarcity and increasing costs of obtaining information on congressional candidates forces the voter to rely on cues such as party, presidential performance, and incumbency status. ' Tufts (1975) presented a case for voting on the basis of past presidential performance. He finds that the vote cast in midterm congressional elections reflects presidential performance, but that this vote is not reflected in the distribution of seat changes in Congress, as demonstrated.by a decrease in the "swing ratio” in midterm years. Fiorina (1977), taking account of the decline of competition for House seats in recent years, provides an excellent critique of earlier efforts to explain incumbency success, and then proceeds to present his own explanation. 29 He begins by observing that a decline in "swing ratio" represents a decline in competitive seats which in turn implies that we can expect little policy change from Con- gress in the future. He argues, however, that before we can discuss incumbency as a factor we need to better under- stand what has led to a decrease in competitive seats in the first place. Fiorina takes issue with Ferejohn with regard to the cue-taking'mode of voting. He states: According to proponents of the behavioral change view party identification traditionally has served as the most important rule of thumb, (for voters). But in recent years the citizenry has become more informed, issue conscious and ideological . . . Ferejohn, Burnham, and others propose that incum- bency voting has filled the void left by weakening party ID: for significant numbers of citizens voting for the incumbent has replaced voting for their party. The preceding argument has a curious ring to it.' On the one hand we are asked to believe that party ID has declined in importance because citizens are increasingly aware and informed. But on the other hand we are to believe that these same citizens increasingly rely on the seemin ly simple-minded rule of voting for incum- bents 1977. p. 177). Having dismissed the cue-taking model, Fiorina then offers his own theory based on his study of two congres- sional districts. His basic arguments are that the resources Congressmen may use for reelection have changed, and that their reelection strategies are more efficient and effective. This condition has been fostered by the growth of the federal bureaucracy in post New Deal America. There are several results of the growth of the bureaucracy according to Fiorina. First, an expanding bureaucracy 30 carries with it an expanding amount of goods and services to be delivered. Second, congressmen are in a middle-man position between the bureaucracy and its clientele--the people. Such a situation allows them to expedite the delivery of goods and services. Third, there is greater opportunity, with increased government involvement in soci- ety, for ”pork barreling,” which in turn aids congressmen in their credit claiming role. The effect of this situation, Fiorina argues, is that U.S. Representatives are spending more time on constit- uency service activities. And, consequently, the view constituents have of their representatives is one of an ombudsman, an expediter. If this is correct, then incumbents have a distinct advantage over their opponents: they can deliver. Fiorina's thesis also fits nicely with past research. As the ombudsmen's role is intensified, party ID becomes less important, because policy is no longer controlled by party in the constituent's mind. Also, studies demonstrating that voters possess little information on the issues have failed to tap information regarding the ombudsman role. Perhaps cue-taking by voters, which is argued for by Forejohn, et. a1., is in reality an awareness of the Congressman's role of provider. Finally, Fiorina observes his theory can also explain the rapid decline of competition found in the mid— sixties. Tb do this he uses Fenno's (1975) 31 conceptualization of "homestyle" (i.e., type of constit- uency service offered and presentation of himself by the representative). Fenno argued that there was an expansion- ist phase and protectionist phase which Congressmen pass through. The protectionist phase is one in which the incumbent, through established homestyle (i.e., longer service), becomes more conservative, tending to rely on past support.3h Fiorina suggests that the Republicans defeated tn 196h and the Democrats defeated in 1966 were not oriented toward a homestyle of constituency service. At the same time, newcomers from both parties recognized the worth of the ombudsman strategy and their seats became correspondingly "safer". The lack of empirical support for the "constitu— ency service" explanation makes it impossible to assess its impact on voting for congressional incumbents. Furthermore, it is not clear that at the state legislative level the conditions necessary for constituency service as Fiorina describes it, are present. While state bureaucracies are undoubtedly expanding, it has yet to be demonstrated that state legislators are in a position to take advantage either of an ”ombudsman" role or of ”pork barreling" activities. waever, recent studies have pursued a related explanation for congressional electoral success. It is increasingly argued that campaign finance is playing a key role in congressional elections. Jacobson (1978) looks at the effects of campaign spending on the 32 congressional vote. Using a simultaneous equation model, he regresses the challenger's vote on the challenger's campaign expenditures, the incumbent's campaign expendi- tures, the challenger's party, and a measure of party competition in the district based on the challenger's percentage of the two-party vote in the last election. Jacobson finds that “it is clearly the challenger's level of spending that has the greatest impact on the outcome of these elections,” and that incumbents, while they can adjust their finances to the challenger's spending level, get little additional support for their money.35 He then conducts a similar regression for those contests in which no incumbent ran. He finds, that for 1972, "the marginal effects of spending are similar for both candidates."36 However, for l97h, ”the Republican candidate's spending makes a much greater difference in the outcome."37 Jacobson contends that this is largely due to the post- Watergate atmosphere and concludes "campaign spending is evidently most useful to candidates suffering severe elec- toral handicaps, no matter what the source."38 An examination of the data presented also shows that incumbents spend.over twice as much in their campaigns in 1972 and l97h as did nonincumbents. It is not clear Just what level of spending is necessary for winning an election. However, if the level of spending by incumbent Congressmen in 1972 and 197h is anyfihere near the appropriate level, it is clear that 33 challengers, in spending half as much, probably fall con- siderably short of the mark. Welch (1976) also finds that differential levels of campaign spending between incumbents and challengers have an effect on the vote at the state as well as the national legislative level. Using data on California and Oregon elections, he reports a definite advantage to incumbents with regard to campaign expendi- tures as well as an "ex officio" incumbency advantage (based on factors such as the franking privilege and greater media coverage).39 Clearly, a wide range of variables are believed to have an important impact on congressional election outcomes. However, not all of these variables fit nicely into one theoretical framework. Explanations which view party com- petition and reapportionment as key elements downplay the importance of incumbent advantages. The emphasis shifts from the individual Officeholder activities to party loy- alties among the electorate and the way in which they are divided among population subgroups. 0n the other hand, distinct activities, such as campaign spending, casework, advertising, and position- taking, tend to cluster in the sense that they are based on the attempt by an incumbent to appeal to the voter as an individual candidate as opposed to a representative of the party. While the party label is in most cases a necessary prerequisite, the incumbent need not emphasize party loyalty in order to win. n . 3h Summary The purpose of this review was to go beyond a mere description of what has been written concerning turnover in state legislatures to cataloging possible factors which could be used to explain turnover more fully. Previous studies show that turnover has not remained static over time. Rather, there is evidence that it has declined in recent years. In addition, levels of turnover are not uni- form throughout the states. There are variations in turnover within states over time and between states for any particular point in time. Finally, there are indica- tions that the decline in turnover in recent years reflects a decline in the number of incumbents voluntarily withdraw- ing from office. The proportion of incumbents suffering electoral defeat has remained relatively constant. Explanations of turnover in state legislatures have dealt with numerous independent variables including the incidence of reapportionment, party competitiveness in the state, the number of opportunities available for higher office, size of the legislative chamber, salary levels and the length of the legislative session. or these, the incidence of reapportionment and the number of opportuni- ties available for higher office receive some empirical support and should be given further consideration. Although most authors agree that the amount of turnover in a state legislature is affected by the attractiveness of legislative service, there is almost no empirical support 35 for this contention. However, previous studies use compen- sation as the primary determinant of the attractiveness of office, ignoring the possibility that other factors, such as office space or legislative staffing, might play a key role. Further, these efforts are limited either with respect to the number of states included in the analysis or with respect to the time period under consideration. Both limitations may distort subsequent analysis. Thus, the effect of the attractiveness of legislative service on turnover in state legislatures has not been adequately tested and, as an independent variable, attractiveness remains plausible from a theoretical standpoint. While the traditional view of turnover was that it resulted primarily from voluntary withdrawal from office, recent data show that in some cases a relatively high percentage of incumbents leaving office do so because of defeat at the polls. I have turned to the literature on U.S. Congressional elections to seek out variables which might be useful in explaining electoral outcomes at the state legislative level as well. Four variables merit further consideration: party competitiveness in the dis- trict, the incidence of reapportionment, advantages due to incumbency and campaign spending. While early studies demonstrated the theoretical and empirical importance of party competition on congres- sional election outcomes, later research discounted its impact. The erosion of party identification among voters 36 and their subsequent need to look elsewhere for voting cues is one possible explanation for the diminished importance of party competition. But the state legislature is not a high visibility office and alternative voting cues may not be present. Thus, party competition merits theoretical consideration when discussing state legislative elections. A similar case can be made with respect to reapportionment. Empirical support for the impact of reapportionment on congressional elections is mixed, but theoretically, with various state officials in control of redistricting, a stronger relationship is expected in state legislative elections. Recent authors studying congressional elections argue that incumbents enjoy-a number of advantages which go hand-in—hand with holding office and enable them to win reelection. These advantages include cheap advertising and the opportunity to engage in position-taking, credit- claiming and constituency service activities. Data on the specific effects of these variables are lacking. However, research does show that incumbency has an impact on the vote when other variables, such as party competition, are controlled. Clearly, a theory of turnover in state legis- latures must take into account the possibility that incumbent advantages operate at the state, as well as the national, level. The effect of campaign spending on election outcomes has also received attention lately. Analyses show that 37 incumbents outspend their challengers and although the evidence is limited, incumbents appear to have an advantage in raising the money necessary to conduct an effective cam- paign. This finding, coupled with an awareness that mass media campaigning has become progressively more expensive and prominent in state legislative races, has lead to the inclusion of campaign finance in my theory of turnover. The literature presented in this chapter suggests a number of variables which are important in a theory of turnover in state legislatures. The level of party compe- tition, the incidence of reapportionment, the attractiveness of the legislature, the number of opportunities for higher office, the existence of activities which incumbents can use to their advantage at election time and the differential ability of incumbents and challengers to finance their campaigns are all variables to be considered. Such consideration is the task of the following chapter. FOOTNOTES 1Charles F. Hyneman, ”Tenure and Turnover of Legis- lative Personnel, ” Annals of the American Acade of Political and Social Science, 195 (January, 1938;, p. 25. 2Belle Zeller (ed.), American State Legislatures: Report of the Committee on American Le islatures: American Political Science Association (New York: Thomas Y. Crowell, 195k), pp. 65-3?. 3Malcolm Jewel and Samuel C. Patterson, The Legis- lative Process in the United States, 2nd ed., (New Yerk: Random House, 19665, p. 120. thid., p. 119 5James D. Barber, The Lawmakers: Recruitment and Ada tion to Le islative Life (Neww Haven: Yale University Press, 1965;, p. 6Ibid., examples of the variables Barber examined include: (1) number of elections for representative won by the minority party in the town, 19u6-58; (2) dwelling units owner-occupied, 1950; (3) urban percentage in the district, 1950; 2h percentage with an income of $5, 000 or more, 1950; 5 percentage of elections in the county won by Democrats, 19h6-58. He notes, "The results were unclear and irreglllar. ”, p. 8-90 7John C. Wahlke, Heinz Eulau, William Buchanan, and LeRey C. Ferguson, The Le islative Systgg: Explorations in Le islative Behavior (New York: John Wiley & Sons, Inc., 1962;, pp. 123-7. The data used involved h72 legislators in office during 1957. Of those committed to the state legislature, only 15% mentioned political considerations, e. g., ”It depends a great deal on the political picture." (p. 126) Of those intending to leave, the reasons were mixed and in addition to reasons related to the benefits of office (such as, compensation, office space, and staffing) other responses include: 18% cited personal reasons and 13% were planning to seek another office. The authors take special note of the lack of political considerations as a Justification for leaving. One difficulty here is that the study deals with ”expectations” as opposed to actual behavior. Perhaps legislators feel that political 38 39 considerations constitute an inadequate response and thus offer an alternative rationale that is more acceptable. 8Duane Lockard, "The State Legislator," in The American Assembl : State Le islatures in American Politics (Englewood Cliffs, N.J.: Prentice-Hall, 1966), p. 103. 9Ibid. There are several ways to approach the relationship between salary and turnover in addition to using high, medium and low categories as employed by Lockard. One method would be to examine the effects on turnover following a significant change in salary. A second method would involve a comparison of legislative salary to salary levels for private careers as a means of determining the relative attractiveness of legislative salaries and the impact on turnover. To my knowledge, the latter method has never been attempted. 1oThe failure to dissect turnover into its component parts (i.e., election defeat, seeking another office, or retirement from public office) is a persistent problem throughout the literature on turnover. Without making these distinctions one can never be sure why a causal factor should affect turnover. 11David Ray, "Membership Stability in Three State Legislatures: 1893-1969," American Political Science Review, 68, No. 1 (March 197 , p. 107. 12Ibid., p. 108. 13The categories identified by Rosenthal are: the political system, electoral system and legislative system. See, Allan Rosenthal, "Turnover in State Legislatures," American Journal of Political Science 18 (August, 1974), 1l‘Ibid” p. 612. Rosenthal also finds compensation, the political opportunity structure, and a measure of the variability of voting to be somewhat important. 15It should be noted this is the period including the Supreme Court decision for "Baker v. Carr" and subse- quent rulings on apportionment. It is during the mid- sixties that turnover increases dramatically in many states. For example, average turnover in 1966 was h7.2%, the highest average for the entire l952-7h period. In Michigan, turn- over averaged 2h.h% for 1952-1976 but, in 1964, the year of reapportionment following the Supreme Court's rulings, it was 52%. 16Rosenthal, op. cit., doesn't include an initial list of variables he felt might be important. As such, it #0 is very difficult to determine what possible combinations of factors might be relevant if viewed from a different perspective. In particular, what indicators of "profes- sionalism" were considered and to what extent? Perhaps, e.g., legislative salary works best when it is viewed in combination with several other factors. 17The "dual nature" of turnover causes special difficulties because it is so difficult to gather data on both the willingness of incumbents to run for reelection and their rate of success in winning reelection over time for a large number of states. 18The variables included in Calvert's analysis are; eviction rate, partisan swing, opportunity for higher office, session length, district size, chamber size, com- pensation, reapportionment and whether the district is single member or multimember. See, Jerry Calvert, "Revolv- ing Doors: Volunteerism in State Legislatures," State Government 52 (Autumn, 1979). Pp. 175-177. 19Dona1d Stokes and Warren Miller, "Party Govern- ment and the Saliency of Congress," Public Opinion Quarterly 26 (Winter, 1962), p. 538. 2oIbid., p. 537. 21Charles 0. Jones, "The Role of the Campaign in Congressional Politics," in M. Kent Jennings and Harmon Zeigler (eds.), The Electoral Process (Englewood Cliffs, N.J.: Prentice-Hall, 1966), p. 26. 221bid., p. 27. 23Although the rate of success by the incumbent party's candidate appeared to decline in later years, this is not a strong indicator of a decline in party voting. It is possible the incumbent failed to run precisely because he sensed a shift in the split of the party vote in the dis- trict and consequently his party's candidate suffered the consequences. 2“Robert S. Erikson, "The Advantage of Incumbency in Congressional Elections," Polity 3 (Spring, 1971), p. #05. 25Ibid. 26John A. Ferejohn, "Deep Vote: Change in Voting Behavior in Congressional Elections," Social Science Work- in Pa or, California Institute of Technology, Fall, 1975), P. #1 27Robert S. Erikson, "Malapportionment, Gerryman- dering, and Party Fortunes in Congressional Elections," American Political Science Review 66 (December, 1972), p. 1236. 28113101., p. 1238. 29Paul R. Abramson, "Generational Change and the Decline of Party Identification in America," American 2211212sl.§g}ence Regisz.70 (June. 1976). p. 4 9. 3°Ibid. 31David Mayhew, "Congressional Elections: The Case of the Vanishing Marginals," Polity 6 (Spring, 197A), P0 299. 32Ibid., p. 298. Mayhew provides data on the dis- tribution of the presidential vote in congressional districts from 1956 through 1972. When comparing the shape of the presidential distributions with congressional Mayhew notes that redistricting should result in a bimodal dis- tribution (indicating more safe seats) for both. In point of fact, the congressional distribution becomes increasingly bimodal over time, while the district presidential vote remains unimodal. Evidently there is something peculiar to congressional races that goes beyond redistricting. 33Ferejohn, op. cit., p. 17. BuRichard Fenno, ”Congressmen in Their Constitu- encies: An Exploration," paper presented at the Annual Meeting of the American Political Science Association, San Francisco, 1976, p. #6. Also see, "U.S. House Members in Their Constituencies,” The Americgn Politiggl Sqignce Review 71 (September, 1977), pp. 883-917. 35Gary C. Jacobson, “The Effects of Campaign Spend- ing in Con essional Elections," American Political Science Review 72 June, 1978), p. #72. 36Ibid. 371b1d., p. A78. 381b1d. 39William P. Welch, ”The Effectiveness of Expendi- tures in State Legislative Races," American Politics 922212212;h (July. 1976). pp. 352-353. CHAPTER 2 A THEORY OF TURNOVER IN STATE LEGISLATURES The intent of this chapter is to develop a theory of turnover in state legislatures which can account for variations between states and within states over time. For the sake of clarity, the following discussion is divided into two parts. First, I will consider those factors which affect an incumbent's decision to withdraw from office voluntarily. The options of leaving politics or seeking a different office are included in this section. Second, I will consider those factors which affect the ability of incumbents to win reelection should they desire to return to the legislature.1 Throughout this chapter hypotheses will be presented which can serve as the basis for an empirical test of the theory in later chapters. The Decision to Run for Reelection Unlike Mayhew and others who study U.S. Congres— sional behavior, students of state legislative behavior are n93.5ustified in conjuring up a vision of incumbents as single-minded seekers after reelection.2 There are cer- tainly large numbers of incumbents in many states who have chosen to pass up the opportunity to run for reelection and #2 1&3 gone on to pursue other goals. Data presented by Calvert (1979, p. 180) show that it is not uncommon for over one-third of the members of a legislature to leave office voluntarily in a given election year. The question is, why? Why isn't reelection a goal for nearly all state legislators in all states? Building on the literature discussed in the previous chapter, the thesis of this sec- tion is that an incumbent's decision to run for reelection is a result of his or her assessment of the attractiveness of the legislature, the feasibility of getting reelected and the availability of higher offices which are more valued than the one currently held. The institutional environment will determine the attractiveness of the legislature. Characteristics of the legislature such as salaries, staff size, distribution of influence among members, etc., define the institutional environment. The feasibility of getting reelected is determined by the electoral environment. The electoral environment is defined by the nature and extent of com- petition in a legislative district. Finally, the avail- ability of higher office is determined by the opportunity structure. The opportunity structure is defined by the degree of competition which is likely to occur for a higher office. The less competition there is for a higher office the greater the opportunity to obtain that office. The institutional environment, electoral environ— ment and opportunity structure impose environmental #4 constraints which affect an incumbent's goals. Schlesinger (1966, pp. 9—10) distinguishes three types of goal-directed behavior: discrete, static and progressive. An individual with discrete ambition plans on holding a particular office for a limited period and then returning to private pursuits. An individual with static ambition is concerned with making a career out of a particular office. Finally, an individual with progressive ambition plans on using a particular office as a stepping stone to higher office. Undoubtedly state legislatures attract all three types. However, the proportion of individuals motivated by dis- crete, static or progressive ambition in a state legisla- ture can be expected to vary from state to state. Sal- aries, office space and staffing vary, as does the competitiveness of elections and the availability of higher office. Candidates run for legislative office initially because of their assessment as to how favorable the insti- tutional environment is to the pursuit of their goals. Once office is obtained, their environment continues to play an important role. Some legislators find their initial assessment to have been somewhat off target and consequently redefine their goals. Even those who find that their initial assessment was correct may soon dis- cover that their environment is susceptible to change. Salaries rise, opportunities for higher office suddenly #5 open up or a redistricting plan turns a comparatively safe district into a highly competitive one and goals shift accordingly. Thus, a legislator's environment will have a decisive impact on his or her decision to run for reelec- tion and a further examination of that environment is in order. Up to the point when an announcement must be made, the willingness of an incumbent to seek reelection is subject to a number of variable conditions. Certainly the goals which prompted him or her to seek office in the first place will have an impact on any decision to run again. The Institutional Environment The institutional environment includes those char- acteristics of the legislature which are likely to be perceived as benefits by its members. It involves two broad classes of benefits, material and psychological. Material benefits include those items traditionally used to distinguish professionalized from amateur legislatures. As was pointed out in the previous chapter, most authors focus on legislative salaries, but office space, expense allowances, pension levels, etc., can be included. Legis- latures which rank high on these factors provide material incentives which can be used to compensate individuals for time spent away from home, family and friends, as well as for adopting a legislative career as an alternative to private pursuits.3 1&6 Psychological benefits, while more difficult to specify, may nonetheless be just as important as material rewards. The reference here is to the degree in which performance of the legislative task yields a sense of personal satisfaction or feelings of individual achieve- ment for the incumbent. The conditions under which a state legislature could foster greater psychological bene- fits as perceived by an incumbent are unclear. However, several plausible possibilities may be considered. One possibility is to turn to these asPects of the institu- tional environment traditionally used to distinguish professionalized from amateur legislatures. In this case, we are dealing with the degree to which the legislature is supported within government. Are legislative services provided, such as bill drafting, budget review, reference facilities, etc.? Is the legislative budget adequate? Are legislators provided sufficient staff, both as indi- viduals and in committee? Does the legislature meet year around (or nearly so) thus indicating the importance of the legislature's role in government? Emphasis in these areas not only distinguishes a legislature as professional- ized, but more important, will enhance the perceptions of incumbents as to the worth of a legislative career. Related to the amount of support provided to the legislature within government is the ability of the legis- lative body to play an active role in government decision 47 making vis-a-vis other institutions, such as the Governor's office. In any occupation, the feeling that your voice will be heard and can have a meaningful impact is essen- tial to feelings of self worth and job satisfaction. If it is the Governor who dominates decision-making, then the likelihood of psychological incentives of this nature are reduced.4 Also of significance to the incumbent's psycho- logical perceptions is the distribution of influence within the legislature. Do all members have a reasonable opportunity to become committee chairmen? Are all members assured that their bills will be heard and taken seriously? Do minority party members have reason to expect that they will one day be in the majority? If these conditions are favorable, the psychological rewards will be greater.5 Finally, the importance of state government as a whole contributes to perceived psychological benefits for the legislator. It would be absurd to argue on the one hand that a sense of satisfaction arises from the impor— tance of the legislative role, while on the other per— ceptions are held that state government itself is not important. Assuming that the role of the legislature is significant, the legislator's belief in state government as a meaningful mechanism through which to address the problems of society can only serve to enhance the psycho- logical benefits of holding office. #8 Consideration of the benefits indicated above is essential to an understanding of state legislative turn— over. If we assume incumbents are rational actors in the sense that they weigh cost vs benefits as part of their decision to run for reelection, the specification of what those benefits are, or might be, is important. In addi- tion, the above discussion identifies at least some costs as well. If material or psychological incentives are lack- ing, their absence may quickly translate into an additional burden to be borne by the incumbent. This point is alluded to by Barber (1965, pp. 6-7): The Work involved in a legislative job-~if it is conscientiously done-~is prodigious and its con- ditions difficult. The scope, volume, and complexity of legislation have increased tremen— dously in the last fifty years, but the number of congressmen has remained about the same and the number of state legislators has decreased. State governments are directly responsible for spending more than twenty billion dollars annually. More and more state legislatures meet in annual rather than biennial sessions. Regu- lar sessions have lengthened and special ses- sions have become more frequent. . . . In 1960, only three states (California, Florida, and Texas) provided private office for legislators. The median biennial pay for the thirty-four states using salary plans in 1962 was $3,000- 4,000. Finally, the benefits available from legislative service have an impact on the kinds of people who are attracted to the legislature initially and may play a decisive role in reshaping their goals once elected. Since the interaction between goals and environment will help determine the proportion of members in a legislature 1&9 who are willing to run for reelection (thus affecting turnover), we need to specify the different outcomes of that interaction which can be expected. Legislatures which offer little by way of material and psychological benefits are likely to attract a large proportion of individuals interested in serving only one or two terms. Motivated by discrete ambition, their goals may include the prestige of serving in a state office, generating publicity for their law practice or business, and/or fulfilling a sense of civic duty. In any event, their goals do not impel them to run for reelection. Legislatures characterized by high benefits will also attract individuals motivated by discrete ambition. However, entry into those legislatures is expected to be more difficult. The value of the office is higher, challengers are more likely to be career-oriented candi- dates and there is a greater likelihood of having to defeat an incumbent. Individuals motivated by static ambition are not likely to seek out a career in a legislature which involves low levels of benefits. Not only will the lack of mater- ial benefits create hardships, but the lack of psycholog- ical benefits can be a crippling blow. Presumable career oriented individuals are interested in the legislature because they want to help shape the future for their state, immerse themselves in the exciting and important process 50 of decision-making, and/or exert influence over their colleagues. They want to be "where the action is." But, a legislature which does not play a central role in decision-making, is not supported with legislative services and facilities, or does not provide sufficient opportun- ities for influence within the institution is unconducive to those goals. Thus, individuals who are interested in a career in public life will seek alternatives to the legis- lature in states where material and psychological benefits are lacking. In states where benefits are not lacking, a greater proportion of the legislature will be career- oriented. It is difficult to predict what impact the attrac— tiveness of the legislature will have on the recruitment of people motivated by progressive ambition. Since these individuals are primarily interested in the state legisla- ture as a stepping stone to higher office, levels of material and psychological benefits are irrelevant to their initial decision to serve. But, what happens when the anticipated opportunity for advancement fails to material- ize? Surely this often occurs given the unpredictable nature of politics. .When the opportunity for higher office is not forthcoming, legislators motivated by progressive ambition are likely to leave legislatures which offer few benefits and pursue alternative strategies. They will become 51 disillusioned at their failure to advance and find it dif- ficult to accept the alternative of a state legislative career that offers low pay, few supportive services and/or little chance for influence. Legislatures that are more attractive can be expected to retain a higher proportion of these incumbents. Some individuals will redefine their goals and be content to develop a career at the state level. Others will find it easier to remain in a legisla- ture offering attractive benefits until a future opportun- ity for higher office comes along. Regardless, a bid for reelection becomes a viable alternative to leaving office. It is unlikely that all legislators' perceptions of legislative life mirror the actual conditions faced while in office. While it is unclear how their prelegis- lative assessments are formed, it is only after legislators have begun serving in the legislature that they feel the full impact of the institutional environment.6 Goals may be redefined as a result. For example, it is undoubtedly the case that a number of individuals with static ambitions find themselves serving in legislatures lacking in material and psychological benefits. When this occurs, many become disillusioned and forgo a reelection effort. Should an opportunity arise, they may run for another office or accept an appointment elsewhere in government. Or, they may simple leave the public sector entirely. ’7 av;- 52 This is not to say that individuals with static ambitions can be found only in attractive legislatures. Many, such as small businessmen, teachers, or farmers, may have private occupations which enable them to serve in unattractive legislatures with little or no loss in income and marginal drawbacks from the time spent in public life. And it could well be the case that what is frustrating and disillusioning to one individual is worthwhile and exciting to another. However, the proportion of career- oriented legislators is expected to be lower in legisla- tures which offer fewer material and psychological benefits. An institutional environment that is characterized by high material and psychological benefits provides incentives to legislators to pursue reelection either to further a career in the legislature or as a viable strategy while waiting for an opportunity for higher office. Legislatures lack— ing those incentives attract legislators who are less willing to run for reelection. These considerations lead to my first hypothesis: H1: Among state legislatures, the proportion of state legislators withdrawing from office voluntarily is inversely related to the degree to which material and psychological benefits are provided in the legislature. The Electoral Environment Despite the importance of benefits to the incumbent with regard to reelection as a goal, considerations such as these are not the only basis for a decision. Rather, the 53 reelection decision will be made in the context of the electoral as well as the institutional environment. At some point the legislator must stop and ask the question; is reelection feasible? There are costs associated with campaigning to be considered and again they may be both material and psychological in nature. Agranoff (1972, p. 27) estimated in 1972 that a state legislative campaign can cost from $10,000 to 20,000 and these expenditures have been increasing over the years. However, financial costs are often covered by contributions and may not be as important as other costs, such as the time which must be spent away from one's business, profes- sion, or family during the campaign. In addition, there are the psychological costs involved in putting oneself before the public and in at least some cases, having to suffer abuse or criticism not to mention the prospect of defeat. It is difficult to estimate how burdensome these costs are in an incumbent's experience. But, as the like- lihood of marshalling the support necessary for running an effective campaign decreases, the costs will become increasingly oppressive. Assuming the candidate's goal is winning (as opposed to taking an ideological stand, "airing the issues,” etc.) the prospect of certain defeat will turn him or her away from the reelection bid as the costs will simply be too great. But, as the likelihood of winning 5h increases, his or her assessment of the cost/benefit ratio will in large part determine the willingness to run. What factors are incumbents likely to use as measures of the likelihood of their chances in an upcoming election? One important datum is bound to be his or her electoral performance in past elections and the candidate‘s assessment of the reasons for success. Prior election victories and, especially, the magnitude of those vic- tories, either in the primary or general election, may well be base line data for predicting success in the future. However, it is difficult to determine just how much weight incumbents give to these considerations. Fenno (1978, PP. lhO-lhl) reports that even congressmen who have won the previous election by substantial margins (70—80 percent) perceive an upcoming election as involving an element of risk. This may be due to the closeness of earlier elec- tions or perhaps to lack of information about the chal- lenger. It is difficult to determine beforehand just how a campaign will develop even if one's opponent is known. However, in many situations, even that information is unavailable until after the primary election. What may be more useful to the incumbent is the nature of his or her prior election victories. Specif- ically, the question will be asked, why did people vote for me? The literature discussed in the previous chapter suggests two alternative responses. First, people may have voted because of party loyalty. Knowing the previous 55 split in the party vote, the incumbent may infer how well he or she might expect to do the next time around. Of course, party voting will not be as useful in gauging future primary election outcomes as intraparty conflict will reduce its reliability. Second, people may have voted for the incumbent because of candidate characteristics independent of party considerations. There is evidence that party loyalties among the electorate are waning. A higher incidence of ticket splitting, the possible increase in issue voting and the growth in the number of voters classifying them- selves as "independents" all point to this conclusion. Consequently, incumbents have an advantage. Incumbents are likely to be better known than their challengers because of past campaigns and their ability to advertise during their term in office. They can promote issues important to the constituencies by introducing bills, offering amendments and voting on legislation. Finally, incumbents are able to do favors for constituents through "casework" activities. For the incumbent who can exploit these activities, there will likely be a reservoir of favorable constituent senti— ment that transcends partisan attachments at election time. The basis of voting behavior can have an important impact on an incumbent's decision to run for reelection. If people are voting out of partisan loyalty, information on prior elections is most useful in situations where party identification is stable and the party split within a 56 district constant over time. In districts dominated by one party, the feasibility of winning becomes irrelevant (with the exception of a possible primary election battle) and the bid for reelection becomes a question of the costs vs. benefits of holding office. Districts producing a competi- tive, two-party balance compound the question of running again as the risks are maximized. More important, the incumbent has little control over the percentage of elec— tors who consider themselves Democrats or Republicans, and consequently, over the outcome of the election. In these situations I would expect a high incidence of voluntary withdrawal from office. When voters use characteristics of the candidates as a cue instead of party, incumbents are faced with past election information that is somewhat less reliable than would otherwise be the case. Even if their expertise at advertising, credit—claiming, position—taking, etc., car- ried the day two years previously, there is little guaran— tee voter loyalty will continue. A well-financed, charis— matic opponent with an effective campaign organization can make past election results virtually useless as a guide to the likelihood of winning this time. Of course, the more competitive the previous election, the greater the risk in mounting a current campaign. However, in competitive districts where a higher proportion of voters rely on candidate characteristics as a voting cue incumbents are less likely to withdraw from 57 office voluntarily than in competitive districts where a higher proportion of voters rely on party loyalty. In districts where candidate characteristics dominate voting incumbents are forced to cultivate the "independent" voter. Party support will be forthcoming with a minimum of effort. In districts where party voting dominates, incumbents cultivate "party" voters. The small percentage of independent voters which can sway the election are unpredictable. This creates a situation in which electoral risk is maximized. In short, districts where voting is candidate-centered provide incumbents with more control over election outcomes than is the case in districts where voting is party-centered. Given greater control, incum- bents are less likely to voluntarily withdraw from office even when information points to the possibility of a competitive election. My second hypothesis is derived from the foregoing discussion. H2: The proportion of voluntary withdrawals from a state legislature is directly related to the proportion of competitive districts represented in that legislature. The competitiveness of districts will provide incumbent's with a good indication of the feasibility of winning. However, the unwillingness of incumbents to seek reelection in competitive districts is mitigated by the nature of voting behavior in the district. When votes are cast because of party considerations, incumbents can 58 do little to sway independent voters in their district and are likely to withdraw voluntarily. When voting is candidate-centered, incumbents cultivate the independent vote and gain more control over election outcomes. Thus, they are likely to run again. Although the vote for state legislator is a mix— ture of party and candidate related factors, the predic- tive value of both to an incumbent assessing his or her reelection chances may decline dramatically following reapportionment. Intuitively, one might expect any change in district boundaries to instill the fear of defeat in the hearts of incumbents. But this will only be the case under certain conditions. Incremental shifts in district boundaries may have little effect on election outcomes, especially if voters rely on party loyalties and party balance is either undisturbed or shifted in favor of the incumbent. 0n the other hand, massive redistricting, such as occurred in most states after the Supreme Court rulings following "Baker v. Carr," would significantly affect the quality of information which could help incumbents assess the feasibility of running again.7 This is especially true if voters use candidate characteristics as the basis of their electoral decision. Whatever goodwill was fostered by the incumbent in the past is lost when his or her district is radically altered and a new block of voters, to whom the incumbent is unknown, is brought in. 59 Candidates relying on a party vote will be affected in a similar manner, although to a lesser extent. Major redis— tricting can be expected to shift the party balance between the old and new district, however, recalculation of the party split as a guide to the chances of winning reelection should be a bit easier and more reliable. These consider- ations lead to the following hypotheses. H3: In election years following legislative reappor- tionment the proportion of incumbents withdraw- ing from office voluntarily in a state legislature will be greater than in years when there is no legislative reapportionment. And: Hh: In a legislative district where district bound- aries have been significantly altered by reapportionment, the likelihood of an incumbent withdrawing from office voluntarily is greater than in a legislative district where district boundaries have been insignificantly altered. The existence of reapportionment is likely to create uncer- tainty for incumbents. Thus, the risk of defeat increases and the likelihood of running for reelection is reduced. This is even more likely when district boundaries are radically altered as a consequence, uncertainty is heightened. Reapportionment will also affect the role played by the availability of higher office in luring incumbents away from the legislature. When district boundaries are radically altered, incumbents with progressive ambition are forced into an untenable position. Since the feasibil- ity of reelection is reduced, incumbents may be unable to 60 wait for an opportunity for advancement. Some will see the situation as hopeless and leave politics altogether. Others, perhaps more committed to moving up, will force the issue and run for higher office despite the lack of a good opportunity. Incumbents faced with the prospect of running for reelection will consider an environment that includes both institutional and electoral factors. Some incumbents will base their decision to run for reelection solely on their assessment of that environment. Others (perhaps most), will also take into account the availability of opportun- ities to seek higher office. What is the opportunity structure expected to look like? How can an "opportunity" be defined? The following section attempts to answer these and related questions. The Opportunity Structure Schlesinger (1966, p. 72) suggests that state leg- islative office is a "base office." A "base office" is an office which can be used to gain entry into a political career, provides experience which will be useful later and serves as a springboard to higher office. Schlesinger adds; "The number and accessibility of state legislative and local elective offices make it likely that many career- ists in politics will at some point have hold such office" (1966, p. 72). But this is not to say that most state legislators will pursue higher office. For that to happen, 61 two conditions are necessary; that most legislators desire higher office and that opportunities for higher office are present. Both conditions are addressed by other authors. Schlesinger (1966, p. 10) states; "A likely assumption is that progressive ambitions dominatelin rela- tion to discrete and static ambition? and are suppressed only when they appear unreasonable in terms of chances [3r opportunities for higher officé7." Rohde (1979, p. 3), analyzing progressive ambition among members of the U.S. House of Representatives, states: We believe, and here explicitly assume, that pro- gressive ambition is held by almost all members of the House. That is, we assume that if a member of the House, on his first day of service, were offered a Senate seat or a governorship without cost or risk, he would take it. I agree, and consequently assume that most state legisla- tors would seek higher office if no cost or risk were involved. Is it likely that seeking a higher office will involve no cost or risk? Probably not. Rather, it is likely that a state legislator will face at least some competition when seeking higher office. The nature of that competition determines the degree to which an oppor- tunity for higher office is present. In this sense we can discriminate between opportunities that are ”re- stricted" or "open" for higher office. A ”restricted" opportunity exists whenever the competition for office is high. As an example, consider 62 the case of a state representative interested in running for the U.S. House. If the incumbent congressman was running for reelection, was popular and was of the same party as the state representative, it would be difficult to argue an opportunity for Congress was truly available. Of course, the state representative could legally run, but in this case he or she probably would not as the cost would be prohibitive and the risk great. 0n the other hand, an "open" opportunity would be present if the state representative sought a congressional seat whose incumbent was retiring, had been popular for many years and chose to endorse the state representative as his replacement. In this instance, the representative is likely to run for Congress as the costs will be manage- able and the risk minimized. The point is, state legis- lators will consider the degree to which an "epen" oppor- tunity for higher office exists, before they try for higher office. Therefore, my next hypothesis is: H5: The likelihood that a state legislative incum- bent will run for higher office is directly related to the degree to which an "open" oppor- tunity for higher office exists. The proportion of "open” opportunities for higher office is expected to vary between states due to struc- tural characteristics as well. There are several reasons for this expectation. First, the ratio of higher office ’ to lower office varies between states. For example, in the New Hampshire General Court there are 2h senate seats 63 9 and #00 house seats. Since all New Hampshire legislators serve two year terms, even if all state senators left office, the ratio of senate seats available to representa— tives would be only .06. Ih.contrast, the Colorado General Assembly has 35 senators and 65 representatives, all serv- ing two year terms.10 The maximum ratio of senate oppor— tunities for representatives would be .5h. Consequently, a larger proportion of representatives would be expected to run for higher office in Colorado than in New Hampshire. Second, the length of term for various offices differs between states as well. In 33 states, senators serve four year terms while representatives serve for two years. If the size of the legislature is held constant, there will be only half as many potential "open” opportun- ities in election years as there will be in states where all legislators have equal terms. Finally, the proportion of "open" opportunities can be expected to vary between states because turnover for different offices varies from state to state. In Kentucky the state constitution prohibits the governor from succeeding himself. In Indiana the governor is limited to two consecutive four year terms. In Michigan, the number of terms the governor may serve is unlimited. Thus, "open" opportunity is partially built into the system in some states. Turnover can occur for other reasons as well. The state senate in some cases may be the dominant 6h path to the U.S. Congress. In other cases, the state senate may be a career in and of itself. States are expected to vary with respect to the proportion of ”open” Opportunities for higher office which are available, and consequently, I offer the following hypothesis: H6: The proportion of state legislative incumbents seeking higher office will be greater in states that offer a high proportion of "open" oppor- tunities than in states which offer a limited proportion of "open" opportunities. State legislators, faced with the decision of whether or not to seek reelection, ponder a number of interrelated variables. Certainly their personal goals will have a decided effect on the course they ultimately decide to take. However, their goals are in large part determined by a continual reassessment of the institu— tional environment, and the desirability of office involves the consideration of factors such as salary, staffing, expense accounts, and the importance of the legislative task. If these benefits are high enough, costs incurred by leaving a private occupation, having to spend time away from one's family, and an extensive workload will be deemed worth it. The electoral environment also plays an important role. Campaigning entails additional costs, and consid- eration must be given to the feasibility of winning reelection. In this context information based on prior 65 election outcomes must be evaluated with special emphasis placed on the reliability of this information. The extent of competition in the district, changes in district bound- aries which can alter the balance of competition, and the nature of that competition in terms of party base or candidate base are interrelated factors which can help the feasibility of winning. However, even if the scenario outlined above proves favorable, the incumbent might forego a reelection bid if an "open" opportunity to run for higher office is present and the risks are reasonable. The theory discussed thus far is illustrated by the model presented in Figure 1. Since this model is only applicable to an incumbent's decision to run for reelection and excludes consideration of election defeats, the total variation in aggregate turnover is unaccounted for. Assuming for the moment that incumbents are rarely defeated, the model is useful in two important respects. First, by conceptualizing turnover as a function of a number of interrelated factors, the model accounts for evidence from earlier studies, which viewed turnover in rather simplistic terms. No relationship was found in a statistical sense between turnover and independent vari— ables hypothesized as important. Thus, for example, Lockard was unable to demonstrate that salary was a deter— minant of turnover in large part, perhaps, because he failed to control for the simultaneous effects of other factors. 66 FIGURE 1. Why Legislators Seek Reelection INSTITUTIONAL ENVIRONMENT: Degree of Professionalism in the Legislature Distribution of Influence Within the Legislature Importance of the Legislature Within Government \ Attractiveness of the Legislature Importance of State Government Opportunity for Higher Office ELECTORAL ENVIRONMENT Previous Electoral Success Competitiveness of the District Decision Stage Leave Politics Perceived Chance of ___3fi__l [Redistrictingw Reelection Perceived Level of Pro-election Support Run for Higher Office Seek Reelection 67 Second, the explanation illustrated by the model in Figure 1 provides a clear rationale for the existence of rather high levels of state legislative turnover preva- lent through the 1950's and the apparent decline in turn- over rates since. Historically, most state legislatures have failed to provide a benefit package attractive enough to induce large numbers of incumbents to bear the costs associated with serving in the legislature and/or running a reelection campaign. However, in recent years, the benefits have increased, in some cases quite dramatically, as many state legislatures have become more "professional- ized" and state government has played an increasingly important role in solving societal problems. Assuming that the feasibility of winning has remained constant and the costs of campaigning have not increased significantly rela- tive to the ability of incumbents to garner additional support, the decline in turnover is consistent with predic- tions which would be obtained on the basis of the model. However, the theory advanced thus far is both imprecise and incomplete. It is imprecise since there are no grounds for deciding whether the model is simply addi- tive or else, more complex. There is little evidence to warrant a particular form of the model at this stage although alternatives will be considered in a later chapter. The theory is incomplete because it fails to con— sider the impact of incumbent defeats on state legislative turnover. Recent data show that the proportion of state 68 legislative incumbents who seek reelection but are defeated varies between states and within states over time (Ray, 1976. pp. h30-h3l and Calvert, 1979, p. 180). To explain variations in turnover due to incumbent defeats, state legislative electoral outcomes are discussed in the following section. Election Outcomes for State Legislators As indicated earlier, the willingness of incumbents to run for reelection provides no guarantee they will in fact win. The central thesis of this section is that once the decision to run for reelection is made, electoral suc- cess is highly probable because of advantages peculiar to incumbency. However, before proceeding it is necessary to briefly discuss why this argument is not structured around party competition as an explanation for election outcomes. The traditional explanation for electoral success is that it is due, for the most part, to the degree of party competition in the district. According to this view, people develop strong party loyalties over time. Assuming party balance in a district remains stable over time the incumbent's chances for reelection depend largely on the degree of party competition in the district. This does not mean that incumbents will always win if they are from districts where one party dominates. They are still sus— ceptible to defeat in the primary where the party label may be meaningless. But, what happens when this scenario changes? 69 As was noted briefly in Chapter 1, there is evidence available which indicates the extent and degree of party identification among the electorate has declined. Although it is unclear from the evidence whether an erosion in party identification applies only to voting for high visibility candidates such as President or Governor, or also to voting for state legislative candidates, it is clear that more people consider themselves independents now than in the past and consequently split their ticket when voting. Tc the extent that this is true, there are fewer people voting out of party loyalty and more basing their vote on candidate characteristics. If the party label and all it connotes can no longer suffice for many voters, other sources of informa- tion become especially important. Television, radio, news- papers, etc., will provide coverage of the candidates and issues. However, state legislative candidates face partic- ularly severe conditions in getting their message across. The emphasis of the media will be on the races for higher office such as President, or Governor, and coverage of legislative candidates may well be restricted to paid advertisements. Also, the media markets are probably quite imperfect for state legislative districts. In many cases state legislative districts are considerably smaller than the area served by the media. Because of this the costs are often prohibitive for candidates. The problem of lack of coverage is especially acute given the 70 realization that the average voter in the United States is increasingly apathetic and uninformed as one moves from presidential to state legislative politics. Given an environment characterized by low informa-‘ tion levels among the electorate, any additional informa- tion is likely to have substantial impact on voting in state legislative contests. Candidates who can raise (and thus, spend) more money, have the edge over their oppon- ents because they can get additional information to the voter. This line of reasoning leads to my next hypothesis. H7: In state legislative elections, the proportion of the vote for a candidate is directly re— lated to their ability to raise money for the campaign. Silberman (1978, pp. lB-lh) suggests congressional incumbents have several advantages: . . . the natural advantage of experience, previous public exposure and a public record. Incumbents possess a relatively high 'recognition factor' com- pared with challengers. The 'disinterested major- ity' is more likely to support an incumbent whom they recognize rather than a challenger about whom they have little information . . . the perquisite of the office, notably, the incumbent's staff offices, almost unlimited free postage, and exper- ience in campaign organization to assist his reelection effort . . . Finally, the incumbent may have an advantage in the ability to raise campaign funds. Do these same factors hold at the state legislative level as well? Probably so, especially in those state which may be classified as having professional legislatures. As state legislatures become more professionalized and state government continues to grow in importance, the 71 visibility of state legislators should increase. For example, as state legislatures become more professionalized, salaries, franking privileges, staff size and the length of the legislative session all increase. In turn, legislators are in a better position to gain public exPosure from activities, such as credit-claiming, advertising, and position-taking (as Mayhew argues). Since the level of professionalism of the legislature varies from state to state, the opportunities for incumbents to use their office for electoral advantage will vary as well. Therefore: H8: Among state legislatures, the proportion of incumbents who are defeated for reelection is inversely related to the degree of pro- fessionalism in the state legislature. Furthermore, the importance of state government, manifested by increased regulation of business, increased state budgets, and an increase in the number of state supported programs in health, education, welfare, employ- ment, and other policy areas, should provide incumbents with additional campaign resources. Interest groups and individuals concerned with influencing policy are likely to offer greater support to incumbents than to their challengers. Incumbents are known quantities and their positions on issues are somewhat predictable. And, ceteris paribus, incumbents will probably win reelection. That incumbents have won in past elections is an indication (albeit an imperfect one) that they are better than average campaigners. In addition, incumbents have a 72 campaign organization that is experienced and can be expected to correct past mistakes. Concerned groups and individuals will support incumbents as a way of improving their odds of getting, at least, something in return. Will interest groups and individuals support all incumbents equally? It is unlikely. Incumbents who have more influence within the legislature are in a preferred position as compared to incumbents with little influence. Since the distribution of influence within the state legb islature varies from state to state (for example, in the number of chairmanships per member) the proportion of incumbents who can translate institutional influence into an electoral advantage varies as well. Thus, my hypothesis is: H ° Among state legislatures, the proportion of incumbents who are defeated for reelection is inversely related to the proportion of incumbents who hold positions of influence within the legislature. It is also likely that incumbents are increasingly able to fulfill the "ombudsman" role as identified by Fiorina (1975). The ombudsman role provides a legislator with an opportunity to serve as the middleman between the bureaucracy and constituents, and to the degree the legis- lator can cut through bureaucratic "red tape" and thus solve constituent problems, he or she can be expected to win additional votes. Opportunities to do so may well be limited at the state level however, as they are dependent on the size of the state bureaucracy as well as the extent 73 of regulatory or programmatic activity engaged in by the bureaucracy. In addition, legislators must be viewed by the citizenry as a viable means of solving frustrations created by the bureaucracy. However, incumbents are cer- tainly in a better position to play a role in these matters than their opponents. Therefore: H10: Among state legislatures, the proportion of incumbents who are defeated is inversely related to the size and scope of the state bureaucracy. The preceding discussion assumes incumbents have a decided electoral advantage over their challengers. However, incumbent advantages will be less important when situations arise in which party-voting is more prevalent One such situation may occur during primary elections. Primary elections can be expected to attract a higher proportion of party-centered voters than general elections. People who consider themselves independents or who have relatively weak partisan attachments are more likely to view the primary as a "party affair" of little concern to them. When incumbents make use of the advan- tages peculiar to holding office, they also shift the focus of their campaign away from the party and toward themselves as individuals. As a consequence, party loyal- ists may become alienated and incumbent success in the primary election less likely. This situation is more probable in states which provide opportunities for incum- bents to engage in activities, such as credit-claiming, 7h position-taking and advertising. From this my eleventh hypothesis follows: H 11: Among state legislatures, the proportion of incumbents who are defeated for reelection in the primary election is directly related to the degree to which the legislature provides opportunities for incumbent advantage. In some states, a higher degree of party-centered voting may be likely in the general election as well as the primary. In states characterized by part-time, citizen legislatures, the opportunities to do case work, claim credit and maintain visibility are reduced. For example, if the legislature meets for only two or three months a year, it is unlikely that "ombudsman" role will be beneficial to incumbents. For the remaining nine to ten months of the year, citizens will turn to other officials as a way of dealing with their frustrations with the bureaucracy. Under these conditions, the party label will be retained as a voting cue. Furthermore, major redistricting in the state will offset many of the advantages an incumbent has built up. The influx of new voters, unfamiliar with the incumbent's name, style, and abilities will create a situation in which more traditional voting rules apply. Once again, the party label is a viable alternative. Finally, events of statewide or national signifi- cance can have an effect on an incumbent's ability to gain electoral advantage from holding office. Even though 75 party loyalties have declined, many voters still rely on the party label when casting their ballots. Events which result in significant shifts in loyalty from one party to the other, can offset the electoral advantage of an incum- bent who happens to be on the wrong side. The relatively high proportion of Republicans who voted for Democratic candidates in 197M following "Watergate" serves as a case in point. The existence of statewide or national events which can shift party loyalties or the incidence of reapportionment represent situations whereby the importance of incumbent advantage in winning elections is reduced. Thus: H12: Among state legislatures, the proportion of incumbents defeated for reelection will be greater following events which significantly affect the balance of party voters in the state. The foregoing discussion leads to a model of turn- over among state legislatures that is more complete than the one previously introduced. The model is illustrated in Figure 2.12 This model proposes that the electoral success of incumbents is a function of the level of party competition in the district, the amount of money they are able to raise and spend, and their ability to take advan- tage of opportunities to claim credit, advertise, serve as ombudsman, and engage in position-taking. As the activities of state government are expanded and the FIGURE 2. Institutional Environment Electoral Environment 76 A Model of State Legislative Turnover Incumbent: Leaves Politics or Runs for Decision Higher Office Stage Opportunity Structure for Higher Office \ Percent Thrnover in the State Incumbent Leglslature Runs for Reelection Electoral Defeat CAMPAIGN ENVIRONMENT Ability of Electoral Success Incumbent to Raise Money w/A 1 Incidence of Perquisites Redistricting of Office \\\\\\ Extent of Incumbent Party Voting Advantage in District Opportunities for: Advertising Position-Taking Credit-Claiming Ombudsman Role 77 legislature becomes professionalized, the importance of party competition will decline and the impact of incumbency increase. Three points may be considered at this juncture. First, the conventional wisdom that variations in state legislature turnover is largely a function of the percentage of incumbents leaving office voluntarily is consistent with this model. But, this is so for reasons which differ from tradition. In the past, incumbents seeking reelection were successful because of the nature of party competition in their districts. However, in recent years there has been a decline in the extent and degree of party identification among voters in many states. Incumbents will win, but only because they use advantages peculiar to holding office. Second, the rationale employed here can help account for the rather high incidence of incumbent defeats in primary elections, especially in the South. The primary election removes the party label as a meaningful voting cue. At the same time, Southern state legislatures have traditionally failed to adopt professional characteristics. Incumbents are faced with a dilemma of having neither the party or perquisites of office at their disposal. It is precisely this situation which presents the greatest chance of defeat. Finally, the point should be made that the model explored thus far can help account for declining turnover in the states in recent years. For the past twenty years, 78 the most significant changes to occur with respect to state legislatures have been those associated with increased pro- fessionalism. As such, the attractiveness of the office is enhanced and the desire for reelection becomes a goal for more legislators. However, unless incumbents are able to win, turnover will remain at prior levels. That incumbents will win, even as party identification among the electorate declines, can be justified on the basis of the model explored earlier. Summggy There are two somewhat distinct processes which help account for turnover; 1.) the desire for incumbents to run for reelection; and 2.) the ability of incumbents to win reelection. The reelection bid may be viewed as a function of the incumbent's assessment of the benefit/cost ratio which results from service in the legislature and his or her assessment of the chances of winning reelection. Incumbents' perceptions of these factors will depend on the importance of state government, the degree to which the legislature has become professionalized, how wall they have done in past elections, and the existence of reappor- tionment. In addition many incumbents are likely to seek higher office should an "open" opportunity arise. Should an incumbent run again, the likelihood of winning is rather high. He or she has an experienced campaign organization, has built up a reservoir of favors, / u — 1 - — I *7 if 79 can raise the money for an effective campaign, and is probably better known than an opponent. This will be the case except when conditions are present which reduce the opportunity for incumbents to take advantage of the benefits of office in states with part-time, citizen leg- islatures or during periods of major redistricting. To what extent does the institutional environment have an impact on turnover in the state legislature? Is party competition as insignificant as I have made it appear? Do incumbents win reelection and why? These and related questions are addressed in the following Chapters. FOOTNOTES 1It is possible that some incumbents are forced out of office by local party ”bosses" through the nomination process. Thus, a third section was considered. However, the number of incumbents leaving office for this reason is not expected to contribute significantly to variations in turnover and this form of turnover was ignored. 2David R. Mayhew, Congres : The Electoral Connec- tion (New Haven: Yale University Press, 1975), p. 5. In point of fact, one difficulty faced when studying state legislative turnover is that so many incumbents do not fall into this category. 3A legislative career need not always be viewed as an alternative to private pursuits. In states with a part- time legislature, holding office is supplemental to an individual's full-time occupation. However, "professional- ized" legislatures typically meet throughout the year and a private career, at least while in office, is difficult to maintain. “Ira Sharkansky, The Maligned States: Policy Accom lishments Problems and 0 ortunities, 2nd ed., (New York: McGraw-Hill, 1978). Sharkansky notes that in many Southern states the Governor does in fact dominate state government decision-making. For an example of how Governors might be ranked according to several formal powers see, Joseph A. Schlesinger, "The Politics of the Executive," in Herbert Jacobs and Kenneth N. Vines (eds.) Politics in the American States, 2nd ed. (Boston: Little, Brown and Co., 1971), pp. 210—237, and Thomas R. Dye, Politics in States and Communities, 3rd ed., (Englewood Cliffs, N.J.: Prentice—Hall, Inc., 1977), pp. 172-180. 5The argument presented here is an attempt to recognize the possibility that the rewards of office from a material standpoint might be quite high, yet, the satis- faction one gains from playing an important role might rate rather low. This is especially true if there is little hope for future success. Surely a Republican leg- isture serving in the Alabama Assembly can entertain little hope of becoming a powerful, influential policy- maker in that body. However, if there were even a glimmer of hope toward achieving influence in the future, real or not, a reason for staying on is provided. 80 81 6The notion that expectations do not necessarily match reality is certainly not new. However, it is impor- tant that it is taken into consideration, because it is likely that many state legislators have never held polit- ical office before and therefore may hold perceptions of political life that are a bit farther from reality than might otherwise be the case. Testing the difference between expectations and reality in this context would be an interesting exercise in and of itself. 7A significant shift in the boundaries of the dis- trict need not prevent a candidate from making an assess~ ment of the party breakdown in the new district. Relying on precinct level data from previous elections, the party split can be reconstructed. However, the reliability of this information would be somewhat less than desired. The incumbent has no idea how much impact he as an individ- ual has on the strength of the party vote. In addition, he and his campaign organization has a much better idea of the strengths and weaknesses of the party vote in the old dis- trict, where his hard core support lies, the precincts he must work a bit harder, etc. In the new district, he is dealing with foreign territory. 8For an excellent discussion of the relationship between ambition and the opportunity structure see, David W. Rohde, "Risk-Bearing and Progressive Ambition: The Case of Members of the United States House of Representatives " American Journal of Political Science 23 (February, 1979 , pp. l-26. Of particular importance to the state legisla— tive level is the possibility that Rohde's argument that given a favorable opportunity, risks are minimized and legislators will run for higher office also applies to state office holders. If so, then the question becomes, how often do "reasonable" opportunities for obtaining higher office occur for state legislators? 9The Book of the States, 1980-81 (Lexington: The Council of State Governments , p. 1OIbid. 11It should be noted here that not all authors agree that party identification among voters has necessarily declined. For an alternative view see, Philip E. Converse, The Dynamics of Party Support: Cohort-Analyzing Party Identification (Beverly Hills: Sage Publications, Inc. 19735. 12For the sake of clarity, the model presented in Figure 1 has been compressed into what is here termed the institutional and electoral environment. CHAPTER 3 TURNOVER IN #7 STATE HOUSES: 1952-78 If the theory developed in Chapter 2 is sound, turnover in the state legislature is a function of a number of factors which affect incumbents' decisions to run for reelection and, in turn, their ability to win. The insti- tutional environment, electoral environment and availabil- ity of opportunities for higher office are presumed to have an impact on the decision to run for reelection. In addi- tion, several aspects of the institutional and electoral environment are expected to affect an incumbent's ability to win. This chapter presents a test of this theory. It is divided into three sections. In the first section, opera- tional definitions of the theoretical concepts are dis- cussed. In addition, the data used in the analysis are described. The second section presents the results of multivariate regression analyses for the 1h election years ill the period 1952-78, averages for the periods 1952—62, l96h-70 and 1972-78, and averages for the entire period. Finally, in the third section I will offer conclusions based on the findings. 82 83 The Data This study utilizes data from forty-seven states (excluding Alaska, Hawaii and Nebraska) for the period 1952-78. The units of analysis are the respective state houses. The data were gathered from appropriate volumes of The Book of the States. I have been hampered in collecting the data by resource limitations in some cases and the unavailability of data in other cases. This has led to several difficul- ties of which the reader should be aware. Ideally, a test of my theory would include data based on interviews with legislators which would, among other things, measure their perceptions of the attractiveness of the house, their chances for reelection and the importance of the house within government. Obviously, this is impossible for the time—span under consideration. Also, it is impractical for forty— seven state houses. As a result, I have no measure of the importance of the house within state government,1 and I have been forced to develop measures for other variables from which legislators' attitudes can only be inferred. In addition, the optimal approach for measuring turnover in the state house would be to partition it according to the proportion of incumbents who were defeated for reelection, voluntarily withdrew from public office or sought another office. With these data, one could observe 81+ possible differences in the relationships between the three kinds of turnover and the independent variables hypothe- sized as important. Unfortunately, data are unavailable for such an analysis. A final difficulty is that I have no data on district-level variables. This is important because turn- over is partially dependent on individual decisions made on the basis of district conditions. For example, competi- tion in the district is hypothesized to have an impact on both the feasibility of running for reelection and the ability to win reelection. While an aggregate measure of state-wide competition will be used, it is based on the party breakdown of seats in the legislature and is not necessarily indicative of conditions in the district. Thus, important relationships may go unnoticed. Despite these difficulties the data cover a wide range of variables for an extended time period. In addi— tion, nearly all states have been included in the analysis which will provide for an extensive test of the theory. In the following subsections the variables will be opera- tionally defined and descriptive data will be presented. The Dependent Variable: Tugpover in the House Turnover is defined as the percentage of new mem— bers in the state house following an election. A member is "new" if he or she did not serve in the house during the legislative session immediately prior to the election under 85 study and if he or she fills a seat that was vacated through voluntary withdrawal or election defeat of an incumbent. In other words, discontinuous service is ignored, and changes in the size of the legislature are taken into account. Table 1 presents the data on average turnover rates for forty-seven state houses for the fourteen elec- tions occurring during the period 1952-78 (data on turn- over for each state for this period are presented in Table Apl, Appendix A). Viewed over the entire period, average turnover rates for state houses have declined. Turnover averaged ho percent or higher throughout the 1950's, dropped below ho percent for most of the 1960's and by the end of the 1970's was under 30 percent. The variance in turnover also declined from 1952 to 1978 indi- cating states tend to cluster around the mean to a greater extent during the latter years studied. Taken together, the decline in the mean and variance indicate that the trend toward lower turnover is not restricted to a few states; rather, declining turnover appears to be a more general phenomenon involving a relatively large number of state houses. Despite the overall reduction in turnover levels fromtl952 to 1978, there are instances where this pattern is broken. The elections of l96h and 1966 are examples. .Average turnover declined in a relatively consistent 86 .poahonOH acumen w: as“: nhdoh ones» :a popaaoxo ma andfinfl504 .tovhonoh noasan :: tea: cheek muons :d popflflouo end Handnnfinofi: pad psdHath .aEunaH< .no>H»wvfionohmoH now cause pohomusuuflon .Hdohla o>d£ noudau Hfloht w: ‘3 w: .3 w: .3 w: a: ma .3 m3 .3 w: .3 .3233 Mo Honadz H.:b :.nm o.n:H a.:HH 3.00H N.:OH m.n:H b.nHN H.:NH m.ahH n.mbH m.mnH n.0wd «.mHN oocdahs> w.w~ m.mN m.:n N.mn N.Nn m.Hn H.h: w.an n.wn >.wm 3.0: m.o: N.:: :.n: ko>ouh58 0M6H0>< whmd whmd shad thd chad mwmd wwmd :wma Nme meH wnmd mnma anon mnmd Aoowdanoohom :flv whmdlmmmfi ”Mdflh ya mflwDo: m9<9m zm>fimlhhmoh ZH MH>OZ¢PH Hln HAQonfidfi Inn ufioofiom 1.0# Tm# 10m road mmdmfiw DEBAQm mom mmmDOm Edgm Zgflm Ipmoh mom mac; n n EON—uh 90 For example, in Washington, although the average turnover for 1952-78 is 30 percent, turnover is #5 percent for 1952, 26 percent for 195h, 10 percent for 197k and 36 percent for 1976. In Wisconsin, turnover for the entire period is 29 percent with 23 percent for 1952, #1 percent for 195h, 31 percent for 1972 and 21 percent for 1978. In Georgia, average turnover is 50 percent, but turnover ranged from a high of 83 percent in l95h to a low of 2h percent in 1976 and 1978. Given variations between states and within states over time, I have averaged turnover (as well as other variables included in this study) for the entire period 1952-78, and three sub-periods: 1952-62, 1964-70 and 1972-78. The election years included in these breakdowns were chosen with reapportionment in mind. The period 1952—62 covers those elections which preceded the Supreme Court ruling of the mid-Sixties estab— lishing the "one—man, one-vote" principle. As such, many states were malapportioned throughout this period. The 1964-70 period includes the elections which were affected by the Supreme Court's rulings. I use 1972 as the begin- ning of the final time period because this is the first election affected by redistricting following the 1970 census. The data for these time periods are presented in Table 2. 91 TABLE 3-2 TURNOVER IN FORTY—SEVEN STATE HOUSES FOR SELECTED TIME PERIODS (in percentages) 1952-62 l96h-7O 1972-78 1952—78 Average Turnover h1.3 38.6 33.0 37.6 Variance 13.0 10.0 5.9 7.2 Number of States* 47 47 47 47 Clearly, average turnover among state houses has declined in recent years. What is necessary at this point is a discussion of those variables that are expected to account for variations in turnover among state houses. Independent Variables: The Institutional Environment The institutional environment includes character- istics of the legislature and state government which can be expected to translate into benefits as perceived by house members. As stated in the previous Chapter (see Figure 1, p. 66) the institutional environment is made up of four components: professionalism of the legislature, distribution of influence within the legislature, the importance of state government and the importance of the legislature within state government. While I have no 92 data on the later, I have constructed measures with regard to the other three components. Professionalism By "professionalism," I mean the extent to which members are compensated, efforts are made to make the leg- islative task more manageable and the legislature is responsive to member goals.2 Presumably, if a "profes— sional" environment exists, legislators are more likely to perceive their policy making role as important and sub- sequently derive satisfaction from it. In addition, handling the often complex task of legislating will be easier and legislators will be able to fulfill supplemental goals necessary for reelection. No single characteristic will suffice to distin- guish a professional from an unprofessional legislature. Rather, professionalism embodies several aspects of the institutional environment which may vary in importance as a source of satisfaction to legislators. For this reason, factor analysis was used to construct a "professionalism index” for each state and each election year. With this method, each variable used in constructing the index is weighted in relation to its relative importance with respect to all the other variables included.3 Five variables were used to construct the "pro- fessionalism index." They were selected on the basis of data availability and theoretical considerations. For 93 some variables, such as staff size, franking privileges and office space, data do not exist. Others, such as house size, were excluded on theoretical grounds. The variables included in the index are: l. Biennial compensation of legislators 2. Annual expenditures for the legislatureh 3. Length of the two-year legislative session in calendar days A. Number of legislative services available 5. NUmber of bills introduced during the busiest year of the two-year session. The data on these five variables are presented in Table 3 (for data on each variable by state, see Appendix A, Tables A-2 through A-6). Compensation is based on the salary paid legisla- tors plus per diem expenses for the two—year session. The level of compensation is indicative of the extent to which legislators are expected to devote their full attention to the legislative task. While it is unlikely that many leg- islators anticipate great wealth as a result of their salaries, they are likely to treat their job in a more serious manner if adequately compensated. Further, the level of compensation can be thought of as a measure of the importance of legislators and, thus, will have an impact on their self—esteem. I assume that it is future compensation which effects an incumbent's decision to run again and I use 9h- TABLE 3-3 VARIABLES INCLUDED IN THE PROFESSIONALISM INDEX FOR FORTY-SEVEN STATES* (Figures are the mean and, in parentheses, the standard deviation) I Legislativeb Lengthc Legislatived s 0111se Year Compensation Expenditures of Session Services Introduced 1952 2970 593920 10h 6.3 (287) (96o)f (99.2) (1.8) NA 195a 3590 611090 92 6.5 NA (379) (816) (91.7) (2.1) 1956 4120 858190 123 6.9 1570 (A16) (1027) (105.1) (2.0) (129) 1958 uu7o 976790 126 7.3 1680 (907) (1128) (108.8) (1.9) (151) 1960 5080 11505h0 138 7.5 1660 (939) (1227) (122.9) (1.8) (192) 1962 60h0 1286730 151 7.8 1790 (595) (1299) (133.3) (115) (150) 196a 7160 1398390 1u0 8.0 1770 (648) (1369) (105.5) (1.6) (155) 1966 9570 1857170 155 8.3 19u0 (83h) (1651) (128.3) (1.5) (175) 1968 12670 2u17h20 1th 8.5 2120 (1054) (1842) (122.3) (1.9) (18A) 1970 15100 3385680 151 8.7 2680 (1222 (2158) (93.9) (1.2) (187) 1972 177u0 uh996h0 189 8.9 2310 (1275) (2834) (191.9) (1.1) (201) 9.5 TABLE 3-3 (c0nt'd.) 197h 1976 1978 20550 6323060 202 9.0 2h20 (1h68) (2801) (199.2) (1.1) (213) 21880 7902730 173 9.5 2&30 (1385) (3279) (83.7) (0.7) (223) 25730 9680330 17h 9.6 2380 (1568) (3537) (77.3) (.7) (219) *Source: For legislative expenditures; "Expenditures on the Legislative Branch," in U.S. Bureau of Census, Compendiumgof State Govern- ment Finance, for the years indicated. All other data came from: The Book of the States, Volumes 9-23. Biennial salary plus per diem expenses. Annual expenditures minus annual compensation In calendar days - figures are approximated for the two-year session Based on a scale from 0-10 with 1 point allocated for each of ten legislative services provided. . Number of bills introduced during the busiest year of the two-year session. This data was not reported until 1956 and, therefore, is not included in the professionalism index for 1952 and l95h. Since the variance is so large, the standard deviation is presented in parentheses. 96 biennial compensation for the period following an election in the index. For example, the professionalism index for 196R includes compensation for the 1965-66 sessions. As the data in Table 3 show, compensation for state legislators has increased steadily from 1952 through 1978. The lowest salary for representatives is $200 in New Hampshire. The highest is $61,600 in California. A word of caution, however, is needed. These figures are for the biennium and include per diem expenses. Few state legislators will become wealthy based on salary alone. "Annual expenditures for the legislature" is used in the index because it indicates the extent to which the legislature has made the legislative task more manageable. This variable includes expenses for legislative salaries, office space, staffing and committees. To avoid duplica- tion, I have subtracted legislator salaries from this measure. Since data on member and committee staff size, office space, etc., are unavailable, legislative expendi- tures are used as a surrogate measure. A larger legisla- tive budget indicates more staffing, facilities and/or services, which are likely to increase manageability. The data in Table 3 show that legislative expendi- tures for the period 1952-78 increased even more dramat- cally than salaries. Average salaries rose 766 percent from 1952 to 1978 while average expenditures increased 1530 percent for the same period. 97 Also contributing to the manageability of the legislative task is the number of legislative support ser- vices provided. To measure this variable, a composite score was created based on ten services available to the states during the period under study. These services include: reference library facilities, bill drafting, statutory revision, preparation of bill and law summaries, recommendations of substantive legislative programs, prep— aration of research reports, spot research and counseling, continuous study of state revenues and expenditures, budgetary review and analysis, and legislative post audits. A legislative service score ranging from 0 to 10 was com- puted by assigning a 1 if a service was provided to the legislature and a 0 if not, and then adding them together. The service score (as well as my annual expenditure measure) provides no indication of quality, but it does provide a measure of the range of services available. From the data presented in Table 3 it is clear that as the 1970's drew to a close, nearly all states provided a large number of legislative services. While the variance is never large throughout the 1952-78 period, by 1978 it is only .6 indicating nearly all states cluster near the maxi- mum and of this measure with an average "services" score of 9.6 out of a possible 10. While manageability makes the legislative task easier, responsiveness enables legislators to fulfill 98 supplemental goals. Legislators must be granted the oppor- tunity to respond to constituents and engage in activities, such as advertising, position-taking and credit-claiming, if they desire reelection (or a solid base from which to run for higher office). Legislatures which are responsive provide these opportunities. To assess the degree to which responsiveness affects legislatures, I have included two variables in the professionalism index. First, the length of the two-year session is included in an attempt to tap the extent to which the legislature is viewed as a full-time job. If legislators serve, in part, because they wish to solve societal problems, it is important that they have the oppor— tunity to handle problems as they arise. If the legisla- ture only meets two or three months a year, such opportuni- ties will be diminished. Second, I measure responsiveness by using the number of bilhsintroduced in the house during the busiest year of the two—year session. The "busiest year" was used because many states meet for a "regular" session one year and a shorter or "special" session the next. During the "special" session rather select criteria are often used to determine the kinds of bills which can be introduced. For example, in Indiana, only budget bills are supposed to be introduced during the special session. Since it is impos— sible to take these select criteria into account, only the "busiest" session was used. 99 In nearly all states there is a time limitation on when bills can be introduced and in other states the number of bills which can be introduced by a member is limited.5 If we assume that members will introduce legislation in relation to the extent in which opportunities arise, then the number of bills introduced will be indicative of the responsiveness of the legislature. Bill introductions are important to a legislator's perceptions that he can engage in significant activities of concern to his or her consti- tuency and will be useful toward position-taking and, perhaps, credit-claiming activities as well.6 The data presented iinable 3 show that on the average, legislators have been kept busier in the latter part of the l952~78 period than was the case earlier. The average number of bills introduced in the busiest session of a two-year legislature has increased from a low of 1570 in 1956 to a high of 2430 in 1976. While the increase in the average length of the two-year session has not risen as consistently as the number of bills introduced, legislatures were meeting on the average nearly 200 calendar days during the Seventies as compared to an average 110 days in the Fifties. One factor affecting both of these variables is whether a state meets in regular annual sessions. In 1956, only 1h states had provisions for regular annual sessions. By 1978, Bh states had such provisions. 100 The Distribution of Influence Within the House By influence I am referring to the ability of a member to have an above average say about matters which are presumably of concern to most members. These matters may include member assignments to important committees, bill referral to committees, scheduling of hearings and rulings on procedural matters. In every state house there is a leadership group which typically includes a speaker, major— ity and minority leaders, whips, and committee chairpersons and ranking members. It is this group which has influence in the sense outlined above. Tb obtain a good measure of the distribution of influence, one would have to take into account the number of leadership positions in the house, the procedures used to fill them and the powers associated with each position. That information is not available. Out of necessity I have defined the "distribution of influence within the house" as the ratio of the number of standing committees to the number of members in the house. This measure has obvious drawbacks. In some states, committees are likely to be less important than in other states. In addition, within a state house there is the possibility that some committees will be more influential than others and this is not accounted for by my measure. Finally, there are probably variations in the length of time chairpersons occupy their position depending on the criteria and procedures for selection. 101 On the other hand, this variable does give some indication of the distribution of influence within the house and is worthy of consideration because of its theoret- ical significance. The data for this variable are presented in Table h. The average number of standing committees per house member has declined slowly, but steadily, from the 1950's through the 1970's. On the average, there were one- half as many opportunities for influence via a committee position in 1971 (.161 per member) than in 1952 (.326 per member). In large part, this decline is a function of the efforts by many states to reduce the number of committees rather than a function of increases in the number of members in the house (see Appendix, Table Ae8, for data on this measure for each state). For example, in 1956 the South Dakota house had 51 standing committees for 75 representa- tives and a score of .680. By 1978, there were only 10 standing committees for 70 representatives in the South Dakota House for a score of .143. Similarly, wyoming had 21 standing committees for 56 representatives (a score of .375) in 1956, but only 12 standing committees for 62 representatives (a score of .193) in 1978. As a legislature streamlines its committee system by reducing the number of standing committees, influence within the house may actually increase for many members. Fewer committees would lead to the dominance of a given 102 TABLE 3-14 DISTRIBUTION OF INFLUENCE WITHIN THE STATE HOUSE FOR FORTY- SEVEN STATES: 1952-1978* 1... M... 32322212.. 1952 .326 .17 195A .295 .163 1956 .277 .158 1958 .281 .158 1960 .271 .lh8 1962 .256 .139 196A .2h6 .136 1966 .238 .127 1968 .197 .085 1970 .191 .087 1972 .175 .078 197A .161 .081 1976 .175 .086 1978 .181 .090 1952-78 .225 .101 ‘fFigures are based on the number of standing committees divided by the membership of the state house. 103 policy domain by one committee. Committee members would specialize and their expertise would be in demand from bureaucrats, lobbyists and fellow legislators.7 This line of reasoning implies that the trend toward fewer committees in state legislatures may be considered as an indication of increased professionalism. Member expertise easily trans- lates into an effort to make the legislative process more manageable. However, this measure was excluded from the professionalism index because the distribution of influence within the house may also have an effect on turnover inde- pendent of professionalism. The Importance of State Government By what standards can the "importance of state government" be judged? To my knowledge this question has not been addressed in the literature.8 But, legislators' perceptions of the importance of government are likely to have an impact on their decision to remain in the house (assuming the house is important). I would argue that the degree to which state government responds to external demands is indicative of its importance. However, quanti- fying government responsiveness presents enormous difficulties. Two approaches were considered to Operationalize the importance of state government. The first was to use the size of the state budget. As the demands placed on government typically involve Spending, the size of the 103 policy domain by one committee. Committee members would specialize and their expertise would be in demand from bureaucrats, lobbyists and fellow legislators.7 This line of reasoning implies that the trend toward fewer committees in state legislatures may be considered as an indication of increased professionalism. Member expertise easily trans- lates into an effort to make the legislative process more manageable. However, this measure was excluded from the professionalism index because the distribution of influence within the house may also have an effect on turnover inde- pendent of professionalism. The Importance of State Government By what standards can the "importance of state government" be judged? To my knowledge this question has not been addressed in the literature.8 But, legislators' perceptions of the importance of government are likely to have an impact on their decision to remain in the house (assuming the house is important). I would argue that the degree to which state government responds to external demands is indicative of its importance. However, quanti- fying government responsiveness presents enormous difficulties. Two approaches were considered to operationalize the importance of state government. The first was to use the size of the state budget. As the demands placed on government typically involve spending, the size of the 10h budget presumably reflects responsiveness. However, Spend- ing levels offer no indications as to the quality of government responsiveness. Further, state government expenditures are in part based on programs of the national government; thus the state budget includes money not totally within the state's control. The second approach I have considered is to use per capita tax revenue as a measure of the importance of state government. Assuming the state spends the money it raises, per capita tax revenues may also be thought of as an indi- cation of state government's responsiveness. In turn, legislators are more likely to perceive state government as important when revenues are large than when they are small. Since tax revenues are less likely than the budget to be tied to national government programs, they represent a better measure of state effort. For this reason, "the importance of state government" is Operationalized as the size of per capita tax revenues for the state in a given year. Since legislators will know (or have a good idea) what the tax rates will be for an upcoming year, revenues for the year following an election will be used to account for variations in turnover. For example, when analyzing turnover from the 1969 election, per capita tax revenues for 1965 are used. This measure has obvious weaknesses. First, revenues may differ between states because of variations 109 budget presumably reflects responsiveness. However,'5pend- ing levels offer no indications as to the quality of government responsiveness. Further, state government expenditures are in part based on programs of the national government; thus the state budget includes money not totally within the state's control. The second approach I have considered is to use per capita tax revenue as a measure of the importance of state government. Assuming the state spends the money it raises, per capita tax revenues may also be thought of as an indi- cation of state government's responsiveness. In turn, legislators are more likely to perceive state government as important when revenues are large than when they are small. Since tax revenues are less likely than the budget to be tied to national government programs, they represent a better measure of state effort. For this reason, "the importance of state government" is operationalized as the size of per capita tax revenues for the state in a given year. Since legislators will know (or have a good idea) what the tax rates will be for an upcoming year, revenues for the year following an election will be used to account for variations in turnover. For example, when analyzing turnover from the 1969 election, per capita tax revenues for 1965 are used. This measure has obvious weaknesses. First, revenues may differ between states because of variations 105 in the tax base. Thus, revenues do not necessarily dif- ferentiate states on the basis of tax effort. Second, states have different needs and per capita tax revenues may not show whether a state has properly addressed those needs. Finally, raising (and spending) large amounts of money have negative connotations for some legislators. If the prevailing philosophy in the state house is, "that government which governs best, governs least," it is likely that low per capita tax revenues will be seen as an indication of responsiveness. These shortcomings do not necessarily invalidate this measure, but they should be kept in mind. The data for per capita state tax revenues are presented in Table 5. Given increases in income levels and inflation during the period 1951-79, the trend toward higher revenues over time exhibited in Table 5 is hardly surprising. However, many states have adjusted tax rates throughout the period (until recently at least) which represents an effort to raise more revenues. These efforts account, in part, for the increased levels of revenue. There is considerable variation between states within this period. Delaware consistently has the highest per capita tax revenue, averaging $362 per person for 1951-79. New Hampshire ranks lowest for this period, averaging $137 per person. 106 TABLE 3- 5 PER CAPITA TAX REVENUES FOR FORTY -S EVEN STATES : 1952-1978 Year Mean 32:22:13. 1953 69.7 15.7 1955 73.9 16.6 1957 88.9 20.2 1959 99.9 21.3 1961 107.2 23.8 1963 120.0 27.5 1965 135.6 30.3 1967 162.u 3h.0 1969 201.0 93.8 1971 295.5 51.7 1973 312.0 62.9 1975 368.0 71.8 1977 955.7 88.6 1979 553.6 110.0 107 Independent Variables: The Electoral Environment Incumbents contemplating reelection must face an electoral environment that will encompass district char- acteristics affecting the costs of campaigning and the chances of winning. Included among these characteristics are the competitiveness of the district and thus the like- lihood of voter support and availability of financial resources and volunteers for campaign activities. While I have no data with regard to perceived support, two measures are available from which we can infer support. The degree of state-wide party competition and the inci- dence of redistricting are addressed. Party Competition For the incumbent contemplating reelection, support from his or her party comes from two sources, campaign resources and votes. The importance of party support will depend on the availability of alternative resources and the degree to which voters rely on partisan attachments when voting. However, these factors take on added impor- tance primarily in relation to the competitiveness (or 9 expected competitiveness) of the election. Competitive- ness is defined according to the distribution of the vote among candidates. The more equal the distribution of the vote among candidates the more competitive the election. Measurement of this variable would be relatively simple 108 if district-level data were available. Unfortunately they are not. Given this situation, I have considered two alter- native approaches to measure competition, both of which rely on aggregate party breakdowns in the state. The first approach was developed by Dawson and Robinson (1963, pp. 265-289). Their measure of party competition was based on the percentage of seats held by the majority party in each house of the legislature and the percentage of the general election vote cast for the winning candi- date for governor. These percentages were averaged over a twenty year period. A second approach was used by Ranney (1971, p. 86), who constructed an index by averaging, over time, the per- centage of seats in each house of the legislature held by Democrats, the percentage of the popular vote for the Democratic candidate for governor, and the percentage of all terms for governor, senate and house in which the Democrats were in control. For my purposes, party control is irrelevant. Therefore, I have excluded, from the Ranney index, the per- centage of all terms for governor, senate and house in which the Democrats were in control. Consequently, the degree of party competition in a state is defined as the percentage of seats in the house and senate held by Democrats and the percentage of the popular vote won by 109 the Democratic gubernatorial candidate. These figures were averaged over a five election period]?0 and the absolute value of fifty minus this average was used as the index of party competition. Thus, index scores can range from 0 to 50 (i.e., perfect two-party competition to one-party dominance). The assumption underlying this procedure is that in states which are competitive according to my measure, a higher proportion of legislative districts will be com— petitive than in states which are not competitive. This assumption may be tenuous. The mere fact that a seat is held by a Democrat or Republican says little about the distribution of the vote between candidates in that dis- trict. For this reason, the index of party competition should be viewed with some caution. The data on the index of party competition among the states are presented in Table 6. There is clearly a trend toward two-party balance as one moves from 1952 through 1978, although the magnitude of the change in average competition from one year to the next is not very large. There is also a rather consistent decline in the variance for this measure indicating the trend toward com- petitive balance involves somewhat more than just a few states. As would be expected, Southern states are the least competitive according to this measure (for party competition by state see Appendix A, Table A99). For 110 TABLE 3—6 INDEX OF PARTY COMPETITIONa FOR FORTY-SEVEN STATES: 1952-1978* Year Mean fiéfiififin 1952 22.h 1h.3 195A 21.1 1h.2 1956 20.# 13.9 1958 20.0 1h.1 1960 19.2 13.7 1962 18.1 1h.1 196A 17.5 13.h 1966 16.5 13.0 1968 15.6 11.3 1970 14.6 11.0 1972 13.8 9.3 197h 13.7 9.9 1976 1h.2 9.1 1978 1h.0 9.7 1952-78 16.8 12.1 *Source: The Book of the States, Volumes 9-23. (a) Figures are based on the following procedure: for each year the percentage of house and senate seats in a state which are Democratic are added to the percentage of the popular vote received by the Democratic gubernatorial candidate. This total is divided by three. This score is then averaged over five election years and the absolute value of SO-average is used. 111 example. Alabama has an average index score of #5 (with 50 being the least competitive) for 1952-78, Georgia, an average of 39, and Mississippi an average of hh. For the entire period, Pennsylvania has the most competitive two- party balance with an average score of 3.5. Reapportionment The optimal approach in defining reapportionment would involve measurement of the actual change in district boundaries in terms of the number and types of voters affected. Since that is not feasible for this study, redis- tricting is operationally defined as the occurrence of redistricting in a state during the two~year period immedi- ately preceding the election in question. A state is coded 1 if redistricting took place and 0 if it did not. The data on redistricting are presented in Table 7. As discussed earlier, a high incidence of reapportionment occurred during the mid-Sixties. While a number of states did not reapportion during the Fifties, many states did so several times during the 1962-66 period. Following the 1970 census, every state had reapportioned by l97h. The 0pportunity_Structure An incumbent's "opportunity structure" can be defined on the basis of whether or not there is a higher office for which he or she legally qualifies, and the degree of competition which is likely to occur while 111 example. Alabama has an average index score of 45 (with 50 being the least competitive) for 1952-78, Georgia, an average of 39, and Mississippi an average of hh. For the entire period, Pennsylvania has the most competitive two— party balance with an average score of 3.5. Reapportionment The optimal approach in defining reapportionment would involve measurement of the actual change in district boundaries in terms of the number and types of voters affected. Since that is not feasible for this study, redis- tricting is operationally defined as the occurrence of redistricting in a state during the two-year period immedi- ately preceding the election in question. A state is coded 1 if redistricting took place and 0 if it did not. The data on redistricting are presented in Table 7. As discussed earlier, a high incidence of reapportionment occurred during the mid-Sixties. While a number of states did not reapportion during the Fifties, many states did so several times during the 1962-66 period. Following the 1970 census, every state had reapportioned by 197A. The Opportunity_Structure An incumbent's "opportunity structure" can be defined on the basis of Whether or not there is a higher office for which he or she legally qualifies, and the degree of competition which is likely to occur while 112 TABLE 3-7 REAPPORTIONMENT IN FORTYFSEVEN STATE LEGISLATURES: 1952-1978* Year Mean Number of States Reapportioninga 1952 .3h 16 195A .17 8 1956 .13 6 1958 .08 a 1960 .oh 2 1962 .55 26 196h .h7 22 1966 .68 32 1968 .19 9 1970 .08 u 1972 .81 38 197k .38 18 1976 .02 1 1978 - 0 *Source: The Book of the States, Volumes 9-23. (a) A state is counted as having reapportioned if reapportion- ment occurred during the two~year period preceding the election year indicated in the table. 113 pursuing that office. An office is "higher" if an incum- bent values that office more than the one currently held because of salary, security, prestige, or some other characteristic. An "open opportunity structure" is pres- ent when an incumbent is legally qualified for an office and competition for it is likely to be non-existent or slight. As competition increases, the opportunity struc- ture "closes." Schlesinger (1966, pp. 37-56) develops a measure of the number of opportunities for office in a state. Schlesinger's measure takes into account the number of offices, the length of term, restrictions on succession and turnover in personnel. This would be a useful measure of opportunities for my study. However, I have no data on personnel turnover for offices which can be considered higher from the perspective of the state house. This lack of data is regrettable because personnel turnover provides at least some indication of competition. If incumbents holding higher offices are continually running for reelec- tion, then competition is presumably higher than when the office is vacated frequently. This is due to the likeli- hood that incumbents have greater name recognition than their challengers and have demonstrated an ability to win in past elections. With this shortcoming in mind, I have defined the "opportunity structure” in a state as the number of state llh senate seats up for election per house member plus the number of U.S. House seats per house member. Since the number of other "higher" offices up for election or appointment is not likely to contribute much variance to my measure, they were excluded.11 These offices include United States senator and statewide offices, such as governor, lieutenant governor, state attorney general, etc. Other offices, such as judge, district attorney, county assessor, etc. were excluded because of measurement dif- ficulties. The assessor of Chicago is undoubtedly a valued office, but the assessor's office for White County in Indiana probably rates quite low in value when compared to state representatives. There is simple no convenient way to incorporate these differences into my measure. A similar argument can be raised in relation to the value of a senate seat vis-a-vis a house seat in some states. Compensation for senators and representatives is the same, and in many states (16 as of 1978) the length of senate and house terms are equal. On the other hand, the size of the membership is smaller in the senate than in the house and this is likely to lead to greater prestige and influence for senators as compared to representatives. In addition, with larger constituencies, senators are in a better position than representatives to build a base of support from which to run for congress or statewide office. 11h senate seats up for election per house member plus the number of U.S. House seats per house member. Since the number of other "higher" offices up for election or appointment is not likely to contribute much variance to my measure, they were excluded.11 These offices include United States senator and statewide offices, such as governor, lieutenant governor, state attorney general, etc. Other offices, such as judge, district attorney, county assessor, etc. were excluded because of measurement dif- ficulties. The assessor of Chicago is undoubtedly a valued office, but the assessor's office for White County in Indiana probably rates quite low in value when compared to state representatives. There is simple no convenient way to incorporate these differences into my measure. A similar argument can be raised in relation to the value of a senate seat vis-a-vis a house seat in some states. Compensation for senators and representatives is the same, and in many states (16 as of 1978) the length of senate and house terms are equal. On the other hand, the size of the membership is smaller in the senate than in the house and this is likely to lead to greater prestige and influence for senators as compared to representatives. In addition, with larger constituencies, senators are in a better position than representatives to build a base of support from which to run for congress or statewide office. 115 While my measure of the opportunity structure in a state does not directly incorporate the level of com- petition for higher office, the number of offices per representative is indicative of the potential number of colleagues an incumbent could face in seeking a higher office. Assuming that representatives' stiffest competi- tion will come from other legislators (aside from facing an incumbent) the more crowded the field the greater the competition. In this sense only is my measure indicative of the proportion of "open" opportunities in a state. Table 8 presents the data on the proportion of Opportunities among the states for the period 1952-78. Although the overall average of this measure has remained rather consistent during the period studied, there is con— siderable variation in the proportion of opportunities for higher office between states and within states over time (see Appendix A, Table A—ll for data on opportunities for higher office by state). For example, the score for Michigan on this measure is .17 in 1968, .52 in 1970, .17 in 1972 and .52 in 197A. Prior to 1968, all state senators in Michigan served two-year terms and, on the average, for every two representatives there was a senate or congres- sional seat up for election. In 1968, Michigan senators began serving four-year terms with all senators up for election every four years. The subsequent reduction in potential opportunities for higher office for Michigan representatives every other election was rather dramatic. 116 TABLE 3-8 STATE REPRESENTATIVES' OPPORTUNITIES FOR HIGHER OFFICE(a) FORTY—SEVEN STATES: 1952-1978* Year Mean 3:312:13. 1952 .336 .163 195h .331 .155 1956 .305 .160 1958 .29h .16h 1960 .314 .151 1962 .320 .158 1964 .299 .1A7 1966 .387 .158 1968 .296 .152 1970 .372 .1u1 1972 .319 .153 197A .326 .162 1976 .315 .161 1978 .326 .166 1952-73 .326 .159 *Source: The Book of the States: Volumes 9-23. (a) Figures are based on the number of state senate and con- gressional seats up for election during the year indicated divided by the membership of the state house. 117 On the other hand, the number of opportunities per house member varies between states as well. Throughout the 1952-78 period, New Hampshire presents each of its representatives with only .07 opportunities for the state senate and congress combined. In contrast, California averages .7# opportunities for the same period. There are clearly large differences between the states and within states over time on this variable. Summggx For the period 1952-78, turnover among state houses has declined in a relatively consistent pattern from over forty percent during the 1950's to under thirty percent by the end of the 1970's. This pattern is broken in l96h— 66 and 1972 when turnover levels again rise to nearly forty percent. Presumably, these anomalies are due to reapportionment. Throughout the period studied there are significant variations in turnover between states and within states over time. It is not uncommon to find turn- over in the 10-20 percent range in some states and over 60 percent in others. A professionalism index, the number of standing committees per house member and per capita tax revenues are used to measure several aSpects of the institutional environment. Compensation, legislative expenditures, length of the session, number of bill introductions and the number of services provided in the house were 118 variables used to construct the professionalism index. Data for these measures show there is considerable varia— tion between states, although, over time, nearly all state legislatures have become more "professionalized." The number of standing committees per house member has, on the average, declined throughout the period 1952- 78. This decline is largely due to efforts to reduce the number of standing committees in the state houses. Per capita tax revenues increased considerably from 1952 through 1978. As is the case for the other variables included in this study, tax revenues vary between states and within states over time. The degree of party competition in the state and the incidence of reapportionment prior to an election are two variables I have used to assess the impact of the electoral environment on turnover in the house. Data show that the degree of party competition has, on the average, increased somewhat during the period 1952-78. This increase appears to be a general phenomenon. Reapportion- ment occurs most often in the mid-Sixties and early- Seventies as might be expected. The "opportunity structure" in a state is defined on the basis of the number of senate and U.S. congressional seats up for election per house member. This measure fluctuates because of the different lengths of the senate term in the states and because some state senates have staggered terms while others do not. 119 Several weaknesses are apparent from the data. First, I have been unable to partition turnover into the percentage of house members voluntarily withdrawing from politics, running for higher office or suffering defeat in a reelection effort. Hence, it is impossible to deter- mine the effects of the independent variables with regard to different types of turnover. Second, I have been unable to develOp a measure of the importance of the state house vis-a-vis other institutions of government. Thus, a variable of theoretical significance has been omitted. Finally, several of the variables used in this study have obvious weaknesses. For example, some tenuous assumptions must be made in using per capita tax revenues to measure the "importance of state government." In addition, a statewide measure of party competition is used to measure what is conceptually district-level competition. Despite these difficulties, until more data can be gathered and more sophisticated measures developed, the data included in this stage of the research provides for a good preliminary test of the theory. W The theory discussed in Chapter 2 posits that turn- over in the state legislature is a function of the institu- tional environment, the electoral environment and opportunities for higher office. To test this proposition, 120 I use multi-variate regression analysis which allows for the simultaneous evaluation of many of the hypotheses presented in Chapter 2. Several alternatives to the simple additive model discussed below were considered. One possibility was that turnover was a function of the interaction of several of the independent variables studied. In particular, I con- sidered the interaction between the measures of profession- alism and influence within the house. A second alternative to the model below was to Specify a curvilinear relations ship between turnover and per capita tax revenues. Neither alternative was supported by analysis. The following equation was estimated for the elections 1952-78: TH = a.+ b PI + b IH + b IG + buPC + b l 2 3 R + b6OS + e 5 where TH is the percentage of new members in the house PI is the state's professionalism index score IH represents influence in the house as measured by the number of standing committees per member IG represents the importance of state government as measured by per capita tax revenues PC is the state's party competition index score R is reapportionment (1 if the state reapportioned, 0 if not) 121 OS represents the state's opportunity structure as measured by the number of state senate and con- gressional seats up for election per house member. "a" is the intercept, the "b's" regression coefficients and "e" the error term. For each state house, turnover is hypothesized to be a function of the degree of profession- alism, the number of opportunities per member for influ- ence within the house, the importance of state government, the degree of party competition, whether or not the state reapportioned, and the number of opportunities per member for higher office. The first three variables are aspects of the institutional environment and all are expected to be inversely related to turnover. Party competition and reapportionment are part of the electoral environment and are exPected to be directly related to turnover. The proportion of opportunities for higher office per member is also expected to be directly related to turnover. The estimates for the equation are presented in Table 9. The hypothesis is only partially supported. The variance explained (R2) by the model indicates a reasonably good fit in several of the years studied (.56 in 1952, .38 in 1956, .ho in 196A, .39 in 1970, .h6 in 1976). However, the model predicts less well in other years, most notice- ably in 1962 and most of the 1970's. The level of professionalism in the state house is the strongest predictor of turnover in relation to the 1112 000:9 hon usedwfldflsoa one oHo>oH concedhfinmfim .whmd ad toneduhonndoh ceases o: .wbmd ad concavhoandoh codes A Mano 3 .ohdoh honao dd nausea 0560 one new Eddies! ad unadnH50A .anGOHUGH and ceases a: flofii check as» wndhsv unedHOOHo on we: dhfiduudoofiz 6:6 Udddhhdx .afldndd< Any .vousdoud ed wooden no oohobflnfl oufiauo one nosoahsh deedaoaha Add Row condo IGOdadsve .oonaOde:UHe we Hobod no. a ended as ceasefitflH t vflOdOdmhooo ceduooHMOH vaudvhdufldam AAV .aGOdOdMMooo downscauem Adv es mo. mo. and. II 61: na.| noa.n ad. moo. on.s mmo.- n~.n mn.~n em.- ones as sea. ea. «so. II all an.1 ass. no. soc. ur.n smmn.1 sn.u tan.su wo.~n when or. ma. mu. com. om. rm.s wa.l ~ma.u no. moo. so.n sno.1 m~.n ow.nn oo.«u seen as ea. moo. moo. as. s~.n wa.\ ara.n 0a.: one.1 so. sno. n~.u on.~u o~.bn «has or tan. so.u muo.n no.1 w~.nn e~.| os~.u as. was. no.1 smo.1 nn.| tom.nu as.s~ ones as can. an. are. an. tom.s oo.l noo.u mo. moo. H«.u ms~.n ~n.| tan.nu so.m~ some or an. «o.1 moo.u as. series su...wna.n so. sue. so. aso.. na.n ~o.ut ou.~n moms as toe. nu. are. me. tnu.na no.1.omo.: Ho. moo. mo. Ame. os.u tr~.wu Ho.am some or as. «a. «we. on. 53.: mu.1.os~.n no. age. ~o.u «Ho.1 w~.s em.~u ~r.s~ «was as an. we. uno. no. s~.s ma.u.aoa.u mo. use. as.1 aca.n as.n tw>.nn «H.nn some es tam. ~H.1 amo.n -.u nH.oan ma.1 asa.n ms. was. aa.u nmo.u mn.n tun.na on.sn ohms as ten. seco.u Hoo.u nu. to~.m mo...owo.n an. same. no.1 mno.u nn.u tnm.en ss.n~ wnma ms tau. «6.: sao.u_ so.n ~s.au Ha...noa.u we. was. nu. mos. ~s.n snn.nu nm.m~ snag as ton. on.1 amen.u ao.n men.1 so.\ oao.t as. was. an. twsu. or.u the.nn ~o.wn «nan unease m on m m on m e um m n so a m an m an am < use» we a «a we a on eH ‘wmwl Hm cognates msumnms .mem=o= msgsm zmsemnssmoh 2H sa>ozmps so mszmo.u :~.a tan.sun mu. tsn~.: NH. one. so. are. w:.u *ws.nn ma.nn «enemas «a A.rm a 9.0m a p.sm a n.sm a n.»w a an.om up 4 scenes mo m on eH mH Hm nauseous msumsma cz< osasoaa .meumnmu .mGOHflfim flflfi flea wmmbbl flfiflfiw ZH>HwIHBmOh 2H ¢H>OZMDH ho WHZOZNDH ho WHZasnszonohe62 to 2 cos sea cod cod Hon see no sea Hes cc” sea Hes oca ooH ooa teases ones es s» as as we we am eh m: we s» mm mm on or :a: use :o«¢004002 zoom are 0a m m n mu m e mm mm as m a s as ea conceded one o:: seducedooz Meow as m as n ea m m o N as m N m n a n ooaeeo hoses: MOO :3: was a as m w as w e s as a ma a o NH a message; HOHOOOOHQ o>s¢2 z za>< and mum :N. mu. QM. we. so. so. we. on. mm. ow. aw. Amwmm s zsssccotohdoz do 2 sea OOH ccs cod 00a OOH cos cod can OCH OCH cos sou ooH ooH acne. me» pm as as so am so as :m or or on up so on no :a: t:6 cowDOOHooz soon was am a r an an as as on :m as em as ms nu as countess ou< as: concocdooz zoom sea a s m m m r cs 6 a Ca me a s es s ceases poses: no.“ :3: saw me a s o as on as as am on an we ms as as assesses HOOOQOOHE O>soq z sa>< mm. ON. an. ms. cm. we. we. so. we. es. mm. mm. rm. lemma s <2H,_. I . I.‘ 'l :. “1 _. :- I 21' .' x . 1 I. l“ . . - . i.. . '.-'l'! .I -_ I ;_.'_._l- . _ _| .' . . '. I‘ . . ' 1 .' .i _, _.. ,.. . _ ., . . " -é.:.. . - . I. i . . - - -_ . ‘ .' 'I . ...' : I ' ' "I "l' .:'n' "'- .‘.' -'- h "1 -' . I 'l. I I . r .. - -. u: - ' , '. I r: 31"1'ID'L ” I i ' “I‘ll. '. J: ' f ""3 ' :- - I ' H! "J'lm'i "11:14;th 5' «w H: ”i -;i:‘ tum 154 create more visibility for more incumbents which is expected to favor them at election time. The specific hypotheses to be tested are as follows: H1: Among state legislatures, the proportion of state legislators withdrawing from office voluntarily is inversely related to the degree to which material and psychological benefits are provided in the legislature. Among the state legislatures, the proportion of incumbents who are defeated for reelec- tion is inversely related to the degree of professionalism in the state legislature. Among state legislatures, the proportion of incumbents who are defeated for reelection in the primary election is directly related to the degree to which the legislature pro- vides opportunities for incumbent advantage. The data in Table 4-3 provide a comparison of the level of professionalism between Indiana and Michigan for the period 1952-78. Throughout the period Indiana's ranking relative to other states remains relatively constant in the 20 to 23 percent range. The Michigan legislature is not highly professionalized in comparison to other states (at least on a consistent basis) until after the 1964 election. Michigan's ranking reaches a low of 43 (out of 47) in 1960. Beginning in 1966 the professionalism index score for the Michigan legislature ranks in the top 7 for the remainder of the period under study. At first glance these data do not support hypoth— esis 1. The proportion of Michigan incumbents withdrawing 155 .sxnsunoz one .fifidiam .dxead< mndbflfloko mosses h: :o woman as Mflfixfidh 9:8: n n a a n h 5 NW wN n: On :N a m #3 ehoow HondH ca...“ ma..." :nJ n34 «n4 mmJ mm. an? an... 5.: mm... an... S4 worn Snedsaoanoououm fl. Wh. +2.. NM; OF. ww— ww. 3W. Nw. OW. mm. W“. :1“. NW% “(ea EUEOH: nN HN ON wH ON NN Na MN 9H MN EN MN ON wH aha—ham okcom Koan mun-I :Ool hHol HOol AN.) Nmol m0... wmol $0.] NNol fin... wN-l wH-l NO. EufifidflOflDDOMOHfl 9N. ON. .1: NM. Oh. ww. @W. :W. Nw. 0w. wm. WW. #n. Nma 5a <32 wblmmmd .mEHu£m«dn chm nausea oEon .ucfiucsoh we emsaoazs 2: no 9: ca :3 c1; c3 cod Hon 9: ilc. ccm ccH Hoa co~ .2309 cmaa ow we :m mm mm mm mm :s we we so do nm pm on :42 oz: MEH::5z wmcfir ac fl an o a m n n o a as he N N s o a a :ofiuooam Houococ ca concede: M:M::dfl onoFr as E em 5 N n n as N n o we as m m s OH ca sunsets one :a soococo: wcacclz whose as i Gama CHH OHM OHM OHH Can Can 0“” 34H OHH CHH Odd 0AA OCH OCH emso: :« mo>fiumacomohaez ac z mend mm vs mm as wm :w so mo 05 00 mm mm aw ow om :owuooaoe: Lou mcaz::= nfizmhsDEoz omflo: 0:4 as R 2 Ease EN. 0N. :N. NH. ON. ww. we. :6. NW. CW. mm. em. :0. Nnmd Z<fl> Zwusu=000hnoz Ho 2 case we be mm No as Nm cm ms :c on we we Na as on :oeoooacca toe u:e:::= afiznhosEoz owns: ezu do i Z ~_m—>< mh. wfi. 3N. NW. on ww. 00. :w. Nw. 00. mm. on. :m. Nna = HF< wh. W5. :F. 6N. ON. ww— WW. 6:0. Nw. Ow. wh— wm. NJM. Nnmfi ~20; 53:on oci 03 cos ooa oos ooa ooH ooa ooa ooa ooa 00H 03 ooa cos noisescmmo‘ao: co 2 co m on: 00 H coH cos 9:. cos com 03 03 03 cos” cos 03 cos sauce was. an om C. so mm no so an on em mm mm we on no a; «:3 :owuuwHemz xeem wmm an N. w mm 5 3 as on :N i on n: n: ma 1 toughen 92 so: floauoeaoom xwwm :3 s a w m w m 0H m. m 0H 3 a. m 0H o 833 3:3: ROM :3.— 3m 3 m a 0 ma 3 oa 3 pm ma 3. we ma as we 33:?” awhoauodm was: Z mfl>< wh. a. :N. .mNN... ON”. ww. wa. ddw. Nw. 0w. mm. WW. 3W. Nnm ”Em; <2HEeoA fi 2 mm>< we. om. an; we. on; we. we. so. we. ow. on. em. an. Nnmw m mzaod fl 2 ma>< we. on. an. mm. om“ we. so. so. me. ow. mm. on. .mn. «we» 55» 35.823: wboanmmma «rez4o rm mm>H8Hu£UHHn ohm maaucu oEom .mnflUESOH mo cm:Mo0a* is :m N: .3 om Nm mm mm an an om mm 3 mm we we mfioeansooz eo 2 cos coH o3 cos cos ooH c3 c5 «5, oi. oi cos no ooa oos :6qu :mm as om. so me E. ms :w 5 3 5 Nb aw i we ow SE 2;. :oquOHwoz Hem uflflfiddx fl 8 m o s 0H m a m m «N m a m N. 3 m 328%.: 92 2:. EOMuooaeod how wzflflnfiz i on n me o m N w o m NH : N m m n n 3:3 3:5: How mcaczfiz R mo S N m me me as 3 m S a NN n me we a 3333 3.33me mnfl>doq R 2 5:2 m? me. it Nu. o? wo. so. :0. Ne. ow. mm. on. em. Nnoa 55; 22056:: woe mm we we cm on em mm m” Nm :m mm m: 3 an an numpooswa no 2 ooH ooH Hos ooH Hos sea 03 03 cos ooH ooa ooH ooa ooH 03 $3.; o mom. 3. am am wk. .1 mm 3 so on Na Nw ww oo Q. on 53 on: ‘ Cesaooaooz LOH Mfiflficdx v 5 3 e m N m: o 5 mm me me m m o NH ma 035800 02 9:; :owuooaww: how mcdcfizz fl 3 m a m we N a o o m N N a N me n 32.5 honed. ROM anc:5z 3 an n s N N a a N n AN 2. 3 N N o m 333$ Hmpoaowsm M:fl>doq R 2 as: w? o? i: N? of we. so. so. No. ow. mm. so. a. Nome 2m» $588sz mumauNan 355E E WB>HE aflonaaofld anon mm Amsmwv mean Assess seems 1:.oae o.oe An.oav o.me en se 2 inflame stoH ‘ Anebes swan opaeaeuao usages .49 MD uohflbdvflomxm :wdanauo Asnsmv eased Asonev comm oousavnuo scans“: on» ha ochflwfivfionxfl anddnado novdvfivfldo HH< Ae.auv e.mn Am.mov e.oe hands o.oa¢u«sa¢o muuuuas use no sasnonum hands A».HHV m.ne nae.mov w.mw oaauusnoo maesau: one you oao> fledaoedfl Hduenoc 0:8 fled¢00Hfi Hdhenew 0:8 nH ESE #02 via o»:o&Eflo:H :H Guam 00CODE=OQH hoflenfiflodH wbmfl .mfi>HF<92fimmflmflfl ho flwbbl Z fl¢< HIM fidfldfi 186 importance of money in the campaign appears to center on not how much is spent, but on how much is spent in relation to one's opposition. Analysis Based on the theoretical discussion in Chapter 2 (p. 'WD), a candidate's proportion of the vote is a func- tion of campaign exPenditures. This premise is based on the assumption that party loyalties have declined over time and many voters have turned to other sources for vot- ing cues, such as the media. Mbney can buy these sources. Hypothesis 7 was thus derived as: H7: In state legislative elections, the preportion of the vote for a candidate is directly related to their ability to raise money for a campaign. My intention is to demonstrate that campaign expenditures account for much of the variation in a candi- date's vote, but in the context of party voting, which may be an important factor. Tb test hypothesis 7, I use multivariate regression analysis employing the ordinary least squares procedure. Equation 1.1 was estimated for the 1978 election in Michigan. WV=a+bWE+bLE+b I+e (1.1) l 2 3WPS + th + b 5 where WV is the winner's percentage of the two party vote WE is the winner's campaign expenditures in thousands of dollars LE is the laser's campaign expenditures in thousands of dollars 187 WPS is the strength of the winner's party in the district P is the winner's party (1 if Democrat, 0 if Republican I is the winner's incumbent status (1 if the winner is an incumbent, 0 if not an incumbent) a is the intercept, the b's are regression coefficients and e is the error term. Both winners' expenditures and losers' expenditures are included because candidates are likely to spend money in reaction to their opponents' level of spending (see Jacobson, 1978). All other vari- ables are included as controls. The data are presented in Table 5-2. The estimates for equation 1.1 provide a good fit with the data as shown by an R2 of .7h. From these data it is evident that campaign expenditures do have a sub- stantial impact on the vote and hypothesis 7 is supported. However, it is equally clear that it is the laser's expenditure level which is of primary importance. The standardized regression coefficient for losers' expendi- tures is -.53 indicating it has far more impact relative to the other variables in the model than winners' expendi- tures with a standardized regression coefficient of .13. Apparently candidates do spend as a reaction to their opponents, but win because their opponents cannot raise sufficient funds. The winning candidate's party strength is of equal importance to his or her vote relative to all other vari- ables as is loser's expenditures. The standardized 188 TABLE 5-2 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS - Standardized Regr9551on . Coefficient Regr9351on Coefficient WV = a 36.0h8 b1WE 0.173 .13 bZLE -o.668* _.53 R2 = .7h* b3WPS 0°529* .52 th 1.462 .07 b5I 1.h99 .06 (N = 93) *Indicates significance at the .05 level. 189 regression coefficient for party strength is .52 and winners can be expected to gain .5 percent of the vote for every 1 percent increase in their party's strength. Neither the winning candidate's party or incumbent status were significant. The failure of incumbency to make a difference was not expected. Incumbents may have an advantage over their challengers but so do other candi— dates who can raise enough money to offset their opponent's spending and who have a favorable party balance in the district. Tb explore this possibility further, equation 1.1 was estimated using elections with no incumbents running. The findings appear in Table 5-3. Incumbent status has been dropped from these equations and in Table 5-3 all variables refer to the incumbent instead of the winner (except CE which equals challenger's spending). The results in Table 5-3 show a substantial differ- ence between models when incumbency is taken into account. For elections with no incumbents in the race, campaign expenditures by the winning candidate were significant at the .05 level whereas they were not when all races were considered. The relative importance of expenditures by winners is quite high with a standardized regression coef- ficient of .34 (compared to -.h8 for losing candidates). In addition, party strength in the district has a greater impact on the vote when non-incumbent races are 190 TABLE 5-3 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS CONTROLLING FOR WHETHER AN INCUMBENT WAS RUNNING A. Elections With Regression gzangzzgzzed No Incumbent Coefficient gr . Coefficient wv = a 17.954 b1WE 0.455* .34 bZLE —o.564* —.48 R2 = .83* bBWPS 0.735* .73 th 1.168 .05 (N = 26) B. Elections With Regression Szanzzzgified An Incumbent Coefficient gr . . Coeff1C1ent IV = a 3701‘7’4 bllE -o.ooh —.oo b2CE -o.631* -.uz R2 = .67* bBIPS 0.557* .56 thP 1.003 .05 (N a 67) *Indicates significance at the .05 level. 191 considered, with a regression coefficient of .735 compared to .529 when all races were included. Party strength in the district is more important relative to the other vari— ables in the equation with a standardized coefficient of .73. In elections where neither candidate is an incumbent, campaign expenditures by both candidates are important, but the relative strength of the two parties in the dis- trict will have a greater effect on the outcome of the election. Table 5-3 shows that the parameters estimated by the equation when elections with an incumbent running are considered, approximate the parameters of the original results in Table 5-2. The estimates for incumbents are comparable to those for all winning candidates since only one incumbent was defeated. As in the original estima- tion, incumbent expenditures are not significant. The variance explained is less (R2 = .67) than that explained by the equation with all candidates included (R? = .7h). Although the impact of challengers' campaign expenditures relative to the other variables included in the equation is dampened somewhat compared to the original estimation (with a standardized regression coefficient of -..lt2 com- pared to -. 53 in the original equation), all other coefficients are similar. These data indicate that incumbent spending has much less of an impact on the vote than spending by the 192 challenger. Party strength in the district is still the most important variable relative to the other variables in the equation when considering races with incumbents running (with a standardized regression coefficient of .56). Hewever, if challengers could raise enough money they might overcome disadvantages in party support in the district.3 One drawback of the analysis thus far is that the ordinary least squares procedure assumes all relationships are in one direction. As Jacobson (1978) found in analyz- ing congressional elections there is good reason to assume that spending not only produces votes, but that the expected vote will affect spending. Assuming the expected vote approximates the actual vote, the relationship between the vote and spending may be reciprocal. If so, the ordinary least squares estimates are biased and inconsis- tent because the independent variables (which are related to the dependent variable in a reciprocal manner) are correlated with the error term (see Johnston, 1972, p. 3&3). The common procedure used to avoid this problem is two-stage least squares. This procedure involves the specification of all variables included in the reciprocal relationship. Since it is not clear that there is a reciprocal relationship between winning candidates' spend- ing and the expected vote (due to less reliable predictions of the vote when no incumbent is running) I analyzed only those elections with incumbents running.)4 193 For elections with incumbents running, the relevant model is: CE = f YRS, IP, IPS, CPR, EV) 2.1 IE = f YRS, IP, IPS, IPR, EV, CE) 2.2 IV a f IE, CE, IPS, IP) 2.3 EV = IV 2.1+ where CE is the challenger's campaign expenditures in thousands of dollars IE is the incumbent's campaign expenditures in thousands of dollars YRS is the number of consecutive years the incumbent has been in the house IP is the incumbent's party (1 if Democrat, 0 if Republican) IPS is the strength of the incumbent's party in the district CPR is 1 if the challenger was opposed in the primary, 0 otherwise , IPR is 1 if the incumbent was opposed in the primary, 0 otherwise IV is the incumbent's percentage of the two—party vote EV is the incumbent's expected vote. A challenger's campaign expenditures are hypothe- sized to be a function of several factors related to the possibility of winning. The decision by contributors to provide money for a challenger's campaign will be based in part on how strong his or her party is in the district. This decision is also expected to be a function of the incumbent's seniority, a factor which could reduce the feasibility of winning. In addition, since challengers are more likely than incumbents to suffer from a lack of name recognition, if they are opposed in the primary they will presumably gain more exposure and, as a result, campaign contributions. Primary campaigns against opposition also _ j? — i ' E _ . 19h demonstrates an ability to win an election which should encourage investors. Finally, the challenger's chances in the election are determined by his or her expected vote in the election. An incumbent's campaign expenditures must be approached from a different perspective. Jacobson (1978, p. #75), analyzing U.S. Congressional elections observes: Incumbents, then, acquire funds only in proportion to the felt necessity to do so. And they can usually get all they need. The variables that determine incumbent spending, therefore, indicate how much the candidate is likely to need. And this, in turn, is primarily a function of the strength of the challenger. Thus a state legislative incumbent's ability to get all the funds he or she needs may be indicated by how little his or her opponents spend (see Table 5-1, p. 185; in 1978 oppo- nents spent an average of only $3,96h) and the strength of the incumbent's party in the district (see Table 5-1, p. 185; in 1978 the average party strength of the incum- bent's party was 60.6 percent). Incumbents will spend money primarily in response to the strength of challengers. Thus, challengers' expenditures are hypothesized as an important determinant of incumbents' expenditures. The threat posed by the challenger will be determined by the seniority of the incumbent, the incumbent's party in the district, whether the incumbent was opposed in the primary election and the expected vote in the general election. Therefore, 195 incumbents' campaign expenditures is hypothesized to be a function of the variables listed in equation 2.2. The two-stage least squares procedure removes the component of the independent variables associated with the error term by obtaining estimates for these variables based on all remaining independent variables in the model. These estimates were derived from the following equations: EstCE = a + b YRS + b IPR + b CPR + thPS + 1 2 3 b51P + e (3.1) EstIE = a + blYRS + bZIPR + bBCPR + thPS + bSIP + e (3.2) The estimated variables (EstCE and EstIE) are then substi- tuted into the original equation in the second stage: IV = a + blEstIE + szstCE + bBIPS + buIP + e (3.3) As Johnston (1972, pp. 380-384) demonstrates, the two-stage least squares estimators are consistent and bias is reduced. The results for equations 3.1, 3.2 and 3.3 are reported in Table S-h. The parameters for challengers' expenditures esti- mated in the first stage equation show that whether a challenger faced opposition in the primary election had the greatest impact on expenditures relative to the other vari- ables included in the model. This is evident from a standardized regression coefficient of .hl compared to -.26 for the incumbent's party strength in the district, 196 TABLE S-h DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS USING TWO STAGE LEAST SQUARES Regression Standardized Coefficient Regression Coefficient First-Stage ESTCE: a lh.26 Equations blYRS -0.05h ..04 (3.1) bzIPR -0.678 -.05 R2=.29* b3CPR 6.983* .hl thPS -o.177* -.26 bSIP -0.521 -.0h (N=67) ESTIE: a 20.395 blYRS -0.319* -.26 (3.2) bzIPR 0.459 .ou R2=.29* bBCPR h.892* .30 thPS —o.177* —.27 b51P 2.2h9 .17 (N=67) Second-Stagec IV: a 36.129 Equation blESTIE 0.078 .03 szSTCE -o.660 —.24 R2=.5h* bBIPS 0.570* .57 bAIP 0.821 .oh *Indicates significance at the .05 level. c2 R and significance were obtained from the second-stage estimates. 197 the only other variable that was significant in equation 3.1. Challengers who faced primary election opposition could be expected to spend nearly $7,000 more during the entire campaign than challengers with no primary election opposition. The number of years seniority, whether the incumbent faced primary election opposition and party were not significant. Challengers are able to raise money depending on their primary election and the degree of party competition in the district. The data in Table 5-h indicate incumbents' expendi- tures (equation 3.2) are a function of three variables. If the challenger had primary election competition, incum- bents could be expected to spend nearly $5,000 more than if there was no primary election opposition to the chal- lenger. Presumably challengers opposed in the primary receive greater name exposure and raise more money thus constituting a greater threat to incumbents. Equally important to incumbent spending based on the standardized regression coefficients were the senior- ity of the incumbent and his or her party strength (coefficients are -.26 and -.27 respectively). Incumbents are expected to spend less when they have more seniority and greater party strength in the district. This is not surprising as they are less likely to face a stiff chal- lenge under these conditions and, thus, there will be less need to spend during the campaign. 198 The estimated parameters for the second-stage equation (3.3) are surprising in that neither candidates' estimated campaign expenditures are significant. However, one reason for this may be that the estimated spending variables are highly correlated at .82. This indicates a problem of multicollinearity which in this instance has resulted in these variables cancelling the effects of each other. To avoid this problem, the second stage equation was run again dropping estimated incumbent expenditures. The results of this procedure appear in Table 5-5. TABLE 5-5 DETERMINANTS OF THE VOTE IN 1978 MICHIGAN HOUSE ELECTIONS USING TWO STAGE LEAST SQUARES: OMITTING ESTIMATED INCUMBENT EXPENDITURES Regression Standardized Coefficient Regression Coefficient IV = a 36.786 2 b20E -o.597* -.22 R = .5A* bBIPS 0.565* .56 thP 1.036 .05 (N = 67) *Indicates significance at the .05 level. The estimated parameters from Table 5-5 show that challengers' expenditures and incumbents' party strength 199 are significant while party is not. Incumbents' party strength has the greatest impact on incumbents' vote relative to the other variables in the equation (with a standardized regression coefficient of .56 compared to -.22 for expenditures and .05 for incumbents party). Fer every 310,000 the challenger spends, the incumbent can be expected to lose nearly 6 percent of the vote as esti- mated by the regression coefficient of -.597. In the 1978 state house elections in Michigan where expenditures weren't often above $20,000 and incumbents won with nearly 70 percent of the vote, one should not be surprised that only 1 incumbent was defeated. Conclusions The hypothesis that a candidate's share of the two party vote in an election is directly related to his or her ability to raise campaign funds was supported, but with some important qualifications. When there is no incumbent in the race, spending by both candidates has a significant impact and of nearly equal magnitude. However, when equations are estimated for elections with an incumbent running, only the challenger's spending is significant in predicting the incumbent's vote. Incumbent spending is not significant in any of the equations estimated. This is evident even when the reciprocal relationship between expenditures and vote is taken into account. 200 In addition, there is evidence that the degree of party competition in the district is the strongest pre— dictor of the vote when other variables are held constant. Party strength is a strong predictor of expenditures as well. With this taken into account it is difficult to argue that incumbents have electoral advantages in Michigan. Incumbents apparently spend in reaction to their chal- lengers. If threatened they will spend more. However, challengers get more votes for their money. Unfortunately, few challengers appear able to push incumbents to the limit of spending. If there were more, we could perhaps see more direct evidence of incumbent advantage. FOOTNOTES 1"Public Act No. 388 of 1976 As Amended" is the formal act which specifies all procedures for reporting campaign contributions and expenditures. Failure to file a statement within 7 days after the filing deadline is subject to a fine of $1,000 or 90 days in jail or both. Zane Democratic incumbent was defeated in the general election. He received only 2h.3 percent of the two party vote. One Republican incumbent was defeated in the primary and his opponent went on to win the general election. Two Democratic incumbents were defeated in the primary election: however, their opponents were both unopposed in the general election and those districts were excluded from this analysis. 3Thirty-nine of the sixty-seven challengers spent less than $3,96h in the 1978 election. Only three spent more than $20,000 and all lost. “When no incumbent is running it may be that spend- ing and expected vote do have a reciprocal relationship based on the strength of the old incumbent in previous elections. Equations based on this premise were estimated and were not significant at the .05 level. 201 CHAPTER 6 CONCLUSIONS AND IMPLICATIONS The purpose of this study was to develop a theory of turnover in state houses and test it. The thesis presented here is that turnover is a function of the insti— tutional environment, electoral environment and opportun- ities for higher office. The analysis presented in this work leads to the conclusion that this theory is supported. Several of the variables hypothesized as important do account for variations in turnover. Turnover in state houses has declined throughout the period 1952-78. 0f the variables examined, the level of professionalism in the state legislature and reappor- tionment have the greatest impact on turnover. But one tends to offset the other from the mid-Sixties through the Seventies. The evidence points to a direct and moderately strong relationship between professionalism and turnover. However, this finding is not as straightforward as it appears. In the South, professionalism has considerable impact on turnover. Whether this is due to historical/ cultural factors such as one party dominance or a norm of 202 203 volunteerism could not be tested with the data. For the non-South, professionalism has considerable impact on turnover. Further, when examining all 47 states included in the analysis, there was some evidence that a "threshold” effect might be taking place. The attractiveness of the institutional environment as represented by professionalism affects turnover up to a certain level. At that point a plateau is reached and a higher level of professionalism has minimal, if any, impact. The trend by many states toward more professional- ism is resulting in a more stable membership in many state houses. Part of the professionalism "package" is longer sessions and more bill introductions. Incumbents may benefit from greater exposure, position-taking and credit- claiming when they run for reelection. Another part of professionalism is the provision of incentives (notably salary) to get incumbents to stay in the legislature. The evidence indicates it works. This study revealed little support for relation- ships between other institutional variables and turnover. This was probably due to the use of poor measures. Future studies would do well to focus on committees and their role as centers of influence within the state house. Reapportionment does have a significant impact on turnover in state houses, but it does not always operate 201+ in the same direction. Based on my findings, reappor- tionment in the Fifties may have helped more incumbents than it hurt. This would lend support to a thesis that reapportionment may be used to protect incumbents. This conclusion must be tempered, however, by the realization that a number of states did not reapportion at all during the Fifties and many of them had consistently high levels of turnover. Following Supreme Court rulings in the mid-Sixties, reapportionment proved devastating to incumbents. Many incumbents who did not leave office voluntarily were defeated. This pattern was repeated in 1972, but the effects of reapportionment were reduced. Clearly, the determining factor in the effect of reapportionment on turnover in the house is the degree to which district boundaries are altered. The level of party competition has an impact on turnover, but it is related to the measure used. Tradi- tionally, studies of legislative turnover have relied on a measure of party competition based on the degree to which one party controls the legislature and governor's office. The evidence presented here indicates this approach is not appropriate. Results were inconsistent and defy explanation. A.much better measure of party competition relies on district level data. Party competition measured at 205 the district level is related to the proportion of incum- bents who leave electoral politics and the proportion of incumbents who are defeated as is evidenced by the data for Indiana and Michigan. Indeed, district level com- petition had the greatest impact relative to other vari- ables on the vote for representative in Michigan in 1978. This finding helps to explain why so many studies examining the relationship between party competition and public policy conclude competition has little impact. Invariably they rely on a statewide measure of competi- tion. This procedure fails to take into consideration the possibility that the majority party's membership may represent highly competitive districts or vice versa. Nor does it consider that a legislator's first concern will be the voters back home. When examining opportunities for higher office, the findings suggest they are related to the proportion of incumbent representatives who run for higher office in Michigan and Indiana. However, the determining factor as to whether a higher office will be sought is if the incum- bent is running. If not, then the proportion of state representatives who run for higher office is greater. If higher officeholders do run for reelection, then a some— what lower proportion of state representatives will try to move up. 206 Campaign expenditures in the 1978 state house elections in Michigan had a significant impact on the vote. The evidence shows incumbent spending is largely a reaction to the threat posed by challengers. This threat was seldom serious in Michigan in 1978. If membership stability is the goal of reformers, Michigan is a state which can serve as their guide, at least with respect to the variables found to be important by this study. As of 1978, Michigan had a highly pro- fessionalized legislature, relatively few competitive districts and few well financed challengers. Incumbents sought reelection often and won easily. Several implications are worth considering here. Price (1975, p. 20) may be correct in that once the mem- bership of the chamber stabilizes, committee stability is sought and then committee seniority. While it is not clear what level of turnover will result in an appropri- ate level of chamber stability and so on, it is clear state houses are headed in that direction. Price (1975, p. 21) suggests "three counter—seniority mechanisms: internal scramble for leadership, outside executive and party interference, and the possibility of strong legis- lative cohesion." Tb the extent that professionalism promotes incumbency advantage it may in turn weaken party voting over time. Legislators as individuals not party representatives would emerge. However, if legislators 207 are able to free themselves from the party, several of these mechanism are removed. That increased professionalism contributes to reduced turnover is apparent from this study. This in turn should result in legislatures which are more cap- able, at least in the sense that they include legislators with more experience. What is unclear is whether low turnover contributes to legislative responsiveness as well. The data from Indiana and Michigan provide some indication that increased professionalism may also con- tribute to a decline in the proportion of competitive districts in a state as Well as the proportion of incum— bents who are defeated. Incumbents are lured into a reelection bid, but the chances that they will be held accountable by a watchful electorate are slight. Indeed, if professionalism provides electoral advantages to incum- bents, all that may be accomplished is a shift from legis- lators who are not responsive because they are willing to leave office to legislators who are not responsive because they know they will be reelected anyway. Unfortunately, an examination of this point of view has not been explored. Much of the literature on state politics focuses on policy decisions made at the state level. Profession- alism is often used as an explanatory variable in analyses of this nature. Yet, turnover in the legislature is ignored. This is intriguing. Efforts to become 208 professionalized deal with structural or procedural factors. Information services are improved, staffing is provided, members can introduce more bills, sessions are annual not biennial and become longer. While these changes will make a legislator's task more manageable or even worthwhile, these changes do not make a difficult vote on a bill any easier to cast. Policy decisions are still value questions of one form or another. And the issue for policy makers is still winning or losing an election. The possibility that an increase in professionalism leads to a decrease in turn- over because incumbents can more easily win reelection would seem to have important ramifications here. Grumm (1971, p. 319) finds that professionalism has a "signif- icant independent impact" on "welfare liberalism." Why? If my reasoning is correct, it is because professionalized states have legislators who are relatively safe from electoral defeat and can "afford" to support redistribu- tion policies. Members of a professionalized legislature need not be as concerned about winning next time around. While this reasoning stretches things a bit, it is Worth future explo rat ion . Clearly this study has left some important issues in doubt. Better measures of influence within a legislature, the importance of the legislature within government and party competition must be devised before 209 variations in turnover can be properly assessed. In addi- tion, further analysis of turnover must include a large number of states with regard to'whether legislators leave politics, run for higher office or are defeated. That such analysis is needed is beyond question. APPENDIX A DATA FOR FORTY-SEVEN STATES: 1952-78 £211) or me on no em as so we no we as we or an cecH me on em me an an at cm as we me an me an dsdataH mm on so on an an as as «N we we as we so oncsaaeH on on an an no on on mm as as as as we as crate an we on no an an an on a: on a: as me we oneuaoo an an an as an as an an on no on an ms or «caucus an an an as as or an or or an an we no or entrance an an an or on .sm an c on an we he as as esoaaoonaeo an on me as mm mm mm me an an mm we we we crusades an no mm no me on as «a me an on an we so . «armaments we on en un an em a be me an on he an or as coosoxh< an we on on an we we on as me or an we as a:ou«h< an sz as H: er a2 an a: on ex um . ez no s2 oscaee< when mead span «and some some wean some «can coma when mnma amen «use Aoomoacoohon an: mhmdlwnmfi 32E E mambo: ankfim zgawlt~5h 2H mares—~59 Hl< "mt—ah. 211 HM MN M: N: NM NM Hm mM N: 5: MM N: M: M: afiddofido fiahoz no an on an an no he on be so an me an am such 3oz em on on an on or an me an as on me me an coaxo: 3oz am mm mm an no on me an or mm mm on mm on stones 302 .3 .3 n: m: an an 0: an FM .3 w: ms .3 e3 cannon—Eon :02 mm no me mm no we me an on we we mm an on oso>oz no as mm on mm an em or as no as on me an encoder no an em me on on me on an mm mm mm en en .ahsccon: as e2 as ez mm m2 we s2 mm oz no m2 we m2 aoaaoonocaz mm mm as or on «a ma we us an em so an em decrease: on re mm no me as an mm mm no we om an om edeason: QH :N wN mm :m wN ON NM MN ON MN NN MN wN navcoflflodundz on s2 ms e2 as mz we e2 we e2 as m2 mm a2 tetanus: an on as we as a: we mm mm on an on mm on team: an an m2 we s2 we s2 mm m2 we e2 on mz we osoaonsoq on no an as an on an am so me an an an an excesses on an an an no on as an on an on an so an nausea when when read name case some come soon «was some wmaa mama smma Nmma AA Acossaesccv Hu< sqn poo: nn on on nn no oo on an no no on nn no ns covenants: no we so as no no as no na so no no no nn cacaeos> no nn on on sn nn on ss os nn an nn on on onoauo> nn mm no nn ms os on an nn ns mm mm os mm as»: oo an oo nn so no nn no ns nn nn on nn nn coups so no mo nn es on on on sn on on on so no ooooosspe nn os an an ns nn nn nn cs ns an nn us as daemon Sodom MN mm a: H: OM MM MM 5N MM m: MM . N: h: N: mflfldohdo fivflom oo on so on no an as no so an no sn on nn season ovens no no oo oo oo so on on no oo so on an no casatanondom no mo nn es os nn no nn on on ns nn en ms douche so nH on an so so no es nn en on an on nn oscsosxo rs ns nu nn Ho an ns on no on on no no no case an nn on an os so on as on nn nn os an on season neocz waH wNMH shad NNMH ONMH wwma wwad :wMH NwMH omMH wMMH wMMH Shad NMMH Atoscsenocv H|< adage 213 oon nno oHo nna osH one on an on on nn nn oo 0o osoH nno nno oso oso sns nHH an as on on on on on on ososssH oon oos sns sos cos cso own one cos cos cos ooH sea son uncanHHH ens ssH no on nn ooH ss Ho na o o o o o csctH one nnH nno sns non won noH os os so on on on Ca osmoooc owo omo nan nsn onn onn nod nn mm mm mm sn am On otnuoah noo oHo oHo con ooH ooa son on oo oo oo oo oo oo choruses ons enH ens ens no es 0s 0s no no no o o o rsosuoosoco ono soo one ona one on so so so as ms on on on otduoaco oHo oao Hso nnn onn nos snn wsn wsH mss msu nsH nsH ooH osshcosnoo con son on ss nn on on on on so so on on on oeossxn< oHo ooH onH end ooH msa on sn an os ns en on nn oscosh< nns nnn nos nan oHH can no on on on on on on no oscsas< wnna onnH snns onnn onns wons oona sons oons eons onns onns snns onna Ao.ooso use onnsuonnu .oneoaam no mambo: semmnbsmoe zH wooe02 nos nos no no os nn nn nn so so sn os ns os ooo>oz oo sn nns oss no so so On os os os os o o snooze: onn soo sso sso nos nos nss nss nss sss ns ns ns ss swscoosz ono soo noo nss oss nss oo ns on On on oo oo oo saasoosoosz son onn ons oos sos ons oss nss on nn nn oo ss on succeeds: son onn ens oos nnn oon nno ono nos nos oos oos oos on guesses: nss non ono non ono ono oso oos oos oss sss sss on on oesoossoooooz nss ooo nsn nnn noo no on no oo oo oo os ss ss enosnnoz nss oss no no on ss on On os os ss os os os case: non non oso nno nos nos nn nn nn on on on on sn onosossos sns ,sns nos nos oos on oo on on on no so so ns sausages soo nos nos oos so on on os cs ss ss ss ss n . oceans onns onns snns onns onns oons oons sons oons oons onns onns snns onns Apoficducoov NI< Qsm<8 ZZIJS os ns nn ns os os ns ns ns os os os n n masses: nss oon snn oso nso oso sso sos sos oo oo oo nn os couscous: nss nos nns on on on on on on on on on on os osnsoos> one: sno sno oss ons ons nn on os nn nn nn nn nn so oceosssoes oso nso ono oos ons ns nn os os os ss ss ss ss usesohs> nns no so nn oo sn oo oo oo so os os os os oacsno> on on on on on on ns ns ns ns ns ns ns ns _ no»: oos oos sns oss nss oss oss oss oss oss on on on on coups ono ,ono ooo oos sn so no ss ss ss ss ss ss os oooocndoe nos no nn oo sn on on on on os os ns os os cocoon sasoo oos oos ooo nso ons on on os os os oo oo oo oo masseuse sosco o o o o o o o o o o o o o o passes steam nnn snn oss oos osn oso oso oso oos oos oos oos oo oo osso>snossoo nsn oso ooo sns nos nos sn so so ns os os os os soothe osn noo ooo ons oos oos nn nn nn nn os os os oo oscsosxo ons onn onn ooo nno nno oos oos oos oos oos oos so so csso nn nn no on nn on oo no ns os ss ss n n cocoon sates onns onns snns onns onns oons oons sons oons oons onnm onns snns onns Acosesencov oa< muons 216 snoo oosn osns noon onno oooo nnno nnns onns noos osos oso onn ooo ates nnoo nosn sonn nnns oosn nnso noos ooss snns non nso ono snn onn oqosoes noono nsooo onnoo nosns oooNs nnoo ooon osss oonn snsn oonn snnn scoo onns oscsssss noso onos ooss onns noss soss snn nnn nos nos onn oon ooo nno chaos nosn snnn onno snnn noos nsos sono nooo nooo nons oons osos noos sso osohcoo ooons nosos oonns snnws oson nnso onon onno onoo snos snns osns ooss nno ooshcsa sooo nsos nsss osns snos sons oon osn ons sns snn non oon oon onotosoo nsno snnn oosn noos onnn onno ssns snns nnns nsns oons snss ooos soo ososeooqnco onnn nons nsss nsoo nsns nsns snos ooos noo nsn oso oon non sno ctoocsco oosoo nnoso oooos ooonn onnoo soooo nnsns nooss snoo oono nnso oons nnnn nson esshcossoo onns nnon sono oons osns nnss nnss onos nno son soo non snn oss oooaesho nnon snss nosn onsn osso snns oon osns non nno oso noo oon snn ooonsh< onos nson snnn sooo ssno onns ooos onns onns son osns onn nns ono esoneso onns onns snns onns onns oons oons sons oons oons onns onns snns onns Ao.ooo.so :sv onnsuonns ”zoseoesm no mambo: meooursmoo 2H mmpeoz onnn onoo noon oson snns nsnn oooo snss noos noos ooos nnos ooos noo nachos 2oz onnn nono snos osns soos onos nnss osn oso ooo non oon nns sos oossoasoo :oz osns soon nnno oono nsns soss soos oon nno snn nsn nns nno nns ooo>oz nnon ooss oooo snns noso onn ono nsn son nns nnn oos ooo sno unease: oonos onoo ooon nsns oson osnn snoo nnno oooo nnos nnos onss soss ooos saucers: nnon onos oosn oooo noos ooo non nnn non nns sns nos nnn ons seesoosoosx nonos sooos onon onnn nonn noos nnno ooso ooos oons oooo nons nsns osos «seconds: snnon nsono onons ssnss sosss oson onsn nnon onno nons onos oons nons nsos esossos: nonoo snsos onons nnsns nsoos nnno nssn oooo soss nson ossn nnnn nsos snoo oeeooasoaooo: oonn noon nnoos sooo osns oono nnoo onss nsns noss osos ooo nno onn ososnne: nonn snon oono sons oons ooos nnns ooss nsn soo oon non nsn oos osso: nnon nnnn osoo snoo oosn osno oons nons snss onso snos osss nnn nss esososscs osso oonn oono nnos nnos noos nso sos ono non ono nns nos nns nooseaou oosn nnnn onns nnoo ssoo sons nnns nnns noos onn soo nno oon nsn noono: onns onns snns onns onns oons oons sons oons oons onns onns snns onns Aocsasacocv ns< ooons 218 nnos nnno nnn sos nos non oon ooo ooo ono ooo ooo nns os ozsscnz osoos nooos nnoo ooso oons nnon nnoo nnns noss soss snos noos nnn nso :snucons: soon nnss ooon sono ooso snns noss nnos ooss ooos onn soo nnn nnn esssous> «no: snnns nsnss oooss nono nonn nson nnon nnoo nnns oons nnos ssns noos ono :cuonssoo: nnoo nono sosn nnnn ooso nnss nos snn ooo soo son non nos oos asusons> sooo ssns nsns ooss onos non nnn nso noo sno sno nso onn oos onosuo> nnno noso soos oess nsos nsn nso nnn onn ssn ono sno nno ono no»: nsono nonns oosos onoos nosn osso nsos sosn ooon ossn nsno oooo oon noo nonos sonn nsss ssos nono nnos onos oon oos nno oon ons nnn osn nno commandos sooo nnos onns nno onn nnn nns non nnn son ono ono ons sns «pesos sesoo ooon nooo ooos nsno onoo nnos onns nnss noos non snn oon oon ooo onsseueo nonoo oson oono noos onss nnn onn son nns ons nos onn noo ons ons theses ooons snosn nooon nsoso nnoos nonos oooss nnoo oonn onsn noon nnon sono snso nnoo onno>snoaaoa snon soon onnn soos nooo osos ssns ooss onns noss soo nso nno ooo nooouo oooo oonn ooos sosn ooon ssno snos ooos noos nnss nnos noss nnss ooss oscsosxo oooos osoos nson onnn nnns nonn ssno snoo nons nooo onoo onos nsss nnos osso ssnn snoo nnso nsns onos ooss ono ono noo ooo oon sos onn nno caches soucz onns onns snns onns onns oons oons sons oons oons onns onns snns onns ”nanoseccov nn< ssmoz oos oos oo oo oo oo so so oo oo oo oo oo oo octane: ono son oon nns ons nns nns nns nos ons ons nss nss nss saucers: oos oos nso oos nos oos nns sns nos sos oo oos nn oos seasoosoosz ons sos oss oss nss nss nss nss oos oos nos sss oos oos anemones: son ssn non onn onn nnn ons nso onn onn nso nos oos oos soossos: onn nss non nsn nns nnn ons oon ons ssn nos nno nns nns ooooosscooooz oos oos oos oos oss oss oos oos oos ooo oos on on oos ososnnez sso ons sss sss nss nss oos oos nos nos nos on so oos osso: nos ons on on on on on on on on on on oo oo «assesses oo so oo on oo on so so so on on so so oo nacsoqos ons nos oso ons nos ons ons nos on on on oo so oo ounces onns onns snns onns onns oons oons sons oons oons onns onns snns onns Asosnsasoov su< osoos 221 so oo os os os os cs ns os os os os os os messes: nns oso nno oso oon ons oso oso oso sso ons oos oos nns qsocccoss nns oos nos nn nn on on ons on on sn oo oo oo escsons> ooo: sss oo oos oos oo oo oo oo oo oo oo oo oo oo scoocssoa: oss oss oos nno on so so so nn so so oo oo oo usesows> sns sos nns ooo oos sos nns oso oso ons nns soo ns nss eecsuo> no oo oo oo oo oo so so oo oo oo oo oo oo so»: nss oss oss oos oss oos oss nns oos oos oos oos oos nns canoe oss nos nss oos nos oos nn nn nn nn nn nn oo nn ooocosspe nos nos nos nos nos nos nos no oo oo on oo oo oo eocsao oosoo soo non snn nno sno os sns sso oso ons nso ons sn oos cassette sesco nns oos sso oos nns so nos oos oos oos oos oos oos oos ceases ooons ono nso ooo soo noo sso oos nns osn ons nn oos noo oos osso>snoeeoa oso oos sns sns sns nns sos sss oos nss oos nss oos on ecooao ooo nso nns noo nns nos sos ons oso nos nos nos oos oos oscsosso noo sso son oso oso nos sos ons osn oos oos sss sss oos osso nn on so so so so so no so so so so so so season sauoz onns onns snns onns onns oons oons sons oons oons onns onns snns onns Aooscsecccv su< osoos 221 so oo os os 0s 0s os ns os os os os os os oasscn: nns oso nno oso oon ons oso oso oso sso ons oos oos nns esoscoos: nns oos nos nn nn on on ons on on sn oo oo oo osnsoos> ooo: sss oo oos oos oo oo oo oo oo oo oo oo oo oo cosmonaut: oss oss oos nno on so so so nn so so oo oo oo osnsoos> sns sos nns ooo oos sos nns oso oso ons nns soo ns nss uscsho> no oo oo oo oo oo so so oo oo oo oo oo oo nooo nss oss oss oos oss oos oss nns oos oos oos oos oos nns coupe oss nos nss oos nos oos nn nn nn nn nn nn oo nn cocooospn nos nos nos nos nos nos nos no oo oo on oo oo oo ceases oosoo soo non snn nno sno os sns sso oso ons nso ons sn oos masseuse Assam nns oos sso oos nns so nos oos oos oos oos oos oos oos cusses ooons ono nso ooo soo noo sso oos nns osn ons nn oos noo oos useo>snossom oso oos sns sns sns nns sos sss oos nss oos nss oos on scoowo ooo nso nns noo nns nos sos ons oso nos nos nos oos oos oscsosso noo sso son oso oso nos sos ons osn oos oos sss sss oos oono nn on so so so so so no so so so so so so season suuoz onns onns snns onns onns oons oons sons oons oons onns onns snns onns Asoscsoccov sn< oso OH OH OH b m OH [\O\O\\OO\®F a b momma OH OH m OH m O\\O\DO\I\O\O\w OH a OH w OH OH OH OH m w w t~ \o O\ ax r~ ts \o o\ r- 1» OH OH OH OH m m o o m o I‘MONFW w o CNN“) N OH 5 w {‘0 O\l~0\ OH m h w m b- b- co 00 non nno m h o \O oosoo nonoo h m w m NONONWF-thfiowml‘ b a30H dfidchH oHOGHHHH onde mHMhoow «VHHofim ohddeQQ 950H¢ooflnoo ovdHOHoo asnwomssao ndondxh< dfiouHH< aEdndH< mfimH whmH :NQH thH ObmH wme wme :me NwaH meH wmmH wmmH :an Nan onosumnos AOH 0H ouoon ESEHKuEv .ZOHHOBAQ #m mambo: mhdem zm>mwleMOE MOE Qmoom mm0H>mmm m>HH02 OOHNOZ >02 hookah #02 ohHunEdm 3oz dvd>oz dddafloz HhflonOHz HmadnoHoodz duoooucHz :deAOHZ ounOofiSOGoodz ofisrax oGHd: acdHuHSOA hxoflpfiou odundu whmH wbmH :hmH NbaH OhmH wme wme :me Nme meH mmmH wan ammH Nan 325353 nu< osn; 252“- OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH OH a OH OH OH OH OH OH OH m m OH o OH OH m OH O\O\t~®O\QO\F\ O H O v-l OH a OH n n w o u b o o b m m m o o o n n s m n b o m m os os n os o o o o o m m o m o os o m m os os o m m m m o n O\\O ao hows w O\\O {‘1‘ m w O\O\O\O\O\\OI~NF\OJQO\\DI~J o O\O\V\\O\O\OI~\OI~\O£wO\\Ol\\D o o ONIfi\O\O\O NW nnno Q0350 o oooxmomnxontsnnhoxsbxo w wcHthz nfiudooudk 5H:HMHH> anon Coaunfinoak dHflHMHH> adofiho> Ada: manna oouoouflpfi caOXGQ nafiow dfiHHOhto naflow UddHOH ovonm aHna>Hhonflom flowoho qundeo 0H£o «oosoo nauoz whmH whaH :hmH meH ObmH mme wwaH :me Nme OmmH man wan dan Nan Avasnsaaoov n|< mqn02 ono onn sns snn osn sno sos sns oos sos oss nss so so ao¢>oz osn nss ssn oou mos mos an nss sos sos oo mo oo no «nonoo: sss snn snn onm sns nns sns mss no on oo mo sn sn snzooosz nos nss osn ooo «no oos sns nss oo oo so on nn on soosoosoosz Nnn noo nsn oos now nsm sos oss nms mss no oo on nn aaoooqus: sno own son oon now onm ons mos nss sms mos nos No no quosnosz nno non son nnn oom omm sns ows mss Nos oo oo oo no nouousnoaauu: ono nsn was onn oom onm sns ons ons nss nos oo on no unasnnu: non ons osn now omm sos ons oss oo no no on no no onsn: onn oss nos osn oon oom oos sos nns oss ons sms sos Nos anusussos oon nns onn son «no noo oss nos oos oo oo no sn ns nxosaqou son oss onn ooo now oos ons oss oos oo so nn on oo aaunao onos onos snos nnos onos ooos ooos soos woos ooos onos onos snos «nos Acoscsacoov on< osm «no: moo son ons nnn oon ooo sno sou nos nos oss nns oss oos :oaoussuu3 nos oos nnn soo soo oos oss nos oo oo oo oo nn sn ascsous> msn nos non onn oon onu oos ons an nss sos so on on unoauo> oon nns snn ssn nsn nos wns oss oos sss oo so nn oo so»: oms oon non onu sos nns nos oss nos so nn nn on sn mango sos non «nu sso nos mos ans nss no no so on so nn ooouonupn onn «on onu sso Nos ons sos so oo so on so so sn «oosoo nnsoo osn nss onn non omm nns mns mus oos oos so on on sn anssouao oosoo onn nns nnn own now own ons nns «as sss oo on sn oo onusns oooso onn sns non non oon Nos nns nns sss mos so oo on on asnu>snunnom osn oss nsn oon oon ooo sos nss sns nss oos mss on mo nooouo sun mss own sow now sos sos sss wns «ms uss nos no no usesasxo ons snn «on oso sos nss sss sos No oo on sn oo oo \ osso sos sos nss sou own oos wss ows nos oo oo on on nn aposao nouoz onos onos snos «nos onos ooos ooos soos woos ooos onos onos snos nnos Anasnsunoov on< msmoouoaooo ooo oomooo xoozs zoss02 NnH.HI :HH.HI «no.3 bnm.l now.l Nw>.l mnn.l shm.l nN:.l mmn.l sns.l nss.l bsw.l wmm.l OOHNQZ 3oz oon. noo.n son. ooo. «no. son. ss~.| ooo.n oon.u onn.u soo.s ono.s nno.s nno.s nounmo aoz oso.s: noo.n smo.¢ noo.s noo.n sno.: son.: oos.n sos.n oss.: ons.s oos.s ooo.n ooo.n oasnuoeam :oz now.l wow.l ono.l wsw.l nno.t Hnw.l oom.t «no.1 Hnn.I mnm.l mos.l Nos.l nno.s hob.l ava>oz onn.: osn.: son.n ooo.n sso.: oon.a osn.s onn.: ons.s oon.n nns.n osn.n ~sn.n snn.u «noono: osn. ooo.n ooo.i sso.u ooo.n NNs. sns.s non.l oos.n osn.l snn.l osn.| now. osn. skdounsz sos.u ono.u nns.u oss.n onn.s nos.u ooo.n nnn.u nns.u ooo.n nn~.u oos.s sos.n oon.u soosnusums: onn. onn. ooo. onn. ons. onm. nns.l osn. nns. ems. nno. onn. sms. osN. aaooondsz son.s onn.s ssn.s sns.s osn.s nnn.s snn. noo.n osn.s sso.a non.u son.n nno.s oso.s :aosnosz noo.s ons.s oos.s oos.s oos.s ons.s nnn. onn.s osn. oso. ooo. soo.s ouo.s oms.m masonsnoauna: oss. onm. noo. ssn. noo.n ns~.u ooo.n sno.a on~.u sss.n ~s~.u ooo.n oos. oss. onasnnnz oon.n oso.u oso.u soo.u son.n ooo.n nns.n oos.u son.s mos.n nnm.n noo.n osn.s ooo.n ossn: sos. osn. nos.n oos. mos.u nno. oo~.u noo.n noo.n nos. nno.n ono.s son. ssn. andsossos nss.n son.s ssn.n onn.| mon.| nns.u non.| oon.n oos.u ooo.n onn.u ono.s ons.u -n.n nxoaaaoo mon.n non.n noo.n ons.u oos.n osn.s noo.n oon.: oon.u nss.| nun.u son.n nso.u noo.u noncwo onos onos snos nnos onos ooos ooos soos ooos ooos onos onos snos «nos soosasacoov nu< osm one: n««. nn«. ooo. noo. «n«. nss.n «s«. soo. oos. sos. n«o.n sss. ons.u so«.u canonsAoa: os«.: nss.u soo.n sn«.u osn.n «so.u son.n nos.s nnn.n s«o.n nss.n nos.n onn.s ooo.n osqsons> son.n «no.1 ooo.n ssn.n son.n ssn.n non.n o«n.n noo.n oon.u non.| oso.n n«n.u son.r puosuo> n«o.sl Noo.sl Ono.l ooo.n oon.l son.n oos.s oon.l ssn.| nos.t mon.| sno.l non.l oos.s :dvp os«. oon. ooo. nnn. oon. n~«. n«n. nsn. sos. onn. soo. oos. «os.n oso.n maxoe ons.s ooo.n on«.u os«.: ons.n oon.: s«n.u oo«.z nss.n sos.a ono.u noo.: ono.n noo.n oonuoqnpn nso.n nno.n onn.s nso.| ssn.n nso.n snn.a nsn.: osn.s «sn.| ssn.: nnn.n snn.a ooo.n ososdo nnsoo nn«.n oss.u sos.n . nos.u n««.: sns.n son.n «on.n «sn.u oon.u osn.n onn.u s«s.u oo«.u anssouao nnsoo sno.u ono.s son.n nno.s sno.n «so.u snn.a on«.u nss.n non.u sns.u «nn.u oon.u non.n onusns ooono ono.s oo«.s ons.s soo.s «os.s osn.s nno.s nsn. «on. nnn. nnn. nsn. soo. oo«.s asaa>snaqa¢o nns. noo. oos.s os«.: «so.u ns«.n ons.s sss.n noo. soo.n ons.s «oo.n noo.n non.: nooouo o«o. noo. nsn.n os~.| ono.u ono. o«n.u ons.s sso.u o«s. ons.n ons.u nno. ssn.| «sooosso «so. nos. oon. os«. n««. nnn. oo«.n «o«.u sn«.u nos.n ooo.n noo.n onn. oos.s osno oon.n noo.n non.n «on.u soo.u ooo.n oss.: nos.| nus.a sn«.| son.n onn.: onn.n non.| «oosoo nauoz wbmH wbmH thH thH ObmH wme wmmH aw¢H NwaH meH whmH wan aan Nan soosasuuoov n|< osmflwl>fim0h “Oh KQQZHZ QWDO: mam mfimBBHZZOO hO mflmZDz wl< HHm<9 2:32 on on on :N :N :N sN ss N: n« n« «n no no No on on on os ss on on Hm On ns ns ms HH HH NH nH nH nH oH oH Nn oH ns ss sn sn sn ss on on «n no so on on on on on NH NH NH 3H sH nH sN 3N sN sN :N NH NH NH HN MN nN nN HN :N am :N n« n« os n« n« o« n« «« n« n« n« o o n o o n o o o o o o nn nn on no no nn on nn on nn nn ns ns ns nm «« os os os os os on NN ON ON mN oN nN HN mN mN on mN nN ON ON nN Hn on O: H: H: on on ns «s ns ss os s« s« s« n« s« on mN on On on on mN wN Nn :3 ss n: o o n n n o o o o o o o s s s n s s m HH HH HH HH o N N OH OH NH NH nH NH nH sH sH :H mH mH NN NN mH NH oH MH oH ns ns ss ss ss ss. ns ss ns ns ns ns ns ns ns os o« on on on on so {I onos ono s snos «nos onos ooos ooos seofi momfl oom~ mnmfi chm «0253.83 o..< osmfin sw..s.n.no.s.mo nuhoz 3H0? :oz oosxoz >02 Nouhwb Boz GHHSmQEd: >02 ova>az 6:69:02 shsonnsz HnanmHnOHz aaononnHz fiosaosz wavmufifiodundz UddHthz oflsdz anasossos hx0=vdom ago /| s :3 «nos 233 os os ns os o« on on on on on on «n on on onsEoo: «n o« o« o« o« o« n« n« n« n« n« n« n« n« :snnoons: ss ss «s ns ns ss s« s« s« n« n« n« n« n« ususons> ooo: o« «« ns ns os os os s« s« nn sn o« o« on nououssna: o« ns ns «« «« sn sn sn sn sn sn sn sn sn osnsous> o o o os os os os n n n n n n ss unoeuo> ns ns «s o« n« n« n« n« n« o« sn n« n« on so»: s« os os ss sn sn o« o« o« o« o« o« o« o« auxin ss ss os os os o ns ns ns ns ns ns ns on oonaonnoa ss ss os o« os nn nn nn sn sn no no oo oo auoxoo nonoo o o n o o o o o o o o o o o dassouao Assam o o o o o o ns ns ns ns ns ns ns ns onusns ooogm on Na NN «N ns NN NN NN nn nn nn nn Nn nn assign—Eon sn sn sn «n «n ns nn on «n «n «n o« n« on 659.3on s« ns ns ss «s ss os ns ns ss os os os os osno ss «s «s «s «s ss ss ns «s os os os ns os «oosoo nuuoz onos onos snos «nos onos ooos ooos soos «oos ooos, onos onos snos «nos Romansunoov o|< osmoz dflavfloz Hkflonosz HandooHuon daoooflndz caudsonz ooaooaflonOIZ onasnuuz oanZ aadHnHSOH hxofivnom odoflda onos onos snos «nos onos ooos ooos soos «oos ooos onos onos snos «nos Acososacoov on< oso «no: s m n n n m w a w m N o s o Gaugudns ss «s ss os n« o« sn «n nn «n on nn nn sn usasous> os os os ss ns ss ns os s« s« n« o« sn sn «nosnwn o n o o o o n n o n n o o o no»: «n on «n so nn nn os ss «s ns os ns ns os coupe ns os ns os «« o« on «n no on nn nn nn on ooouoqcpn «s o «s ns os os ns os ns ns os s« on nn «oosoo nsooo on o« om nn on «s ns on On On on on On On d:ssohdo fiasco o« «« os ns ss ns ss os o n n n o n causes ooono s s s n n N N N N N s s s o assabeosEom o o o n n s n n o ss ss ns «« o« nooono o« o« os os os «« o« o« on on o« o« o« n« «aosasoo o o o o os o o o o o «s ns ns ns osso ss «s os ns ss ns os ss ns n« on no on os season spuoz onos onos snos «nos onos ooos ooos soos «oos ooos onos onos snos «nos sooscsucoov oa< osmoz ns ss os os o os o o o o n o «:35: ns ss «s ss ss ns «s «s «s o o o susouns: n« s« o« «« os n« s« os os os os os soosmmscus: nn sn on oo on on ns on nn on os nn 30:55.: sn sn o« on so o« ns «s ss os ss o fiosnosz on so mo No nN ON on o: on on N: Nn ouvonflnodunuz nn o« o« «« n« «« on s« os ns os ns «.518»: os os os os os ns os os ns ns ns os ozsaz n« n« s« «« o« ns os os os ns o« os sausages ss «s «s os «s «s ss os «s os os o noono—So «s «s «s os os ss 3 o o o os o manna: onos onos snos «nos onos ooos ooos soos «oos ooos onos onos nowansvcoov Osl< asm ooo: nm on... oN HN oH os NH ss ss ss sH sH deeming» s« ns os ns os ns ns «s os ss ss «s nsnsous> o o n s s n m n N o o o vac—Fuss o n o o N o n o o o o o 839 on so n« o« s« o« os os ns ns ss ss coups on o« o« n« «« s« o« ns o« os o« n« wounonupa N N o o o N o o o Os o o «assoc nonoo ns ss ns os os s« ss ss n «s ss «s cussouoo nnsoo n« n« n« n« o« os n« os os ss ss ss causes ooonm «n on .sn on «s no no o« on no n« n« asn¢>snunaom o« s« n« os os ns ss os o« ns ns ss nooono o ss o o ss os «s os os ss ss os osoncsso ss «s s« ss ns ss ss ss ns os ss ss osso «s ss os ss «s os o o o o o o auosao nauoz onos onos snos «nos onos ooos ooos soos «oos ooos onos onos Acosaspcoov osns osm