‘WST (-n‘ .ocv-.~--~r “‘1- fivn “1-- 95‘1”..- {:01 . , I v-n-n. -p- - ”uni..- ‘ I . Ipsm*“. III” "I. I |..--. I 'II I II IIfif I 1 .I . ,:I‘:.v;:'-.‘Ir,= ‘II'I I INN“ III “.7 II' l1‘111’IIIII I .{TIII III, ii; .. I I {I I) III IIIIII.I IIIIII I'f I:!‘IIIt 3.11.113)!“ I I I” I. .1 II I 1‘ III Ii IIJIIIflIHI 41:31:14; . 31:5“ I- ' III ”Jr "0! {WI (II. I, “VII I'II‘I’E: II I' .. I3 J {I .1; I", W. .11 . 5‘ '1 I 1 .II I' ' I t - . IKIIW ‘.. ‘ '_' IIII'VII‘II'III" 1| ‘ II'I' . I nip II. . I‘IIHI I I WM 1 v ' 'I ‘I‘ I :I‘ I‘Q’ffl‘:,‘If: I I 3 II 'I I '- , . I ".‘,'Il1|I-'II; 11:! I I ‘: I‘:~ II I HI I II'IW. II'I ‘ I II’II'I'IIIII I; In,” IIII II'IqtoIpH II . ,”I|I I I I,” III‘II" MI 1111'” I I I|IIIJ I. III I II “1111\(‘II d“ 1". I11)“ I” (””I I: 51,1. "IIIIII‘ 0'11 I1'Il 1‘ 1 A" III‘JI IIIIIII II ’“IHIJ ”I .““hand”flmm WIWNh WWWWIIM~WMHMIIW WMIUflImIIWHd II II III‘III’ {IIIWI .Ik I ’“NUV‘J IIII III {LBVWIII JIIK I' 11m, . ”III 11:.E .71 ~ ..._.; L“ . .. A... . 'a'ufuatt ‘Wm. ' I ”MM. 9. .. .. . M; a .. -~ m a .~ .1 -4 ow- I'I IIIII'I' '. ;I' II': IIIII ' I . : v I: .t-I - IlngIIfiIIWII'II I I§IIIIIII1IwIIIW11 II; I‘I'IIIII .IIIZ III ”.111". If! {11; "Ali” ’I,IIJ§ . dIIIII “W4“ _I I‘.’ IIII‘ . IIV IWWHII “2" , . ”I W 'IWWWFUWI .6 IIII‘II -:- ._ ~.. I “If II‘II.~LI';‘.IIIEI -: “If «1 \r’ (a 1 U8 Q Q Q nCHtGAN STAT QQQQQQQQQQQQ QQQQQ; QQQQ 3 1293 @qu Q. This is to certify that the dissertation entitled TESTING DISENGAGEMENT THEORY AS AN EXPLANATION OF POLITICAL INACTIVITY presented by Rick E. Rollenhagen has been accepted towards fulfillment of the requirements for Doctoral Political Science degree in Mfl/ M Major professor Date August 13, 1982 MS U i: an Affirmative Action/ Equal Opportunity Institution 0-12771 MSU LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. - T It; - {A TESTING DISENGAGEMENT THEORY AS AN EXPLANATION OF POLITICAL INACTIVITY By Rick E. Rollenhagen 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 TESTING DISENGAGEMENT THEORY AS AN EXPLANATION OF POLITICAL INACTIVITY By Rick E. Rollenhagen The study of political socialization has devoted little attention of the political behavior of elderly Americans primarily because of the prominence of disengagement theory in the litera- ture. Proponents of this theory claim that as individuals grow older they gradually withdraw from political activity. This analysis employs the 1956-58-60 and 1972-74-76 University of Michigan Survey Research Center panel studies of the American electorate to test this theory. The theoretical perspective here contends that age is a surrogate concept designed to assess circumstances and events which occur throughout the life cycle and serve as indicators of disengagement. In particular, three components influence political disengagement. Old age disengagement from politics is a function of sociological factors, key events or circumstances, and explicitly political factors. Methodologically, the analysis formulates conceptual typologies for two- and three-wave panel study designs. These Rick E. Rollenhagen typologies provide for both direct and indirect assessments of whether or not older people disengage from politics at greater levels than those in other age groups. In addition, the project assesses the disengagement thesis in two separate time periods and considers voting and political campaign activity separately. The empirical findings showed that in 1956-60 there was no strong evidence of old age disengagement from voting but some evidence of old age disengagement from political campaign activity. In 1972-76, on the other hand, there was some evidence of elderly disengagement from voting but not from political campaigns. The relationship between age cohort membership and politi- cal disengagement was also examined with controls for sex and level of education. A large part of the diverse pattern in old age disengagement from voting and political campaigns between 1956-60 and 1972-76 can be accounted for by the increased tendency of elderly females to disengage from voting between 1972 and 1976. In addition, retirement from the labor force and contact by a political party had little impact on old age disengagement. The final part of the analysis pitted the disengagement theory against an alternative argument, continuity theory. The findings suggested that political behavior across time is re- markably durable for the electorate in general and for older people in particular. ACKNOWLEDGMENTS The analyses presented here all utilized the 1956-58-60 and 1972-74-76 panel studies of the American electorate conducted by the Survey Research Center and the Center for Political Studies of the University of Michigan. Of course, neither the Consortium nor the original collectors of the data bear any responsibility for my analyses or conclusions. A number of people were helpful in the completion of this study, and a few deserve special mention. First, I must thank the members of my dissertation committee for their contributions to this study. Paul Abramson, my chairperson, has been a constant source of inspiration and encouragement throughout the duration of the entire project. His mentorship was especially helpful in those moments when completion of the study seemed insurmountable. He has always taken an active interest in this and other projects of mine and his contributions are continually over and beyond the call of duty. I am also grateful to Cleo Cherryholmes and Ada Finifter for their comments and suggestions at various stages of the project. Second, I am also grateful to other scholars who have com- mented on various portions of the study. These include Bernard Finifter, Denton Morrison, John Aldrich, and John Hunter. ii Third, and finally, much thanks are due to my family and friends who were a constant source of encouragement. Although they may not understand what I do as a political scientist, they ap- preciate fully my pursuit of that endeavor. I, of course, accept full responsibility for the entire analysis. iii TABLE OF CONTENTS LIST OF TABLES. . . . . . . . . . . . . . . . . LIST OF FIGURES . . . . . . . . . . . . . Chapter 1 THE STUDY OF OLD AGE AND POLITICAL BEHAVIOR. . . . Introduction . . . . . . . . . . . . . . . . . . Analytical Discussions and Previous Tests of Disengagement Theory . . . . Discussion . . . . . . . . . . . . . . . . . . . Plan of Analysis . . . . . . . . . . . . . . . . NOTES TO CHAPTER 1 . . . . . . . . . . . . . . . 2 TESTING DISENGAGEMENT THEORY AS AN EXPLANATION OF POLITICAL INACTIVITY--CONSIDERATIONS OF THEORY, DESIGN AND MEASUREMENT . . . . . . . . . . . . . Political Participation. . . . . . . . . . . The 1956—60 and 1972-76 Time Periods . . . . Panel Study Analysis of Changing Levels of Political Activity . . . . . . . . . . . . . . Two-Wave Analysis: Assessing Changing Levels of Over Time Political Activity in Presi- dential Elections, 1956-60 and 1972-76 . . . Three-Wave Analysis: Assessing Changing Levels of Over Time Political Activity at Two Year Intervals. . . . . . . . . . . . . . . . . . NOTES TO CHAPTER 2 . . . . . . . . . . . . . . . 3 DISENGAGEMENT FROM VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-60 AND 1972-76. . . . . . . . Voting . . . . . . . . . . . . . . . . . . . . . Political Campaign Activity. . . . . . . . . . . Attempts to Influence the Vote of Others . . . Giving Money to Political Candidates . . . . . Attending Political Meetings or Rallies. . . . Wearing a Campaign Button or Placing a Sticker on the Car . . . . . . . . . . . . . . . iv Page viii xii 10 12 15 16 16 18 21 25 29 33 34 35 39 4O 44 47 50 Chapter A Closer Look at the Differences Between Voting and Political Campaign Activity. . . Discussion and Conclusions. . . . . . . . . . . NOTES TO CHAPTER 3 . . . . . . . . . . . 4 THE IMPACT OF COMPOSITIONAL FACTORS ON POLITICAL DISENGAGEMENT . . . . . . . . . . . . Variations in Political Disengagement by Sex. . Voting. . . . . . . . . . . . . . . . . . . Political Campaign Activity . . . . . . . . Variations in Political Disengagement by Level of Education. . . . . . . . . . . Voting. . . . . . . . . . . . . . . . . . Political Campaign Activity . . . . . . . Discussion. . . . . . . . . . . . . . . . . . NOTES TO CHAPTER 4. . . . . . . . . . . . 5 ASSESSING THE IMPACT OF CIRCUMSTANCES AND EVENTS ON POLITICAL DISENGAGEMENT. . . . . . . Retirement and Political Disengagement. . . Institutions and Political Disengagement: The Impact of the Political Party . . Discussion and Conclusions. . . . . . . . . . . 6 CONTINUITY AND CHANGE IN POLITICAL ACTIVITY ACROSS A SERIES OF THREE SUCCESSIVE ELECT IONS I O O O O O O O O C O O O O O 0 Continuity and Change in Voting Behavior. Continuity and Change in Influence Attempts . . conClus ions 0 O O O O O O O O O O O O O O O O 0 NOTES TO CHAPTER 6. . . . . . . . . . . . . . . 7 REFLECTIONS OF THE STUDY OF OLD AGE AND POLITICAL BEHAVIOR. . . . . . . . . . . . . . Summary of Key Findings . . . . . . . . . . . Theoretical Implications and Avenues for Future Research . . . . . . . . . . . . . . . Policy Implications . . . . . . . . . . . . . . LIST OF REFERENCES 0 O O O O O O O O O O O O O O O O O O Page 53 62 66 67 68 68 78 85 86 94 102 108 . 109 . 110 117 124 . 127 130 144 157 159 160 160 164 168 . 170 APPENDIX 3 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-60 and 1972-76. 0 o o o o o 0 Appendix A3-1.--Each Political Activity Item Across Time by Age Cohort, 1956-60 . . . . . Appendix A3-2.--Each Political Activity Item Across Time by Age Cohort, 1972-76 . . . . Appendix A3-3.--Index of Political Campaign Ac- tivity Across Time by Age Cohort, 1956-60. Appendix A3—4.—-Index of Political Campaign Ac- tivity Across Time by Age Cohort, 1972-76. 4 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, SEX, AND LEVEL OF FORMAL EDUCATION, 1956-60 AND 1972-760 0 o o o o o o o o o o o 0 Appendix A4-1.--Voting Across Time by Age Cohort and Gender, 1956-60 and 1972—76 . . . Appendix A4-2.--Voting Across Time by Age Cohort and Level of Formal Education, 1956—60 and 1972-76. . . . . . . . . . . . . Appendix A4-3.--Politica1 Campaign Activity Across Time by Age Cohort and Sex, 1956-60 and 1972-76. a o o o o o o o o o o o o o 0 Appendix A4-4.—-Politica1 Campaign Activity Across Time by Age Cohort and Level of Formal Education, 1956-60 and 1972-76. . . . 5 FREQUENCY DISTRIBUTIONS FOR TWOHWAVE ANALYSIS BY AGE COHORT CHANGE IN EMPLOYMENT STATUS, AND PARTY CONTACT STATUS, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-60 and 1972-76 . . . . 6 FREQUENCY DISTRIBUTIONS AND TRANSITION MATRICES FOR THREE-WAVE ANALYSIS BY AGE COHORT, VOTING AND ATTEMPTS TO INFLUENCE THE VOTE OF OTHERS, 1956-60 and 1972-76. . . . . . . . . . . . . . Appendix A6-1.--First Order Transition Matrices for Changing Levels of Voting Across Time, by Age Cohort, 1956-60 . . . . . . . . . . . vi Page 173 174 179 184 185 186 187 191 195 199 203 209 210 APPENDICES Page 6 Appendix A6—2.--First Order Transition Matrices for Changing Levels of Voting Across Time by Age COhOI‘t, 1972-74-760 o o o o o o o o o o o o 215 Appendix A6- 3. --Second Order Transition Matrices for Changing Levels of Voting Across Time by Age Cohort, 1956- 58-60. . . . . . . . . . . . 220 Appendix A6-4.--Second Order Transition Matrices for Changing Levels of Voting Across Time by Age Cohort, 1972-74-76. . . . . . . . . . . . . 225 Appendix A6-5.--Across Cohort Tests for Equality of Second Order Processes for Continual Activ- ity in Voting, 1956-58-60 and 1972-74-76. . . . 230 Appendix A6-6.--Across Cohort Tests for Equality of Second Order Processes for Continual Inactiv- ity from Voting, 1956-58-60 and 1972-74-76. . . 232 Appendix A6-7.--First Order Transition Matrices for Changing Levels of Influence Attempts Across Time by Age Cohort, 1956-58-60 . . . . . 234 Appendix A6-8.--First Order Transition Matrices for Changing Levels of Influence Attempts Across Time by Age Cohort, 1972-74-76 . . . . . 239 Appendix A6—9.--Second Order Transition Matrices for Changing Levels of Influence Attempts Across Time by Age Cohort, 1956-58-60 . . . . . 244 Appendix A6-10.--Second Order Transition Matrices for Changing Levels of Influence Attempts Across Time by Age Cohort, 1972-74-76 . . . . . 249 Appendix A6-ll.--Across Cohort Tests for Equality of Second Order Processes for Continual Activ- ity in Influence Attempts, 1956-58-60 and 1972-74-76. . . . . . . . . . . . . . . . . . . 254 Appendix A6-12.--Across Cohort Tests for Equality of Second Order Processes for Continual In- activity from Influence Attempts, 1956-58-60 and 1972-74-76. . . . . . . . . . . . . . . . . 256 vii LIST OF TABLES Table Page 2-1 Proportion Who Were Active in 1956, 1960, 1972, and 1976 Cross-Sectional Samples. . . . . . . 3-1 Change Patterns Over Time by Age Cohort for the Act of Voting, 1956-60 and 1972-76. . . . . . . . 35 3-2 Change Patterns Over Time by Age Cohort for the Act of Attempting to Influence the Vote of Others, 1956-60 and 19 72-76 0 o o o o o o o o o o 41 3-3 Change Patterns Over Time by Age Cohort for the Act of Giving Money to Political Candidates, 1956-60 and 1972-76 0 o o o o o o o o o o o o o o 45 3-4 Change Patterns Over Time by Age Cohort for the Act of Attending Political Meetings or Rallies, 1956-60 and 1972-76 . . . . . . . . . . . . . . . 48 3-5 Change Patterns Over Time by Age Cohort for the Act of Wearing a Campaign Button or Placing a Sticker on the Car, 1956-60 and 1972-76 . . . . . 51 3-6 Percentage Who Disengaged, Among Those Who Were Active at t=1 from Voting and the Percent Who Disengaged from Political Campaign Activity by Age Cohort, 1956-60 and 1972-76 . . . . . . . . . 57 3-7 Percentage Who Were Continuously Inactive, Among Those Who Were Inactive at t=1, From Voting and the Percentage Who Were Continuously Inactive From Political Campaign Activity by Age Cohort, 1956-60 and 1972-76 . . . . . . . . . . . . . . . 59 3-8 Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged by Age Co- hort, Voting and Political Campaign Activity, 1956-60 and 1972-76 . . . . . . . . . . . . . . . 62 4-1 Sex by Age Cohort, 1956-60 and 1972-76. . . . . . . 69 4-2 Percentage Who Disengaged Among Those Who Were Active at t=1 from Voting by Age Cohort and Sex, 1956-60 and 1972-76 . . . . . . . . . . . . . . . 71 viii Table 4-4 4—5 4-7 4—8 4-9 4-10 4-11 4-12 4-13 Page Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Voting by Age Cohort and Sex, 1956-60 and 1972-76. . . . . . 73 Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Voting by Age Cohort and Sex, 1956-60 and 1972-76 . . . . . 76 Percentage Who Disengaged Among Those Who Were Active at t=1 from Political Campaign Activity by Age Cohort and Sex, 1956-60 and 1972-76 . . . . 78 Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Political Campaign Activity by Age Cohort and Sex, 1956-60 and 1972-76. . . . . . . . . . . . . . . . 81 Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Poli- tical Campaign Activity by Age Cohort and Sex, 1956-60 and 1972-76. . . . . . . . . . . . . . . . 84 Level of Education by Age Cohort, 1956-60 and 1972-760 0 o o o o o o o o o o o o o o o o o o o o 87 Percentage Who Disengaged Among Those Who Were Active at t=1 from Voting by Age Cohort and Level of Education, 1956-60 and 1972-76. . . . . . 88 Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Voting by Age Cohort and Level of Education, 1956-60 and 1972-76. . . . . . . . . . . . . . . . . . . . . . 91 Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Voting by Age Cohort and Level of Education, 1956-60 and 1972-76. . . . . . . . . . . . . . . . . . . . 93 Percentage Who Disengaged Among Those Who Were Active at t=1 from Political Campaign Activity by Age Cohort and Level of Education, 1956-60 and 1972-76. . . . . . . . . . . . . . . . . . . . 95 Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Political Campaign Activity by Age Cohort and Level of Edu- cation, 1956-60 and 1972-76. . . . . . . . . . . . 97 ix Table 4-14 5—3 5-4 5-5 5-6 6-1 6-2 6-5 Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Poli- tical Campaign Activity by Age Cohort and Level of Education, 1956-60 and 1972-76. . . . . Change in Employment Status by Age Cohort, 1956-60 and 19 72- 760 o o o o o o o o o o 0 Percentage Who Disengaged from Voting Among Oldest Cohort by Change in Employment Status, 1956-60 and 1972-76. . . . . . . . . . . . Percentage Who Disengaged from Political Cam- paigns Among Oldest Cohort by Change in Em- ployment Status, 1956-60 and 1972-76 . . . Change in Party Contact Status by Age Cohort, 1956-60 and 1972-76. . . . . . . . . . . . Percentage Who Disengaged from Voting by Age Cohort and Change in Party Contact Status, 1956-60 and 19 72-760 0 o o o o o o o o o 0 Percentage Who Disengaged from Political Campaign Activity by Age Cohort and Change in Party Con- tact Status, 1956-60 and 1972-76 . . . . . Comparison of Expected (Under First Order Assump- tion) Voting Turnout at t=3 with Actual Voting Turnout at t=3, by Age Cohort, 1956-58-60 and 1972-74’76 a o o o o o o o o o o o o o o 0 Summary of Results for Hypotheses Tests of Con- stant Transition Matrix (First Order Process) for Voting, by Age Cohort, 1956-58-60 and 19 72-74—76 0 o I o o o o I o o o o o o o 0 Summary of Results for Hypotheses Tests for First vs. Second Order Markov Processes for Voting, by Age Cohort, 1956-58—60 and 1972-74-76 . . . . Continual Activity in Voting by Age Cohort, 1956—58-60 and 1972-74-760 o o o o o o o 0 Summary of Results for Hypotheses Tests for Simi- lar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Activity in Voting, 1956-58-60 and 1972-74-76. . Page 101 112 113 115 119 121 122 133 134 137 138 141 Table Page 6-6 Continual Inactivity From Voting by Age Cohort, 1956-58-60 and 1972-74-76. . . . . . . . . . . . 142 6-7 Summary of Results for Hypotheses Tests for Similar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Inactiv- ity from Voting, 1956—58-60 and 1972-74-76 . . . . 144 6-8 Comparison of Expected (Under First Order Assump- tion) Influence Attempts at t=3 with Actual In- fluence Attempts at t=3, by Age Cohort, 1956-58-60 and 1972-74-76. . . . . . . . . . . . . 148 6-9 Summary of Results for Hypotheses Tests of Constant Transition Matrix (First Order Process) for In- fluence Attempts, by Age Cohort, 1956-58-60 and 1972-74-76 . . . . . . . . . . . . . . . . . . . . 149 6-10 Summary of Results for Hypotheses Tests for First vs. Second Order Markov Processes for Influence At- tempts, by Age Cohort, 1956-58-60 and 1972-74-76 . . . . . . . . . . . . . . . . . . . . 150 6-11 Continual Activity in Influence Attempts by Age Cohort, 1956-58-60 and 1972-74-76. . . . . . . . . 152 6-12 Summary of Results for Hypotheses Tests for Similar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Activity in Influence Attempts, 1956-58-60 and 1972-74-76 . . . . . . . . . . . . . . . . . . . . 154 6-13 Continual Inactivity in Influence Attempts by Age COhOrt, 1956-58-60 and 1972-74-760 o o o o o o o o 155 6-14 Summary of Results for Hypotheses Tests for Simi- lar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Inactivity from Influence Attempts, 1956-58-60 and 1972-74-76 . . . . . . . . . . . . . . . . . . 157 xi Figure 2-1 2-2 2-3 3-1A 3-lB 3-1C 3-1D 3-1E 3-2A 3-2B 3-2C 3-2D 3-2E LIST OF FIGURES A Typology of Possible Combinations of Changing Levels of Political Activity for Two-Wave Panel Data. 0 O O O O O O O O O O O O O O O O 0 Transition States for Participation for Three-wave AnaIYSiS o o o o o o o o o o o o o A Typology of Continuity States for Three-Wave AnaIYSiS O O O O O O O O O O O O O O O O O O O 0 Voting Across Time by Age Cohort, 1956-60 . . . Attempts to Influence the Vote of Others Across Time by Age Cohort, 1956-60 . . . . . . . . . . Giving Money to Political Candidates Across Time by Age COhOI’t, 1956-60 0 o o o o 0 Attending Political Meetings or Rallies Across Time by Age Cohort, 1956-60 . . . . . . . . Wearing a Campaign Button or Placing a Sticker on the Car Across Time by Age Cohort, 1956-60 0 o o o o o o o o o o o o . Voting Across Time by Age Cohort, 1972-76 . . . Attempts to Influence the Vote of Others Across Time by Age COhOI‘t, 1972-76 0 o o o o o o o o 0 Giving Money to Political Candidates Across Time by Age Cohort, 1972-76 . . . . . . . . . . Attending Political Meetings or Rallies Across Time by Age COhort, 1972-76 0 o o o o o o o o o Wearing a Campaign Button or Placing a Sticker on the Car Across Time by Age Cohort, 19 72-76 D O O O I O O O O O O O O O O O O O O 0 Index of Political Campaign Activity Across Time by Age Cohort, 1956-60. . . . . . . . . . . . . xii Page 22 30 31 174 175 176 177 178 179 180 181 182 183 184 Figure 3-4 4-1A 4-1B 4-1C 4-1D 4-2A 4-2B 4-2C 4-2D 4-3A 4-3B 4-3C 4-3D 4-4A 4-4B 4-4C Index of Political Campaign Activity Across Time by Age Cohort, 1972-76. . . O O O O O 0 Voting Across Time by Age Cohort, Males, 1956-60. Voting Across Time 1956-60. Voting Across Time 1972-76. Voting Across Time 1972-76. Voting Across Time 1956-60. Voting Across Time tion, 1956-600 0 o o o o o o o 0 Voting Across Time 1972-76. Voting Across Time tion, 1972-76. a o o o o o o o a Political Cohort, Political Cohort, Political Cohort, Political Cohort, Political Cohort, Political Cohort, Political Cohort, Campaign Activity Across Males, 1956-60 a o o o 0 Campaign Activity Across Females, 1956—60 0 o o 0 Campaign Activity Across Males, 1972-76 0 o o o 0 Campaign Activity Across Females, 1972-76 . . . . Campaign Activity Across Low Education, 1956-60 . Campaign Activity Across High Education, 1956-60. Campaign Activity Across Low Education, 1972-76 . xiii by Age Cohort, Females, by Age Cohort, Males, by Age Cohort, Females, by Age Cohort, High Educa- by Age Cohort, High Educa- Time by Age Time by Age Time by Age Time by Age Time by Age Time by Age Time by Age by Age Cohort, Low Education, by Age Cohort, Low Education, Page 185 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 Figure 4-4D 5-1 5-2 5-3 5-4 6-1A 6-lB 6-1C 6-1D 6-1E 6-2A 6-2B 6-2C 6-2D Political Campaign Activity Across Time by Age Cohort, High Education, 1972-76. . . . . Transition Matrices for Voting Across Time for Change in Employment Status, Groups, Ages 61 and Above, 1956-60 and 1972-76 . . . . . Transition Matrices for Political Campaign Activity Across Time for Change in Employment Status Groups, Ages 61 and Above, 1956-60 and 1972-76 0 o o o o o o o o o o o o o o Tests for Equal Transition Matrices for Voting Across Time Between Party Contact Status Groups, Ages 31-40 and 41-50, 1956-60 and 1972-'76. o o o o o o o o o o o o o o o o Tests for Equality of Transition Matrices for Political Campaign Activity Across Time Between 51-60, Party Contact Groups, Ages 21—30, 41-50, and 61 and Above, 1972-76. . . . . . . . First Order Transition Matrices for Voting Time, Ages 21-30, 1956-58-60 0 o o o o 0 First Order Transition Matrices for Voting Time, Ages 31-40, 1956-58-60 . . . . . . First Order Transition Matrices for Voting Time, Ages 41-50, 1956-58-60 . . . . . First Order Transition Matrices for Voting Time, Ages 51-60, 1956-58_60 o o o o o 0 First Order Transition Matrices for Voting Time, Ages 61 and Above, 1956-58—60. . . First Order Transition Matrices for Voting Time, Ages 21-30, 1972-74-76 . . . . . . First Order Transition Matrices for Voting Time, Ages 31-40, 1972-74-76 . . . . . . First Order Transition Matrices for Voting Time, Ages 41-50, 1972-74-76 0 o o o o 0 First Order Transition Matrices for Voting Time, Ages 51-60, 1972-74-76 . . . . . xiv Across Across Across Across Across Across Page 202 204 205 206 208 210 211 212 213 214 215 216 217 218 Figure 6-2E 6-3A 6-3B 6-3C 6-3D 6-3E 6-4A 6-4B 6-4C 6-4D 6-4E 6-5A 6-5B 6-6A 6-6B First Order Transition Matrices for Voting Across Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages Second Order Time, Ages 61 and Above, 1972-74-76. . . . . Transition Matrix 21-30, 1956-58-60 Transition Matrix 31-40, 1956-58-60 Transition Matrix for 41-50, 1956-58-60 . Transition Matrix 51-60, 1956-58-60 . Transition Matrix for Voting Across Voting Across Voting Across Voting Across Voting Across 61 and Above, 1956-58-60. . . . . Transition Matrix 21-30, 1972-74-76 Transition Matrix 31-40, 1972-74-76 Transition Matrix 41-50, 1972-74-76 Transition Matrix 51-60, 1972-74-76 . Transition Matrix for for for for Voting Across Voting Across Voting Across Voting Across Voting Across 61 and Above, 1972-74-76. . Across Cohort Tests for Equality of Second Order Processes for Continual Activity in Voting, 1956-58-60 Across Cohort Tests for Equality of Second Order Processes for Continual Activity in Voting, 1972-74-76 Across Cohort Tests for Equality of Second Order Processes for Continual Activity from Voting, 1956-58-60 Across Cohort Tests for Equality of Second Order Processes for Continual Inactivity from Voting, 1972-74-76 Page 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 Figure 6-7A 6-7B 6-7C 6-7D 6-7E 6-8A 6-88 6-8C 6-8D 6-8E 6-9A 6-9B First Order Transition Matrices for Attempts Across Time, Ages 21-30, 1956-58-60. First Order Transition Matrices for Attempts Across Time, Ages 31-40, 1956-58—60. First Order Transition Matrices for Attempts Across Time, Ages 41-50, 1956-58-60. First Order Transition Matrices for Attempts Across Time, Ages 51-60, 1956-58-60. First Order Transition Matrices for Attempts Across Time, Ages 61 and 1956-58-60. First Order Transition Matrices for Attempts Across Time, Ages 21—30, 1972-74-76. First Order Transition Matrices for Attempts Across Time, Ages 31-40, 1972-74-76. First Order Transition Matrices for Attempts Across Time, Ages 41-50, 1972-74-76. First Order Transition Matrices for Attempts Across Time, Ages 51-60, 1972-74-76. First Order Transition Matrices for Attempts Across Time, Ages 61 and 1972-74-76. Influence Influence Influence Influence Influence Above, Influence Influence Influence Influence Influence Above, Second Order Transition Matrix for Influence Attempts Across Time, Ages 21-30, 1956-58-60. Second Order Transition Matrix for Influence Attempts Across Time, Ages 31—40, 1956-58—60. xvi Page 234 235 236 237 238 239 240 241 242 243 244 245 Figure 6-9C 6-9D 6-9E 6-10A 6-10B 6-1OC 6-10D 6-10E 6-11A 6-11B 6-12A 6-128 Second Order Transition Attempts Across Time, 1956-58-‘600 o o o o 0 Second Order Transition Attempts Across Time, 1956-58-60. 0 o o o 0 Second Order Transition Attempts Across Time, 1956-58-60. . . . . . Second Order Transition Attempts Across Time, 1972-74-760 o o o o 0 Second Order Transition Attempts Across Time, 1972-74-760 o o o o 0 Second Order Transition Attempts Across Time, 1972-74-760 o o o o 0 Second Order Transition Attempts Across Time, 1972—74-760 o o o o 0 Second Order Transition Attempts Across Time, 1972-74-760 o o o o o Across Cohort Tests for Matrix for Influence Ages 41-50, Matrix for Influence Ages 51-60, Matrix for Influence Ages 61 and Above, Matrix for Influence Ages 21-30, Matrix for Influence Ages 31-40, Matrix for Influence Ages 41-50, Matrix for Influence Ages 51-60, Matrix for Influence Ages 61 and Above, Equality of Second Order Processes for Continual Activity in Influence Attempts, 1956-58-60. Across Cohort Tests for Equality of Second Order Processes for Continual Activity in Influence Attempts, 1972-74-76. Across Cohort Tests for Equality of Second Order Processes for Continual Inactivity from Influence Attempts, 1956-58-60. . . . . . . . Across Cohort Tests for Equality of Second Order Processes for Continual Inactivity from Influence Attempts, 1972-74-76. . . . . . . . xvii Page 246 247 248 249 250 251 252 253 254 255 256 257 CHAPTER 1 THE STUDY OF OLD AGE AND POLITICAL BEHAVIOR Introduction The problem of the political consequences of the transition to old age is important for several reasons. First, the number of the aged (often defined as those of age 65 and above, or retired) has been increasing. Second, the dependency ratio--the number of those of age 65 and above divided by the number of those in the work force age group (ages 18 to 64)--has been increasing. Third, the aged of today are a much more highly educated cohort than their predecessors. The proportion of those 65 and above who completed high school rose from 22 percent in 1956 to 30 percent in 1972, and movement of younger and even better educated cohorts into old age will result in an increase in education levels for the future elderly population.1 Significantly, previous studies suggest that education is an important correlate of political activity and involvement (Milbrath and Goel, 1977; Verba and Nie, 1972). Fourth, and finally, these demographic trends have the potential to establish old age-based demands as a prominent feature of the political agenda. Neal E. Cutler (1977, p. 1013) has noted that, "dependency ratios . . . become increasingly important as one con- siders the demands which the older population is likely to impress upon the rest of society. Medical care, transportation, housing, and basic economic well-being all represent needs which the sup- portive sectors of society will be called upon to finance." Although the relationship of age to political behavior and attitudes has become an important area of study in political science, sparse attention has been devoted to the political socialization of elderly Americans. As Sigel and Hoskins (1977, p. 271) note, the aged "seem far less interesting to the scholar than the pre-schooler or high schooler, and our knowledge about older adults' political socialization therefore is quite incom- plete." Probably one of the key reasons for this neglect of the aged is the prominence of disengagement theory in the literature. Proponents of this theory claim that as individuals grow older they gradually withdraw from social and political activity. If this thesis is valid, the inactivity of the elderly will diminish greatly the potential impact of the growth of the elderly popula— tion. The disengagement theory is both theoretically and intui- tively appealing. Previous studies have demonstrated that the cross-sectional realtionship between age and political participa- tion is curvilinear. Political activity levels are low for the young, increase for the middle-aged, and then decrease for the old. Many scholars argue that this pattern is the result of life-cycle effects. (See, among others, Campbell, 35Ha1., 1960, p. 493-498; IMilbrath and Goel, 1977, p. 114-116.) As Milbrath and Goel (1977, p. 215) contend, "the variation of participation with age is per- haps best explained by position in the life cycle." Intuitively, such a curvilinear relationship appears plausible. Given the onset of physical, mental, and psychological infirmities that accompany old age, it is not surprising that the elderly would withdraw from social and political activity. More- over, vis-a-vis society, the aged are treated as a marginal group who have outlived their usefulness. Especially in the economic sphere, older people in the United States lose their adult roles. For example, the assignment of low status to the elderly has been recognized by corporate policies which encourage early retirement. In addition, the elderly themselves may voluntarily and purpose- fully disengage from social and political activity so that they may devote time to leisure activities or a "righly deserved" rest and avoid the costs in time and effort which would be required for social and political activity. Analytical Discussions and Previous Tests of Disengagement Theory Analytical discussions of the disengagement theory (See, for example, Cumming and Henry, 1961; Rose, 1964; Hochschild, 1975) view disengagement as a mutual process that occurs between society and the individual. As Cumming and Henry (1961, p. 211) state, "disengagement is an inevitable process in which many of the relationships between a person and other members of society are severed, and those remaining are altered in quality." The original statement of the disengagement theory expressed in Cumming and Henry (1961) implies that age itself can affect pat- terns of activity and behavior. An attempt is made to specify age as a key independent variable. Cumming and Henry (1961, p. 23) state that the disengagement theory attempts to answer the question, "haw'much can we really tell about a person if we know only how old he is?" Specifically, as Cumming and Henry define aging, it is atmarked decline in "interpersonal activity." They claim that, aging is "an inevitable mutual withdrawal or disengagement, re- sulting in decreased interaction between the aging person and others in the social system he belongs to." (p. 14) There is a certain amount of ambiguity in the way age is defined as a concept in Cumming and Henry's seminal work. Appar- ently, the conceptual perspective of age offered by Cumming and Henry is that there are specific circumstances and events which indicate empirically the process of disengagement. They give Particular attention to the loss of certain roles among people moving into old age. According to Cumming and Henry, retirement frona the labor force and the death of a spouse are the key in- €11Gators of the process of disengagement. Finally, they emphasize that: the concept of disengagement does include ill health and attrition of income that accompany old age.2 This conception of disengagement implies that it is not old age per se but the loss of certain roles and relationships which aceOmpany old age that defines disengagement. Moreover, since disengagement is in part indicated by retirement from the labor force, the individual does not really disengage as long as he or she remains active in economic type roles. That is, people who remain economically active do not really disengage in the Cumming-Henry sense until they become mentally or physically feeble. Critics of Cumming and Henry's seminal statement of the disengagement theory have suggested a distinction should be made between personal, or individual, disengagement and "societal" dis- engagement (See N. Cutler, 1977). Personal disengagement is that which is due to the individual process of aging and includes death, the onset of psychological and physiological disabilities, and voluntary, purposeful disengagement. On the other hand, societal disengagement is that "related to the demands and constraints set by the social structure." (Maddox, 1963, p. 202) These social 1 structural factors can lead to disengagement that is independent of the individual process of aging. Corporate policies which force .an.individual to retire at a certain age or after a certain period (Bf service are a key example of this type of disengagement. Theoretically, one would not expect the impact of aging on I>c>1itica1 participation and involvement to be affected by a form (>15 societal disengagement. As Neal Cutler (1977, p. 1017) points Out, "unlike other areas of social activity, no formal or legal I‘llles require people to withdraw from political activity at a given age." Moreover, political institutions, such as political Parties and interest groups, may actively solicit and encourage Political activity, not disengagement, among the aged. Assuming older people are considered a valued resource by people within political institutions, such individuals will not exclude the elderly. Given that older people are becoming a larger segment of the American population, together with the fact that they are more dependent on governmental programs than other population segments, one would expect individuals within political institutions to solicit their support. As Cumming and Henry (1961, p. 215) con- tend, "if the individual is ready for disengagement before society is, and if he has disengaged himself 'prematurely,' then society may try to re-engage him." Studies which have examined the relationship between age and political participation have focused almost exclusively on voting turnout and psychological involvement in politics. Gergen and Back (1966) through an analysis of Gallup surveys found that aging individuals were more likely to give "no opinion," "don't know," or other neutral responses to attitudinal survey items, .and claimed that these findings provided support for the disen- gagement theory. Glenn and Crimes (1968) utilized a combined ssearies of cross-sectional Gallup surveys to demonstrate that levels ()1? reported voting turnout in presidential elections are low for tZILe young, increase for the middle-aged and decrease for the old. litatmever, when controls for level of education and sex were intro- Citlczeed the level of turnout for the aged was not significantly less t:tléir1 that for other age groups. In fact, older people of both seXes and among all educational levels were more likely than yoLiltlger people to be interested in politics. These results for political interest, moreover, were corroborated by Glenn (1969) through his analysis of opinion items from Gallup polls. Glenn found that "persons aging into advanced maturity" did not become less interested in national and international affairs. Glenn and Grimes (1968) advanced an important theoretical position to explain why older pe0ple are more attentive to public affairs and interested in politics than are those in younger age g groups. They contend that there are certain characteristics of l the adult life cycle which can account for increased attention to politics and public affairs as people grow older. Specifically, it is argued that older people are less distracted than younger people by, among other things, interpersonal relationships and earning a livelihood. Furthermore, greater attention to public affairs and interest in politics may compensate for the loss of interpersonal relationships and occupational activity. As Glenn and Crimes (1968) contend, "greater opinionation and knowledge about events covered by the mass media may be the result of, rather ‘than evidence against, one kind of societal disengagement." (p. 29) Although this compensation or distraction hypothesis may liaave utility for explaining psychological involvement in politics aarnong the aged, more recent studies have suggested that it has LIJIJnitations for accounting for the relationship between age and I)C>1;itical behavior. Wolfinger and Rosenstone (1980) specifically address this hypothesis in their analysis of the cross-sectional 1 IPGEJLzationship between voting turnout and age in the 1972 presiden- t:j-al election. Through an analysis of the 1972 Current Population L ; Survey conducted by the U.S. Bureau of the Census, they demonstrate empirically that voting turnout fails to decline with age among better-educated, male, and married older people. Theoretically, they contend that the distraction or compensation hypothesis im- plies that these types of people would be least likely to vote. They assert that "our findings provide no evidence for the pro- position that people with less free time or more 'other activit- ies and interests' are less likely to vote. Indeed, the opposite is the case. Whatever we know about the relative amount of free time available to different demographic categories, it is in- variably the case that turnout is higher in groups with_1g§§ free time." (p. 49) Nie, Verba, and Kim's (1974) five nation study is the best attempt to test the disengagement theory for forms of political activity other than voting. Using a summary measure of conven- tional political activity (excluding the act of voting) they show that participation diminishes with old age. Across all age groups, the relationship between age and political participation is cur- xzilinear. This curvilinear relationship, moreover, persists even vwfien controls for sex and education are introduced. Furthermore, Nie, Verba, and Kim demonstrate that, among the elderly, "those taint) remain in the active work force are more active politically 131163:1 those who have retired." (p. 340) The theoretical justification which Nie, Verba, and Kim F’ITCDNIide for the different levels of political activity among older Pecple who retired and those who did not is based on the disengagement thesis. They argue that "old age brings socio- logical withdrawal as individuals retire from active employment. It brings in addition physical infirmities and fatigue that lower the rate of political activity." (1974, p. 333) This theoretical perspective used by Nie, Verba, and Kim is useful in that it provides a better specification of the disengage- ment thesis by identifying circumstances and events (i.e., retire- ment) which accompany aging. As Hochschild (1975, p. 563) has argued, "it is not aging per se which determines disengagement, but a combination of factors associated with aging (for example, poor health) and other factors associated with the nature of society and one's location in it which together influence disengagement or engagement." Methodologically, the studies cited above all utilize either a single time point cross-sectional survey or combined cross-sectional surveys taken from many points in time. Major dif- ficulties with these designs include the problem of partitioning or aseparating age, cohort, and period effects, especially with the asingle time point design. Panel study designs, which measure the assime individuals at two or more points in time, are no less immune 12¢) this problem than are cross-sectional designs. However, an important advantage of the panel study is that it allows one to 11<3L1£ss-sectional design, unlike the panel study, does not allow for tirléi measurement of individual change, only aggregate change, and 10 cannot directly measure the continued engagement or eventual dis- engagement of the elderly individual. There have been a paucity of studies that used panel survey designs as a method to test the disengagement theory and none have used explicitly political variables. Babchuk and Booth (1969), based on a panel analysis of survey items for voluntary association membership, found no evidence of disengagement until the individ- uals were near the age of 70, and even at this age many panel finalists were still involved. However, only a footnote is de- voted to nonresponse and panel mortality, and no explanation is given telling how this may affect their conclusions. Stephen J. Cutler (1977) found no evidence of disengagement based on survey data on voluntary association membership taken from two separate panels. None of these studies, however, used national random panel samples, but very selective ones.4 This seriously detracts from the external validity or generalizability of these studies (See Campbell and Stanley, 1963, pp. 5-6 and pp. 16—22; Cook and Campbell, 1979, pp. 70-80) The political behavior and attitudes of those in aselective panel samples may differ systematically from the national population . Discussion The above review of previous analytical formulations and empirical tests has suggested that there is ambiguity in the notion (’15 (iisengagement itself and political disengagement in particular. There is diversity in the theoretical perspectives as well as in 11 the findings of previous studies. In Chapter 2 below we shall of fer a more rigorous conceptualization of the disengagement thesis by formulating alternative indicators of across time changing levels of political participation. One of the features previous studies of the disengagement thesis have in common is the importance they assign to identifying the circumstances and events which attend aging and contribute to disengagement. Specifically, we can identify three sets of types of circumstances and events that have an impact on the probability of disengagement from political activity. First, compositional differences, most prominently sex and level of formal education, distinguish older people from those in younger age groups. Given that females have a longer life span than males, we expect the former group to exceed the latter in number among older people. Moreover, older people have lower levels of formal education com- Pared to those in younger age groups. To the extent to which sex and level of education are related to changes in political activity ac-‘-?l'-‘Oss time, controls must be introduced for these factors in an anal—Yais of the relationship between age and change. Second, socioeconomic factors, most notably retirement from the Work force, may influence the relationship between aging and political activity. Although only one study has examined the im- Pact of retirement of political activity (Nie, Verba, and Kim, 1974) , we expect older people in the work force to in general be In Ore active in other spheres of activity, including political participation, than those who are no longer employed. 12 Third, and finally, explicitly political variables may in- fluence the likelihood of disengagement from political activity among older people. A particular candidate or prominent issue may draw large proportions of older people, relative to other age groups, into political participation. Moreover, individuals in political institutions may actively solicit and encourage political participation among the aged. In fact, given the increasing number of aged in American society, we would expect people in political institutions to attempt to mobilize older people. The goal of this study will be to examine changing levels of political participation across time among those in different birth cohorts in two separate time periods. More importantly, we will attempt to identify a set of circumstances and events which discriminate between those who disengage from politics and those Who do not for all age groups in general and for older people in Particular. Plan of Analysis The theoretical perspective and research design of the Study are the major topics addressed in Chapter 2. A signal feature of this chapter is the development of a conceptual typology of Various patterns of changing levels of political activity across time - Moreover, an argument is presented for examining the disen- gagefllent thesis in two separate time periods and for the utility of making a distinction between voting and other forms of political c ampa 1gn activity. 13 Chapter 3 examines empirical estimates of alternative in- dicators of changing political activity across age cohorts through an analysis of the University of Michigan Survey Research Center 1956-60 and 1972-76 panel studies of the American electorate. The analysis demonstrates that age-related patterns of change are dis- tinct both across time periods and for different types of political activities. In Chapter 4, we examine the impact of compositional dif- ferences across age groups on patterns of change in political pa’rticipation. Our analyses will suggest that sex and level of formal education have a dramatic impact on the relationship between age and political disengagement. In fact, we will show that among older people with relatively high levels of education the proba- bility of disengagement from political activity over time ap- Pl‘oaches zero. Chapter 5 focuses on an examination of the circumstances and events which discriminate between people who disengage from Political activity and those that do not among both individuals in get181‘al and older people in particular. For older people in par- ticUlar we will assess the dynamic impact of retirement from the labor force on disengagement from political activity, and fully emtI)l0:Lt the panel nature of our design. For individuals in getiel‘al, we will consider the dynamic impact of political party Contact on disengagement from politics. Thus, we capture both ind ivIidual and institutional circumstances and events which are p OS1th to affect political disengagement. 14 Chapter 6 will assess the utility of testing the disen- gagement theory with panel data from three points in time. We will explore in detail the notion of continuous activity or non-activity in this chapter, and directly pit the disengagement thesis against an alternative argument which contends that the nature and amount of an older person's political participation is primarily an exten- sion of a pattern begun in middle age or earlier. Finally, Chapter 7 will summarize the major findings of c>ur study. Of particular importance here will be a discussion of ‘tlie theoretical implications of our results for the study of localitical participation in general and aging and political parti- (2:113ation in particular. Moreover, an assessment will be made of t:11e2 policy consequences of disengagement theory and political I>é£:?t:icipation, or lack thereof, among the aged. NOTES TO CHAPTER 1 1These percentages were obtained through my analyses of the University of Michigan Survey Research Center presidential election surveys of the American electorate. 2Cumming and Henry excluded any individual in their sample who was in "poor" physical or mental health and those who had suf- ficient income for "independence." That is, their sample was over- represented by older people with a relatively good health and financial situation. Compositional factors, such as sex and level of formal 3 eeciucation, may be distributed unequally across people of different The panel design has the ability to automatically control ages. for these factors since the same specific individuals are surveyed at two or more points in time. The combined cross-sectional design does not allow for this automatic control of these factors since it is very unlikely that any of the individuals surveyed at Time 1 will be in the sample at Time 2. Babchuk and Booth (1969) used a sample taken in 1961 from the population of one midwestern state. Stephen J. Cutler (1977) used two samples: one from a Duke University study on aging which Was confined to the immediate geographical area surrounding the S Chool, and one from a single suburb of Cleveland, Ohio. This 8‘-—1't>1.1‘r:ban sample, moreover, contained only the names of members of a local health insurance association. 15 CHAPTER 2 TESTING DISENGAGEMENT THEORY AS AN EXPLANATION OF POLITICAL INACTIVITY—-CONSIDERATIONS OF THEORY, DESIGN, AND MEASUREMENT In this chapter we will develop an analytical framework for testing disengagement theory as an explanation of changing levels of political activity across time. The chapter will focus on three issues. First, a distinction will be made between voting and other forms of political campaign activity. Second, the utility of testing the disengagement theory in two separate time periods will be examined. Third, and most importantly, we will develop alter- native indicators of over time changing levels of political activ- ity based on conceptual typologies for two- and three-wave panel S til—ldy designs. Political Participation Previous research on political participation has suggested that participation can be viewed either unidimensionally or multi- dimensionally. The early work of Milbrath (1965) claimed that participation is unidimensional since the various forms of par- t icipation can be ordered on a continuum of level of intensity or tiegl‘ee of effort required for the act. Later studies (Verba and Nie, 1972; Verba, Nie, and Kim, 1971, 1978; Nie, Verba, and Kim, 1 974; Milbrath and Goel, 1977), on the other hand, contend that 16 17 there are different dimensions of political participation. The best conceptual and empirical demonstration that participation is multidimensional is offered in Verba, Nie and Kim's (1978) Partic- ipation and Political Equality. In this cross-national study, they provide a detailed methodological analysis of the constructs used to formulate the various modes or dimensions, and the amount of conflict involved in the participatory act and the ease of the act for the citizen, among other things, are used to specify each mode. In this analysis we shall consider two distinct modes of political participation--voting and political campaign activity. Verba, Nie, and Kim (1971, 1978) have demonstrated with cross-sec- tional data that voting is a separate mode of participation and conceptually distinct from political campaign activity. They show that two distinct dimensions emerge empirically when voting and other forms of campaign activity are factor analyzed. Moreover, they present results from several other studies which have found voting and political campaign activity to be two different dimen- sions of political participation (See Appendix A, pp. 331-339, Verba, Nie, and Kim, 1978). The five indicators of political participation to be studied in this analysis are voting, giving money to political candidates, wearing a campaign button or placing a sticker on the Car, attending political meetings or rallies, and attempts to in- fTLuence the vote of others.1 Like Verba, Nie, and Kim (1978), we shall view voting as a separate dimension of activity from the Other four campaign acts. First, overall levels of participation 18 are markedly higher for voting than they are for influence at— tempts (See Table 2—1 below). Second, and more importantly, we will demonstrate (See Chapter 3 below) that across time patterns of change are markedly different for voting than they are for non-voting forms of political participation, including influence attempts. TABLE 2-l.--Proportion Who Were Active in 1956, 1960, 1972, and 1976 Cross-Sectional Samples. Change Change From From 1956 1960 1956-60 1972 1976 1972-76 A. Vote .73 .79 +.06 .73 .72 —.01 B. Influence .28 .34 +.O6 .32 .37 +.05 C. Button .16 .21 +.05 .14 .08 -.06 D. Meetings .07 .08 +.Ol .09 .06 -.03 E. Money .10 .12 +.02 .10 .16 +.06 NOTE: The numbers upon which these percentages are based vary slightly from item to item. The lowest number upon which they are based is as follows: 1956, 1961; 1960, 1822; 1972, 2188; 1976, 2394 (weighted N). The 1956-60 and 1972-76 Time Periods This analysis is the first attempt to test the disengage- lnent theory in two time periods with national multistage probability Ilanel samples. As we know from studies on trends in political ‘Participation, there are periods when participation raises sharply in response to, among other things, particular political events, 19 whereas in others it may sharply decline. This is particularly important for testing this theory since the level of political activity may be very different in each time period. Moreover, changing levels of political activity may show different dynamics over the two time periods. Table 2—1 shows the proportion of the population who partic- ipated in each type of political activity for cross-sectional, na- tional random samples taken in 1956, 1960, 1972, and 1976. For three of the five political activities the aggregate proportion who participated increased by at least five percent between 1956 and 1960. However, over the 1972-76 period the aggregate proportion who were politically active decreased for three of the five activ- ities. Moreover, the SRC data for reported voting turnout between 1956-60 and 1972-76 are parallel to real world trends observed from election statistics. According to Bureau of Census estimates, voting turnout rose from 59.3 percent in 1956 to 62.8 percent in 1960, but fell from 55.5 percent in 1972 to 54.4 percent in 1976 (Statistical Abstract of the United States). Given these distinctively different aggregate patterns of changing levels of political activity for the two time periods, it is possible to assess the consequences of such patterns for the validity of the disengagement theory. The predictions of the theory may be accurate for certain time periods but not for others. Tunis is an issue we shall return to in more detail in the analysis Of the panel samples. 20 The basic problem with measuring changing levels of polit- ical activity across time is that certain circumstances and events which occur between the initial and final points of observation may influence the patterns of change (See Campbell and Stanley, 1963; Cook and Campbell, 1979). That is, we may find patterns of polit- ical disengagement that have little to do with aging itself, but rather are the result of other factors, such as a particular candi— date, an important issue, or a particular political event. More- over, these period effects could impact differently upon different age groups. This problem is even more prominent when assessing change in two separate time periods as we do here. It is difficult to identify any particular election spe- cific events in both the 1956-60 and 1972-76 time periods that would have differentially effected the patterns of over time change in political participation among any particular age group, in- cluding the elderly. Across periods, however, there are indeed differences in certain circumstances that may affect age-related patterns of political disengagement. Most notably, the electorate in general and older people in particular were better educated in 1972-76 than they were in 1956-60. Moreover, in the 1972—76 period, almost one-third of the population were contacted by a political party, whereas in 1956-60 only one-fifth were contacted.2 We:account for the impact of changing levels of education in (Shapter 4, and for the impact of institutional change--specifi- cally party contact efforts--in Chapter 5. 21 The panel nature of our design provides for the separation of effects due to aging from compositional and other effects. That is, we can control for the impact of compositional factors, such as level of formal education, and political factors, such as being contacted by a political party, on the relationship between aging and political disengagement. This will allow us to assess the importance of changes between 1956-60 and 1972-76 in certain socio- logical and political variables and their relationship to disengage— ment from political activity (See Chapters 4 and 5 below). Panel Study Analysis of Changing Levels of Political Activity Panel study analysis is uniquely appropriate for studying the durability or stability of individual political behavior over time. It provides the design and data base for an empirical assess- ment of the predictions of the disengagement theory, and allows one to directly measure the continued engagement or eventual disen- gagement of the elderly individual. Specifically, the panel nature of the research design and data can be fully exploited by focusing on change in political behavior as a dependent variable. The par- ticular nature of the dependent variable can be specified by a conceptual typology which classifies the over time changing levels of political activity for individuals. Figure 2—1 shows that four distinct, but not independent, patterns can be discerned by (:ross-tabulating activity or nonactivity at time one with that at time two. 22 Time 1 Did Not Participated Participate . Continuously Newly Engaged Partic1pated Engaged (B) (A) Time 2 Did Not Disengaged Continuously Participate (C) Inactive (D) Figure 2-l.-—A Typology of Possible Combinations of Changing Levels of Political Activity for Two-Wave Panel Data. 23 Given that the disengagement theory seeks to explain and predict the changing levels of political activity of those moving into old age, some type of a baseline is needed to compare the elderly respondents with all others. Thus, the 1956-60 panel re— spondents will be partitioned into ten year age groups according to their age in 1956; and the 1972-76 panel respondents according to their age in 1972. This analysis will focus on six age cohorts for each panel, and the cohorts range from ages 21-30 to ages 61 and above. Throughout the analysis we shall refer to the elderly as that cohort which includes individuals of ages 61 and above. This particular division was chosen for two reasons. First, all of the individuals in this cohort will have reached age 65 by the second time point of measurement, and thus will have attained the conven- tional definition of old age. Second, this particular division of the oldest cohort allows for sufficient cases both for examining the zero order relationships between age cohort membership and disengagement from political activity and, moreover, for examining the impact of compositional, political, and situational factors (sex, level of formal education, party contact, and retirement from the labor force) on the relationship between age cohort membership and political disengagement. For both time periods, the conceptual typology will be estimated for all age cohorts. This allows for a test of the con- tention that those individuals in the oldest age group are more likely to disengage from politics than are those in younger age 24 groups. Moreover, it provides for the analysis of change among individuals within each age cohort. One problem with panel analysis is that assumptions and hypotheses must be made about the causal lag or the amount of time required for the change of theoretical interest to occur (See Davis, 1978). Although the time interval of our design is insufficient if disengagement takes longer than four years, this analysis in- directly accounts for this problem, first of all, by focusing on activities that are explicitly linked with political campaigns. The election campaign is an event that occurs at relatively fixed points in time, and participation in campaigns is subject to these periods. Therefore, the first part of our analysis of change (Chapters 3, 4, and 5) will focus on the 1956 and 1960 presidential elections employing the 1956-60 panel, and on the 1972 and 1976 presidential elections utilizing the 1972-76 panel. Second, for the 1956-58-60 and 1972-74-76 panels, we will analyze over time activity change for the acts of voting and attempting to influence the vote of others for three elections over two year intervals.3 This analysis, to be discussed further in Chapter 6, will extend the conceptual typology and allow for a more detailed, although indirect, examination of the time interval required to test the predictions of disengagement theory. Third, and most importantly, from the disengagement theory and the conceptual typology we will derive hypotheses that posit certain relationships between age cohort membership and patterns of over time changing levels of political activity. An integral part of this task is to derive 25 indicators of political disengagement from the typology, and to these we now turn. Two-Wave Analysis: Assessing Changing Levels of Over Time Political Activity in Presidential Elections, 1956-60 and 1972-76 There are a variety of distinct indicators that can be deve10ped from this typology (See Figure 2-1), but only a subset can address the issues posed by the disengagement thesis. For example, four indicators could be deve10ped by considering the source of individual political activity at time 2. If the question of in- terest is how many of those who participated at time 2 were non-par- ticipants at time 1, the indicator of interest would be B/(A + B). On the other hand, if the issue of interest is how many of those who did not participate at time 2 were participant at time 1, the indi- cator would be C/(C + D). And, finally, if we wish to assess how many of those who did not participate at time 2 were non-partici- pants at time 1, the indicator of interest would be D/(C + D). Additionally, four indicators can be deve10ped from the typology by considering the proportion of the total electorate who manifested each of the four over time states. These indicators can be formed simply by percentaging through the entire table. First, if the issue of concern is how many of the entire electorate were active at both time 1 and time 2, the indicator of interest is A/(A + B + C + D). Second, if we are concerned with how many people in the entire electorate started up political activity after abstaining from participation in a previous election, the indicator 26 is B/(A + B + C + D). Third, the indicator C/(A + B + C + D) assesses how many of the entire electorate dropped out of partici- pation after participating in a previous election. Fourth, and finally, D/(A + B + C + D) considers how many of the entire electorate completely abstained from participating at two time points. We have discussed some of the wide variety of indicators of change across time that can be derived from the typology in Figure 2-1. We are, however, only concerned with those particular in- dicators that allow for an assessment of the disengagement thesis. These specific indicators will all focus on the destination states at time 2 for those in different states at time 1. In what follows, we develop three different indicators which allow for both a direct and an indirect assessment of the theory. This analysis will test hypotheses for three different in- dicators of political disengagement derived from the typology. In general, all of the indicators are created by examining the pat— terns of change within each age cohort in each time period and require the construction and estimation of a separate score for each cohort. Our hypotheses, for the most part, involve compari- sons in each period of the various indicators across age cohorts. For example, the first, and most direct, indicator of political disengagement is to examine (within each cohort) the actual extent of withdrawal from political activity among those initially active. According to the typology, the proportion who disengaged or with- drew from political activity is simply category C divided by the 27 sum of categories A and C. Given the predictions of the disengage- ment theory, we hypothesize that C/(A + C) should be larger for the oldest cohort (ages 61 and above) than for all others (cohorts of age 21-30 through age 51-60, both inclusive). Put simply, we ex— pect individuals in the oldest cohort to be more likely to drop out of political activity than individuals in younger cohorts. Our first indicator does not account for people who may have already drOpped out of political activity before the initial point of measurement. That is, it is possible to contend that political activity is an intermittent type of behavior for individ- uals—-that is, they may engage, disengage, and re-engage throughout the life cycle. Given the predictions specified by the disengage- ment theory, one would expect individuals in the older cohorts to remain continuously inactive over time. In other words, once they become inactive (even if before the first time point of measurement) they are expected to remain that way. Moreover, it would be very unlikely for older people to start up political activity after a period of inactivity. The second indicator of political disengage— ment is thus the proportion of an age cohort who were continuously inactive, or from the typology D/(B + D). The anticipation of the disengagement thesis is that the magnitude of D/(B + D) should be markedly greater for the elderly than that for all other cohorts. A third indicator of political disengagement is the assess- umnt of the difference between the proportion who disengaged and the proportion who re-engaged within each age cohort. According to the typology, the difference for each cohort can be expressed as: 28 C/(A + C) — B/(B + D) Our hypothesis is that the imbalance between the disengaged and the newly engaged should be greatest, and in favor of the former group, for the most elderly cohort. Positive values of this difference indicate that the proportion of an age cohort who dis— engaged is greater than the proportion who newly engaged; and, negative values, that the proportion who newly engaged exceeded the proportion who disengaged. These three indicators, together with the specific hypoth- eses about the magnitude and sign of their values across age co- horts, provide the means to exploit the panel analysis of over time changing levels of political activity and allow for an indirect way of solving the causal lag problem. The first indicator is the best direct measure of disengagement since it allows for the comparison of the relative dropout rates across cohorts. However, this in- dicator fails to consider people who may have dropped out of politics in a previous election and who may never again become active. This group is captured by our second indicator of dis- engagement by comparing across cohorts the proportion who were continuously disengaged. Finally, the difference indicator allows for the assessment of the balance in proportions between those with opposing patterns of change in activity levels. 29 Three-Wave Analysis: Assessing Changing Levels of Over Time Political Activity at Two Year Intervals Information about reported voting behavior and attempts to influence the vote of others is available at all three time points of both the 1956-58-60 and 1972-74-76 panels. The three wave data allow for a more detailed analysis of changing levels of political activity as an intermittent type of behavior and as a continuous or discontinuous form of behavior. In this part of the analysis (See Chapter 6), we will consider the hypothesis that older people are less continuous, and more discontinuous, in their political activ- ity than are those in younger age groups. The full array of transition states for participation over three time points is shown in Figure 2-2. As with our previous typology, this array will be estimated for each birth cohort. Our major task is to consider activity levels across cohorts for the first and fourth rows (cells I, II, VII, and VIII). These rows include all of those individuals who did not change their activity between time 1 and time 2. Here the comparison focuses on cohort differences in the proportion who remain either continuously active or continuously inactive at time 3. Specifically, Chapter 6 em- ploys the probability model of a Markov process to test the argu- 'ments of the disengagement theory. According to the predictions of the disengagement theory, ‘we expect older people to have higher levels of continual inactivity compared to those in younger age groups over a series of elections. .A typology derived from the eight celled array assesses the 30 Time-3 Did Not Time—l Time—2 Participate Participate Participated Participated I II Did Not Participate Participated III IV Did Not Participated Participate V VI Did Not Did Not Participate Participate VII VIII 2-2.——Transition States for Participation for Three-Wave Analysis. 31 participatory status at time 3 among those who were either continu- ously inactive or continuously active between time 1 and time 2. This typology is shown in Figure 2-3. Active Time 3 Participa- tion Status Inactive Figure 2-3.--A Typology Continuity Status Between Time 1 and Time 2 Active-Active Inactive-Inactive of Continuity States for Three-Wave Analysis. 32 Our hypotheses here predict that individuals in the oldest cohort will have a proportionately larger value in the D cell and a proportionately smaller value in the A cell as compared to those in younger cohorts. According to the typology, A/(A + C) for the oldest cohort should be less than that for the other age groups. Moreover, D/(B + D) for the oldest group should be greater than that for all other cohorts. Our analyses of both two- and three-wave changing levels of political activity will allow for a rigorous and creative test of the disengagement theory. The signal feature of this analysis is its ability to both directly and indirectly assess the predictions of the theory through the construction of typologies from which hypotheses about change across cohorts can be derived. We now turn to the empirical estimates of these indicators of change. NOTES TO CHAPTER 2 1These particular five political activity items were chosen, first of all, on the grounds that previous literature (most notably, Verba, Nie, and Kim, 1978) has demonstrated that they form two dis— tinct modes of participation. Second, they comprise the only set that was comparable at each time point in both the 1956-60 Univer- sity of Michigan Survey Research Center panel studies of the American electorate. 2My own analyses of the University of Michigan Survey Re- search Center studies of the American electorate show that 16 per- cent of the population were contacted by a party in 1956, 23 per- cent in 1960, 32 percent in 1972, and 33 percent in 1976. The im- pact of this increase between 1956-60 and 1972-76 in the percentage of the electorate who were contacted by a party on disengagement will be examined in detail in Chapter 5 below. 3Unfortunately, we cannot perform this analysis for all five activities since the 1958 wave of the 1956-58-60 panel in- cluded only two of the activity items employed in this analysis-- the vote item and the attempts to influence the vote of other item. 33 CHAPTER 3 DISENGAGEMENT FROM VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-60 AND 1972-76 The above discussion demonstrated that the disengagement theory should be tested with a variety of alternative, but not strictly independent, indicators of over time changing levels of political activity. It is the task of this chapter to empiri- cally estimate these indicators for the acts of voting and other forms of political campaign activity through an analysis of the University of Michigan Survey Research Center 1956-60 and 1972-76 panel studies of the American electorate. We anticipate from our discussion in Chapter 2 above that the patterns of across time change for the two sets of participatory acts should be different both across age cohorts and time periods. Our results below will demonstrate that in 1956-60 there was modest support for old age disengagement from both voting and political campaign activity. In 1972-76, on the other hand, our analyses suggest a different pattern. That is, the old in 1972—76 were slightly more likely than those most in other age groups to disengage from voting but no more likely to do so for political campaign activity. The final part of this chapter attempts to develop explanations for these divergent patterns across the two time periods.1 34 35 Voting The patterns of over time changing levels of voting for all age groups are distinctively different in the two time periods. Table 3-1A shows the results of all three of the two-wave indi- cators of political disengagement by age cohort for the act of voting. Among all age groups disengagement from voting was more prevalent in 1972-76 than it was in 1956-60. In 1956-60, 4.2 per- cent of the electorate who were initially active in voting dis- engaged as compared to 9.2 percent in 1972-76 (See Table 3-1A). TABLE 3-1.--Change Patterns Over Time by Age Cohort for the Act of Voting, 1956-60 and 1972-76. R for Age Group in All Age Years:a 21-30 31-40 41—50 51-60 61+ Groups A. Percentage who disengaged, among those who were active at t=1: 1956-60 5.0% 5.4% 2.7% 2.8% 5.0% 4.2% (N) (180) (260) (224) (141) (140) (945) 1972-76 15.1% 5.6% 8.4% 5.5% 9.9% 9.2% (N) (218) (177) (190) (163) (161) (909) B. Percentage continuously inactive, among those who were in- active at t=1: 1956-60 52.9% 53.6% 54.0% 61.8% 71.1% 56.6% (N) (104) ( 97) ( 63) ( 34> ( 45) (343) 1972-76 44.9% 59.6% 65.5% 74.4% 72.9% 61.1% (N) ( 78) ( 47) ( 29) ( 39) ( 59) (252) C. Difference between the percentage of disengagers and the percentage of new engagers: 1956-60 -42.1% -4l.0% -43.3% -35.4% -23.9% -39.2% 1972-76 -40.0% -34.8% -26.1% -20.1% -17.2% -29.5% aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 36 The greater incidence of disengagement from voting in the 1972-76 period was anticipated from our analysis of the cross-sec- tional samples in Chapter 2. Part of this difference between the two time periods can probably be accounted for by changes other than disengagement among older people. Table 3-1A shows that in 1956-60 only 5.0 percent of those in the most elderly group (ages 61 and above) disengaged from voting. Across all age groups in 1956-60, the percentage who disengaged from voting is flat across the 21-30 and 31-40 year old age groups, decreases about 2 percent for the 41-50 and 51-60 year old groups, and then increases again to 5.0 percent for those in the oldest group. The pattern of disengagement from voting (See Table 3-1A) over the 1972-76 period also shows that the young (ages 21-30) and the old (ages 61 and over) are the most likely age groups to dis- engage. The most marked differences between age cohorts are those between the youngest group and all others. However, the most pro- nounced percentage increase in disengagement between the two time periods was that for the youngest age cohort. Only 5.0 percent of the young disengaged from voting between 1956 and 1960 as Opposed to 15.1 percent in 1972-76. The percentage of the oldest group who disengaged from voting over the 1972—76 period is almost twice that for the 31-40 and 51-60 year old age groups, but barely exceeds the percentage who disengaged among the 41-50 year old group. However, among the 31—40 through the oldest age groups, it is those in the oldest group who are most likely to disengage from voting. The 37 difference between the 1972-76 and 1956-60 time periods is marked, moreover, for those in the oldest age cohort. In 1956-60, older people were not unique in comparison to those in other age groups in their disengagement from voting, but in 1972—76 they were one of the most likely groups to drop out of voting. However, even in 1972-76, the proportionate level of disengagement from voting among the aged is indeed not massive compared to that for those in other cohorts. Table 3-1B shows the percentage of those initially inactive who were continuously inactive from voting by age cohort in each time period. Except for the youngest cohort, continual inactivity was more frequent in 1972-76 than it was in 1956-60. In 1956-60, 56.6 percent of those inactive in 1956 were inactive in 1960, whereas in 1972-76, 61.1 percent of the 1972 inactives remained inactive in 1976. This is consistent with the results of the first indicator of political disengagement. Continual non-voting in 1956-60 is much more prevalent among the oldest group than it is among the other age cohorts. Across age cohorts the distribution of the percentage who were continual non-voters rises from the youngest through the oldest age groups and provides support for the predictions of the disengage- ment thesis. In 1972-76, moreover, the distribution of the per- centage who were continual non-voters slopes upward from the youngest through the oldest age cohorts. The most pronounced upward shift in continual inactivity is the almost 15 percentage point difference between the young (ages 21-30) and early 38 middle-aged (ages 31-40) cohorts. The pattern of continual non-vot- ing levels off across the middle-aged groups (ages 31-40 and 41-50), increases rather sharply (about 10 percent) for the 51-60 year old group, and finally flattens out for the oldest group. In general, the results of Table 3-lB suggest that old age continual non—voting is characteristic of both time periods. In 1972—76, however, the elderly were not unique in their relatively high levels of continual inactivity. High levels of continual in— activity are also characteristic of those in late middle age (ages 51-60). Finally, Table 3-lC shows the distribution by age cohort of the differences between the percentage who disengaged from voting and the percentage who voted after abstaining in a previous elec- tion in each time period. For all age groups in each time period the percentage of new engagers is much greater than the percentage of disengagers. In 1956—60, this difference for the entire set of age groups was a negative 39.2 percent, and in 1972-76, a negative 29.5 percent. Once again, these results demonstrate that disen- gagement from voting was proportionately more prevalent in 1972—76 than it was in 1956-60. All three indicators of political dis- engagement support this conclusion. There are some marked differences on this indicator across age cohorts in the 1956-60 period (See Table 3-1C). The pattern of the differences decreases from the youngest through the oldest age groups. In the 1972-76 period, moreover, the difference between the percentage of disengagers and the percentage of new 39 engagers in the oldest age group is less than that for all other age cohorts. Among those in the oldest group the percentage of disengagers exceeds the percentage of new engagers by only 17.2 percent. On the other hand, among the youngest age cohort there is a very wide gap (a difference of 40.0 percent) between the dis- engagers and the new engagers. The percentage difference declines steadily from the youngest through the oldest group in 1972-76. Once again, these results suggest that disengagement from voting is more likely for those in the oldest group as compared to the other age cohorts in the 1972-76 period. However, we can also see that the difference is very small for those in the late middle age (ages 51—60) in 1972-76. The results of all three indicators of political disengage— ment suggest that the patterns of over time changing levels of voting across age cohorts are very similar in the two periods. Although overall levels of disengagement from voting were pro- portionately more frequent in 1972-76 than in 1956-60, older people were in general one of the age groups most likely to dis- engage in each time period. However, the elderly are not that unique in their disengagement from voting in each period. On all three indicators the differences between the aged and all remain- ing groups are not marked. Political Campaign Activity The patterns of change over time for other forms of political campaign activity are quite different from those for 4O voting. We shall first examine the relationship between age co- hort membership and changing levels of political participation separately for each activity and then compare the trends in voting to a summary measure of political campaign activity. This analy- sis will demonstrate that in the 1956-60 period older people were somewhat more likely than those in other age groups to disengage from both voting and political campaign activity. In 1972-76, on the other hand, our analyses will demonstrate that older people are less likely than those in other age cohorts to disengage from political campaign activity, but are slightly more likely to disengage from voting. Attempts to Influence the Vote of Others Table 3-2 shows the results for the three indicators of political disengagement for the act of attempting to influence the vote of others. As Table 3-2A shows, the extent of disengage— ment from vote influence attempts is markedly greater than that for voting in both time periods. In 1956-60, 45.6 percent of the entire set of age cohorts who were initially active disengaged from attempts to influence the vote of others, and in 1972-76, 39.1 percent of those initially active disengaged. In 1956-60, there are no distinct differences between age cohorts in the percentage who disengaged from vote influence at- tempts (See Table 3-2A). The extent of disengagement is relatively low for those in the 21-30 and 31-40 year old groups (43.4 and 43.5 percent respectively), rises to about 50 percent for the 41-50 41 TABLE 3-2.--Change Patterns Over Time by Age Cohort for the Act of Attempting to Influence the Vote of Others, 1956-60 and 1972-76. R for Age Group in All Age Years:a 21-30 31-40 41-50 51-60 61+ Groups A. Percentage who disengaged, among those who were active at t=1: 1956-60 43.4% 43.5% 48.7% 45.1% 49.2% 45.6% (N) ( 83) (108) ( 78) ( 51) ( 59) (379) 1972-76 40.3% 41.0% 40.4% 37.7% 33.9% 39.1% (N) (119) ( 83) ( 89) ( 69) ( 59) (419) B. Percentage continuously inactive, among those who were inac- tive at t=1: 1956-60 74.5% 76.3% 74.6% 75.8% 77.3% 75.6% (N) (200) (249) (209) (124) (128) (910) 1972-76 69.5% 80.9% 73.8% 74.4% 78.7% 75.3% (N) (177) (141) (130) (133) (160) (741) C. Difference between the percentage of disengagers and the percentage of new engagers: 1956-60 +17.9% +19.8% +23.3% +20.9% +26.5% +21.2% 1972-76 + 9.8% +21.9% +14.2% +12.l% +12.7% +14.4% aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. year old group, and levels off for the two oldest groups. The relationship between age cohort membership and influence attempts in 1972-76, on the other hand, shows that the elderly are the least likely age group to disengage. The pattern of the percentage who disengaged from influence attempts is relatively high and flat across the first three age groups but declines to a low of 33.9 percent for the oldest group. Older people were somewhat more 42 likely than those in other age cohorts to disengage from influence attempts in 1956-60, but slightly less likely to do so in 1972-76. Table 3-2B shows the percentage of those initially inactive who were continuously inactive from influence attempts by age co- hort in both time periods. These results also show that continual inactivity from influence attempts is proportionately greater in both time periods than that from voting. Among all age groups in 1956-60, 75.6 percent of the 1956 inactives were continuously inactive from attempts to influence the vote of others, and in 1972-76, 75.3 percent of the 1972 inactives were continuously in- active. These percentages are much greater than those for voting in each time period. There are no marked differences across age cohorts in the percentage who were continuously inactive from influence attempts in the 1956-60 period. About three-quarters of the initially in- active in each age cohort were continuously inactive from this act in 1956-60. In 1972-76, on the other hand, the pattern of the percentages is much more uneven across age cohorts. Although 78.7 percent of those in the oldest group were continually inactive, this percentage is not much different from those for the three middle-aged groups (ages 31-40, 41-50, and 51-60). The lowest percentage is that for the youngest age group. These results suggest that in neither period are the aged more likely than those in other age groups to be continually inactive from the act of attempting to influence the vote of others. 43 Table 3-2C shows the distribution by age cohort of the dif- ference between the percentage of disengagers and the percentage of new engagers. Among all age groups in each time period the per- centage of disengagers exceeded that of new engagers. In the 1956-60 period, this difference was a positive 21.2 percent, and in 1972-76, a positive 14.4 percent. The results of this indicator for influence attempts are once again distinct from those for voting in both time periods. In the 1956-60 period, it is among individuals in the old- est group where the balance between disengagers and new engagers is greater (and in favor of the disengagers) than that for all other age cohorts. The distribution of these differences in 1956—60 is flat across the 21-30 through 51—60 year old groups but increases somewhat for the oldest group to a positive 26.5 per- cent. In 1972-76, on the other hand, the difference of +12.7 per— cent for the oldest age group is exceeded in magnitude by the 31-40 and 41-50 year old groups. In fact, in the 1972—76 period, the value of this difference declines in magnitude across the 31-40 year old cohort through the oldest group. Moreover, across time periods, the value of this difference decreased in magnitude for all age cohorts (except the 31-40 year old group), and the largest decrease was that for the oldest group. The decline in the difference (that is, in positive magnitude) for the oldest age cohort was over 12 percent across time periods. From 1956-60 to 1972-76, older people were becoming less likely than those in other age groups to disengage from influence attempts while at 44 the same time they were becoming more likely to disengage from voting. Giving Money to Political Candidates Table 3-3 shows the results of the three indicators of political disengagement for the act of giving money to political candidates. As was the case for the act of attempting to influ- ence the vote of others, the extent of disengagement for the act of giving money is significantly greater than that for voting. About 50 percent of those in all age groups combined who were initially active in 1956 disengaged from the act of giving money in 1956-60, and 51.4 percent of those who were initially active in 1972 did so in 1972-76 (See Table 3—3A). Although overall levels of disengagement were similar in each period, there are differences between the periods in the way disengagement relates to age cohort membership for the act of giving money to political candidates. As Table 3-3A shows, in 1956-60, the greatest proportionate incidence of disengagement is among those in the 21-30 and 31-40 year old groups. The percentage drops by about 10 percent to 51.2 percent for the 41-50 year old group, declines another 10 percent to 40.7 percent for the 51-60 year old group, and then increases again to about 50 percent for those in the oldest group. In 1972-76, on the other hand, it is those in the youngest and oldest age groups who are most likely to disengage from the act of giving money. About 60 percent of the individuals in the youngest and oldest groups gave money to a 45 candidate in 1972 but not in 1976. Among those in the three middle-aged groups (ages 31-40, 41—50, and 51-60), however, the percentage never exceeds 50 percent. TABLE 3-3.--Change Patterns Over Time by Age Cohort for the Act of Giving Money to Political Candicates, 1956-60 and 1972—76. i for Age Group in All Age Years:3 21-30 31—40 41-50 51-60 61+ Groups A. Percentage who disengaged, among those who were active at t=1: 1956-60 55.0% 60.0% 51.2% 40.7% 50.0% 51.6% (N) ( 20) ( 30) ( 41) ( 27) ( 10) (128) 1972-76 63.3% 46.7% 47.1% 42.3% 59.1% 51.4% (N) ( 30) ( 30) ( 34) ( 26) ( 22) (142) B. Percentage continuously inactive, among those who were in- active at t=1: 1956—60 91.7% 92.9% 90.7% 91.2% 96.6% 92.5% (N) (264) (325) (246) (147) (177) (1159) 1972-76 97.0% 92.1% 90.4% 95.5% 95.3% 94.3% (N) (233) (178) (167) (157) (190) (925) C. Difference between the percentage of disengagers and the per- centage of new engagers: 1956-60 +46.7% +52.9% +4l.9% +3l.9% +46.6% +44.l% 1972-76 +60.3% +38.8% +37.5% +37.8% +54.4% +45.7% aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Table 3-3B shows that the pattern of the percentages of those initially inactive who were continuously inactive from the act of giving money to political candidates is flat across cohorts in both time periods. In fact, over 90 percent of the initially 46 inactive in all age groups combined were continuously inactive from this act in each time period. In 1956-60, there is a small in- crease in this percentage when the oldest group is compared to all others. On the other hand, in 1972-76, those in the oldest age group are no more likely than those in other cohorts to be con- tinuously inactive. The age group most likely to be continuously inactive over the 1972-76 period is the youngest one. The distribution of the differences across age cohorts between the percentage of disengagers and the percentage of new engagers from the act of giving money is shown in Table 3-3C. Once again, for all age cohorts, both separately and combined, this difference is large and positive in both time periods. In the 1956-60 period, this difference was a +44.l percent, and in 1972-76, a +45.7 percent for all age groups combined. In 1956—60, the difference for those in the oldest group is not markedly dissimilar from that for those in all other age groups except the 51-60 year olds. In 1972-76, however, the dis- tribution of these differences across age cohorts follows the expected curvilinear pattern. The difference is at its highest point of +60.3 percent for the youngest group, declines in magni- tude over 20 percent and is flat across the three middle-aged groups, and then increases in magnitude to +54.4 percent for those in the oldest group (See Table 3-3C). 47 Attending Political Meetings or Rallies Table 3—4 shows the patterns across age cohorts of the various disengagement indicators for the act of attending politi— cal meetings or rallies. Disengagement from this act, like that from other forms of political campaign activity, is much more likely among all age groups in both time periods than is disengage- ment from voting. In 1956-60, 68.5 percent of those initially active in all age groups combined disengaged from attending poli- tical meetings or rallies, and in 1972-76, 71.9 percent did so (See Table 3-4A). In 1956-60, Table 3-4A shows that individuals in the old- est age group were one of the most likely cohorts to disengage from attending political meetings or rallies. The distribution in 1956-60 of the percentage by age cohort who disengaged is flat from the youngest group through the 41-50 year old cohort, but rise sharply for those in the 51-60 year old and the oldest age groups. In 1972-76, on the other hand, this pattern is reversed, and the proportionately greatest incidence of disengagement is among those in the youngest age group. Almost 90 percent of the initially active in the youngest cohort disengaged from attending political meetings or rallies over the 1972-76 period. The pro— portionate incidence of disengagement declines about 20 percent for the 31-40 year old group, and the pattern of the percentages is relatively flat across all remaining age groups. The likelihood of disengagement for the oldest age cohort, moreover, declined from 1956-60 to 1972-76. In 1956-60, 71.4 percent of the initially 48 actives in the oldest group disengaged as compared to only 66.7 percent in 1972-76. TABLE 3-4.--Change Patterns Over Time by Age Cohort for the Act of Attending Political Meetings or Rallies, 1956-60 and 1972-76. R for Age Group in All Age Years:3 21-30 31-40 41-50 51—60 61+ Groups A. Percentage who disengaged, among those who were active at t=1: 1956-60 63.6% 65.5% 68.0% 76.9% 71.4% 68.5% (N) ( ll) ( 29) ( 25) ( 13) ( l4) ( 92) 1972-76 89.2% 69.0% 62.5% 61.1% 66.7% 71.9% (N) ( 37) ( 29) ( 32) ( 18) ( 12) (128) B. Percentage continuously inactive, among those who were in— active at t=1: 1956-60 93.4% 96.6% 94.3% 87.7% 94.8% 93.9% (N) (273) (325) (261) (162) (173) (1194) 1972-76 96.5% 96.4% 95.7% 96.2% 95.2% 96.0% (N) (258) (195) (187) (184) (207) (1031) C. Difference between the percentage of disengagers and the per- centage of new engagers: 1956-60 +57.0% +62.l% +62.3% +64.6% +66.2% +62.4% 1972-76 +85.7% +65.4% +58.2% +57.3% +6l.9% +67.9% 8Age in 1956 for 1956-60 panel; age in 1972 for 1972—76 panel. Table 3—3B shows the distribution by age cohort of the per- centage of those initially inactive who were continuously inactive from the act of attending political meetings or rallies. For all age groups combined, continuous inactivity from the act of attend- ing political meetings or rallies is proportionately much more 49 frequent than it is from voting in both time periods. In both periods, over 90 percent of the individuals who were initially inactive in each time period were continuously inactive over time from this act. In 1956-60, those in the oldest age group were no more likely than those in other age groups to be continuously inactive; the differences across age cohorts are very small (See Table 3-4B). Similarly, in the 1972-76 period, the distribution of the per- centages who were continuously inactive is flat across all age cohorts. Those in the oldest group were no more likely than those in the other cohorts to be continuously inactive from attending political meetings or rallies in 1972-76. The differences between the percentage of disengagers and the percentage of new engagers from the act of attending political meetings or rallies is shown in Table 3-4C. As was the case for other forms of political campaign activity, the distribution of these differences is distinct from that for the act of voting. In 1956-60, the percentage difference for all age groups combined was a +62.4 percent, and in 1972-76, a +67.9 percent. The relationship between age cohort membership and the difference indicator is similar in each period. In 1956-60, the difference of a +66.2 percent for those in the oldest group is only somewhat greater than that for all other age cohorts. The distribution of this difference indicator is flat across all other age cohorts in 1956-60. In 1972-76, the largest difference is that for those in the youngest age group. The magnitude of this 50 difference drops at least 20 percent when the youngest cohort is compared to all others. The value of the difference for those in the oldest age group, however, is not much different from that of the three middle-aged groups. Finally, the differences for those in the youngest and oldest groups were markedly different across the two time periods. The imbalance between new engagers and dis- engagers for the youngest age cohort shifted in favor of the dis- engagers from 1956-60 to 1972-76. On the other hand, for the oldest age cohort, the balance between the newly engaged and the disengaged shifted from 1956-60 to 1972-76 in favor of the new engagers. Wearing a Campaign Button or Placing a Sticker on the Car The results of our analyses of over time change for the act of wearing a campaign button or placing a sticker on the car are shown in Table 3-5. Once again, disengagement from this act is proportionately more frequent than that from voting in each time period. As Table 3-5A shows, in 1956-60, 53.7 percent of those initially active in all age groups combined disengaged from wearing a campaign button or placing a sticker on the car, and in 1972-76, 79.7 percent of those initially active did so. The changes between 1956-60 and 1972-76 in levels of dis- engagement from the act of wearing a campaign button or placing a sticker on the car are much more dramatic than those for the other types of political campaign activity. This massive increase in disengagement among all age groups combined from this particular 51 act can probably be accounted for by the impact of the Federal Elections Campaign Act of 1974 which imposed ceilings on total ex- penditures in presidential campaigns. As Aldrich and Rohde (1978) have argued, this act reduced the use of such campaign materials (i.e., buttons and bumper stickers) in the 1976 presidential election campaign. TABLE 3—5.--Change Patterns Over Time by Age Cohort for the Act of Wearing a Campaign Button or Placing a Sticker on the Car, 1956-60 and 1972-76. i for Age Group in All Age Years:a 21-30 31-40 41—50 51-60 61+ Groups A. Percentage who disengaged, among those who were active at t=1: 1956-60 52.2% 45.5% 61.0% 60.7% 55.6% 53.7% (N) ( 46) ( 55) ( 41) ( 28) ( 18) (188) 1972-76 87.1% 78.8% 74.4% 75.0% 70.6% 79.7% (N) ( 70) ( 33) ( 43) ( 24) ( 17) (187) B. Percentage continuously inactive, among those who were inactive at t=1: 1956-60 81.9% 80.3% 84.4% 79.6% 89.8% 82.9% (N) (238) (300) (243) (147) (167) (1095) 1972-76 93.4% 94.2% 93.2% 93.7% 94.5% 93.8% (N) (226) (191) (176) (174) (201) (968) C. Difference between the percentage of disengagers and the per- centage of new engagers: 1956-6O +34.l% +28.8% +45.4% +40.3% +45.4% +36.6% 1972-76 +80.5% +73.0% +67.6% +68.7% +65.1% +73.5% 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 52 In 1956-60, the distribution of the percentage of those initially active who disengaged (See Table 3-5A) is quite uneven across the age cohorts, and the greatest proportionate incidence of disengagement is among those in the 41-50 year old group, not among the oldest group. In 1972-76, on the other hand, disengage- ment from the act of wearing a campaign button or placing a sticker on the car is most prevalent among those in the youngest age group. Moreover, those in the oldest age group are least likely to dis— engage from this act in 1972-76. The pr0portionate incidence of disengagement across the age cohorts declines sharply to a low point of 70.6 percent for the oldest group. Although continual inactivity from wearing a campaign but— ton or placing a sticker on the car was proportionately more fre- quent in 1972-76 than it was in 1956-60; Table 3-SB shows that there are no marked differences across age cohorts in both time periods. This is consistent with our results for the other forms of political campaign activity. Finally, Table 3-5C shows the differences between the per- centage of disengagers and the percentage of new engagers for the act of wearing a campaign button or placing a sticker on the car. Once again, the values of this difference are large and positive for all age groups in each time period. However, there are marked differences between periods in the way the difference indicator is related to age cohort membership. In 1956-60, the pattern of these differences across age cohorts is uneven but the value of the difference for those in the 53 oldest age group is not unique from those in other age groups. In fact, the only marked differences are those between the 21-30 and 31—40 year old groups and all others. In 1972-76, on the other hand, it is the youngest group which has the greatest proportionate imbalance between disengagers and new engagers. The difference for the youngest age group is a +80.5 percent, a figure which markedly exceeds that for all other age cohorts. The oldest group, in fact, has the smallest positive difference in the 1972-76 period. A Closer Look at the Differences Between Voting and Political Campaign Activity In the 1956—60 period, our results provided some evidence of elderly disengagement from the act of voting; and for the other forms of political campaign activity (except the act of giving money to political candidates), our indicators suggested that those in the oldest group were slightly more likely than those in other cohorts to disengage. Our results for 1972-76, on the other hand, showed a strikingly different pattern of old age disengagement from political activity. In 1972—76, older people were slightly more likely than those in other age groups to disengage from voting but less likely than those in other age cohorts to disengage from political campaign activity. Our results suggest that, in both time periods, the analy- sis of over time changing levels of political participation should consider voting and political campaign activity separately. In general, disengagement from political campaign activity is 54 proportionately more frequent than that from voting among all age groups in general and older people in particular. Moreover, the analysis demonstrated the utility of examining over time change in different periods of time. The dynamics of changing levels of political participation over the 1956-60 and 1972-76 time periods varied both for the electorate in general and for older people in particular. The basic problem posed by our results is to explain the different patterns of over time change for the oldest age group in each period. Why, in the 1972-76 period were older people in comparison to those in most other age groups more likely to dis- engage from voting but less likely to do so for political campaign activity? The familiar argument that political campaign activity requires more individual initiative and effort than voting may account for the patterns across age cohorts in 1956-60, but clearly cannot account for the patterns of change in 1972—76. That is, older people may tend to drop out at greater levels than those in younger age groups from more difficult political acts but not from easier acts such as voting. However, the nature of the period itself--or, more specifically, certain circumstances and events associated with aging in each period--may be responsible for these patterns of changing levels of political participation across time. We shall explore the impact of such circumstances and events on political disengagement in Chapters 4 and 5. A comparison of across time changes in voting to those for a summary measure of political campaign activity will set the stage 55 for the analysis of our problem. In general, our results above demonstrated that the relationship between age cohort membership and political campaign activity was similar within each period for the acts of attempting to influence the vote of others, attending political meetings or rallies, and wearing a campaign button or placing a sticker on the car. That is, in 1956-60, the analyses of across time change by age cohort for each of these acts pro- vided evidence of disengagement for those in the oldest group. In 1972-76, on the other hand, our analyses provided no evidence of disengagement for older people in comparison to those in other age cohorts for these three campaign acts. For the act of giving money to political candidates, our results suggested that, in the 1972-76 period, those in the oldest group were more likely than those in other age groups to disengage. Since the across time changes by age cohort for the act of giving money exhibit different patterns than the other three types of political campaign activity, it shall be excluded from our summary measure. The summary measure of political campaign activity will be a dichotomized variable constructed from the individual responses to the activities of attempting to influence the vote of others, attending political meetings or rallies, and wearing a campaign button or placing a sticker on the car. An individual is con- sidered to be a participant if he or she performed at least one of these activities. This summary measure will allow us to directly pit the relationship between age cohort membership and voting against that for political campaign activity. 56 Table 3-6 compares the percentage who disengaged from voting and the percentage who disengaged from political campaign activity in each time period. One of the most striking features of these results for all age groups combined is the similarity in the direction of change of disengagement from voting and political campaign activity between 1956-60 and 1972-76. The probability of disengagement from voting among all age groups combined increased from 4.2 percent in 1956-60 to 9.2 percent in 1972-76. Similarly, the probability of disengagement from political campaign activity increased about 5 percent from 40.3 percent in 1956-60 to 44.7 per- cent in 1972-76. Our results for the summary measure of political campaign activity are consistent with those presented above for each indi- vidual campaign act (except the act of giving money to political candidates). Those in the oldest age group are slightly more likely than those in other age groups to disengage from campaign activity in 1956-60, but not in the 1972-76 time period. And, for the act of voting, those in the oldest group are slightly more likely than those in most other age cohorts to disengage in both 1956-60 and 1972-76. For the act of voting, it is possible that some type of period effect2 is responsible for the increase in levels of dis- engagement between 1956-60 and 1972-76. The changes in the like- lihood of disengagement from voting between 1956-60 and 1972-76 are similar in direction, but not in magnitude, across the entire range of age cohorts. The likelihood of disengagement from voting 57 increased from 1956—60 to 1972-76 by 10.1 percent for the 21—30 year old group, 0.2 percent for the 31-40 year olds, 5.7 for the 41-50 year olds, 2.7 percent for the 51-60 year olds, and 4.9 per- cent for those 61 years old or more. TABLE 3-6.--Percentage Who Disengaged, Among Those Who Were Active at t=1 from Voting and the Percent Who Disengaged from Political Campaign Activity by Age Cohort, 1956-60 and 1972-76. i for Age Group All Age in Years:8 21-30 31-40 41-50 51-60 61+ Groups A. 1956—60 Voting 5.0% 5.4% 2.7% 2.8% 5.0% 4.2% (N) (180) (260) (224) (141) (140) (945) Political Campaign Activity 38.0% 37.3% 43.6% 39.4% 45.7% 40.3% (N) (108) (142) (110) ( 71) ( 70) (501) B. 1972-76 Voting 15.1% 5.6% 8.4% 5.5% 9.9% 9.2% (N) (218) (177) (190) (163) (161) (909) Political Campaign Activity 49.1% 48.6% 38.3% 42.5% 41.8% 44.7% (N) (165) (109) (115) ( 87) ( 79) (555) aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. The increase in the proportionate incidence of disengage- ment from political campaign activity among all age groups combined between 1956-60 and 1972-76 is not, on the other hand, mirrored by each age cohort. Those in the 21-30, 31-40, and 51-60 year old groups contributed to the increase in the incidence of disengagement 58 from political campaign activity between 1956-60 and 1972-76. The likelihood of disengagement from political campaign activity in- creased from 1956—60 to 1972—76 by 11.1 percent for the 21-30 year old group, 11.3 percent for the 31-40 year olds, and 3.1 percent for the 51-60 year olds. On the other hand, the probability of disengagement from political campaign activity for the oldest cohort decreased by 3.9 percent between 1956-60 and 1972-76. This is the reverse of the direction of change between 1956-60 and 1972-76 in disengagement from voting for the aged. While the likelihood of disengagement for most age groups from political campaign activity was increasing from 1956-60 to 1972—76, it was decreasing for the aged. Table 3-7 compares the percentage who were continuously inactive from voting with that who were continuously inactive from political campaign activity in both time periods. Like the re- sults for the first indicator of political disengagement, the pro- portionate incidence of continual inactivity from both voting and political campaigns increased between 1956-60 and 1972-76. In 1956-60, 56.6 percent of those initially inactive in all age groups combined abstained from voting in both presidential elections, and in 1972—76, 61.1 percent of the initially inactives did so. Similarly, the incidence of continual inactivity from political campaigns increased about 5 percent from 69.4 percent in 1956-60 to 75.0 percent in 1972-76. In general, the incidence of continual inactivity in- creased from 1956-60 to 1972-76 for each age cohort. For the act 59 of voting, the only cohort for which continual inactivity did not increase was the youngest one. On the other hand, the proportion- ate incidence of continual inactivity from voting increased from 1956-60 to 1972-76 by 6.0 percent for the 31-40 year olds, 11.5 percent for the 41-50 year olds, 12.6 percent for the 51-60 year olds, and 1.8 percent for the oldest age cohort. The same type of change patterns across periods emerge for political campaign activ- ity. That is, for the youngest cohort the prOportionate incidence of continual inactivity decreases between 1956-60 and 1972-76, whereas for those in other age groups it increases. TABLE 3-7.--Percentage Who Were Continuously Inactive, Among Those Who Were Inactive at t=1 from Voting and the Percentage Who Were Continuously Inactive from Political Campaign Activity by Age Cohort, 1956—60 and 1972-76. i for Age Group All Age in Yearsza 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Voting 52.9% 53.6% 54.0% 61.8% 71.1% 56.6% (N) (104) ( 97) ( 63) ( 34) ( 45) (343) Political Campaign Activity 68.4% 69.1% 69.6% 64.0% 75.8% 69.4% (N) (190) (230) (184) (114) (132) (850) B. 1972-76 Voting 44.9% 59.6% 65.5% 74.4% 72.9% 61.1% (N) ( 78) ( 47) ( 29) ( 39) ( 59) (252) Political Campaign Activity 67.3% 78.2% 77.6% 74.2% 77.7% 75.0% (N) (147) (133) (125) (132) (166) (703) aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 6O Across the entire range of age cohorts, those in the oldest group are one of the most likely to be continuously inactive from voting in each time period. The differences in continual inactivity from voting between the oldest age cohort and all others are greatest in the 1956—60 period. The distribution is flat across the first three age groups, increases about 7 percent to 61.8 percent for the 51-60 year olds, and then increases almost 10 percent to 71.1 per- cent for those in the oldest group. In 1972-76, on the other hand, the proportionate incidence of continual inactivity from voting increases steadily from the youngest group through the 51-60 year old cohort reaching a high of 74.4 percent, but then declines slightly to 72.9 percent for those in the oldest group. In both time periods, the aged are indeed one of the groups most likely to abstain from voting in two successive presidential elections. The pattern of continual inactivity across age cohorts for political campaign activity, however, differs between the two time periods. In the 1956-60 period, the distribution of the percent— ages who were continuously inactive is flat across the first three age groups, declines slightly to 64.0 percent for the 51-60 year olds, and then increases over 10 percent to 75.8 percent for those in the oldest cohort. In 1972-76, on the other hand, the rela- tively low level of continual inactivity among those in the young- est age group represents the only departure from an otherwise flat pattern across the remaining age cohorts. These results show that in 1956-60 the old were one of the groups most likely to be con— tinually inactive from political campaigns, whereas in 1972-76, 61 their level of continual inactivity is no different from any other age cohort except the young. Table 3-8 shows the difference between the percentage of disengagers and the percentage of new engagers from voting and political campaign activity in each time period. These results show that the direction of change between 1956-60 and 1972-76 for both voting and political campaign activity is similar. That is, for both voting and political campaign activity the imbalance be— tween disengagers and new engagers shifted in favor of the dis- engagers between 1956-60 and 1972-76. Across the range of age cohorts, Table 3-8 shows that there is some evidence of old age disengagement from voting in each time period. For political campaign activity, on the other hand, the old are more likely to disengage (and less likely to re-engage) than are those in other age groups in the 1956-60 period but not in 1972-76. The value of the difference in 1956-60 is positive and flat across the first two cohorts, increases to a +13.2 percent for those in the 41-50 year old group, drops to a low of +3.4 percent for the 51—60 year olds, and then increases to a high of +21.5 per- cent for the oldest cohort. In 1972-76, on the other hand, the pattern of the differences reaches a high of +26.8 percent for the 31-40 year old group, but then flattens out across the 41-50 through the oldest age cohort. 62 TABLE 3-8.--Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged by Age Cohort, Voting and Political Campaign Activity, 1956-60 and 1972-76. X for Age Group All Age in Years:3 21-30 31-40 41-50 51—60 61+ Groups A. 1956-60 Voting —42.1% -41.0% -43.3% -35.4% -23.9% -39.2% Political Cam- paign Activ- + 6.4% + 6.4% +13.2% + 3.4% +21.5% + 9.7% ity B. 1972—76 Voting —40.0% -34.8% -26.1% -20.1% -l7.2% -25.9% Political Cam- paign Activ- +16.4% +26.8% +15.9% +16.7% +19.5% +19.7% ity 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Discussion and Conclusions Our analyses here have attempted to capture the notion of old age disengagement from political activity in a variety of ways. Our best and most direct indicator of political disengagement--a comparison across cohorts of the percentage of those initially active who actually withdrew from political participation-~pro- vided, at best, modest support for the disengagement thesis. In both time periods, contrary to the argument of the disengagement thesis, the aged did not have uniquely high proportionate levels of disengagement from either voting or political campaign as com- pared to those in other age cohorts. In fact, the pattern of 63 disengagement across cohorts is rather flat for both types of activity in both time periods. Our second, indirect indicator of political disengagement-— a comparison across cohorts of the percentage of those initially inactive who remained inactive over time-—also provided little sup- port for the disengagement thesis. In 1956-60, for both voting and political campaign activity, there was a marked increase in con- tinual inactivity for the elderly across the entire range of age cohorts; but, in 1972-76, across the cohorts (except the youngest group, which had notoriously low levels of continual inactivity), the pattern of continual inactivity from both voting and campaigns was quite flat. Apparently, continual inactivity from politics is a general feature of the electorate. Those who are non-partici- pants in one election rarely participate in proceeding ones; and older people are not unique in this respect.3 Finally, we compare the balance across cohorts in the per- centage who dropped out of politics and the percentage who became active in politics after refraining from participation in a prev- ious election. In 1956-60, our results for both voting and poli- tical campaign activity showed that the elderly were quite unique compared to other cohorts in their greater tendency to disengage and lesser tendency to re-engage, or become newly engaged,in politics. In 1972-76, on the other hand, the elderly were not unique in this respect, and the pattern of the difference indica- tor for both voting and political campaign activity is quite flat across the 41-50 through the oldest cohort. 64 One problem which emerges in the above analysis is the decline in disengagement from political campaigns (that is, direct withdrawal) among the elderly between 1956-60 and 1972-76 that occurred without an accompanying decline in disengagement from voting. In general, the aged were no more likely than those in other age groups to disengage from voting in both the 1956-60 and 1972-76 time periods. However, for political campaign activity, the aged were slightly more likely than those in other age groups to disengage between 1956 and 1960, but slightly less likely to do so compared to those in other cohorts between 1972 and 1976. In addition, among those in the oldest group, disengagement from voting increased between 1956-60 and 1972-76, but disengagement from campaign activity declined. For all other cohorts (except the 41-50 year old group), increased levels of disengagement from voting between 1956-60 and 1972-76 were accompanied by increased levels of disengagement from political campaigns (See Table 3-6). It is difficult to conceive of election-specific or short-term factors which would have the effect of simultaneously decreasing the likelihood of old age disengagement from political campaign activity and of increasing the likelihood of old age dis— engagement from voting. Presumably, any configuration of short—term factors (such as the particular issues or candidates) in 1976 would have contributed to smaller levels of elderly political disengage— ment between 1972 and 1976 from both voting and political campaign activity. 65 It is also difficult to identify long-term, secular changes that would between 1956-60 and 1972-76 simultaneously decrease the likelihood of old age disengagement from political campaign activ— ity and increase the likelihood of old age disengagement from voting. For example, increased levels of disengagement from voting may possibly be accounted for by the weakening of partisan loyal- ties and the decline in feelings of "external" political efficacy (See Abramson and Aldrich, 1982). However, we would expect these attitudinal trends to have a similar impact on change in political campaign activity. Additionally, it would be difficult to account for this diverse pattern of change in old age disengagement from voting and campaigns by the long—term rise in educational levels among the elderly between 1956-60 and 1972-76. Increasing levels of education would presumably decrease the probability of disen— gagement from both voting and political campaign activity. In Chapters 4 and 5 we will attempt to identify a set of factors which might possibly account for across time changes in political participation among the electorate in general and older people in particular. Specifically, we will assess the impact of sociological variables, such as level of education and gender, political variables (being contacted by a political party), and situational variables (most notably, retirement from the labor force) on the relationship between age cohort membership and changing levels of political activity across time. To these analy- SES we HOW turn. NOTES TO CHAPTER 3 1For the convenience of the reader, the frequency distri- butions for our analyses here are shown in Appendices A3—l through A3-4 o 2A period effect is defined as that which has an impact on the attitudes and behavior of all members of a population regardless of age cohort membership. Certain configurations of candidates and issues could presumably have such an effect on changing levels of voting across time. 3The reader should keep in mind that this analysis mea- sures reported voting turnout. Entrance into and out of the electorate may be greater if a better sample of nonvoters were available. 66 CHAPTER 4 THE IMPACT OF COMPOSITIONAL FACTORS ON POLITICAL DISENGAGEMENT Our discussion of previous analytical and empirical studies of the disengagement thesis in Chapter 1 above suggested that level of education and sex may have a significant impact on the probability of political disengagement for individuals in general and older people in particular. In this chapter we will show that there are indeed marked differences across age cohorts, in both 1956-60 and 1972-76, in level of formal education and in the proportion of males and females. Moreover, we utilize an important advantage of panel data by separating compositional ef- fects from the effect of age cohort membership on disengagement from political activity.1 Our strategy is to assess the separate impact of sex and level of education on the relationship between age cohort member— ship and political disengagement.2 Our analysis will suggest that level of education has the most powerful impact on the likelihood of disengagement from political activity for both individuals in general and older pe0ple in particular in each time period. We will continue to pit the across time change patterns for voting against those for the summary measure of political campaign activ— ity introduced in Chapter 3. This strategy is adopted since, as 67 68 demonstrated in Chapter 3, the across time changes for the summary measure are similar to those for the individual political campaign activity items. Variations in Political Disengagement By Sex Table 4-1 shows the percentage of men and women in each age cohort in both the 1956-60 and 1972—76 time periods. Among all age groups combined in the 1956-60 time period, 45.1 percent were males and 54.9 percent were females. Similarly in 1972-76, there are more females than males among all age groups combined. Among all age groups, 44.5 percent were males and 55.5 percent were females. There are differences between the two periods, however, in the percentage of males and females in each age group. In 1956-60, there were roughly equal numbers of men and women in the oldest age cohort. On the other hand, in 1972-76, 61.8 percent of those in the oldest group were females and 38.2 percent were males. How- ever, this gender balance in 1972—76 is largely a function of sampling error.3 Data from the United States Census Bureau (1970) show that the balance between sexes among the aged were very simi- lar in the two time periods. According to these data, among those 60 and above 43.4 percent were males and 56.6 percent were females in 1960, and in 1970, 46.0 percent of this age group were males and 54.0 percent were females. Voting The percentage who disengaged from voting by age cohort and sex is shown in Table 4-2. For both sexes, disengagement from 69 voting was much more prevalent in 1972-76 as compared to 1956-60. Moreover, in each period, among all age groups combined, females are slightly more likely to disengage from voting than are males. In 1956-60, 3.1 percent of the males disengaged from voting and 5.5 percent of the females did so. On the other hand, in 1972-76, 8.8 percent of the males disengaged as compared to 10.4 percent of the females. TABLE 4—1.--Sex by Age Cohort, 1956-60 and 1972-76. i for Age Group All Age in Years:a 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Male 36.6% 43.5% 50.3% 49.2% 49.0% 45.1% Female 63.4% 56.5% 49.7% 50.8% 51.0% 54.9% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) (298) (372) (294) (185) (202) (1351) B. 1972-76 Male 50.7% 43.3% 44.5% 43.6% 38.2% 44.5% Female 49.3% 56.7% 55.5% 56.4% 61.8% 55.5% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) (296) (224) (218) (202) (220) (1160) aAge in 1956 for 1956—60 panel; age in 1972 for 1972—76 panel. Although the differences between sexes in each period are very small, the relationship between age cohort membership and disengagement from voting is strikingly different both across sex groups and across periods. Among males in 1956-60, it is those in 70 the youngest and oldest groups who are most likely to disengage from voting. Disengagement from voting among males in 1956-60 was most prevalent for the young, declines to 2.4 percent for the 31-40 and 41-50 year olds, and increases to 5.1 percent for those in the oldest group. On the other hand, among females in 1956-60, those in the oldest group are no more likely to disengage from voting than are those in the other age cohorts. Although 4.8 per- cent of initially active females in the oldest group disengaged, this percentage is not markedly different from that for females in the other age groups. Finally, there are no large differences in disengagement from voting in 1956-60 between older males and older females. The percentage difference between males and females in the oldest age group is only 0.3 percent. Older males, however, contributed to a larger preportion of the total pool of disengagers from voting between 1956 and 1960 than did females. Males account for 57.1 percent of the total pool of old age disengagers, whereas females account for 42.9 percent of this pool.4 In 1972-76, a quite different pattern across age cohorts emerges. Among males in 1972—76, those in the oldest group are one of the least likely groups to disengage from voting. The highest proportionate incidence of disengagement from voting is among young males. Almost one-fifth of the males in the youngest age group who voted in 1972 did not vote in 1976. This percentage declines dramatically for the 31-40 year old group, and the pattern of the percentages is flat across the remaining age cohorts. On the other hand, among females in 1972-76, it is those in the youngest and 71 TABLE 4-2.--Percentage Who Disengaged Among Those Who Were Active at t=1 from Voting by Age Cohort and Sex, 1956-60 and 1972-76. R for Age Group All Age in Years:3 21-30 31-40 41—50 51-60 61+ Groups A. 1956-60 Male 6.3% 2.4% 2.4% 0.0% 5.1% 3.2% (N) ( 79) (124) (123) ( 72) ( 78) (476) Female 4.0% 8.1% 3.0% 5.8% 4.8% 5.3% (N) (101) (136) (101) ( 69) ( 62) (469) B. 1972-76 Male 18.0% 3.9% 8.0% 4.0% 5.6% 8.8% (N) (111) ( 76) ( 88) ( 75) ( 71) (421) Female 12.1% 6.9% 9.0% 6.8% 13.3% 9.7% (N) (107) (101) (100) ( 88) ( 90) (486) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. oldest age groups who are most likely to disengage. Moreover, in 1972-76, there are marked differences between sexes among those in the youngest and oldest age cohorts. Among individuals in the youngest age group, males are more likely than females to disen- gage from voting. likely than females to disengage. Among older people, however, males are less In 1956-60, older males were more likely than older females to disengage from voting, but in 1972-76, older females are more likely than older males to do so. In 1972-76, moreover, females account for a substantially larger proportion of the total pool of aged disengagers from voting than do males. Females account for 75.0 percent of the total pool of 72 aged disengagers, and males account for only 25.0 percent of this pool. Much of the increase in disengagement from voting among older people between 1956-60 and 1972-76 can thus be accounted for by elderly females. Table 4-3 shows the percentage of those initially inactive who were continual non-voters by age cohort and sex in each time period. Among all age groups combined, continual non-voting was more likely in 1972-76 than in 1956-60 for members of both sexes. In 1956-60, 51.4 percent of initially inactive males and 59.0 per- cent of initially inactive females were continual non-voters. On the other hand, in 1972-76, 55.8 percent of the initially inactive males were continual non-voters as compared to 64.3 percent of the initially inactive females. The relationship between age cohort membership and con- tinual non-voting differs both across sexes and across time per- iods. In 1956-60, there is no evidence of old age disengagement among men. Continual inactivity is most frequent among younger men, drOps off sharply for the 31-40 and 41-50 year old groups, increases again for the 51-60 year old group, and then declines slightly to 53.8 percent for older men. Among females, on the other hand, continual inactivity rises over 30 percent from 47.4 percent for the youngest group to almost 80 percent for older females. Continual inactivity rises steadily across the entire range (that is, from youngest through oldest) of female cohorts 73 TABLE 4-3.--Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Voting by Age Cohort and Sex, 1956-60 and 1972-76. R for Age Group All Age in Years: 21-30 31-40 41-50 51-60 61+ Groups A. 1956-6O Male 67.9% 41.9% 38.1% 56.3% 53.8% 51.4% (N) ( 28) ( 31) ( 21) ( 16) ( 13) (109) Female 47.4% 59.1% 61.9% 66.7% 78.1% 59.0% (N) ( 76) ( 66) ( 42) ( l8) ( 32) (234) B. 1972-76 Male 46.2% 57.1% 77.8% 76.9% 46.2% 55.8% (N) ( 39) ( 21) ( 9) ( l3) ( 13) ( 95) Female 43.6% 61.5% 60.0% 73.1% 80.4% 64.3% (N) ( 39) ( 26) ( 20) ( 26) ( 46) (157) aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. In 1956-60, with each age cohort, excepting the youngest one, females are more likely than males to be continual non-voters. The largest gap between males and females, moreover, is within the oldest age group. In fact, the prominent tendency of older women to be continual non-voters between 1956 and 1960 accounts for a great deal of the age cohort related pattern we found above in Chapter 3. Older females account for 78.1 percent of the total pool of continual non-voters among the oldest cohort in 1956-60, Fdhereas males account for only 21.8 percent of this pool. A large amount of the trend in continual inactivity between 1956 and 74 1960 among older people can thus be accounted for by the over time behavior of older females. In 1972-76, there are also distinct differences between men and women across age cohorts in the percentage of continual non—voters. Continual inactivity is at one of its lowest levels for young men, increases steadily for those in the 31-40 and 41—50 year old groups, levels off for the 51-60 year olds, and then de- clines over 30 percent to 46.2 percent for old men. The pattern of continual inactivity across female age cohorts, on the other hand, is consistent with the expectations of the disengagement thesis. Continual inactivity increases steadily from the youngest through the oldest female cohorts and reaches a high of 80.4 percent for older women. It is again apparent that older females are primarily responsible for the increased levels of old age continual non-voting between 1956-60 and 1972—76 which were shown in Chapter 3. In 1972-76, among those in all age groups combined who were inactive in 1972, females are much more likely than males to also abstain from voting in the 1976 presidential election. The dif- ference between males and females is most pronounced, moreover, for the oldest age group. There is over a 30 percent difference between the percentage of older men and that of older women who were con- tinual non-voters. In fact, women account for 86.0 percent of the total pool of continual non-voters among the oldest age group in 1972-76, whereas men account for only 14.0 percent of this pool. Once again, most of the incidence of continual non—voting among 75 older people can be accounted for by the across time behavior of elderly women. Table 4—4 shows the distribution by age cohort and sex of the difference between the percentage of disengagers and the per- centage of new engagers for the act of voting in each time period. In each period, the balance between disengagers and new engagers is negative and more in favor of the latter group for males as com- pared to females. In 1956-60, this difference for males is a -45.5 percent, and for females, a -35.7 percent. In 1972-76, moreover, the difference for males is a —33.8 percent, and for females, a -26.1 percent. Among males in 1956-60, there is an uneven pattern in this difference measure across all age cohorts. The lowest value of this difference is that of a -25.8 percent for the youngest group. The value of the difference, moreover, decreases almost 30 percent for the 31-40 year old group, is flat across the 31-40 and 41-50 year old groups, and falls off to just over a negative 40 percent for those in the two oldest cohorts. The results suggest that among males in 1956-60, it is the young rather than the old who are most likely to disengage from voting. 0n the other hand, the distribution of this percentage difference across age cohorts for females shows that the difference is lowest in negative magnitude for those in the oldest age group. For females the imbalance between the disengagers and new engagers is greatest, and in favor of the latter group, among the youngest cohort. Across the entire range of female age cohorts, the value of this difference increases 76 steadily from the youngest through the oldest group. The pro- portionate imbalance between disengagers and new engagers is still in favor of the latter group among elderly females but not to the extent that it is for women in the other age groups. TABLE 4-4.--Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Voting by Age Cohort and Sex, 1956-60 and 1972-76. i for Age Groupa All Age in Years: 21-30 31-40 41-50 51—60 61+ Groups A. 1956-60 Male -25-870 -550770 -5905% -43-870 .4101% “45.5% Female -48.6% -32.8% -35.1% -27.5% -17 1% -35.7% B. 1972—76 Male -35.8% -39.0% -l4.2% -l9.l% -48.2% -33.8% Female -44.3% -3l.6% -3l.0% -20.l% - 6.3% -26.1% aAge in 1956 from 1956-60 panel; age in 1972 for 1972-76 panel. Our results for 1972-76 (See Table 4-4B) show a similar pattern across age cohorts for each sex. Among males, the dif- ference of a -48.2 percent for the oldest group is much greater than that for all other age groups. Across the entire range of male age cohorts, the difference is largest for the 21-30 and 31-40 year old groups and declines dramatically for those in the 41—50 and 51-60 year old groups. On the other hand, for females in 1972-76, the imbalance between disengagers and new engagers is greatest and in favor of the latter group, among those in the 21—30 77 year old cohort. This difference increases steadily across female cohorts from the youngest through the oldest groups. As was the case in 1956-60, among older females new engagers proportionately are greater than disengagers but not to the extent as for those in younger age groups. In both 1956-60 and 1972-76, the balance between disen- gagers and new engagers is greater for older males than it is for older females. In 1956-60, the difference for older males was a -41.1 percent, and for females, a -l7.1 percent. Similarly, in 1972-76, the difference for older males of a -48.2 percent is much larger than that for females of a -6.3 percent. One of the most interesting features of these results is the differences between the two time periods in the patterns of disengagement for older men and older women. In 1956-60, older females were at a disadvantage compared to older males only in their lesser tendency to re-engage in voting. That is, older women were no more likely to dr0p out of voting than older males, but were much less likely to vote in 1960 after abstaining in 1956. In the 1972-76 period, on the other hand, older women were at a dis- advantage compared to older men for all three types of across time change in voting. Older women were less likely than older men to vote in 1976 after abstaining in 1972 and, in addition, more likely to drop out of voting between 1972 and 1976. Moreover, in 1972-76, older females account for a substantially greater proportion of the total pool of aged disengagers and continual non-voters than do older men. 78 Political Campaign Activity Table 4-5 shows the percentage who disengaged from poli- tical campaign activity (that is, according to the dichotomous measure introduced in Chapter 3) by age cohort and sex for each time period. In both periods, females are more likely to disen- gage than males. In 1956-60, among those initially active in all age groups combined, 43.2 percent of females disengaged and 37.3 percent of males did so. Similarly, in 1972-76, 43.3 percent of initially active females and 37.8 percent of initially active males disengaged. TABLE 4-5.--Percentage Who Disengaged Among Those Who Were Active at t=1 from Political Campaign Activity by Age Cohort and Sex, 1956-60 and 1972-76. i for Age Group All Age in Years: 21-30 31—40 41-50 51-60 61+ Groups A. 1956—60 Male 22.0% 32.9% 45.3% 44.4% 48.7% 37.3% (N) ( 50) ( 82) ( 69) ( 36) ( 39) (276) Female 51.7% 43.3% 43.9% 34.3% 41.9% 44.0% (N) ( 58) ( 60) ( 41) ( 35) ( 31) (225) B. 1972-76 Male 46.9% 43.1% 25.9% 33.3% 31.0% 37.8% (N) ( 81) ( 51) ( 54) ( 39) ( 29) (254) Female 44.6% 47.1% 43.4% 41.5% 38.1% 43.3% (N) ( 74) ( 51) ( 53) ( 41) ( 42) (261) 3Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 79 In contrast to the between-period differences for males and females we found for the act of voting, overall levels of disen- gagement were similar across periods for both men and women. Both sexes were more likely to disengage from voting in 1972-76 than they were in 1956-60. However, the likelihood of disengagement from political campaign activity for both males and females was almost the same in 1972-76 as it was in 1956-60. There are differences between males and females in the relationship between age cohort membership and disengagement from political campaign activity in each time period. In 1956-60, the pattern across male age cohorts supports the predictions of the disengagement theory. The percentage who disengaged increases steadily from the youngest through oldest male age cohorts. Only 22.0 percent of the initially active young males disengaged com- pared to 48.7 percent of initially active older males. Moreover, the percentage of older males is much larger than that for all other male age groups. The pattern for females in the 1956-60 period, on the other hand, provides no support for the expectations of the disengagement theory. Women in the youngest age group are the most likely among all females to disengage. Older women, on the other hand, are no more likely to disengage from political cam- paign activity than any other female age group with the exception of the 51-60 year olds. Moreover, older females are less likely to disengage than are older males. Only 41.9 percent of initially active elderly females disengaged as compared to 48.7 percent of initially active elderly males. Moreover, unlike the act of 80 voting, older males compose a substantially larger proportion of the total pool of aged disengagers from political campaign activity than do older females. Older men account for 59.4 percent of this pool, whereas older females comprise only 40.6 percent of the pool. In 1972-76, a quite different pattern emerges across male and female age cohorts. Among males in 1972-76, the percentage who disengaged from political campaign activity declines from the young- est through the oldest age cohort. In fact, older males are less likely than those in any other age group, except those in the 41-50 year old cohort, to disengage. Older women, in addition, are less likely than those in all other female age groups to disengage from political campaign activity in 1972-76. Finally, unlike 1956-60, in 1972-76 older women comprise a substantially larger proportion of the total pool of aged disengagers from campaigns as compared to older men. In 1972-76, older men account for 36.0 percent of this pool, whereas older women account for 64.0 percent of the pool. Across the two time periods, the percentage change in dis— engagement from political campaign activity is much greater for older men than it is for older women. Both groups were less likely to disengage from campaign activity in 1972-76 than they were in 1956-60, but the decrease in the likelihood of disengagement is much larger for older men. For older men the percentage of ini— tially actives who disengaged declines almost 20 percentage points (from 48.7 percent to 31.0 percent) from 1956-60 to 1972-76, whereas the decline for older females is only 3.8 percent (from 41.9 percent to 38.1 percent). 81 Table 4-6 shows the percentage of those who were initially inactive who never participated in political campaigns by age co- hort and sex in each time period. Among all age groups combined, continual inactivity was more likely in 1972-76 than in 1956—60 for members of both sexes. In 1956-60, 65.8 percent of the initially inactive males were continuously inactive and 71.8 percent of the initially inactive females were. In 1972-76, on the other hand, 69.1 percent of initially inactive men were continually inactive as compared to 75.2 percent of the initially inactive women. TABLE 4-6.--Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Political Campaign Activ- ity by Age Cohort and Sex, 1956-60 and 1972-76. R for Age Group All Age in Years:8 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Male 66.1% 62.5% 68.4% 65.5% 66.7% 65.8% (N) ( 59) ( 80) ( 79) ( 55) ( 60) (333) Female 69.5% 72.7% 70.5% 62.7% 83.3% 71.8% (N) (131) (150) (105) ( 59) ( 72) (517) B. 1972-76 Male 63.8% 71.7% 60.5% 73.5% 76.4% 69.1% (N) ( 69) ( 46) ( 43) ( 49) ( 55) (262) Female 68.1% 78.9% 83.8% 71.2% 74.5% 75.2% (N) ( 72) ( 76) ( 68) ( 73) ( 94) (383) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 82 Despite these similarities among all age groups combined, there are differences both between sexes and between periods in the way age cohort membership relates to continual inactivity from political campaigns. In 1956-60, the pattern of continual inactiv- ity across the male age cohorts provides no evidence to support the predictions of the disengagement thesis. The distribution of the percentages across male age cohorts is flat. Among females in 1956—60, on the other hand, the pattern of percentages across age cohorts supports the predictions of the disengagement theory. The pattern is flat across the first three female age cohorts, declines to a low of 62.7 percent for those in the 51-60 year old group, and then increases over 20 percent to 83.3 percent for older women. For most age cohorts, except those in the 51-60 year old group, continual inactivity is proportionately more frequent among females than it is among males. Moreover, the gap between the sexes is largest for older people. Among older females, 83.3 per- cent were continually inactive as compared to only 66.7 percent of older males. In fact, older females are responsible for a large part of the age cohort related pattern of continual inactivity from campaigns which we found above in Chapter 3. In 1956—60, older women accounted for 60.0 percent of the total pool of continually inactives among the aged, whereas older men represented only 40.0 percent of this pool. The patterns of change for 1972-76 are markedly different from those for 1956-60. Across the male age cohorts, there is an uneven pattern of percentages, and older men are most likely to 83 abstain from participation in political campaigns in both 1972 and 1976. On the other hand, older women are one of the least likely female age cohorts to remain continually inactive from campaigns. Across the entire range of female age cohorts, the proportion con— tinually inactive rises steadily to a high of 83.8 percent for those in the 41-50 year old group, declines over 10 percent to 71.2 percent for the 51-60 year olds, and then increases slightly to 74.5 percent for older people. Moreover, among the oldest age cohort the proportionate incidence of continual inactivity for males increased by about 10 percent between 1956-60 and 1972-76 but de- creased by about 10 percent between the two periods for females. Although there are marked differences between periods in the across time political campaign behavior of older men and women, the latter group still accounted for a clear majority of the total pool of continual inactives among the elderly in 1972-76. Older women account for 62.5 percent of the total pool of elderly con- tinually inactives, whereas older men account for only 37.5 per- cent of this pool. Table 4-7 shows the difference between the percentage of disengagers and the percentage of new engagers for political cam- paign activity in each time period. In each period, the imbalance between disengagers and new engagers is greater, and in favor of the latter group, for males than it is for females. Among males in 1956-60, this difference (-4.0 percent) is negative and indicates that new engagement is proportionately more frequent than disen- gagement. Among females in 1956-60, on the other hand, the 84 percentage difference of a +ll.4 percent demonstrates that disen- gagers are proportionately more frequent than new engagers. In 1972-76, this difference is positive for both sexes, although the imbalance between disengagers and new engagers is greatest and in favor of the former group for females. TABLE 4—7.--Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Political Cam- paign Activity by Age Cohort and Sex, 1956-60 and R for Age Group All Age in Years:a 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Male -ll.9% - 4.6% +ll.9% +9.9% +15.4% - 4.0% Female +21.2% +16.0% +14.4% -3.0% +25.2% +ll.4% B. 1972-76 Male +lO.7% +14.8% —13.6% + 6.8% + 7.4% + 6.9% Female +12.7% +26.0% +27.2% +12.7% +12.6% +18.5% 3Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. In each period, there are also distinctive patterns across age cohorts on this difference indicator. Among males in 1956-60, those in the 21—30 and 31-40 year old groups have a negative dif- ference, whereas this difference is positive for those in the other three age groups. However, the imbalance is greatest and in :favor of disengagers for older males. Among females in 1956-60, 85 it is also those in the oldest age group who have the greatest positive difference. In the 1972-76 period, older women and men did not have the greatest proportionate imbalance between new engagers and disen- gagers compared to those in other age groups. Although this dif- ference is a +7.4 percent for older men, it is not much different from that for any other age group except the 41-50 year olds. Older women, in fact, have one of the smallest positive differences in 1972-76. The difference of a +12.6 percent for older females is much smaller in magnitude than those for the 31-40 and 41-50 year old groups and not markedly different from those for the 21-30 and 51—60 year old groups. The likelihood of disengagement from political campaign activity decreased from 1956—60 and 1972-76 among both older male and female cohorts. However, in 1956-60, older females were less likely than older males to actually withdraw from campaign activity, whereas in 1972-76, they were more likely than males to do so. Unlike the pattern found for voting, moreover, both older males and females from 1956-60 and 1972-76 became less likely to disengage from campaign activity compared to their younger counterparts. Variations in Political Diseggagement By Level of Education Differences in level of formal education across age cohorts lmay also account for some of the variation in over time changing 'levels of political participation. Previous studies have demon- strated that political participation is positively related to 86 level of formal education (See Chapter 1). It would also be reasonable to expect that changing levels of political participa- tion across time would also be related to education. If level of education is indeed related to patterns of change across time, then compositional differences across age groups may be responsible in part for the patterns of age cohort related change. Table 4-8 shows that in both time periods there are dis— tinct differences across age cohorts in the amount of formal edu— cation. In both 1956-60 and 1972-76, there is a very strong re- lationship between age cohort membership and level of formal education. In 1956-60, over three-fourths of those in the oldest age group had less than a high school education, a percentage much greater that for all other age groups. In 1972-76, moreover, those with less than a high school education are clearly in the majority for the oldest age cohort. The aged were, however, much better educated in 1972-76 than they were in 1956-60. In 1972-76, almost 40 percent of those 61 and above had completed at least high school and only 22.5 percent had done so in 1956-60. The dramatic dif- ferences in the level of formal education between individuals in the oldest group and those in other age groups may account for much of the political disengagement characteristic of older peOple. Voting Table 4-9 shows the percentage of those initially active ‘who disengaged from voting by age cohort and level of education. In both time periods, those with less than a high school education 87 are more likely than those who have completed high school or more to disengage from voting. In 1956-60, only 2.4 percent of those with education of high school or more disengaged as compared to 6.9 per- cent of those with less than a high school education. In 1972-76, the difference between these educational groups is even larger. Among those who completed high school or more in 1972-76, 7.4 per— cent disengaged from voting, whereas over 15 percent of those whose level of education was less than high school did so. High levels of education clearly decrease the likelihood of disengagement from voting among all age groups. TABLE 4-8.--Level of Education by Age Cohort, 1956-60 and 1972-76. i for Age Group All Age in Years:8 21-30 31-40 41—50 51-60 61+ Groups A. 1956-60 Less than high school 33.2% 39.8% 50.7% 63.8% 77.5% 49.6% High school or more 66.8% 60.2% 49.3% 36.2% 22.5% 50.4% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) (298) (372) (294) (185) (200) (1349) B. 1972-76 Less than high school 10.8% 23.7% 30.8% 46.1% 61.4% 33.5% High school or more 89.2% 76.3% 69.2% 53.9% 38.6% 66.5% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) (278) (219) (214) (204) (233) (1148) aAge in 1956 for 1956-60 panel; age in 1972 for 1972—76 lianel. 88 TABLE 4-9.--Percentage Who Disengaged Among Those Who Were Active at t=1 from Voting by Age Cohort and Level of Educa- tion, 1956-60 and 1972-76. i for Age Group All Age in Years: 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Less than high school 11.4% 10.1% 3.0% 4.8% 7.1% 6.7% (N) ( 44) ( 79) ( 99) ( 83) ( 98) (403) High school or more 2.9% 3.3% 2.4% 0.0% 0.0% 2.4% (N) (136) (181) (125) ( 58) ( 41) (541) B. 1972-76 Less than high school 25.0% 23.1% 11.6% 10.7% 16.1% 15.2% (N) ( 12) ( 26) ( 43) ( 56) ( 87) (224) High school or more 15.2% 2.3% 6.3% 3.2% 1.6% 7.1% (N) (178) (131) (126) ( 95) ( 64) (594) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Level of formal education also has a marked impact on the relationship between age cohort membership and disengagement from voting. horts is quite uneven. In 1956-60, the pattern across the less educated age co— Disengagement from voting is most prevalent among those in the two youngest age groups, declines to a low of 4.8 percent for those in the 41-50 year old group, and then in- creases slightly for the two oldest groups. Among the better educated group, however, the probability of disengagement from voting actually decreases from the youngest through the oldest 89 age cohorts. In fact, absolutely none of the individuals in the two oldest groups who completed high school or more disengaged from voting over the 1956-60 period! Less educated individuals accounted for the entire pool of aged disengagers from voting in 1956-60. Level of formal education, moreover, has a distinct impact on the relationship between age cohort membership and disengagement from voting in the 1972-76 period. Among those with less than a high school education, the pattern of the percentage who disen- gaged from voting across the age cohorts is uneven. Those in the two youngest groups are most likely to disengage from voting. The percentage who disengaged from voting among the less educated group declines sharply for the 41-50 year olds, levels off for the 51-60 year old group, and increases slightly to 16.1 percent for the oldest group. Among those who have completed high school or more, however, those in the oldest age cohort are one of the least likely age groups to disengage from voting. As Table 4-9 shows, only 2.9 percent of better educated older people disengaged from voting between 1972 and 1976. This percentage is well below the average for all age groups combined of 7.4 percent and that of 14.3 percent for those in the youngest age cohort. This suggests that the less educated older people are in part responsible for the pattern we found between age cohort membership and disengagement from voting in 1972-76 in Chapter 3. In fact, less educated people accounted for 93.3 percent of the total pool of aged disengagers 90 from voting in 1972-76, whereas better educated people accounted for only 6.7 percent of this pool. Table 4-10 shows the percentage of those initially inactive who were continual non-voters by age cohort and level of formal education in each time period. Among all age groups combined, con- tinual non-voting was more likely in 1972-76 than in 1956-60 for both educational groups. In 1956-60, 64.4 percent of the less edu- cated were continual non-voters as compared to only 40.0 percent of the better educated. In 1972—76, on the other hand, 74.0 percent of the initially inactive with less than a high school education were continual non-voters as compared to only 47.9 percent of those with at least a high school education. In the 1956-60 period, level of formal education has a marked impact on the relationship between age cohort membership and continual non-voting. The distribution of the percentages across the less educated cohorts provides clear support for the predic- tions of the disengagement thesis. The proportionate incidence of continual inactivity decreases from the youngest cohort through the 31-40 year old group, and then increases steadily to a high of 74.4 percent for less educated older people. On the other hand, the pattern of continual activity from voting between 1956 and 1960 among better educated cohorts is flat across the youngest through the 51-60 year old group, and then declines to zero for the (aged. Although the small number of cases for older people makes inferences about this specific group tenuous, the general pattern is clear. Level of formal education has a dramatic impact on the 91 over time political campaign behavior of individuals in general and older people in particular. TABLE 4-lO.--Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Voting by Age Cohort and Level of Education, 1956-60 and 1972-76. R for Age Group All Age in Years: 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Less than high school 66.7% 58.7% 60.0% 65.5% 74.4% 64.4% (N) ( 48) ( 63) ( 50) ( 29) ( 43) (233) High school or more 41.1% 44.1% 30.8% 40.0% 0.0% 40.0% (N) ( 56) ( 34) ( 13) ( 5) ( 2) (110) B- _1_9_7_2;Zé Less than high school 52.9% 70.0% 72.2% 82.2% 80.0% 74.0% (N) ( l7) ( 20) ( 18) ( 28) ( 40) (123) High school or more 39.3% 53.8% 55.6% 55.6% 57.9% 47.9% (N) ( 56) ( 26) ( 9) ( 9) ( 19) (119) aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. The greater tendency of the less educated to be continually inactive from voting between 1956 and 1960 accounts for a large ‘portion of the age cohort related pattern we found in Chapter 3. iLess educated individuals account for the entire pool of con- tinually inactives from voting among the oldest age group. 92 In 1972-76, there are also distinct differences between less and better educated age cohorts in the percentages of con- tinual non-voters. Unlike 1956-60, in 1972-76 the pattern of con- tinual inactivity among the less educated age groups is greatest for both the 51-60 and the oldest age groups. Among the better educated, on the other hand, continual inactivity is at its lowest level for the youngest age group, increases to 53.8 percent for the 31-40 year olds, and flattens out across the remaining age groups, including the elderly. Across all age cohorts in 1972-76, the initially inactive less educated are much more likely than the initially inactive better educated to abstain from voting in both presidential elec- tions. Once again, among the oldest age group the less educated account for a substantially larger portion of the total pool of continually inactives than do the better educated. The less educated elderly account for 74.4 percent of this pool, whereas the better educated elderly account for only 25.6 percent of the pool. Table 4—11 shows the difference between the percentage of disengagers and the percentage of new engagers for the act of voting by age cohort and level of education in each time period. Among both educational groups there are distinct differences between the two time periods in the imbalance between the disen— gagers and the new engagers. Among both the better and less edu- cated groups, the imbalance between disengagers and new engagers shifted in favor of the former group from 1956-60 to 1972—76. 93 Similarly, among those with high school or more, the difference decreased from a negative 58.0 percent in 1956-60 to a negative 43.3 percent in 1972-76. TABLE 4-ll.--Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Voting by Age Cohort and Level of Education, 1956-60 and 1972-76. i for Age Group All Age in Years:8 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 Less than high school -21.9% -31.2% —37.0% -29.7% -18.5% —28.7% High school or more -56.0% -52.6% -66.8% -60.0% -100.0% -58.0% B. 1972-76 Less than high school -22.1% - 6.9% -l6.3% - 7.2% - 3.9% -10.1% High school or more -45.5% -43.9% -38.1% -4l.2% -40.5% -45.0% 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. In 1956-60, the lowest difference among the less educated group is that for those in the oldest age cohort of a negative 18.5 percent. The pattern of this difference across age cohorts is, as anticipated, curvilinear. The imbalance between disengagers and new engagers becomes increasingly in favor of the new engagers from the 21-30 year old cohort through the 41-50 year old group. On the other hand, the imbalance becomes increasingly in favor of the dis- «engagers from the 41-50 year old cohort through the oldest age 94 group. Among the better educated group in 1956-60, however, the data form anything but a curvilinear pattern across the age cohorts. In fact, the proportionate imbalance between disengagers and new engagers among better educated older people in 1956-60 is com- pletely in favor of the latter group. If a better educated older person did not vote in 1956, they did indeed do so in 1960. Among those in other age cohorts who completed high school or more, how— ever, this percentage difference never exceeds a negative 70 percent. Political Campaign Activity Table 4-12 shows the percentage of those initially active who disengaged from political campaign activity by age cohort and level of education in each time period. Among both the better and less educated groups the proportionate incidence of disengagement from campaign activity increased between 1956-60 and 1972-76. In 1956-60, among those who were initially active, 50.2 percent of those with less than a high school education disengaged from cam— paign activity, and in 1972-76, 57.0 percent did so. Similarly, among the better educated group, 32.9 percent of the initially active disengaged between 1956 and 1960, whereas between 1972 and 1976 over 40 percent of the initially active did so. In each time period, there are marked differences in dis- engagement from political campaign activity between the less and better educated groups. In both periods, the percentages of dis- engagers among those with high school or more is over 15 percent less than that for the less educated group. 95 Higher levels of edu- cation clearly decrease the probability of disengagement from political campaign activity. Moreover, higher levels of formal edu— cation decrease the probability of disengagement among those in each age cohort. TABLE 4-12.--Percentage Who Disengaged Among Those Who Were Active at t=1 from Political Campaign Activity by Age Cohort and Level of Education, 1956-60 and 1972-76. Age Group A. in Years: 1956-60 Less than high school (N) High school or more (N) 1972—76 Less than high school (N) High school or more (N) 21-30 31-40 41-50 51.9% 50.0% 50.0% ( 27) ( 50) ( 48) 33.3% 30.4% 38.7% ( 81) ( 92) ( 62) 64.3% 63.2% 67.9% ( 14) ( l9) ( 28) 47.4% 45.2% 31.4% (133) ( 73) ( 70) 51-60 48.7% ( 39) 28.1% ( 32) 55.6% ( 27) 37.7% ( 53) 61+ 51.0% ( 49) 33.3% ( 21) 45.0% ( 40) 37.8% ( 37) R for All Age Groups 50.2% (213) 32.9% (288) 57.0% (128) 41.5% (366) 3Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. In 1956-60, level of education has little impact on the relationship between age cohort membership and disengagement from political campaign activity. Among those with less than a high 96 school education, there are no large differences across age cohorts in the percentage who disengaged. About 50 percent of those in each age group who were initially active disengaged from political cam- paign activity between 1956 and 1960. Similarly, among the better educated group in 1956-60, the pattern of the percentages across age cohorts is flat. The relatively high level of disengagement among better educated 41-50 year olds is the only striking departure from the otherwise flat pattern. In 1972-76, the percentage of those initially active who disengaged from political campaign activity decreased across age cohorts among both the less and better educated groups. Among those with less than a high school education the percentage ini- tially active who disengaged is relatively flat across the 21—30 to the 41-50 year old age groups, drops over 10 percent to 55.6 per- cent for those in the 51-60 year old cohort, and finally drops 10 percent to 45.0 percent for those in the oldest group. The extent of disengagement among the better educated cohorts is less than that among the less educated among all age groups. However, among the better educated, the aged are one of the cohorts least likely to disengage. The percentage who disengaged among those with a high school education or more in 1972-76 is at its highest point of 47.4 percent for the youngest group but then declines steadily across all other age groups. Table 4-13 shows the percentage of those initially inactive ‘who never participated in political campaigns by age cohort and level of formal education in each time period. As was the case for 97 TABLE 4-13.--Percentage Who Were Continuously Inactive Among Those Who Were Inactive at t=1 from Political Campaign Ac- tivity by Age Cohort and Level of Education, 1956-60 and 1972-76. Age Group A. In Years: 1956-6O Less than high school (N) High school or more (N) 1972-76 Less than high school (N) High school or more (N) 21-30 31-40 41-50 51-60 76.4% 79.6% 73.3% 68.4% ( 72) ( 98) (101) ( 79) 63.6% 61.4% 65.1% 54.3% (118) (132) ( 83) ( 35) 75.0% 87.9% 89.5% 82.1% ( l6) ( 33) ( 38) ( 67) 67.0% 77.7% 71.8% 66.7% (115) ( 94) ( 78) ( 57) 61+ 81.1% (106) 50.0% ( 24) 81.6% (103) 75.5% ( 53) i for All Age Groups 76.1% (456) 61.5% (392) 83.3% (257) 71.5% (397) aAge in 1956 for 1956-60 panel; age in 1972 for 1972—76 panel. voting, among all age groups combined who were initially inactive, continual inactivity was more prevalent in 1972-76 than in 1956-60 for both educational groups. Moreover, in each period the less educated were more likely than the better educated to abstain from political campaign activity in two successive presidential elec— tions. In 1956-60, 76.1 percent of the initially inactive less educated were continually inactive compared to 61.5 percent of the initially inactive better educated. Similarly, in 1972-76, 83.3 98 percent of the initially inactive with less than a high school education were continually inactive as compared to 71.5 percent of the initially inactive with at least a high school education. In the 1956-60 period, the predictions of the disengagement thesis are supported by the age cohort related pattern of continual inactivity across the less educated groups. Among those with less than a high school education, the prOportionate incidence of con— tinual inactivity from campaigns is at a high of 81.1 percent for older people. On the other hand, the pattern across the better educated age groups shows that the elderly are the least likely cohort to be continuously inactive. Across the entire range of those cohorts with at least a high school education, the pattern of the percentages is flat from the youngest group through the 41-50 year old cohort, declines to 54.3 percent for the 51-60 year olds, and then declines further to a low of 50.0 percent for the aged. The greater tendency of the less educated to abstain from campaign activity in both the 1956 and 1960 presidential elections accounts for much of the age cohort related pattern of continual inactivity shown in Chapter 3. Level of education has the most marked impact on the across time behavior of the elderly. The difference between the less and better educated elderly in the percentage who never participated in campaigns is over 30 percent, a difference greater than that between educational groups in all other age cohorts. In fact, in 1956-60, those with less than a Irigh school education accounted for 87.8 percent of the total pool 99 of older people who never participated in political campaigns, whereas the better educated accounted for only 12.2 percent of this pool. In 1972—76, there are also distinct differences between less and better educated age cohorts in the percentages of those initially inactive who never participated in political campaigns. However, the pattern of the percentages across both the less and better educated groups offers no clear support for the disengage- ment theory. For those with less than a high school education, continual inactivity from political campaigns is at its lowest level for those in the youngest age group, increases slightly for the 41-50 year olds, and then declines to 81.6 percent for older people. Among those with at least a high school education, the distribution of these percentages across age cohorts is much more uneven. Continual inactivity rises about 10 percent from the youngest group through the 31-40 year old cohort, declines to 71.8 percent for the 41-50 year olds, declines further to 66.7 percent for the 51-60 year old group, and then increases to 75.5 percent for the elderly. Across all age cohorts in 1972-76, the less educated are more likely than the better educated to abstain from participating in campaigns in both the 1972 and 1976 presidential elections. Once again, among the oldest age group those with less than a high school education account for a larger portion of the pool of con- tinual inactivities. The less educated elderly account for 67.7 100 percent of this pool, whereas the better educated elderly account for only 32.3 percent of the pool. Table 4—14 shows the difference between the percentage of disengagers and the percentage of new engagers for political cam- paign activity in each period. The contrasts between educational groups and between periods are marked. In 1956-60, among less educated cohorts this difference was a +21.l percent. Among the better educated, on the other hand, new engagers proportionately exceeded disengagers by 10.9 percent in 1956-60. Among both educational groups in 1972-76 the balance between disengagers and new engagers is in favor of the former group. Among the better educated in 1972-76, this difference was a +13.0 percent, and for the less educated group, a +40.3 percent. In the 1956-60 period, less educated older people have the largest positive difference. Across all age cohorts with less than a high school education, the distribution of the percentage dif- ferences fits the curvilinear pattern specified by the disengage— ment theory. The pattern is flat across the first two cohorts, declines across the 41-50 and 51-60 year old groups, and then in— creases in positive magnitude by 15 percent to a +32.1 percent for those in the oldest age cohort. Among better educated groups, on the other hand, those in the oldest cohort have one of the largest negative differences. High levels of education decrease the probability of disengagement and increase the probability of re-engagement from political campaign activity among all age co- horts. In fact, a comparison of the two educational groups within 101 each cohort shows that (with the exception of the 41-50 year old group) the difference is negative for the better educated and posi- tive for the less educated. The biggest contrast between educa- tional groups, moreover, is among those in the oldest age cohort. There is a 48.8 percent difference between the less and better educated elderly (-l6.7 percent minus +32.1 percent), whereas the percentage difference for all age groups combined is only 32.1 percent (-10.9 percent minus +21.1 percent). TABLE 4-14.--Difference Between the Percentage Who Disengaged and the Percentage Who Newly Engaged from Political Cam— paign Activity by Age Cohort and Level of Education, 1956-60 and 1972-76. i for Age Groupa All Age in Years: 21-30 31—40 41-50 51-60 61+ Groups Less than high school +28.3% +29.6% +23.3% +17.1% +32.l% +21.2% High school or more - 3.1% - 8.2% + 3.8% -17.6% -16.7% -10.9% B. 1972-76 Less than high school +39.3% +51.l% +57.4% +37.7% +26.6% +40.3% High school or more +14.4% +22.9% + 3.2% + 4.4% +13.3% +13.0% aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 p-anel . 102 In 1972-76, level of formal education also had a signifi— cant impact on the imbalance between disengagers and new engagers, but not to the extent that it did in 1956-60. Among those with less than a high school education, the value of the difference increases in magnitude across the 21-30 through the 41-50 year old groups. Moreover, from the 41-50 year old group through the oldest age cohort the value of this difference decreases in magnitude. In fact, among the less educated the value of the difference for the oldest age group of a +26.6 percent is much less in magnitude than that for any other age group. Among the better educated groups in 1972-76, the distribution of these percentage differences across age cohorts, on the other hand, fits the curvilinear pat- tern specified by the disengagement thesis. The difference rises from a +14.4 percent for the youngest group to +22.7 percent for the 31-40 year old group, falls to a +3.2 percent for the 41-50 year olds, flattens out, and then increases to a +13.3 percent for the oldest groups. Discussion The above analyses of the impact of level of education and sex on the relationship between age cohort membership and political disengagement support several conclusions about the across time political behavior of individuals in general and older people in particular. These compositional factors had a distinct impact on change across time in both the 1956-60 and 1972-76 time periods. 103 Our results for the impact of sex on changing levels of political participation across time support the conclusion that females are more likely than males to disengage regardless of how change in participation is measured. Moreover, this statement applies (with two exceptions to be discussed below) both to voting and political campaign activity and to each time period. More specifically, females are more likely than males to abstain from voting after having voted in a previous election, and they are also more likely than males to do so for political campaign activity. In addition, females are more likely than males to abstain com- pletely from voting and political campaign activity in two succes- sive elections. Finally, women are more likely to disengage and less likely to newly engage than are men for both voting and political campaign activity. When coupled with findings from cross-sectional studies of political participation which show that men are more likely than women to participate (for a comprehensive review see Milbrath and Goel, 1977), our results are hardly encouraging for those who desire to see more equal levels of political activity among men and women. Even more discouraging is our demonstration that the sexual imbalance among individuals in general in participation across time has persisted in two distinct time periods. Our findings for older people in particular are consistent with the sex differences in patterns of change across time for individuals in general with two exceptions. In 1956-60, among those who were of age 61 or more, men were more likely than women 104 to disengage from voting, and they were also more likely than women to do so for political campaign activity. This feature of the results is surprising since women in the older group for the 1956-60 panel, as well as those in the 1972-76 panel, were born prior to the ratification of the suffrage amendment for women in 1920. In fact, those 61 and above in 1956 were born in 1895 or earlier, and those 61 and above in 1972 were born in 1911 or earlier. If we accept the argument that the influence of this formative period would be to inhibit political participation among older women, it is dif- ficult to account for the higher proportionate incidence of dis- engagement among older men as compared to older women in the 1956-60 period. This explanation, however, can account for the differences in political disengagement between older men and women in the 1972-76 period. Our analyses of sex differences for 1956-60 also demon— strated, however, that older women are indeed at a disadvantage compared to older men for the other two indicators of over time change-~continual inactivity and the difference between the per- centage who disengaged and that who newly engaged. The difference between older males and older females for continual inactivity from both voting and political campaigns was massive compared to that for individuals in general. Moreover, analyses of the difference between the percentage who disengaged and that who newly engaged showed that older women are much less likely than older men to re-engage in voting or political campaign activity after abstaining :from such participation in a previous election. Apparently, the 105 impact of the pre-ratification period in America inhibited older women in 1956-60 from engaging in political participation, although it did not diminish political participation for older women who were initially politically involved in 1956. Our analyses of the changes between 1956—60 and 1972-76 in the proportionate extent of the actual withdrawal, or disengagement, of older men and older women from political campaigns demonstrated that older men were primarily responsible for the decline in the likelihood of disengagement from campaigns among older peOple. The decrease in the likelihood of disengagement from political campaigns between 1956-60 and 1972-76 for older men was over five times greater than the decrease for older women. In addition, the de- crease from 1956-60 to 1972-76 in the likelihood of continual in- activity from political campaigns among the aged was the result of the pattern for older men. Our results for the impact of level of formal education on changing levels of political participation across time clearly demonstrated that the less educated are more likely than the better educated to disengage, or actually withdraw, from both voting and political campaign activity. The less educated are also much more likely than the better educated to be continually inactive. Finally, the less educated are much less likely than the better educated to begin voting or participating in campaigns after ab- staining from such participation in a previous election. These results apply to both time periods. 106 High levels of education certainly decreased the probability of disengagement from voting among the aged. In fact, our analyses demonstrated for both time periods that among better educated individuals older people were the least likely cohort to disengage from voting. Moreover, within the class of better educated people, the old are no more likely to be continually inactive than are those in younger age groups for the act of voting. As the aged popula— tion becomes increasingly better educated, then (ceteris paribus) they will not be at a disadvantage compared to their educational counterparts in other age groups in their changing levels of voting across time. The less educated elderly, however, are more likely than those in younger age groups to disengage from voting and to be continual non-voters. Our results for political campaign activity showed a much different impact of education on the relationship between age co- hort membership and changing levels of activity. Most of the decline in the proportionate incidence of old age disengagement from campaigns between 1956-60 and 1972-76 is the result of a marked decrease in disengagement among lggg educated older people. Among less educated individuals in 1972—76, older people were least likely of all age cohorts to disengage from political cam- paigns. On the other hand, within the class of better educated peOple in 1972-76, older people were more likely to disengage from campaigns compared to those in some age groups but less likely tn) do so compared to those in other age groups. Similarly, in 1972-76, less educated older people were less likely than those in 107 most other age cohorts to never participate in political campaigns, whereas among better educated groups they were more likely to be continually inactive. In conclusion, our results confirm findings from cross-sec- tional analyses (See Chapter 1) about the impact of sex and level of formal education on political participation. We have, addi- tionally, shown that these sociological factors have a marked impact on changing levels of political activity across time. Now it is our task to fully exploit the panel feature of the design by considering the impact of change in individual circumstances (retirement from the labor force) and institutional circumstances (being contacted by a political party) on disengagement from political activity. To these analyses we now turn. NOTES TO CHAPTER 4 1For the convenience of the reader, the frequency distri- butions for our analyses here are shown in Appendices A4-l through A4-4 0 2Unfortunately, the combined impact of these variables could not be examined due to a lack of sufficient cases for analy— sis. RI' . . 3There is reason to suspect that the Survey Research Center oversamples females. This is due to the fact that within each primary sampling unit a sample of private households is obtained, and actual respondents are selected from these households. Fe- males are more likely than males to be at home, and consequently are oversampled. 4These are found by, first, multiplying the percentage who disengaged by the base N for each sex. This product yields the N for the pool of disengagers. Thus, for the oldest group in 1956-60, the total pool of disengagers is: (Percentage Who Disengaged) X (Base N) = N for Pool Males (5.1%) X (78) = 4 Females (4.8%) X (62) = 3 Total Pool N. . . . . . . . . . . . . . . . . . . . . . . . 7 The contribution of each sex to this pool is simply found by dividing the N for each sex by the total pool N. Thus, the con- tribution of males to this pool is 4/7 or 57.1 percent, and for females, 3/7 or 42.9 percent. This same procedure is used throughout this chapter. 108 CHAPTER 5 ASSESSING THE IMPACT OF CIRCUMSTANCES AND EVENTS ON POLITICAL DISENGAGEMENT In Chapter 4 we demonstrated how sex and level of formal education influenced the relationship between age cohort member- ship and political disengagement. The emphasis was placed on com— paring the proportionate incidence of political disengagement across cohorts. In this chapter our focus shifts from examining variations in disengagement across cohorts to an analysis of two key factors which contribute to variation in political disengage- ment within cohorts. Specifically, we shall, first, examine the dynamic impact of a socioeconomic factor, change in employment status or retirement, on disengagement from political activity among the aged in particular. Second, we will assess the impact of political institutions, the political parties, on disengagement from politics among individuals in general and older people in particular. Unfortunately, only the separate impact of each factor can be analyzed, since there are too few cases within each age cohort to conduct adequate multivariate analyses. In this chapter and Chapter 6, we shall employ tests of significance for first and second order Markov probability pro- cesses. In Chapters 3 and 4, no such tests were performed since Vie were interested in general patterns of political disengagement 109 llO across the entire range of age cohorts. In addition, we utilized precise hypothesized patterns to evaluate the relationship between age and disengagement from political activity. Here and in Chapter 6, the focus shifts primarily to the analysis of change within cohorts, and we shall show how Markov probability models are ideally suited to this task. Retirement and Political Disengagement Retirement or withdrawal from the labor force is a signal feature of social and economic disengagement and particularly prominent among older people. In fact, as discussed in Chapter 1 above, retirement was a key indicator of societal disengagement in Cumming and Henry's (1961) seminal work. Moreover, Nie, Verba, and Kim (1974) found retirement to be a key predictor of political activity among older people in their cross-sectional analysis of political participation in the United States. There are also compelling substantive reasons for con- sidering the impact of retirement on disengagement from political activity. For most older people, withdrawal from the labor force implies that they will take advantage of the benefits to be derived from governmental programs, most notably Social Security and Medicare. Given these benefits retired older people receive, we ‘vould anticipate that they should withdraw from political activity eat a greater extent than those older people who remain in the labor fierce. Moreover, it is reasonable to assume that political disen- gagement would accompany disengagement from the economic sphere. ‘- 111 We shall examine the utility of retirement as a predictor of changing levels of political activity across time. The panel nature of our design allows us, moreover, to examine the impact of retirement on political disengagement in a truly dynamic sense. That is, we can compare the incidence of political disengagement among those who were working at both time points of measurement with that among those who were working at the first time point of measurement but had retired by the second time point of measure- ment. This allows for a direct assessment of whether withdrawal from the economic sphere of society (that is, the labor force) is accompanied by withdrawal from the political sphere of society (that is, voting and political campaign activity). Table 5-1 shows the nature of change in employment status by age cohort in both the 1956-60 and 1972-76 time periods. In both time periods, retirement from the labor force is a signal feature of old age. In 1956-60, 54.5 percent of older people who worked in 1956 had retired by 1960, and in 1972-76, 56.8 percent of the aged who were working in 1972 had retired by 1976. The continuity of labor force participation for the remaining cohorts, on the other hand, is remarkable. In the 1972-76 period, however, we do see evidence of the increasing trend of early retirement for those in late middle age (that is, the cohort of ages 51—60). To the extent that withdrawal from the labor force is accomplished by ‘withdrawal from political activity, we can anticipate that poli- tical disengagement will increasingly become characteristic of those in late middle age. r 112 TABLE 5-1.--Change in Employment Status by Age Cohort, 1956-60 and 1972-76. i for Age Groupa All Age in Years: 21-30 31-40 41-50 51-60 61+ Groups A. 1956-6O Worked in 1956; 100.0% 99.5% 98.0% 94.4% 45.5% 92.7% worked in 1960 Worked in 1956; 0.0% 0.5% 2.0% 5.6% 54.5% 7.3% retired in 1960 Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) (129) (216) (196) (108) ( 77) (726) B. 1972-76 Worked in 1972; 100.0% 100.0% 99.4% 83.3% 43.2% 92.9% worked in 1976 Worked in 1972; 0.0% 0.0% 0.6% 16.7% 56.8% 7.1% retired in 1976 Total 100.0% 100,0% 100.0% 100.0% 100.0% 100.0% (N) (205) (161) (156) (150) ( 44) (716) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Table 5-2 shows the percentages of those in the oldest cohort who disengaged from voting by change in employment status in both time periods. In 1956-60, withdrawal from the labor force was accompanied by withdrawal from voting. Among those who dropped out of the labor force by 1960, 11.8 percent also withdrew from voting, whereas among those who remained in the labor force over the period absolutely no one disengaged from voting. 113 TABLE 5-2.--Percentage Who Disengaged from Voting Among Oldest Cohort by Change in Employment Status, 1956-60 and Time Period and Nature of Percentage Who Disengaged Change in Employment Status From Voting A. 1956-60 1. Worked in 1956; worked in 1960 0.0% (N) (28) 2. Worked in 1956; retired in 1960 11.8% (N) (34) B. 1972-76 1. Worked in 1972; worked in 1976 7.7% (N) (13) 2. Worked in 1972; retired in 1976 5.9% (N) (17) In the 1972-76 period, on the other hand, the pattern is reversed, and those who retired over the period are less likely than those who continued to work to disengage from voting. Among those who worked in both 1972 and 1976, 7.7 percent disengaged from voting, and among those who had retired only 5.9 percent did so. A more precise test that the transition probabilities for voting across time are not equal for the two change in employment status groups can be expressed as: 114 . w,t=l w,t=l H0. Vw,t=2 is equal to Vr,t=2 w,t=l w,t=l HA Vw,t=2 is not equal to V1,,t=2 Where V is the matrix of transition probabilities for voting behavior across time. t=1 refers to either the 1956 or 1972 presidential election; t=2, to either the 1960 or 1976 presidential election. And, w represents those who work, whereas r represents those who are re- tired. These matrices are shown in Figure A5-1 in Appendix 5. Since we are interested in the comparative instance of disengagement from voting, we shall compare the participant rows at t=1 across the two change in employment status groups. This comparison yields a two-by—two matrix in each period, and x2 with 1 degree of freedom is the appropriate test statistic (See Goodman, 1962; Markus, 1979). The nggl in 1956-60 is 3.534, and in 1972-76, .036, neither of which allow us to reject HO (assuming an alpha level of at least .05). Apparently, the argument of the disengagement thesis has limited utility in accounting for changes in aged voting behavior across time. That is, withdrawal from the work force among older people is not accompanied by withdrawal from voting in both time periods. There are no significant differences between the voting change patterns of the employment status groups. Table 5-3 shows the percentage of those in the oldest age cohort who disengaged from political campaign activity by change in employment status in each time period. In the 1956-60 period, Frithdrawal from the labor force appears to be accompanied by 115 disengagement from political campaigns. Among those who dropped out of the labor force between 1956 and 1960, 86.7 percent also disengaged from campaigns, whereas only 66.7 percent of those who worked at both time points did so. TABLE 5-3.--Percentage Who Disengaged from Political Campaigns Among Oldest Cohort by Change in Employment Status, 1956-60 and 1972-76. Time Period and Nature of Percentage Who Disengaged from Change in Employment Status Political Campaign Activigy A. 1956-60 1. Worked in 1956; worked in 1960 66.7% (N) (12) 2. Worked in 1956; retired in 1960 86.7% (N) (15) B- _1_9__72:_7_6_ 1. Worked in 1972; worked in 1976 50.0% (N) ( 6) 2. Worked in 1972; retired in 1976 50.0% (N) (12) In the 1972-76 period, there is absolutely no difference between the change in employment status groups in disengagement from campaigns. Exactly one half of those in each group withdrew from campaigns between 1972 and 1976. This is similar to what we found for disengagement from voting: in the 1956-60 period with- drawal from the labor force appears to be accompanied by 116 withdrawal from political activity (but not significantly), and in 1972-76, it is not. Once again, we can perform a more precise test to consider whether or not the transition probabilities for campaign activity across time are equal for the two change in employment status groups. The hypotheses can be expressed as: W t=1 W t=1 H : P ’ ' ’ O w,t=2 18 equal to Pr,t=2 . w,t=l w,t=l HA. Pw,t=2 is pp£_equal to Pr,t=2 Where P is the matrix of transition probabilities for political campaign activity across time. t=1 refers to either the 1956 or 1972 presidential election; t=2, to either the 1960 or 1976 presidential election. And, w represents those who work, whereas r represents those who are retired. These matrices are shown in Figure A5-2 in Appendix 5. Once again, the comparison of interest is that of the participant rows at t=1 across the two change in employment status groups. The comparison yields a two-by-two matrix in each period, and x2 with 1 degree of freedom is the apprOpriate test statistic. The x§f=1 in 1956-60 is 1.54, and in 1972-76, identically zero, and neither of these values, of course, allow us to reject H0 (assuming alpha equals .05). Once again, the argument of the disengagement thesis has limited utility in accounting for changing levels of political campaign activity across time. Although there was a difference between the employment status groups in 1956—60, our more precise test shows that it is not significant. 117 Institutions and Political Disengagement: The Impact of the Political Party The activity or event of being contacted by a political party requires no individual initiative or commitment. Participa- tion in this type of activity requires that a social organization, the political party, undertake the initiative. This is a direct measure of a societal type of political activity since the indi- vidual must be sought out by society to qualify as a participant in being contacted by a political party. Moreover, the society-- specifically, the political party--can effectively exclude peOple by simply ignoring or discounting them. However, it may also actively solicit and encourage their support. Older people can, in principle and possibly in fact, be engaged in politics if political institutions like the party solicit their support or demands through the act of contacting. Previous studies of political participation have demon— strated that party contact does indeed contribute to enhanced levels of individual political activity. As Milbrath and Goel (1977, p. 136) point out, "one of the most important stimulants to political participation is personal contact by a party worker . . It has been found repeatedly that persons contacted by party workers were more likely to vote and also to participate in gladiatorial activities." There are suggestive theoretical reasons for examining the impact of institutions on individual political activity. Verba, Nie, and Kim (1978) formulated a conceptual framework for the 118 analysis of participation which emphasized institutionally based processes of political mobilization. According to the logic of their argument, political participation is a joint function of individual and institutional factors. Moreover, voting and poli- tical campaign activity are two modes of participation which are particularly susceptible to institutional stimulation, especially by political parties (Verba, Nie, and Kim, 1978, pp. 73-74). The impact of the political party on political participa- tion is also important for substantive reasons. The Democratic and Republican parties recognize both the potential and actual political importance of older peOple. This recognition has con- tributed to the development of special appeals and organizational devices, such as Senior Democrats or Republicans, which are designed to stimulate political participation among the elderly. In fact, Anne Lewis, the political director of the Democratic National Committee has asserted that, "the elderly are our single most important national constituency." (See Light, 1981, p. 2343.) We are interested in the dynamic impact of party contact on disengagement from political activity, since the primary issue is whether or not institutional withdrawal from the individual is accompanied by withdrawal from politics by that individual. Thus, Table 5-4 shows change in party contact status by age cohort in ‘both the 1956-60 and 1972-76 time periods. In each period, parties are more likely to contact people in both elections than they are to withdraw from contacting people. This applies to each cohort in both time periods. In addition, parties were much more diligent in 119 their contact efforts in 1972-76 as compared to 1956-60. In 1956—60, the percentage contacted in both presidential elections never ex— ceeds 70 percent, whereas in 1972-76, continual party contact never falls below 86.5 percent. To the extent to which continual party contact decreases the probability of individual disengagement from politics, its impact will be greater in 1972-76 than it was in 1956-60. TABLE 5-4.--Change in Party Contact Status by Age Cohort, 1956-60 and 1972-76. X for Age Groupa All Age in Years: 21-30 31-40 41-50 51-60 61+ Groups A. 1956-6O Contacted in 1956; contacted in 1960 70.0% 65.9% 52.5% 57.8% 54.0% 60.4% Contacted in 1956; not contacted in 1960 30.0% 34.1% 47.8% 42.2% 46.0% 39.6% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) ( 50) ( 91) ( 67) ( 45) ( 50) (303) B. 1972-76 Contacted in 1972; contacted in 1976 86.5% 91.2% 87.7% 90.6% 89.0% 88.8% Contacted in 1972; not contacted in 1976 13.5% 8.8% 12.3% 9.4% 11.0% 11.2% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% (N) (223) (170) (155) (139) (172) (859) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 120 Table 5-5 shows the percentage who disengaged from voting by age cohort and change in party contact status in each time per- iod. These results show that for most of the age cohorts in each time period contact by a political party has absolutely no impact on individual disengagement from voting. For example, in 1956-60, party contact decreases the probability of disengagement from voting only among those in the 31-40 and 41-50 year old cohorts. For all other cohorts, none of the people in either party contact group disengaged from voting between 1956 and 1960. Similarly, in 1972-76, party contact decreased the probability of disengagement from voting only among those in the 31-40 and 41-50 year old groups. In fact, among the remaining age cohorts party contact actually increased the probability of disengagement from voting between 1972 and 1976. Even for the 31-40 and 41-50 year old groups whose probability of disengagement from voting across time was decreased by continuous party contact, a test of equal transition matrices across party contact groups reveals no significant differences in the transition probabilities for across time voting. (See Figure A5-3 in Appendix 5 for these transition matrices.) The value of de=l for the 31—40 year old group in 1956-60 is 2.720, and in 1972-76, a .693. Neither value allows us to reject the hypothesis of equal transitional probabilities between party contact groups (that is, assuming an alpha value of at least .05). Similarly, the value of de=l for the 41-50 year old cohort in 1956-60 is a 2.160, and in 1972-76, a .035. Thus, in 1972-76, we also fail to 121 reject the equal transition hypothesis between party contact groups for the 41-50 year old cohort. Party contact has absolutely no impact on disengagement from voting irrespective of age cohort membership and time period. TABLE 5-5.--Percentage Who Disengaged from Voting by Age Cohort and Change in Party Contact Status, 1956-60 and 1972-76. i for Age Group All Age in Yearsza 21-30 31—40 41-50 51-60 61+ Groups A. 1956-60 1. Contacted in 1956; contacted in 1960 0.0% 2.5% 0.0% 0.0% 0.0% 1.0% (N) (17) (40) (27) (13) (ll) (96) 2. Contacted in 1956; not contacted in 1960 0.0% 13.0% 7.7% 0.0% 0.0% 5.2% (N) (10) (23) (26) (18) (19) (96) B. 1972-76 1. Contacted in 1972; contacted in 1976 15.9% 3.5% 9.2% 2.1% 8.3% 8-2% (N) (132) (113) (98) (96) (96) (535) 2. Contacted in 1972; not contacted in 1976 12.0% 8.3% 10.5% 0.0% 6.3% 6.0% (N) (25) (12) (19) (11) (16) (83) aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Table 5-6 shows the percentage who disengaged from poli- tical campaigns by age cohort and change in party contact status 122 in both time periods. In the 1956-60 period, party contact de- creased the probability of disengagement from political campaigns only for those in the 41-50 and oldest cohorts. However, the im— pact of the party on disengagement from campaigns was incredibly small for both of these cohorts, and the difference between contact groups within each cohort never exceeds 3 percent. For the re— maining cohorts, party contact actually increased the probability of disengagement from political campaigns. TABLE 5-6.--Percentage Who Disengaged from Political Campaign Activ- ity by Age Cohort and Change in Party Contact Status, 1956-60 and 1972-76. i for Age Groupa All Age in Years: 21-30 31-40 41-50 51-60 61+ Groups A. 1956-60 1. Contacted in 1956; contacted in 1960 75.0% 55.6% 75.0% 84.6% 90.0% 72.1% (N) ( l6) ( 27) ( 20) ( 13) ( 10) ( 86) 2. Contacted in 1956; not contacted in 1960 25.0% 50.0% 77.8% 77.8% 92.3% 72.2% (N) ( 4) ( 10) ( 18) ( 9) ( 13) ( 54) B. 1972-76 1. Contacted in 1972; contacted in 1976 49.1% 47.9% 40.9% 40.7% 39.2% 44.6% (n) (116) ( 71) ( 66) ( 59) ( 51) (363) 2. Contacted in 1972; not contacted in 1976 57.1% 28.6% 50.0% 71.4% 50.0% 52.2% (N) ( l4) ( 7) ( 12) ( 7) ( 6) ( 46) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972—76 panel. 123 As our results above demonstrated, parties became more diligent in 1972-76 in their contact efforts, and Table 5-6 shows that this enhanced diligence apparently contributed to a lower probability of disengagement from political campaigns among all age cohorts except those in the 31-40 year old group. Among those in all other cohorts, those who were contacted in both 1972 and 1976 were less likely to disengage from campaigns than were those who were contacted in 1972 but not in 1976. There is at least a 9 per- cent difference between party contact groups within any given age cohort. Once again, a more precise test of whether party contact made any difference in individual disengagement from political campaigns can be performed by comparing the matrix of transition probabilities for each group. The hypotheses can be expressed as: . c,t=l . c,t=l HO. Pc,t=2 is equal to Pnc,t=2 . c,t=l . c,t=l HA. Pc,t=2 is not equal to Pnc,t=2 Where P is the matrix of transition probabilities for political campaign activity across time in the 1972-76 period. t=1 refers to the 1972 presidential election; t=2, to the 1972 presidential election. And, c repre- sents the group contacted by a party, whereas nc rep- resents the groups not contacted by a party. The matrix for each cohort (except those in the 31-40 year old group) is shown in Figure A5-4 of Appendix 5. Once again, x2 with 1 degree of freedom is the appropriate test statistic. The value of x§f=1 is .320 for the 21-30 year old group, .341 for the 41r50 year olds, 2.390 for the 51—60 year olds, 124 and .260 for those of ages 61 and above. None of these results are statistically significant (assuming an alpha of at least .05), and do not allow us to reject H 0. Discussion and Conclusions Our analyses in this chapter lend much doubt to previous notions in the literature about circumstances and events which contribute to disengagement from political activity. Moreover, the panel feature of our design has allowed us to conduct direct tests of the impact of changes in individual and institutional circumstances on political disengagement. We have shown that the hypothesis purporting that with- drawal from the labor force is accomplished by withdrawal from political activity has little utility in accounting for the across time behavior of the elderly. Retirement from the labor force made little or no difference in both the 1956-60 and 1972-76 time periods in the propensity of older people to disengage from voting or political campaign activity. There are no significant dif- ferences in either period in the transition probabilities between the group who remained in the labor force over the period and those who withdrew from the labor force over the period. Theoretically, our results for the impact of retirement on old age disengagement from political activity cast doubt on the compensation or distraction hypothesis originally proposed by Glenn and Grimes (1968). As they state (1968, p. 564): 125 An increase in political interest may compensate for loss of Opportunity or inclination to attend to other interests . . . A related hypothesis is that political interest is inversely related to degree of involvement with personal problems and ambitions, the exigencies of day-to-day living, and non-political interests . . . When the husband retires, both male and female have all too little with which to be preoccupied. Our analyses lead us to view this hypothesis negatively at least insofar as retirement can be viewed as a reduction of dis- tracting influences. It seems reasonable to assume that retirement does indeed lead to increased levels of free time, ceteris paribus (for example, good health). However, it does not contribute to lower levels of disengagement from political activity. Our results have also shown that the political party has virtually no impact on the propensity of disengagement from voting and political campaign activity among individuals in general, and older people in particular, in both the 1956—60 and 1972-76 time periods. There were no marked or significant differences in proportionate disengagement between those who were contacted by a party at both elections and those who were contacted at one elec- tion but not at the proceeding one. Moreover, even though parties were more diligent in their contact efforts in 1972-76, this en- hanced diligence has no pay—off in terms of reducing the likelihood of disengagement from political activity among individuals. Theoretically, our notion about the impact of institutions, specifically political parties, on political disengagement, adds to the framework of Verba, Nie, and Kim's (1978) distinction between individual and institutional processes of political 126 mobilization. The argument posed here was that institutions can influence the relationship between age and changing levels of political participation across time. Although empirically our results lent no support to this notion, further work is needed in this area. Specifically, future analyses should look at the impact of other institutions, such as interest groups, on political dis- engagement. Furthermore, the impact of special units of the political party, such as senior citizens' committees, should be studied. Future research on the impact of the circumstances and events which attend aging must, in addition to specifying theoreti- cal issues, be conducted with improved research designs. In particular, these issues could be addressed through the use of specialized samples. Such samples could assess the impact of a wide variety of circumstances and events on political disengage- ment. For example, the impact of the geographical location of the elderly on political disengagement could be assessed by comparing the participation levels of older people living in specific loca- tions or areas which are primarily populated by the elderly with those of older peeple living in areas with a more balanced mix of ages. CHAPTER 6 CONTINUITY AND CHANGE IN POLITICAL ACTIVITY ACROSS A SERIES OF THREE SUCCESSIVE ELECTIONS In this chapter we shall investigate the nature of changing levels of political activity across a series of three successive elections.1 The panel nature of our design is uniquely equipped to investigate issues of importance both to the study of political participation in general and that among older peOple in particular. Specifically, we can assess the extent to which the characteristics of specific elections influence changing levels of political activ- ity across time. Additionally, we can assess the issue of whether levels of political activity across time are continuous or inter- mittent and discontinuous. Of particular concern, finally, is whether or not older people are less continuous in their political activity across time than are those in younger age groups. Previous studies of political participation have demon- strated that people are more likely to participate in elections perceived as important (See, among others, Campbell, 1960), and that elections for higher office, such as the presidency, encourage higher levels of participation than those for lower offices, such as congressional elections (Milbrath and Goel, 1977). As Milbrath 127 128 and Goel (1977, p. 139) contend, "national elections are nearly always perceived as more important than local elections, and turn- out is nearly always higher for national elections." Although similar data for political campaign activity are nonexistent, statistics for voting turnout among the United States population in the 1956-58—60 and 1972-74-76 time periods show that participation is indeed higher in the presidential elections than in the off-year elections. According to official estimates of turnout, 59.3 percent of the population voted in 1956, 43.0 per- cent in 1958, and 62.8 percent in 1960. Similarly, 55.5 percent of the population voted in 1972, 36.1 percent in 1974, and 54.4 percent in 1976.2 With the panel design one can examine the actual extent of individual change across one presidential election, the subsequent off—year election, and the next proceeding presidential election. Our expectation is that individual change in participation between one presidential and the next off-year election seriously under- estimates the extent and nature of individual change between the off-year election and the subsequent presidential election. More- over, although the absolute level of political participation across time will vary across age cohorts, we anticipate that this hypothesis is one which applies to the electorate in general. We shall also entertain hypotheses about the durability of political participation across the three elections in each time period. The anticipation is that political behavior should be almost constant across time for those who do indeed participate in 129 off-year elections. As Campbell and his colleagues (1960, p. 92) argued, "it is plausible to think of voting as a type of conduct that is somewhat habitual and to suppose that as the individual develops a general orientation toward politics he comes to in- corporate either voting or non-voting as part of his normal behavior." It is also plausible to assume that this argument may also apply to political campaign activity. The disengagement thesis contends that this durability of political behavior across time is not characteristic of older people. However, others argue that participation across time should be as durable for older people as it is for the young and middle-aged. They propose a life-long persistence model of behav- ior, and contend that "the amount and style of an aged person's participation are an extension of an emerging pattern begun in childhood and shaped in earlier adulthood" (Videbeck and Knox, 1965, p. 29; also, see Sigel and Hoskin, 1977). That is, according to this continuity thesis, if one was politically active and in- volved in middle age, or younger, then he or she will continue to be politically active and involved in later life; those who were not active and involved to begin with, will not be in old age. Our analysis will address these theoretical arguments pri- marily with the probability model of a Markov process (See Ross, 1972; Markus, 1979). The panel feature of our design, together with this probability model, allows for a rigorous and appropriate methodological framework to test the above hypotheses. 130 Continuity and Change in Voting Behavior Continuity and change in voting behavior across a series of three elections can be assessed with the probability model of a Markov process, and our panel design is uniquely equipped to em- ploy this type of probability model. We can examine, first, the hypothesis that participation at time t is strictly a function of participation at t-l (that is, the process is first order Markovian); and, second, that participation at time t is a joint function of participation at times t-l and t-2 (that is, that the process is second order Markovian). As discussed above, our hypothesis specifically asserts that voting behavior across one presidential election, the sub- sequent off-year election, and the proceeding presidential election cannot adequately be viewed as a first order Markov process. That is, participation in voting in a presidential election predicts voting participation in an off-year election less well than voting in an off—year election predicts participation in the proceeding presidential election. Our results show that among those in each age cohort who voted in the 1956 presidential election, 69 percent of those 21—30, 77 percent of those 31-40, 83 percent of those 41-50, 90 percent of those 51-60, and 88 percent of those 61 and above voted in the 1958 off-year election. On the other hand, among those in each cohort who voted in the 1958 off-year election, 94 percent of those 21-30, 97 percent of those 31-40, 98 percent of those 41-50, 98 percent of those 51-60, and 99 percent of those 61 and above 131 voted in the 1960 presidential election. Similarly, among those in each age cohort who voted in the 1972 presidential election, 67 percent of those 21-30, 76 percent of those 31-40, 83 percent of those 41—50, 87 percent of those 51-60, and 83 percent of those 61 and above voted in the 1974 off-year election. On the other hand, among those who voted in the 1974 off-year election, 93 per- cent of those 21-30, 96 percent of those 31-40, 96 percent of those 41-50, 97 percent of those 51-60, and 92 percent of those 61 and above voted in the 1976 presidential election. (See Appendices A6—l and A6-2 for the transition matrices for each age cohort in each period.) The support for our hypothesis about cross-election voting is remarkable across the entire set of age cohorts in each time period. The extent to which a first order Markov process is in- adequate for predicting changing levels of voting across time can be tested more precisely (that is, with known probabilities of error), first, by comparing the predictions for voting behavior at t+2 under the assumption of a first order process with the per- centage voting behavior actually observed at t+2; and, second, by testing the hypothesis that the transition matrix for voting be- havior between t and t+l is equal to that between t+l and t+2. This second test can be formally expressed as: H0: V from t=1 to t=2 is equal to V from t=2 to t=3 HA: V from t=1 to t=2 is not equal to V from t=2 to t=3 132 Where V is equal to a matrix of transition probabilities. V for each group in each time period is shown in Appen— dices A6-l and A6-2. In addition, t=1 refers to either the 1956 or the 1972 presidential election; t=2 to the 1958 or 1974 off-year election; and t=3, to the 1960 or 1976 presidential election. Table 6-1 shows the expected proportionate voting turnout at t=3 under the assumption of a first order Markov process, and the actual voting turnout at t=3 for each age cohort in both time periods. These results show that the assumption of a constant transition matrix for change in voting behavior between 1956, 1958, and l960--as well as between 1972, 1974, and l976--is unreasonable for those in all age cohorts. The essential problem with this assumption of a constant transition matrix V is that it severely underestimates voting turnout in the presidential election follow- ing the off-year election for those in each age cohort in both time periods. As Table 6—1 shows, the discrepancies between expected and actual turnout are marked and never fall below 10 percent. The x2 statistic can be calculated for each age cohort in each period to test the hypothesis of a constant transition matrix for changing levels of voting across time. The complete transition matrices for each cohort in both periods are shown in Appendices A6-l and A6-2. For example, for the oldest age cohort in 1956-58-60, we calculate x2 for each section of Figure A6-1E in Appendix A6-1 (that is, for the participant section and the non-participant section). Each section is a two-by-two table and yields a x2 with 1 degree of freedom. The appropriate test statis— tic is the sum of the two x: for each section, and is a x2 with 2 133 degrees of freedom. For the oldest group in 1956-58-60, X2df=l for the participant section is 85.06 and x2 for the non-participant sec- tion is 5.94, both being significant at the .05 alpha level. More- over, the combined x2 with df=2 is 91.00, which is also beyond any reasonable significance level (alpha less than .01). Thus, the as- sumption of a constant transition matrix for the oldest cohort is not tenable, and we can confidently reject HO. TABLE 6-1.--Comparison of Expected (Under First Order Assumption) Voting Turnout at t=3 with Actual Voting Turnout at t=3, by Age Cohort, 1956-58-60 and 1972-74-76. A. 1956-58-60 Expected Actual ,Age Group in Years:8 1960 1960 Expected—Observed 21-30 .43 .77 -.34 31-40 .53 .82 -.29 41-50 .72 .88 —.16 51-60 .76 .86 -.10 61+ .65 .79 —.14 B. 1972-74-76 Expected Actual Age Group in Years:3 1976 1976 Expected—Observed 21-30 .47 .77 -.30 31-40 .56 .83 -.27 41-50 .68 .84 -.16 51-60 .68 .82 -.14 61+ .60 .73 -.13 aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Table 6-2 shows the results of this test for a first order process for each cohort in each time period. These results clearly demonstrate the failure of the constant transition matrix assumption 134 to describe changing levels of voting for each cohort in each time period. All values of x2 are significant at least at the .01 alpha level. TABLE 6-2.--Summary of Results for Hypotheses Tests of Constant Transition Matrix (First Order Process) for Voting, by Age Cohort, 1956-58-60 and 1972-74-76. A. 1956-58-60 Age Group in Years:8 Test Statistic3(x§f=2) Decision Rule 21-30 69.73** Reject H0 31-40 72.68** Reject H0 41-50 34.57** Reject HO 51-60 9.55** Reject H0 61+ 90.99** Reject HO B. 1972-74-76 Age Group in Years:a Test Statistic (x§f=2) Decision Rule 21-30 60.49** Reject HO 31-40 43.45** Reject H0 41-50 19.53** Reject H0 51-60 15.l7** Reject HO 61+ 13.03** Reject HO aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. A more reasonable contention for changing levels of voting across the three elections in each time period is that participa- tion at t=3 (that is, 1960 or 1976) is a joint function of partici- pation at t=1 and t=2, being the assumption that the process is second rather than first order Markovian. That is, we can better predict if one will vote or abstain in 1960 (or 1976) if we know 135 whether or not they voted or abstained in 1956 and 1958 (or in 1972 and 1974). Figure A6-3E in Appendix A6—3 shows the second order transition matrix in 1956-58—60 for those in the oldest age cohort. An examination of these proportions clearly shows that the transi- tion probabilities linking voting behavior in 1958 and 1960 vary with 1956 voting behavior. The probability, for example, that someone in the oldest age cohort voted in both 1958 and 1960 is unity if that individual also voted in 1956. This probability, on the other hand, is only .75 if the respondent abstained from voting in 1956. A more formal test that changing levels of voting across time is second as Opposed to first order Markovian can be expressed as: H0: The transition process is first order. HA: The transition process is second order. Figures displaying the second order transitions matrices for voting for each age cohort in both time periods are presented in Appendices A6-3 and A6-4. Once again, the x2 statistic can be used to test the above hypothesis that the transition process for change in voting across time is second as opposed to first order. The matrix is simply divided to form two subtables, a x2 value with df=l is computed for each subtable, and the two x: with df=l are summed to yield a testlx2 with df=2. For our oldest group in 1956-58-60, the de=l for the upper half of the matrix is 32.58, 136 the de=l for the lower half of the matrix is 11.83, and the com- bined value of x2 is 44.41 (df=2). These values are significant beyond an alpha of .01, and thus allow the rejection of the null hypothesis that the process is first order Markovian. Table 6-3 shows the results of the hypotheses tests pitting a first order interpretation against a second order interpretation of the transition process for voting across time for each cohort in both periods. For each age group in each time period the null hypothesis purporting that the process is first order is rejected. The consistency with which we can reject the null across all age groups is remarkable. All x2 values are significant at at least the .05 alpha level. Given that our analysis has demonstrated support for the hypothesis that voting behavior at t=3 is a joint function of voting behavior at t=1 and t=2 (where t=1, t=2, and t=3 are defined as above), our next task is to consider the hypoth- esis that older people are less continuous in their voting behavior than are those in younger age groups. According to the disengage- ment thesis, the expectation is that older peOple are less likely than those in younger age groups to be continuously active in voting over a series of elections. Specifically, we expect that the percentage of the oldest age cohort who were active in each of the three elections will be markedly smaller than the percentages for younger age cohorts who were continually active over these elections. In addition, we anticipate that continual inactivity should be greater among the oldest group than among all others. 137 TABLE 6-3.--Summary of Results for Hypotheses Tests for First vs. Second Order Markov Processes for Voting, by Age Cohort, 1956-58-60 and 1972-74-76. A. 1956-58-60 Age Groupjin Years:a Test Statistic3(x§f=2) Decision Rule 21-30 53.15** Reject HO 31-40 32.28** Reject HO 41-50 38.43** Reject H0 51-60 15.58** Reject HO 61+ 44.4l** Reject HO B. 1972-74-76 Age Groupgin Years:a Test Statistic (x§f=2) Decision Rule 21-30 8.81* Reject H0 31-40 41.39** Reject HO 41-50 16.82** Reject H0 51-60 33.7l** Reject HO 61+ 28.90** Reject H0 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Table 6-4 shows the results for continual non-voting over a series of three elections by age cohort in both time periods. In 1956-58-60, the pattern of the percentages across the range of age cohorts does not support the disengagement thesis. In fact, all of the older people who voted in both 1956 and 1958 voted again in 1960! Moreover, reported voting behavior across time is in- credibly constant for those in all age cohorts. The percentage of those continually active with any given age cohort never falls below 97.4 percent. Participation in an off-year election clearly contributes to stable voting across an entire series of elections in 1956-58-60. 138 TABLE 6-4.--Continua1 Activity in Voting by Age Cohort, 1956-58-60 and 1972-74-76. R for Age Groupa All Age in Years: 21-30 31—40 41-50 51-60 61+ Groups A. Percent active in 1960 who were active in both 1956 and 1958 97.5% 97.4% 98.2% 98.3% 100.0% 98.2% (N) (118) (190) (170) (118) (112) (708) B. Percent active in 1976 who were active in both 1972 and 1976 93.1% 98.5% 96.1% 98.5% 93.8% 96.0% (N) (144) (133) (154) (136) (130) (697) 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. As Table 6-4 shows, there is also no evidence in 1972-74-76 to support the disengagement hypothesis that older people are less likely compared to those in other age groups to vote across the entire series of elections. Among older people who voted in both 1972 and 1974, 93.8 percent also voted in 1976. Moreover, as was the case in 1956-58-60, reported voting behavior across time in 1972-74-76 is incredibly durable among those who participated in the off-year election irrespective of age cohort membership. For no age group does the percentage fall below 93.1 percent. That voting behavior among participants in off-year elections is so robust in two distinct time periods clearly suggests that changing levels of voting across time are primarily a joint function of participation in the two preceding elections. 139 A more precise and exact assessment of the disengagement hypothesis can be performed through a series of tests which con- sider whether the second order transition matrix for the oldest cohort is pairwise different and greater than that for each other age cohort. Specifically, we are interested in comparing the transition probabilities for the continually active row of the second order matrix of the oldest age cohort pairwise with all others. Our research hypothesis will be that of the disengagement thesis--viz., that older people are less continuous in their voting as compared to those in each other age group. This hypoth- esis will be pitted against a continuity hypothesis which prOposes that there are no differences between the oldest group and all others in their continual activity in voting across time. More formally, these rival hypotheses can be expressed as: HO: Vca for those of ages 61 and above is pairwise no different for each other cohort-—the continuity hypothesis. HA: Vca for those of age 61 and above is pairwise dif- ferentzuuiless than the transition probabilities of Vca for each other cohort--the disengagement hypothesis. Where Vca is the continually active row of the second order transition matrix. In addition, the transition matrices for this test are shown in Appendix A6-5. Table 6-5 shows the results of these hypotheses tests for continual activity in voting behavior over three successive elec- tions in both the 1956-58-60 and 1972-74-76 time periods. In 1956-58-60, the continuity hypothesis cannot be rejected in favor 140 of the disengagement hypothesis for each pairwise comparison. This is consistent with the results of our test in Table 6-4 above. In 1972-74-76, however, for the pairwise comparisons between the old- est group and the 31-40 and 51-60 year olds the continuity hypoth- esis can be rejected in favor of the disengagement hypothesis. This result is due to the unusual durability of voting across time for the 31—40 and 51-60 year old groups. Although 93.8 percent of older people voted across the entire series of elections, 98.5 per- cent of those in the 31-40 and 51-60 year old cohorts did so. Voting behavior across time is incredibly stable for individuals in general and for older people in particular. Although the elderly were among the most durable voters in 1956-58-60, Table 6-6 shows that the percentage incidence of con- tinual non-voting is at its highest level among the elderly. Among those of age 61 and above who abstained from voting in both 1956 and 1958, 82.9 percent also abstained from voting in 1960. Con- tinual non-voting across the entire series of three elections increases gradually from the youngest cohort through the 51-60 year old group but then levels off across the two oldest cohorts. In the 1972-74-76 period, the percentage incidence of con- tinual non-voting is certainly not greatest among older people (See Table 6-6). Although 81.2 percent of those of age 61 and above who did not vote in both 1972 and 1974 also did not vote in 1976, this percentage is slightly less than those for the 41-50 and 51-60 year old groups. Continual non-voting in 1972-74—76 is clearly not a unique feature of old age; these results do not 141 furnish any support for the disengagement thesis. Finally, as was the case in 1956-58-60, those in the two youngest groups are most unlikely to continue abstaining from voting over a series of elections. TABLE 6-5.--Summary of Results for Hypotheses Tests for Similar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Activity in Voting, 1956-58-60 and 1972-74-76. A. 1956-58-60 Comparison of Vca for oldest cohort with that for: Age Cohort in Years:a Test Statistid3(x3f=1) Decision Rule 21-30 2.88 Do not reject HO 31-40 2.99 Do not reject H0 41-50 1.99 Do not reject HO 51-60 1.90 Do not reject HO B. 1972-74-76 Comparison of Vca for oldest cohort with that for: Age Cohort in Years:3 Test Statistic (de=l) Decision Rule 21-30 0.07 Do not reject HO 31-40 3.89* Reject H 41-50 .76 Do not reject H0 51-60 4.02* Reject HO 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Once again, a more precise test of the disengagement hypothesis can be performed by comparing the transition probabil- ities for the continually inactive row of the second order matrix 142 for the oldest group pairwise with all others. The disengagement hypothesis will be that older people are more likely to be con— tinually inactive in their voting behavior as compared to those in each other age cohort. The rival hypothesis contends that there are no differences between the oldest group and all others in their continual non-voting across time. These hypotheses can be formally expressed as: HO: VCi for those of ages 61 and above is pairwise no different from Vci of each age cohort--the continuityfhypothesis. H : V . for those of ages 61 and above is pairwise dif- ferent from, and greater than, the transition probabilities for Vci for each other age co- hort--the disengagement hypothesis. Where Vci is the continually inactive row of the second order transition matrix. In addition, the transition matrices for this test are shown in Appendix A6-6. TABLE 6-6.--Continual Inactivity from Voting by Age Cohort, 1956-58-60 and 1972-74-76. R for Age Groupa All Age in Years: 21-30 31—40 41-50 51-60 61+ Groups A. Percentage inactive in 1960 who were inactive in both 1956 & 1958 59.8% 61.1% 73.0% 82.6% 82.9% 67.6% (N) ( 82) ( 67) ( 37) ( 23) ( 35) (244) B. Percentage inactive in 1976 who were inactive in both 1972 & 1974 56.7% 64.9% 81.8% 83.9% 81.2% 71.2% (N) ( 60) ( 37) ( 22) ( 31) ( 48) (198) 3Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 143 Table 6—7 shows the results of these hypotheses tests for continual non-voting in both time periods. In the 1956-58-60 period, for the pairwise comparisons between the oldest group and the 21-30 and 31-40 year olds the continuity hypothesis can be rejected in favor of the disengagement hypothesis. If we refer back to Table 6-6, the 82.9 percent of older people who were scored as continually inactive from voting in 1956-58-60 is clearly greater than the 59.8 percent of the 21-30 and the 61.1 percent of the 31-40 year old groups who did so. However, continual inactiv- ity from voting in 1956-58-60 is not a unique feature of old age since levels of continual inactivity are also very high for those in the 51-60 year old group (See Table 6—6 above). In the 1972-74-76 period, we can reject the continuity hypothesis in favor of the disengagement hypothesis only for the comparison between the oldest cohort and the youngest one. The level of pr0portionate continual inactivity in the oldest group is markedly higher than that for the young. As shown in Table 6—6 above, only 56.7 percent of the 21-30 year olds who did not vote in 1972 abstained from voting across the series of elections, whereas over 80 percent of older people who did not vote in 1972 did so. Once again, however, high levels of continual non-voting are not unique for older people in 1972-74-76. Over 80 percent of those in the 41-50 and 51-60 year old age groups were also continual non-voters in this period. 144 TABLE 6—7.--Summary of Results for Hypotheses Tests for Similar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Inactivity From Voting, 1956-58-60 and 1972-74-76. A. 1956-58-60 Comparison of Vc Age Cohort in Years:3 21-30 31-40 41-50 51-60 B. 1972-74-76 Comparison of V . c1 Age Cohort in Years:a 5.90* 4.99* 1.02 0.00 Test Statistic (x§f=1) 21-30 31-40 41-50 51-60 7.37** 2.92 0.00 0.09 for oldest cohort with that for: 3 Test Statistic (de=l) Decision Rule Reject H Reject H Do not reject H Do not reject HO for oldest cohort with that for: Decision Rule Reject H Do not reject H Do not reject H Do not reject H0 aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Continuigy and Chagge in Influence Attempts Once again, we can examine continuity and change for the political campaign act of attempting to influence the vote of others across a series of three successive elections with the probability model of a Markov process. We will entertain the status of a first order process, and check the discrepancies, 145 if any, between the predicted and actual results. And, we shall pit a first order process against a second order one. Our hypotheses, once again, are that participation in the act of attempting to influence the vote of others cannot be ade- quately viewed as a first order Markovian process across one presidential election, the subsequent off-year election, and the proceeding presidential election. That is, participation in the act of attempting to influence another's vote in a presidential election predicts participation in the proceeding off-year elec- tion less well than participation in this act in an off-year election predicts participation in the proceeding presidential election. Our results show that among those in each age cohort who participated in the 1956 presidential election, 44 percent of those 21-30, 31 percent of those 31-40, 32 percent of those 41-50, 43 percent of those 51-60, and 29 percent of those of ages 61 and above participated in the 1958 off—year election. On the other hand, among those in each cohort who participated in the 1958 off-year election, 40 percent of those 21-30, 64 percent of those 31-40, 65 percent of those 41-50, 64 percent of those 51-60, and 62 percent of those of ages 61 and above participated in the 1960 presidential election. Similarly, among those in each age cohort who participated in the 1972 presidential election, 29 percent of those 21-30, 37 percent of those 31-40, 33 percent of those 41-50, 46 percent of those 51-60, and 48 percent of those of ages 61 and above participated in the 1974 off-year election. On the other 146 hand, among those in each age cohort who participated in the 1974 off-year election, 74 percent of those 21—30, 56 percent of those 31-40, 55 percent of those 41-50, 77 percent of those 51-60, and 76 percent of those of ages 61 and above participated in the 1976 presidential election. The support for our hypothesis about cross-election parti- cipation in influence attempts across the entire set of age cohorts (except the 21—30 year olds in 1956-58-60) in both time periods is remarkable. Additional tests to assess the extent to which a first order Markov process for predicting change in levels of influence attempts across time can be performed, firstly, by comparing the predictions for the act of influence attempts at t+2 under a first order process with the proportionate influence attempts actually observed at t+2; and, secondly, by testing the hypothesis that the transition matrix for influence attempts between t and t+1 is equal to that between t+1 and t+2. The second test can be formally expressed as: H0: I from t=1 to t=2 is equal to I from t=2 to t=3 HA: I from t=1 to t=2 is not equal to I from t=2 to t=3 Where I is equal to a matrix of transition probabilities. I for each cohort in each time period is shown in Appen- dices A6-7 and A6-8. In addition, t=1 refers to either the 1956 or 1972 presidential elections; t=2 to the 1958 or 1974 off-year election; and, t=3 to the 1960 or 1976 presidential election. Table 6-8 shows the expected proportionate influence at— tempts at t=3 under the assumption of a first order process, and 147 the actual proportionate influence attempts at t=3 for each age cohort in each time period. These results indeed show that the assumption of a constant transition matrix for change in influence attempts between 1956, 1958, and l960--as well as between 1972, 1974, and l976--is not reasonable for those in all age cohorts. As was the case for voting, the essential problem with the assump- tion of a constant transition matrix is that it severely under- estimates proportionate influence attempts in the presidential election proceeding the off-year election for those in each age cohort in both time periods. The discrepancies between expected and actual proportionate influence attempts are marked and never fall below 10 percent (See Table 6-8). The x2 statistic can be calculated for each age cohort in each time period to test the hypothesis of a constant transition matrix for changing levels across time of attempts to influence the vote of others. The transition matrices for each cohort in both periods are shown in Appendices A6-7 and A6-8, and Table 6-9 shows the results for each age group. For example, the value of Xéfsz is 20.05 for the oldest age group in 1956-58-60; and this value is significant at at least the .01 alpha level. In addition, Table 6-9 shows that the assumption of a constant transition matrix for all cohorts in both time periods can be rejected in favor of the alternative hypothesis. Changing levels of attempts to in- fluence the vote of others clearly cannot be described by a first order Markovian process. 148 TABLE 6-8.--Comparison of Expected (Under First Order Assumption) Influence Attempts at t=3 with Actual Influence At- tempts at t=3, by Age Cohort, 1956-58-60 and 1972-74-76. A. 1956-58-60 a Expected Actual (Age Group in Years: 1960 1960 Expected-Observed 21-30 .13 .32 -.19 31-40 .16 .34 -.18 41-50 .16 .33 -.17 51-60 .17 .35 -.18 61+ .13 .33 -.20 B. 1972-74—76 3 Expected Actual Age Group in Years: 1976 1976 Expected-Actual 21-30 .14 .43 -.29 31—40 .23 .33 -.10 41-50 .15 .40 -.25 51-60 .17 .39 -.22 61+ .09 .34 -.25 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Figure A6-9E in Appendix A6-9 shows the second order transition matrix in 1956-58-60 for those in the oldest age cohort. An examination of these proportions clearly shows that the transi- tion probabilities linking influence attempts in 1958 and 1960 vary with 1956 attempts to influence the vote of others. For example, the probability that someone in the oldest age cohort attempted to influence another's vote in both 1958 and 1960 is .73 if that individual also attempted to influence someone's vote in 1956. This probability is only .45 if the individual in the oldest group did not attempt to influence another person's vote in 1956. 149 TABLE 6-9.--Summary of Results for Hypotheses Tests of Constant Transition Matrix (First Order Process) for Influence Attempts, by Age Cohort, 1956-58-60 and 1972-74-76. A. 1956-58-60 jAge Group in Years:3 Test Statistic3(xéf=2) Decision Rule 21-30 39.41** Reject HO 31-40 3l.02** Reject H0 41-50 23.53** Reject H0 51-60 l4.75** Reject H0 61+ 20.05** Reject H0 B. 1972-74-76 Age Groupgin Years:a Test Statistic (x§f=2) Decision Rule 21-30 63.24** Reject H0 31-40 7.61* Reject H0 41-50 32.35** Reject HO 51—60 25.76** Reject HO 61+ 49.68** Reject HO 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. The more formal test that changing levels of influence attempts across time is second as opposed to first order Markovian can be expressed as: H0: The transition process is first order. HA: The transition process is second order. Figures displaying the second order transition matrices for influence attempts for each age cohort are presented in Appendices A6-9 and A6-10, and the test statistics for each group (that is, x2 with 2 degrees of freedom) are shown in Table 6-10. 150 Once again, the x2 statistic is employed to test the above hypoth- eses for each cohort in each period. As Table 6-10 shows, for each age cohort in each time period the null hypothesis purporting that the process is first order is rejected in favor of the alternative hypothesis purporting a second order process. Once again, the con- sistency with which we can reject the null hypothesis across all age cohorts is striking. All values are significant at at least the .05 alpha level. TABLE 6—lO.--Summary of Results for Hypotheses Tests for First vs. Second Order Markov Processes for Influence Attempts, by Age Cohort, 1956-58-60 and 1972-74-76. A. 1956-58-60 Age Group in Years:8 Test Statistic3( 2 _ ) Decision Rule de-z 21—30 l7.70** Reject HO 31-40 20.97** Reject HO 41-50 7.77* Reject H0 51-60 9.12* Reject HO 61+ 13.37** Reject H0 B. 1972-74-76 Age Group in Years:8 Test Statistic (x§f=2) Decision Rule 21-30 17.40 Reject HO 31—40 38.06 Reject HO 41-50 21.74 Reject H0 51—60 9.38 Reject HO 61+ 23.98 Reject H0 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. 151 Given our support for the contention that attempts to in- fluence another's vote at t=3 is a joint function of influence at— tempts at t=1 and t=2 (where t=1, t=2, and t=3 are defined as above), our task is now to consider the hypothesis that older peeple are less continuous in their attempts to influence others than are those in younger age groups. Once again, the expectation is that older peOple are less likely than those in younger age groups to be continually active in influence attempts over a series of three successive elections. Thus, we expect that the proportion of the oldest age group who were active in each of the three elections will be markedly smaller than the proportions of younger age groups who were continually active in influence at- tempts over these elections. Finally, it is anticipated by the disengagement thesis that proportionate continual inactivity should be greater among those in the oldest group than among those in all others. Table 6-11 shows the empirical results with which to assess the argument of disengagement theory for continual activity in attempts to influence the vote of others over a series of three successive elections in both time periods. In 1956-58—60, there is no support for the contention that proportionate continual activity is lower among the aged as compared to those in other age cohorts. In fact, among older people who attempted to influence another's vote in both 1956 and 1958, 73.3 percent still did so in 1960. This percentage is, moreover, higher than that for any other group except the 31—40 year olds. Similarly, in 1972-74-76, there 152 is no support for the disengagement hypothesis that older peOple are less continuous in their influence attempts compared to those in other groups. As Table 6-11 shows, 84.2 percent of older people who attempted to influence the vote of others in both 1972 and 1974 also did so in 1976. This percentage is barely different from that for any other cohort, and markedly higher than that for those in the 41-50 year old group. TABLE 6-ll.--Continual Activity in Influence Attempts by Age Cohort, 1956-58-60 and 1972-74-76. R for Age Group All Age in Years:8 21-30 31-40 41-50 51-60 61+ Groups A. Percent active in 1960 who were active in both 1956 & 1958 34.1% 83.9% 65.2% 66.7% 73.3% 61.1% (N) ( 41) ( 31) ( 23) ( 21) ( 15) (131) B. Percent active in 1976 who were active in both 1972 & 1974 85.3% 86.7% 66.7% 83.9% 84.2% 81.3% (N) ( 34) ( 30) ( 30) ( 31) ( 19) (144) aAge in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. A more precise and exact test of the disengagement hypoth- esis can be performed by a series of tests of whether the second order transition matrix for the oldest cohort is different pairwise from that for each other age cohort. Specifically, we are interested in comparing the transition probabilities for the 153 continually active row of the second order matrix of the oldest age cohort pairwise with all others. Our research hypothesis will be that of the disengagement thesis--viz., that older people are less continuous in their influence attempts compared to those in each other age group. This hypothesis will be pitted against a con- tinuity hypothesis which purports that there are no differences between the oldest group and all others in their continual activity in influence attempts across time. More formally, these hypotheses can be expressed as: HO: I for those of ages 61 and above is pairwise no ca different from Ica of each other cohort--the continuity hypothesis. HA: Ica for those of ages 61 and above is pairwise different from and less than the transition probabilities of Ica for each other cohort-- the disengagement hypothesis. NOTE: The transition matrices for this test are shown in Appendix A6-ll. Table 6-12 shows the results of these hypotheses tests for continual activity in influence attempts in both time periods. The failure to reject the continuity hypothesis for each pairwise com- parison of the oldest group with all others (except the one between the oldest group and the 21-30 year olds in 1956-58-60) is remark- able and consistent with our comparisons of proportionate continual activity in influence attempts across cohorts in Table 6-11 above. Moreover, in the only comparison where we could reject H it is 0’ due to the fact that proportionate continual activity is markedly less among the 21-30 year olds in 1956-58-60 than it is among older 154 people. Continual activity in political campaigns is clearly a function of previous participation, and older people are not unique in this respect in either time period. TABLE 6-12.--Summary of Results for Hypotheses Tests for Similar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Activity in In- fluence Attempts, 1956-58—60 and 1972—74-76. A. 1956—58-60 Comparision of Ica for oldest cohort with that for: Age Cohort in Years:a Test Statistid3(x§f=l) Decision Rule 21—30 6.84** Reject H 31-40 0.72 Do not reject H 41—50 0.12 Do not reject HO 51-60 0.18 Do not reject HO B. 1972-74-76 Comparison of Ica for oldest cohort with that for: Age Cohort in Years:a Test Statistic (de=l) Decision Rule 21-30 0.01 Do not reject HO 31-40 0.06 Do not reject HO 41-50 1.84 Do not reject HO 51-60 0.00 Do not reject HO aAge in 1956 for 1956-60 panel; age in 1972 for 1972—76 panel. The argument that continual inactivity should be greater among the aged than among all other age groups is another hypoth- esis of the disengagement theory. Table 6-13 shows the results for proportionate continual inactivity for influence attempts among 155 each age cohort across three successive elections in both time periods. In both 1956—58-60 and 1972-74—76, there is no evidence to support the disengagement hypothesis. In both periods, the distribution of proportionate continual inactivity is flat across the entire set of age cohorts. Once people become, or were ini- tially, continually inactive from influence attempts, they tend to remain inactive. Almost 80 percent of those in each cohort in each time period were continually inactive from this form of political campaign activity. TABLE 6-13.—-Continual Inactivity in Influence Attempts by Age Cohort, 1956-58-60 and 1972-74-76. R for Age Group All Age in Years:3 21-30 31-40 41-50 51-60 61+ Groups A. Percentage inactive in 1960 who were inactive in both 1956 & 1958 75.9% 77.8% 77.4% 78.4% 80.2% 77.7% (N) (174) (194) (164) (102) (101) (735) B. Percentage inactive in 1976 who were inactive in both 1972 & 1974 72.0% 83.5% 74.6% 76.9% 79.3% 77.1% (N) (157) (121) (114) (117) (150) (659) 3Age in 1956 for 1956-60 panel; age in 1972 for 1972—76 panel. Once again, a more precise test of the disengagement hypothesis can be performed by comparing the transition proba- bilities for the continually inactive row of the second order 156 matrix for the oldest cohort pairwise with all others. The dis- engagement hypothesis is that older peeple are more likely to be continually inactive in their influence attempts as compared to those in each other age group. The test of this hypothesis will be pitted against a rival hypothesis that there are no differences between the oldest group and all others in the continual inactiv- ity from influence attempts across time. These hypotheses can be expressed as: HO: Ica for those of ages 61 and above is pairwise no different from Ici of each other age cohort--the continuity hypothesis. H : I for those of ages 61 and above is pairwise different from and greater than the transi— tion probabilities of Ici for each other cohort--the disengagement hypothesis. NOTE: The transition matrices for this test are shown in Appendix A6-12. Table 6-14 shows the results of these hypotheses tests for continual inactivity from influence attempts in both time periods. As was the case for continual activity, we can easily reject the disengagement hypothesis for each pairwise comparison of the oldest cohort with all others. This is consistent with our comparisions of proportionate continual inactivity from influence attempts across the range of cohorts in Table 6-13 above. Continual in- activity from influence attempts is strikingly similar for all age cohorts in both time periods. 157 TABLE 6-l4.--Summary of Results for Hypotheses Tests for Similar Second Order Transition Matrices Between Oldest Age Cohort and All Others for Continual Inactivity From Influence Attempts, 1956-58-60 and 1972-74-76. A. 1956-58-60 Comparison of ICi for oldest cohort with that for: Age Cohort in Years:a Test Statistic3(x§f=l) Decision Rule 21-30 0.69 Do not reject HO 31—40 0.22 Do not reject HO 41—50 0.28 Do not reject HO 51-60 0.10 Do not reject HO B. 1972-74-76 Comparison of Ici for oldest cohort with that for: Age Cohort in Years:3 Test Statistic (de=l) Decision Rule 21-30 2.25 Do not reject HO 31-40 0.52 Do not reject HO 41-50 0.84 Do not reject HO 51-60 0.22 Do not reject HO 8Age in 1956 for 1956-60 panel; age in 1972 for 1972-76 panel. Conclusions In this chapter our analyses have provided important findings for the study of political participation in general and aging and political participation in particular. 0f general interest to those who study political participation, we have directly (that is, through panel analysis) found that reported for the American electorate. political behavior across time is remarkably durable and stable Moreover, those who participate in 158 off—year elections are the most stable in their political activity across time. These conclusions apply to both voting and political campaign activity, and to both the 1956-58-60 and 1972-74—76 time periods. Our analyses here also cast substantial doubt on the utility of the disengagement thesis as an explanation of changing levels of political activity across time. We directly pitted the disengagement argument against an alternative argument, continuity theory, which purported that the nature and amount of an older person's political participation is primarily an extension of a pattern deve10ped earlier in life. Our results provided support for the continuity, not the disengagement hypothesis. Older people are not unique from those in other age cohorts in the durability of their political activity across time. NOTES TO CHAPTER 6 1Unfortunately, we cannot perform this analysis for all five activities since the 1958 wave of the 1956-58-60 panel in- cluded only two of the political activities employed in this analysis—-the attempts to influence the vote of others item and the vote item. 2See Statistical Abstract pf the United States for the years 1956, 1960, 1972, and 1976. 3*Significant at .05 level; **Significant at .01 level. 159 CHAPTER 7 REFLECTIONS ON THE STUDY OF OLD AGE AND POLITICAL BEHAVIOR The above analysis has accomplished a good deal of both theoretical and empirical work for the study of political partici- pation in general and old age and political participation in par- ticular. Our final task is threefold. We shall, first, summarize the key findings of the above chapters. Second, we will spell out the theoretical implications of the analysis and suggest pathways for future research on aging and political behavior. Third, and finally, the policy implications of the study of old age and political behavior will be discussed. Summary of Key Findings In Chapter 1 we reviewed previous studies of the disengage- ment thesis and focused on its applications to the study of politi- cal participation. Of particular importance was the analysis of the ambiguity of the concept of age as expressed in previous studies. We suggested that age is really a surrogate concept designed to assess circumstances and events which occur throughout the life cycle and serve as specific indicators of disengagement in a general sense. 160 161 Our contribution to the use of old age as a concept which contributes to disengagement from political activity was the argu- ment that three components, or circumstances and events, influence political disengagement. Old age disengagement from politics was viewed as a function of sociological or compositional factors (sex and level of formal education), key events or circumstances (re- tirement from the labor force), and explicitly political factors (being contacted by a political party). In short, disengagement from political activity was viewed to be a configuration of both individual and institutional factors. Chapter 2 had as its signal feature the development of an adequate research design to test the disengagement thesis. An important component of this chapter was the formulation of con- ceptual typologies for two- and three-wave panel study designs. These typologies provided for both direct and indirect assessments of whether or not older peOple disengage from politics at greater levels than those in other age groups. In short, under the dis- engagement thesis we expected older people to be more likely to dr0p out of, more likely to remain inactive, and less likely to start up political activity as compared to those in other age cohorts. The other important contributions of Chapter 2 were the arguments for assessing the disengagement thesis in two separate time periods and for considering voting and other forms of political campaign activity separately. Aggregate levels of changing activity differed between 1956-60 and 1972-76, and we 162 anticipated that individual change would also show variation be- tween these two periods. More importantly, an examination of these two periods gave us the Opportunity to examine the impact of changes between periods in circumstances and events which accompany aging and influence political disengagement. The most important changes between the periods were that levels of formal education and party contact efforts were higher in 1972-76 than they were in 1956-60. To the extent that changes in these particular circumstances and events influence political disengagement, our two period design allows us tO detect it. Finally, our initial analyses in this chapter demonstrated that voting and other forms Of political campaign activity should be viewed separately. Our empirical work began in Chapter 3. We found only modest support for the disengagement thesis when changing levels of political activity were examined across two successive elections in each time period. The patterns of change differed for voting and the four types of political campaign activity were examined. In particular, we found that the patterns of across time change for three Of the four campaign acts (the exception being the act Of giving money to political candidates) were similar enough to form a summary measure Of political campaign activity. In 1956-60, there was no strong evidence Of old age disengagement from voting but some evidence Of early disengagement from political campaign activity. In 1972-76, on the other hand, there was some evidence of elderly disengagement from voting but not from political 163 campaigns. Possible explanations for these diverse patterns were assessed in the concluding section of Chapter 3. In Chapter 4 we examined the relationship between age co- hort membership and disengagement from politics with controls introduced for the impact of sex and level of formal education. This chapter showed that a large part of the diverse pattern in old age disengagement from voting and political campaigns between 1956-60 and 1972-76 can be accounted for by the increased tendency of elderly females to disengage from voting between 1972 and 1976. Level of formal education could not account for this pattern since it was the less educated elderly who were least likely to disen- gage from voting (compared to those in other age cohorts) between 1972 and 1976. Chapter 5 assessed, in a truly dynamic sense, the impact of retirement from the labor force on disengagement from politics among older people in particular, and the impact of being con- tacted by a political party on disengagement among the electorate in general. Surprisingly, these two key circumstances or events had no impact on disengagement from either voting or political campaign activity in both time periods. Our findings cast much doubt on the compensation or distraction hypotheses (Glenn and Grimes, 1968) and on the ability of political institutions, like the political party, to prevent or inhibit individual withdrawal from political activity. Processes of institutional mobilization (see Verba, Nie and Kim, 1978) made no difference in changing 164 levels of political activity across time irrespective of age cohort membership and time period. Finally, in Chapter 6, we thrust the final and lethal knife into the heart of the disengagement thesis by pitting it against an alternative argument, continuity theory. We found that for the electorate in general and older peOple in particular political behavior is remarkably durable across time. Moreover, those who participate in Off-year elections are the most durable political actors. In short, older peOple were not unique from those in other age groups in their continual activity or inactiv- ity from politics over a series of three successive elections. Theoretical Implications and Avenues for Future Research In this study we have established that the disengagement thesis has little utility in accounting for changing levels of political activity across time. Essentially, we found continuity theory to be an attractive alternative to the disengagement theory. Once again, the American electorate, including older people, is remarkably stable in their political behavior across time. How- ever, continuity theory, unlike disengagement theory, has limited utility in accounting for those changes which indeed do occur in political behavior from one election to the next. There is a larger theoretical and empirical task to be done in the study of Old age and political behavior. Specifically, we must further attempt to identify the specific configuration of 165 circumstances and events which contribute to surges and declines in political activity among the aged and among the general electorate. Some of the most fruitful work in the study of changing levels of political behavior has focused on how attitudes can contribute to changes in political activity and other key political attitudes. Abramson and Aldrich (1982), for example, find that about three-quarters of the decline in voting turnout between 1960 and 1980 can be accounted for by the weakening of partisanship and the decline in feelings of "external" political efficacy. In addition, Rollenhagen (1982) has found that about one-half of the decline between 1960 and 1980 in concern about the election outcome--an important correlate of turnout and a prominent feature of rational choice models of voting--can also be accounted for by these two attitudinal trends. The key theoretical perspective of the analyses of Rollenhagen (1982) and Abramson and Aldrich (1982) is that the attitudinal and behavioral trends in party identification, "exter- nal" efficacy, extent of concern about the electoral outcome, and voting turnout are a product of the actions and behavior of people within political institutions. In short, changes in the macro political system, such as the decline in the governing abilities of parties and the political system in general, contribute to attitudinal and behavioral change at the micro or mass level. Moreover, as noted above in Chapter 3, Aldrich and Rohde (1978) have shown that changes in election financing laws have had a major impact on political behavior at the micro level. 166 It is imperative that political scientists resurrect the study of institutions and macro-level political change and, further- more, demonstrate how these changes and actions influence indiv- idual political behavior and attitudes among both older people and the electorate in general. This need has been stated most vividly by William Riker (1980) who argues that: . . . we cannot leave out the force of institutions. The people whose values and tastes are influential live in a world of conventions both about language and values themselves. These conventions are in turn condensed into institutions, which are simply rules about behavior, especially about making decisions . . . So interpersonal rules, that is, institutions, must affect social outcomes just as much as personal values. (p. 432) In the above analysis we have indeed attempted to account for, both theoretically and empirically, the impact of political institutions on changing levels of political activity across time. We found that one such institution, the political party, failed to influence such changes at least through the act of contacting. We must, however, extend this type of institutional analysis in future research on aging and political behavior. Of particular importance in the study of the impact Of institutions on the political behavior of the elderly is the identification of the particular context which contributes to age-based stratification and conflict (See N. Cutler, 1977; Foner, 1974). A signal feature to study here is the types of policies produced by institutions, and the impact of such policies on political behavior among the aged. For example, the policies of Detroit's Mayor Young for the destruction of many homes occupied 167 by the aged contributed to a surge of elderly political activity in that area. More generally, the massive assault by the Reagan administration on policies and programs designed to help the dis- advantaged (including, but not limited to, the elderly) may con- tribute tO increased levels of elderly political participation in the early 19803. And, most significantly, the impending crises facing the Social Security system (Ginsberg, 1982; Congressional Quarterly, 1981) should encourage increased levels of political participation and interest among the aged of the 19803. Although institutions and political events should be assigned more importance in the study of old age and political behavior, we should not neglect individual level circumstances, events, and situations which contribute to political participation. We have already shown that the aged portion of the population is much better educated than the elderly of the past, and movement of younger and even better educated cohorts into Old age will result in an increasingly better educated older population. And, although we found no evidence that retirement from the workforce is related empirically to change in political activity, its effect should continue to be examined, especially given the narrowing numerical gap between Social Security beneficiaries and contributors (See Chapter 1). The intersection of individual circumstances, retire- ment from the workforce, with institutional failure to solve the impending Social Security crisis may lead to massive increases in political participation, including political protest, by the aged. 168 There are also a wide variety of other individual circum- stances and events which presumably would influence changing levels of political participation among the aged. For example, the migra- tion of older people from Northern regions to the South and South- west may influence the extent and nature of their political parti- cipation. As a recent issue of Congressional Quarterly (1981) noted, the migration of the aged to the South and Southwest has altered the partisan balance of congressional districts in Florida and Arizona. We must also examine how migration impacts the political behavior and attitudes of the aged themselves. Policy Implications Assuming the political participation and attitudes of the aged matters (and impressionistic evidence Of elite behavior on old age policy issues suggests that it does), then the study of old age and political behavior indeed has policy implications. The aged are, in fact, prime beneficiaries of governmental pro- grams (Social Security and Medicare being the most prominent), and such programs are long—term and well-established in laws and policies. Given the great dependency of the aged on governmental programs, we expect political elites to solicit or, at the very least, not to lose their support. The political importance of the aged is, and has been, recognized by political elites in government. The current reluc— tance of President Reagan and members of the Congress to make needed revisions in Social Security and health care programs reflects this 169 recognition. This importance rings with striking clarity when combined with the fact (See Congressional anrterly, 1981) that the majority of the aged have voted for the Republican candidate in seven of the eight presidential elections between 1952 and 1980 (all but 1964). Political institutions, particularly political parties, have and will continue to solicit the political support Of older Americans. Both parties have special devices for this solicita- tion which include, but are not limited to, special organizations and mailings. Increasingly large numbers of older people, together with presumably increasing demands, will no doubt make the aged a group to be reckoned with by the political system. LIST OF REFERENCES LIST OF REFERENCES Abramson, Paul R. and John H. Aldrich (1982). "The Decline Of Electoral Participation in America." American Political Science Review, September, forthcoming. Aldrich, John H. and David W. Rohde (1978). "Law and Political Behavior: A Note on Some Effects of Federal Campaign Finance Reform on the Electorate," Unpublished Manuscript, Michigan State University. Babchuk, Nicholas and Alan Booth (1969). "Voluntary Association Membership: A Longitudinal Analysis." American Socio- logical Review 34:31-45. Campbell, Angus (1960). "Surge and Decline: A Study of Electoral Change." Public Opinion Quarterly, 24:397-418. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes (1960). The American Voter, New York: Wiley. Campbell, Donald T. and Julian C. Stanley (1963). Experimental and Quasi-Experimental Designs for Research. Chicago: Rand McNally. Congressional Quarterly (1981). Special Report, "The Aging of America," 28 November, 2329-2346. Cook, Thomas D. and Donald T. Campbell (1979). Quasi-Experimenta- tion. Chicago: Rand McNally. Cumming, E. and W.E. Henry (1961). Growing Old: The Process of Disengagement. New York: Basic Books. Cutler, Neal E. (1977). "Demographic, Social-Psychological, and Political Factors in the Politics of Aging: A Foundation for Research in 'Political Gerontology.'" American Political Science Review 71:1011-1025. Cutler, Stephen J. (1977). "Aging and Voluntary Association Participation." Journal of Gerontology 32:470—479. 170 171 Davis, James A. (1978). "Studying Categorical Data Over Time." Social Science Research 7:151-179. Foner, Anne (1974). "Age Stratification and Age Conflict in Political Life." American Sociolggical Review 39: 187-196. Gergen, Kenneth J., and Kurt W. Back (1966). "Communication in the Interview and the Disengaged Respondent." Public Opinion Quarterly 30:385-398. Glenn, Norval D. (1969). "Aging, Disengagement, and Opinionation." Public Opinion Quarterly 33:17-33. Glenn, Norval D., and Michael Grimes (1968). "Aging, Voting, and Political Interest." American Sociological Review 33: 563-575. Goodman, Leo A. (1962). "Statistical Methods of Analyzing Pro- cesses of Change." Journal of Sociology 68:57-78. Hochschild, Arlie Russell (1975). "Disengagement Theory: A Critique and Proposal." American Sociological Review 40: 553-569. Light, Larry (1981). "Democrats Hoping to Break Traditional GOP Loyalties of Voters Age 65 and Older." Congressional Quarterly 28, November, 2343-2346. Maddox, George L. (1963). "Activity and Morale: A Longitudinal Survey of Selected Elderly Subjects." Social Forces 42: 195-204. Markus, Gregory B. (1979). Analyzinnganel Data. Sage University Paper Series on Quantitative Applications in the Social Sciences, Series No. 07-018. Milbrath, Lester W. (1965). Political Participation. Chicago: Rand McNally. Milbrath, Lester W. and M.L. Goel (1977). Political Participation, 2nd ed. Chicago: Rand McNally. Nie, Norman, Sidney Verba, and Jae-on Kim (1974). "Participation and the Life Cycle." Comparative Politics 6:319-340. Riker, William H. (1980). "Implications from the Disequilibrium of Majority Rule for the Study of Institutions." American Political Science Review 74:432-446. 172 Rollenhagen, Rick E. (1982). "Explaining Variation in Concern About the Outcome of Presidential Elections, 1960—1980." Unpublished Manuscript, Michigan State University. Rose, Arnold M. (1964). "A Current Theoretical Issue in Social Gerontology." The Gerontologist 4:46-50. Rose, Sheldon M. (1972). Introduction to Probability Models. New York: Academic Press. Sigel, Roberta, and Marilyn Hoskins (1977). "Perspectives on Adult Political Socialization-—Areas of Research," in Stanley Renshon, ed., Handbook of Political Socialization. New York: The Free Press. U.S. Bureau of the Census. Current ngulation Reports. Verba, Sidney, and Norman H. Nie (1972). Participation in America. New York: Harper and Row. Verba, Sidney, Norman H. Nie, and Jae-on Kim (1971). The Modes Of Democratic Participation: A Cross-National Analysis. Beverly Hills: Sage Papers in Comparative Politics, NO. 01-013. Verba, Sidney, Norman H. Nie, and Jae-on Kim (1978). Participa- tion and Political Equality. Cambridge University Press. Videbeck, Richard, and Alan Knox (1964). "Alternative Participa- tory Responses to Aging," in Arnold M. Rose and Warren M. Peterson, eds., Older People and Their Social World. Philadelphia: F.A. Davis, 37-48. Wolfinger, Raymond E. and Steven J. Rosenstone (1980). Who Votes? New Haven: Yale University Press. APPENDIX 3 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-60 AND 1972-76 APPENDIX 3 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-6O AND 1972-76 Introduction This Appendix includes Appendices A3-l through A3-4. Throughout the Appendix, "P" indicates that the individual parti- cipated; and "not-P" that the individual did not participate. In addition, age cohort membership for the 1956-60 panel respondents is determined by age in 1956; and cohort membership for the 1972-76 panel respondents is determined by age in 1972. 173 APPENDIX A3-1 EACH POLITICAL ACTIVITY ITEM ACROSS TIME BY AGE COHORT, 1956-6O 1960 Not-P 1960 Not-P 1960 Not-P Figure A3-lA.--Voting Across Time by Age Cohort, 1956—60. 174 Ages 21-30 1956 P Not-P 171 49 220 P 1960 9 55 64 Not-P 180 104 284 Ages 41-50 1956 P Not-P 198 29 227 P 1960 6 34 40 Not-P 204 63 267 1 Ages 61+ 1956 P Not-P 133 4 13 146 P 1960 140 45 185 Ages 31-40 1956 P Not-P 246 45 291 14 52 66 260 97 357 Ages 51-60 1956 P Not-P "1 137 13 150 4 21 25 141 34 175 All Age Groups Combined 1956 P Not-P 905 149 1054 40 194 234 945 343 1288 " A 1960 Not-P 1960 Not-P 1960 Not-P Figure A3-lB.-—Attempts to Influence the Vote of Others 175 Ages 21-30 1956 P Not-P 47 51 98 P 1960 36 149 185 Not-P 83 A 200 283 Ages 41-50 1956 P Not-P 40 53 93 P 1960 38 156 194 Not-P 78 209 287 1 Ages 61+ 1956 P Not-P 30 I 29 59 P 1960 29 99 128 Not-P 59 128 187 by Age Cohort, 1956-60. Ages 31-40 1956 P Not-P 61 59 120 47 190 237 108 249 357 Ages 51-60 1956 P Not-P 28 30 58 23 94 117 51 124 175 All Age Groups Combined 1956 P Not-P 206 222 428 173 688 861 379 910 1289 Across Time 1960 Not-P 1960 Not-P 1960 Not-P Ages 21-30 1956 P Not-P 9 22 31 11 242 253 20 264 284 Ages 41-50 1956 P Not-P :1! 20 23 43 21 223 244 41 246 287 Ages 61+ 1956 P Not-P 5 6 ll 5 171 176 10 177"”'T87 176 1960 Not-P 1960 Not-P Ages 31-40 1956 P Not-P 12 23 35 18 302 320 30 325 355 Ages 51—60 1956 P Not-P l6 13 29 11 134 145 27 147 174 All Age Groups Combined 1960 Not-P 1956 P Not-P 62 87 149 66 1072 1138 128 11594' 1 7 Figure A3-lC.--Giving Money to Political Candidates Across Time by Age Cohort, 1956-60. 1960 Not-P 1960 Not-P 1960 Not-P Ages 21-30 P Not-P 4 18 22 7 255 262 11 273 284 Ages 41-50 1956 P Not-P :‘fl 8 15 23 17 246 263 25 261 L 286 Ages 61+ 1956 P Not-P 4 9 13 10 164 174 14 173 187 Ages 31—40 1956 P Not-P P 10 ll 21 1960 Not-P 19 314 333 29 325 354 Ages 51-60 1956 P Not-P 4— P 3 20 23 1960 Not-P 10 142 152 13 162 175 All Age Groups Combined 1956 P Not-P p 29 73 102 1960 Not-P 63 1121 1184 92 1194 1286 Figure A3—1D.—-Attending Political Meetings or Rallies Across Time by Age Cohort, 1956-60. 178 Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 22 43 65 P 30 59 89 1960 1960 Not-P 24 195 219 Not-P 25 241 266 46 I 238 284 55 300 355 Ages 41-50 Ages 51-60 ‘ 1956 1956 P Not-P P Not-P "1 P 16 38 54 P ‘ ll 30 41 1960 1960 Not-P 25 205 230 Not-P 17 117 134 41 243 284 28 147 175 A893 61+ All Age Groups Combined 1956 1956 P Not-P P Not-P P 8 4 17 25 P 87 187 274 1960 1960 Not-P 10 150 160 Not—P 101 908 1009 18 167 185 188 1095 1283 Figure A3-lE.—-Wearing a Campaign Button or Placing a Sticker on the Car Across Time by Age Cohort, 1956-60. APPENDIX A3-2 EACH POLITICAL ACTIVITY ITEM ACROSS TIME BY AGE COHORT, 1972—76 179 Ages 21-30 Ages 31-40 1972 1972 P Not-P P Not-P 1 P 185 43 228 P 167 19 186 1976 1976 Not-P Not-P 33 35 68 10 28 38 218 78 296 177 47 224 Ages 41-50 Ages 51—60 ' 1972 P Not-P P Not-P P 174 10 184 p 154 10 164 1976 1976 Not-P 16 19 35 Not-P 9 29 38 190 29 219 163 39 202 A835 61+ All Age Groups Combined 1972 1972 P Not-P P Not-P P 145 ‘ 16 161 P 825 98 923 1976 1976 Not-P 16 43 59 Not-P 84 154 238 161 59 :220 909 252 1161 Figure A3e2A.--Voting Across Time by Age Cohort, 1972-76. 1976 Not-P 1976 Not-P 1976 Not-P 180 Ages 21-30 1972 P Not-P 71 54 125 48 123 171 119 177 296 Ages 41-50 1972 P Not-P 53 34 87 36 96 132 89 130 219 Ages 61+ 1972 P Not-P 39 34 73 20 126 146 59 160 219 1976 Not-P 1976 Not-P Ages 31-40 1972 P Not-P 49 27 76 34 114 148 83 141 224 Ages 51-60 1972 P Not-P 43 34 77 26 99 125 69 133 202 All Age Groups Combined 1976 Not-P 1972 P Not-P 255 183 438 164 558 722 419 741 1160 Figure A3-2B.--Attempts to Influence the Vote of Others Across Time by Age Cohort, 1972—76. 181 Ages 21-30 Ages 31-40 1972 1972 P Not-P P Not-P P 11 7 18 P 16 14 30 1976 1976 Not-P Not-P 19 226 245 14 164 178 30 233 263 30 178 208 Ages 41-50 Ages 51-60 I 1972 1972 P Not-P P Not-P P 18 16 34 p 15 7 22 1976 1976 Not-P 16 151 167 Not-P 11 150 161 34 167 201 26 157 183 l Ages 61+ All Age Groups Combined 1972 1972 P Not-P P Not-P P 9 I 9 18 P 69 53 122 1976 1976 Not—P 13 181 194 Not-P 73 872 945 22 190 212 142 925 1067 Figure A3-2C.--Giving Money to Political Candidates Across Time by Age Cohort, 1972-76. 1976 Not-P 1976 Not-P 1976 Not-P Ages 21-30 1972 P Not-P 4 9 13 33 249 282 37+ 258 295 Ages 41-50 1972 P Not-P 12 8 20 20 179 199 32 187 219 Ages 61+ 1972 P Not-P 4 A 10 14 8 197 205 12 207 219 182 Ages 31-40 1972 P . Not-P p 9 7 16 1976 Not-P 20 188 208 29 195 224 Ages 51-60 1972 P Not-P p 7 7 14 1976 Not-P 11 177 188 18 184 202 All Age Groups Combined 1972 P Not-P P 36 41 77 1976 % Not-P I 92 990 1082 128 1031 1159 Figure A3-2D.-—Attending Political Meetings or Rallies Across Time by Age Cohort, 1972-76. 1976 Not-P 1976 Not-P 1976 Not-P Figure A3-2E Ages 21-30 1972 P Not-P 9 15 24 61 211 272 70‘ 226 296 Ages 41-50 1972 P Not-P ll 12 23 32 164 196 43 176 219 Ages 61+ 1972 P Not-P 5 ll 16 12 190 202 17 201 218 .--Wearing a Campaign Button or Placing a Sticker on the Car Across Time by Age Cohort, 1972-76. 183 1976 Not-P 1976 Not-P Ages 31-40 1972 P . Not-P 7 ll 18 26 180 206 33 191 224 Ages 51-60 1972 P Not-P 6 ll l7 18 163 181 24 174 198 All Age Groups Combined 1976 Not-P 1972 P Not-P 38 60 98 149 908 1057 187 968 1155 APPENDIX A3-3 INDEX OF POLITICAL CAMPAIGN ACTIVITY ACROSS TIME BY AGE COHORT, 1956-6O 184 Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 67 60 127 P 89 71 160 1960 1960 Not-P 41 130 171 Not-P 53 159 212 10877 190 298 142 230 372 Ages 41-50 Ages 51-60 1956 1956 P Not-P P Not-P ‘1-H P ' 62 56 118 P 43 41 84 1960 1960 Not-P 48 128 176 NOt-P 28 73 101 110 184 294 71 114 185 L A893 61+ All Age Groups Combined 1956 1956 P Not-P P Not-P P 38 l 32 70 P 299 260 559 1960 1960 Not-P 32 100 132 Not-P 202 l. 590 792 __L l 70 132 202 501 850 1351 Figure A3—3.--Index Of Political Campaign Activity Across Time by Age Cohort, 1956-60. APPENDIX A3-4 INDEX OF POLITICAL CAMPAIGN ACTIVITY ACROSS TIME BY AGE COHORT, 1972-76 1976 Not-P 1976 Not-P 1976 Not-P Figure A3-4 Ages 21-30 1972 P Not-P 84 48 132 81 99 180 165+ 147 312 Ages 41-50 1972 P Not-P 71 28 99 44 97 141 115 125 240 Ages 61+ 1972 P Not-P 46 I 37 83 33 129 162 79 166 245 .--Index of Political Campaign Activity 185 Age Cohort, 1972-76. Ages 31-40 1972 P . Not-P P 56 29 85 1976 Not-P 53 104 157 109 133 242 Ages 51-60 1972 P Not-P P 50 34 84 1976 Not-P 37 98 135 87 132 219 All Age Groups Combined 1976 Not-P P Not-P 527 248 775 176 307 483 703 555 1258 Across Time by APPENDIX 4 FREQUENCY DISTRIBUTIONS FOR TWO—WAVE ANALYSIS BY AGE COHORT, SEX, AND LEVEL OF FORMAL EDUCATION, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-6O AND 1972-76 APPENDIX 4 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, SEX, AND LEVEL OF FORMAL EDUCATION, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-6O AND 1972-76 Introduction This Appendix includes Appendices A4-1 through A4—4. Throughout the Appendix, "P" indicates that the individual parti- cipated, and "Not-P" that the individual did not participate. In addition, level of formal education is coded dichotomously. "Low" indicates less than high school, and "High" indicates high school or more. Finally, age cohort membership for the 1956—60 panel respondents is determined by age in 1956; and cohort membership for the 1972-76 panel respondents is determined by age in 1972. 186 APPENDIX A4-1 VOTING ACROSS TIME BY AGE COHORT AND GENDER, 1956-6O AND 1972-76 187 Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 74 9 83 P 121 18 139 1960 1960 Not-P 5 19 24 NOt-P 3 13 16 79 A 28 107 124 31 155 Ages 41-50 Ages 51-60 1956 1956 P Not-P P Not-P P 120 13 133 P 72 7 79 1960 1960 Not-P 3 8 11 NOt-P 0 9 9 123 21 144 72 16 188 A893 51+ All Age Groups Combined 1956 1956 P Not-P P Not-P P 74 6 80 P 461 53 514 1960 1960 NOt-P 4 7 11 NOt'P 15 56 71 78 13 91 476 109 585 Figure A4—1A.--Voting Across Time by Age Cohort, Males, 1956-60. 188 .- _ Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 97 40 137 P 125 27 152 1960 1960 Not-P 4 36 40 Not-P 11 39 50 101 7 76 177 136 66 202 Ages 41-50 Ages 51-60 1956 1956 P Not-P P Not-P _:Z. g P 98 16 114 P 65 6 71 1960 1960 Not-P 3 26 29 Not-P 4 12 16 101 42 143 69 18 87 L A893 61+ All Age Groups Combined 1956 1956 P Not-P P Not-P P 59 I 7 66 P 444 96 540 1960 1960 Not-P 3 25 28 Not-P 25 138 163 62 32 94 469 234 703 Figure A4-lB.--Voting Across Time by Age Cohort, Females, 1956—60. 1976 Not-P 1976 Not-P 1976 Not-P Figure A4-lC.--Voting Across Time by Age Cohort, Ages 21-30 1972 P Not-P 91 21 112 20 18 38 11177 39 150 Ages 41-50 1972 P Not-P 81 2 83 7 7 14 88 9 97 Ages 61+ 1972 P Not-P 67 ‘ 7 74 4 6 10 71 13 84 189 1976 Not-P 1976 Not-P Ages 31-40 1972 P Not-P 73 9 82 3 12 15 76 21 97 Ages 51-60 1972 P Not-P 72 3 75 3 10 13 75 13 88 All Age Groups Combined 1976 Not-P 1972 P Not-P 384 42 426 37 53 90 421 95 516 Males, 1972-76. 190 Ages 21-30 1972 P Not-P P 94 22 116 1976 Not-P 13 17 30 107 39 146 Ages 41-50 1972 P Not—P P 91 8 99 1976 Not-P 9 12 21 100 20 120 Ages 61+ 1972 P Not-P P 78 9 87 1976 Not-P 12 37 49 90 46 136 1976 1976 1976 Not-P Not-P Ages 31-40 1972 P . Not-P 94 10 104 7 16 23 101 26 127 Ages 51-60 1972 P Not-P 82 7 89 6 19 25 88 26 114 All Age Groups Combined Not-P 1972 P Not-P 439 56 495 47 101 148 486 157 643 Figure A4-lD.--Voting Across Time by Age Cohort, Females, 1972-76. APPENDIX A4-2 VOTING ACROSS TIME BY AGE COHORT AND LEVEL OF FORMAL EDUCATION, 1956-6O AND 1972-76 1960 Not-P 1960 Not-P 1960 Not-P Figure A4-2A.--Voting Across Time by Age Cohort, Low Education, Ages 21-30 Ages 31—40 1956 P Not-P 71 26 97 8 37 45 79 63 142 Ages 51-60 1956 P Not-P 79 10 89 4 19 23 83 29 112 All Age Groups Combined 1956 P Not-P 39 16 55 P 1960 5 32 37 Not-P 44 48 92 Ages 41-50 1956 P Not-P :T 96 20 116 P 1960 3 3o 33 Not-P 99 50 149 Ages 61+ P Not-P 91 11 102 P 1960 7 32 39 Not-P 98 43 141 1956-60. 1956 P Not-P 376 83 459 27 150 177 403 233 636 192 Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 132 33 165 P 175 19 194 1960 1960 Not-P 4 23 27 Not-P 6 15 21 136 I 56 192 181 34 215 Ages 41-50 Ages 51-60 1956 1956 P Not-P P Not-P :‘fi P 122 9 131 P .58 3 61 1960 1960 Not-P 3 4 7 Not-P O 2 2 125 13 {138 58 5 63 A893 61+ All Age Groups Combined 1956 1956 P Not-P P Not-P p 41 2 43 P 528 66 594 1960 1960 Not-P O O O Not-P 13 44 57 41 2 43 541 110 651 Figure A4-2B.--Voting Across Time by Age Cohort, High Education, 1956-60. Ages 21-30 Ages 31-40 1972 1972 P Not-P P Not-P P 9 8 17 P 20 6 26 1976 1976 Not-P Not-P 3 9 12 6 14 20 12 17 29 26 20 46 Ages 41—50 Ages 51-60 1972 P Not-P P Not-P P 38 5 43 P 50 5 55 1976 1976 Not-P 5 13 18 Not-P 6 23 29 43 18 61 56 28 84 Ages 61+ All Age Groups Combined 1972 1972 P Not-P 7 P Not-P P 73 4 8 81 P 3 190 32 222 1976 1976 Not—P 14 32 46 Not-P 34 91 125 87 40 127 224 123 347 Figure A4—2C.—-Voting Across Time by Age Cohort, Low Education, 1976 Not-P 1976 Not-P 1976 Not-P Ages 21-30 1972 P Not-P 151 34 185 27 22 49 178A 56 234 Ages 41-50 1972 P Not-P 118 4 122 8 5 13 126 9 135 Ages 61+ 1972 P Not-P 63 . 8 71 l 11 12 64 19 83 Ages 31-40 1972 P . Not-P p 128 12 140 1976 Not-P 3 14 17 131 26 157 Ages 51-60 1972 P Not-P P 92 4 96 1976 Not-P 3 5 8 95 9 104 All Age Groups Combined 1972 P Not-P p 552 62 614 1976 Not-P 42 57 99 594 119 713 Figure A4-2D.--Voting Across Time by Age Cohort, High Education, 1972-76. APPENDIX A4-3 POLITICAL CAMPAIGN ACTIVITY ACROSS TIME BY AGE COHORT AND GENDER, 1956-6O AND 1972-76 195 Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 39 20 59 P 55 3O 85 1960 1960 Not—P ll 39 50 Not-P 27 50 77 50 59 109 82 80 162 Ages 41-50 Ages 51-60 1956 1956 P Not-P ’ P Not-P P J 39 25 64 P 20 19 39 1960 1960 Not-P 3O 54 84 Not-P 16 36 52 69 79 i 148 36 55 91 A895 61+ All Age Groups Combined 1956 1956 P Not-P P Not-P P 20 J 20 40 P 173 114 287 1960 1960 Not-P 19 4O 59 Not-P 103 219 322 39 60 99 276 333 609 Figure A4-3A.--Political Campaign Activity Across Time by Age Co— hort, Males, 1956—60. 1960 Not-P 1960 Not-P 1960 Not-P 196 Ages 21-30 1956 P Not-P 28 4o 68 P 1960 30 91 121 Not-P 58 131 189 Ages 41-50 1956 P Not-P 23 31 54 P 1960 18 74 92 Not-P 41 105 146 t Ages 61+ 1956 P Not-P 18 4 12 30 P 1960 13 60 73 Not-P 31 72 103 Ages 31-40 1956 P Not-P 34 41 75 26 109 135 60 150 210 Ages 51-60 1956 P Nog;% 23 22 45 12 37 49 35 59 94 All Age Groups Combined 1956 P Not-P 126 146 272 99 371 470 225 517 742 Figure A4-3B.--Political Campaign Activity Across Time by Age Co- hort, Females, 1956-60. 1976 Not-P 1976 Not-P 1976 Not-P Figure A4-3C Ages 21-30 1972 P Not-P 43 25 68 38 44 82 81 69 150 Ages 41-50 1972 P Not-P 40 17 57 14 26 40 54 43 87 Ages 61+ 1972 P Not-P 20 l 13 33 9 42 51 29 55 84 197 Ages 31-40 1972 P a Not-P p 29 13 42 1976 Not-P 22 33 55 51 46 97 Ages 51-60 1972 P Not-P p 26 13 39 1976 Not-P 13 36 49 39 49 88 All Age Groups Combined 1972 P Not-P P 158 81 239 1976 Ngt_P 96 181 277 254 262 616 .-—Political Campaign Activity Across Time by Age Co- hort, Males, 1972-76. 1976 Not-P 1976 Not-P 1976 Not-P Figure A4-3D Ages 21-30 1972 P Not-P 41 23 64 33 49 82 74 72 146 Ages 41-50 1972 P Not-P 30 11 41 23 57 80 53 68 121 Ages 61+ 1972 P Not-P 26 ‘ 24 50 16 70 86 42 94 136 198 1976 1976 1976 Not-P Not-P Ages 31-40 1972 P Not-P 27 16 43 24 60 84 51 76 127 Ages 51-60 1972 P Not-P 24 21 45 17 52 69 41 73 114 All Age Groups Combined NOt-P 1972 P Not-P 148 95 243 113 288 401 261 383 644 .--Political Campaign Activity Across Time by Age Co— hort, Females, 1972-76. APPENDIX A4-4 POLITICAL CAMPAIGN ACTIVITY ACROSS TIME BY AGE COHORT AND LEVEL OF FORMAL EDUCATION, 1956-6O AND 1972-76 199 Ages 21-30 Ages 31-40 1956 1956 P Not-P P Not-P P 13 17 4 30 P 25 20 45 1960 1960 Not-P 14 55 69 Not-P 25 78 103 27 72 99 50 98 148 Ages 41-50 Ages 51-60 1956 1956 P Not-P P Not-P P J 24 27 51 P 20 25 45 1960 1960 Not-P 24 74 98 Not-P 19 54 73 48 101 i 149 39 79 118 A893 61+ All Age Groups Combined 1956 1956 P Not-P P Not-P p 24 4 20 44 P 106 109 215 1960 1960 NOt-P 25 86 111 Not-P 107 347 454 49 106 155 213 456 669 Figure A4-4A.--Political Campaign Activity Across Time by Age Co- hort, Low Education, 1956—60. 1960 Not-P 1960 Not-P 1960 Not-P Ages 21-30 1956 P Not-P 54 43 97 27 75 102 81 118 199 Ages 41-50 1956 P Not-P 38 29 67 24 54 78 62 83 t 145 Ages 61+ 1956 P Not-P 14 I 12 26 7 12 19 21 24 45 1960 Not-P 1960 Not-P Ages 31-40 1956 P Not-P 64 51 115 28 81 109 92 132 224 Ages 51-60 1956 P Not-P 23 16 39 9 19 28 32 35 67 All Age Groups Combined 1960 Not-P 1956 P Not-P 193 151 344 95 241 336 288 392 680 Figure A4-4B.--Political Campaign Activity Across Time by Age Co- hort, High Education, 1956-60. 201 Ages 21-30 Ages 31-40 1972 1972 P Not-P P , Not-P P 5 4 9 p 7 4 11 1976 1976 Not-P Not-P 9 12 21 12 29 41 l4 16 30 19 33 52 Ages 41-50 Ages 51-60 1972 1972 P Not-P A P Not-P P 9 4 13 p 12 12 24 1976 1976 Not-P 19 34 53 Not-P 15 55 70 28 38 66 27 67 94 Ages 61+ All Age Groups Combined 1972 1972 P Not-P P Not-P P 22 ‘ 19 41 P 55 43 98 1976 1976 Not-P 18 84 102 Not-P 73 214 287 40 103 143 128 257 385 Figure A4-4C.--Politica1 Campaign Activity Across Time by Age Co- hort, Low Education, 1972-76. 202 Ages 21-30 Ages 31-40 1972 1972 P Not-P P . Not-P P 70 38 108 P 40 21 61 1976 1976 Not-P Not-P 63 77 140 33 73 106 133 115 148 73 94 167 Ages 41-50 Ages 51-60 1972 1972 P Not-P P Not-P p 48 22 70 P 33 19 52 1976 1976 Not-P 22 56 78 Not-P 20 38 58 70 78 L148 53 57 110 Ages 61+ All Age Groups Combined 1972 1972 P Not-P P Not-P P 23 l 13 36 P 214 113 327 1976 1976 Not-P 14 40 54 Not-P 152 284 436 37 53 90 366 397 763 Figure A4-4D.--Political Campaign Activity Across Time by Age Co- hort, High Education, 1972-76. APPENDIX 5 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, CHANGE IN EMPLOYMENT STATUS, AND PARTY CONTACT STATUS, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-6O AND 1972-76 APPENDIX 5 FREQUENCY DISTRIBUTIONS FOR TWO-WAVE ANALYSIS BY AGE COHORT, CHANGE IN EMPLOYMENT STATUS, AND PARTY CONTACT STATUS, VOTING AND POLITICAL CAMPAIGN ACTIVITY, 1956-6O AND 1972-76 Introduction Throughout this Appendix, the following abbreviations are used. First, "P" indicates that the individual participated; and "Not-P," that the individual did not participate. Second, "W" indicates that the person worked; and "R" that the person is re- tired. Third, "C" that the person was not contacted. In addition, for the chi square statistics, a single asterisk indicates that the value is significant at the .05 level, and the double asterisk indicates that the value is significant at the .01 level. Finally, age cohort membership for the 1956-60 panel re- spondents is determined by age in 1956; and cohort membership for the 1972-76 panel respondents is determined by age in 1972. 203 204 Figure A5-l.--Transition Matrices for Voting Across Time for Change in Employment Status Groups, Ages 61 and Above, 1956-60 and 1972-76. A. 1956-60 1960 P Not-P w56; W60 P 28 0 28 I956 W56; R60 P 30 4 34 2 :- xdf=1 3.534 58 4 62 B. 1972-76 1976 P Not-P w72; W76 P 12 1 13 1972 1 W72; R76 P 16 l 17 X§£=l = .036 28 2 3o 205 1960 P Not-P w56; W60 P 4 8 12 1956 w56; R60 P 2 13 15 2 = de=l 1.530 6 21 18 B. 1972-76 1976 P NOt-P N72; N76 P 3 3 6 1972 W72; R76 P 6 6 12 2 = 0 9 9 18 x8f=l Figure A5-2.--Transition Matrices for Political Campaign Activity Across Time for Change in Employment Status Groups, Ages 61 and Above, 1956-60 and 1972-76. - A. 1956-60 1. Ages 31-40 C56; C60 P 1956 C56; 060 Not-P x2 = 2 720 df=l ' 2. Ages 41-50 056;C60 P 1956 C56; C60 Not-P de=l = 2.160 206 1960 P Not-P 39 l 40 20 3 23 59 4 63 1960 P Not-P 27 O 27 24 2 26 51 2 53 Figure A5-3.--Tests for Equal Transition Matrices for Voting Across Time Between Party Contact Status Groups, Ages 31-40 and 41-50, 1956-60 and 1972-76. A. 1972-76 1. Ages 31-40 C72; C76 P 1972 C72; C76 Not-P 2 _ de=l - .693 2. Ages 41-50 C72; C76 P 1972 C72; C76 Not-P X§f=1 = .035 Figure A5-3.--Continued. 207 1976 P Not-P 39 l 40 20 3 23 59 4 63 1976 P Not-P 89 9 98 17 2 19 106 11 117 208 .onumnma .+Ho one oonam .omuHa .omlam nmw< .naaono nonunoo sauna nomaaom mans mmouo< mufi>auo< cwfimasmu Hmowuwflom MOM mmowuumz :OfiuwwcmuH mo huwamsvm pom mumeII.QIm< ouswwm mm mm «M o m m an em am i -8 a cmma mm mm we NH o c we mm mm mluoz m ma com. u Huowx mluoz omommmo NNmH m emomwmo +Hc mow< HQM. " Hnvax anuoz emomNao mama m omommmo omnaa naw< co mN mm m m N an «N mm ¥[ mluoz m ommfi OMH no no «a w o baa mm mm mluoz m enma oam.~ u Huowx aunoz eaomwno Nana m omommmo oeuam mama ONM. " HHMWX auaoz eeommao Nana a eaommmo omIHN mmw< APPENDIX 6 FREQUENCY DISTRIBUTIONS AND TRANSITION MATRICES FOR THREE-WAVE ANALYSIS BY AGE COHORT, VOTING AND ATTEMPTS TO INFLUENCE THE VOTE OF OTHERS, 1956-6O AND 1972—76 APPENDIX 6 FREQUENCY DISTRIBUTIONS AND TRANSITION MATRICES FOR THREE-WAVE ANALYSIS BY AGE COHORT, VOTING AND ATTEMPTS TO INFLUENCE THE VOTE OF OTHERS, 1956-6O AND 1972-76 Introduction This Appendix includes Appendices A6-l through A6-12. Throughout the Appendix, "P" indicates that the individual parti- pated; and "Not—P" that the individual did not participate. In addition, for the chi square statistics, a single asterisk indicates that the value is significant at the .05 level, and a double asterisk indicates that the value is significant at the .01 level. Finally, age cohort membership for the 1956-58-60 panel respondents is determined by age in 1956; and cohort membership for the 1972-74-76 panel respondents is determined by age in 1972. 209 APPENDIX A6-1 FIRST ORDER TRANSITION MATRICES FOR VOTING ACROSS TIME BY AGE COHORT, 1956-58-60 210 Response at t + 1 P Not-P t = 1 P at t t = 2 (1958) 127 8 135 xzdf=1 = 26.698** Response at t + 1 P Not-P t - l (1956) 17 82 99 Not-P at t t== 2 98 136 234 x§f=l = 43.036** 2 2 = 2 = ** XP at; df=1 + XNot-P at t; df=1 de=2 69°734 Figure A6-lA.--First Order Transition Matrices for Voting Across Time, Ages 21-30, 1956-58-60. 211 Response at t + l P Not-P t - 1 (1956) 190 56 246 P at t t - 2 (1958) 195 7 202 X6f=l = 34.196** Response at t + 1 P Not-P (1956) Not-P at t t P 2 73 50 123 (1958) 85 117 202 de=l = 38.480** 2 2 = 2 = ** XP at t; df=1 + XNot-P at t; df=1 de=2 72-576 Figure A6-lB.--First Order Transition Matrices for Voting Across Time, Ages 31-40, 1956-58-60. - 212 Response at t + 1 P Not-P t a 1 (1956) 170 35 205 P at t t - 2 (1958) 184 4 188 x2df=l = 24.516** Response at t + 1 P Not-P t = l (1956) 18 37 55 Not-P at t t=- 2 (1958) 44 28 72 62 65 1 127 nggl = 10.052** 2 2 = 2 = ink XP at c; df=1 + XNot-P at c; df=1 xdflz 34.568 Figure A6-lC.-—First Order Transition Matrices for Voting Across Time, Ages 41-50, 1956-58-60. t-l (1956) P at t t=2 (1958) X2df=1 = 8.197** t - 1 (1956) Not-P at t t'- 2 (1958) 2 _ de=1 - 1.355 2 2 = 2 XP at t; df=l + XP at t; df=l Xd£=2 213 Response at t + 1 P Not-P 118 13 131 125 2 127 243 15 258 Response at t + 1 P Not-P 9 23 32 15 21 36 24 44 68 = 9.552** Figure A6-lD.--First Order Transition Matrices for Voting Across Time, Ages 51-60, 1956-58-60. t I 1 (1972) P at t t = 2 (1974) 2 = t I 1 (1972) Not-P at t t = 2 (1974) 2 = x df=l 5.936* 214 Response at t + 1 P Not-P 112 115 115 1 227 116 Response at t + 1 P Not-P 4 35 16 34 20 69 2 2 _ 2 _ X P at t; df=1 + XNot-P at t; df=1 ‘ de=2 ‘ 90-996** 227 116 343 39 50 89 Figure A6-lE.--First Order Transition Matrices for Voting Across Time, Ages 61 and Above, 1956-58-60. APPENDIX A6-2 FIRST ORDER TRANSITION MATRICES FOR VOTING ACROSS TIME BY AGE COHORT, 1972-74-76 t = 1 (1972) P at t t = 2 (1974) 2 = AA de=l 37.366 t = l (1972) Not-P at t t - 2 (1974) 2 = AA de=l 23.124 215 Response at t + 1 P Not-P 144 71 215 151 11 162 295 82 377 Response at t + l P Not-P 18 60 78 75 56 131 93 116 209 = 60.490** 2 2 = 2 XP at t; df=l + XNot—P at t; df=l Xd£=2 Figure A6-2A.--First Order Transition Matrices for Voting Across Time, Ages 21-30, 1972-74—76. t t 1 (1972) P at t t . 2 (1974) 2 = AA de=l 25.249 t - l (1972) Not-P at t t = 2 (1974) 2 = AA xdf=l 18.198 2 2 = 2 XP at t; df=l + XNot-P at c; df=l Xd£=2 216 Response at t + 1 P Not-P 133 41 174 137 5 142 270 46 316 Response at t + 1 P Not-P 9 37 46 46 32 78 55 69 124 = 43.447** Figure A6-ZB.--First Order Transition Matrices for Voting Across Time, Ages 31-40, 1972-74-76. 217 Response at t + 1 P Not-P t ' 1 154 32 186 (1972) P at t t - 2 (1974) 5 6 xzdf=l = 14.310** Response at t + 1 P Not-P t E l 7 22 29 (1972) Not-P at t (1974) x§f=1 = 5.220** 2 2 = 2 = AA XP at t; df=1 + xNOt-P at t; df=1 de=2 19.530 Figure A6-2C.--First Order Transition Matrices for Voting Across Time, Ages 41-50, 1972-74-76. 218 Response at t + 1 P Not-P (1972) P at t t I 2 139 4 143 (1974) 2 = *k de=1 10.973 Response at t + 1 P Not-P t I 1 7 31 38 (1972) 1 Not-P at t t I 2 20 32 52 (1974) 2 = * de=l 4.199 2 2 = 2 = AA XP at t; df=1 + XNot-P at t; df=1 de=2 15-172 Figure A6-2D.--First Order Transition Matrices for Voting Across Times, Ages 51-60, 1972-74-76. t I 1 (1972) P at t t = 2 (1974) 2 = * x df=1 5.783 t . 1 (1972) Not-P at t t I 2 (1974) 2 = AA de=l 7.250 2 2 = 2 XP at t; df=1 + XNot-P at t; df=1 de=2 219 Response at t + 1 P Not-P 130 27 157 129 11 140 259 38 297 Response at t + l P Not-P 10 48 58 29 46 75 39 94 133 = l3.033** Figure A6—2E.--First Order Transition Matrices for Voting Across Time, Ages 61 and Above, 1972-74-76. APPENDIX A6-3 SECOND ORDER TRANSITION MATRICES FOR VOTING ACROSS TIME BY AGE COHORT, 1956-58-60 220 1960 1229. 1958 P Not_P P P 115 3 Not-P P 12 5 P Not—P 48 5 Not-P Not-P 33 49 X§f=1 for uPPer half of matrix 19.179** for lower half of matrix 33.967** 2 xdf=1 2 — de=2 combined - 53.146AA Figure A6-3A.--Second Order Transition Matrix for Time, Ages 21-30, 1956-58-60. Voting Across 118 17 53 82 221 1960 £52 1958 F NO“, P P 185 5 NOt-P P 10 2 P Not-P 47 9 I Not-P Not—P 26 41 def=1 for upper half of matrix = 6.553* def=1 for lower half of matrix = 25.728** 2 X = ** df=2 combined 32.281 Figure A6-3B.--Second Order Transition Matrix for Time, Ages 31-40, 1956-58-60. Voting Across 190 12 56 67 222 1960 1229. 122§ P Not-P P P 167 3 Not-P P 17 l P Not-P 34 1 Not-P Not-P 10 27 X§f=1 for upper half of matrix = 1.232 = 37.200** xéf=1 for lower half of matrix x§f=2 combined = 38.432** Figure A6—3C.--Second Order Transition Matrix for Voting Across Time, Ages 41—50, 1956-58-60. 170 18 35 37 223 3229 1228 P Not-P P P 116 2 Not-P P 9 0 P Not-P ll 2 Not-P Not—P 4 l9 2 de=1 for upper half of matrix .153 X§f=l for lower half of matrix 15.423** Xéf=2 combined = 15.576** Figure A6-3D.--Second Order Transition Matrix for Voting Across Time, Ages 51-60, 1956-58-60. 118 13 23 224 1960 L922 1958 P No”, P P 112 O Not-P P 3 1 P Not-P 10 5 Not-P Not-P 6 2 de=1 for Upper half of matrix 32.578** 2 Xaf=1 for lower half of matrix 11.834** Xéf=2 combined = 44,412** Figure A6—3E.--Second Order Transition Matrix for Voting Across Time, Ages 61 and Above, 1956—58-60. 112 15 35 APPENDIX A6-4 SECOND ORDER TRANSITION MATRICES FOR VOTING ACROSS TIME BY AGE COHORT, 1972-74-76 225 1976 1972 1974 P Not-P P P 134 10 Not-P P 17 1 P Not-P 49 22 Not-P Not-P 26 34 2 . xdf=1 for upper half of matrix .0484 ng for lower half of matrix =1 8.760** xéf=2 combined = 8.8084** Figure A6-4A.—-Second Order Transition Matrix for Voting Across Time, Ages 21-30, 1972-74-76. 144 18 71 60 226 1976 .121; 1974 P Not-P P P 131 2 P Not-P 33 8 NOt’P NOt'P 13 24 xéf=l for Upper half of matrix = x§f=1 for lower half of matrix = X§f=2 combined = 41,394** 24.862** 16.532** Figure A6-4B.--Second Order Transition Matrix for Voting Across Time. Ages 31-40, 1972-74-76. 133 41 37 227 19 76 19 72 1974 P Not-P P P 148 6 Not-P P 6 1 P Not-P 23 9 Not—P Not-P 4 18 a )82f=1 for upper half of matrix 1.782 )82f=1 for lower half of matrix 15.036** >82f=2 combined = 16.818** Figure A6-4C.--Second Order Transition Matrix for Voting Across Time, Ages 41-50, 1972-74-76. 154 32 22 228 _1_9_72_ AL”. P Not-P P P 134 2 NOt-P P 5 2 P Not-P 15 6 Not-P Not-P 5 26 X§f=1 for upper half of matrix = 17.554** X§f=1 for lower half of matrix = 16.160** 2 = *9: de=2 combined 33.714 Figure A6-4D.--Second Order Transition Matrix for Voting Across Time, Ages 51-60, 1972-74-76. 136 21 31 '1. 229 1976 1972 1974 P Not-P P P 122 8 Not-P P 7 3 P Not-P 20 7 Not-P Not-P 9 39 2 xaf=1 for upper half of matrix 7.231** x§f=1 for lower half of matrix 21.67l** x§f=2 combined = 28.902** Figure A6-4E.--Second Order Transition Matrix for Voting Across Time, Ages 61 and Above, 1972-74-76. 130 10 27 48 APPENDIX A6-5 ACROSS COHORT TESTS FOR EQUALITY OF SECOND ORDER MATRICES FOR CONTINUAL VOTING, 1956-58-60 AND 1972-74-76 .oolmmlcmmH .wcwuo> Hosafiuaoo now wmowuumz nacho vcooom mo zufiamsvm pow mummy unocoo mwouo Hmscfiucoo pow mmofiuumz umcuo waooom mo hufiawsvm HON mummy ocN OH omN oma N qma OMH w NNH M mluoz m 2 Quad moN OH mmN mma N Hma omH w NNH mluoz m onmfi .owlcmleaH uuosou mwouoIcoz Hmocauaoo now moofiuumz uwwuo vaooom mo >uHHmsvm now momma uponou mmouolcoz Hmscfiucou now mmofiuumz porno wcoomm mo mafiamovm HOW mummy uuosoo mmouoauu< on Ha mm am a «H ma 6 Ha muuoz m cows 66 a em Hm m cm ma 6 Ha muuoz a coma owH. fl HHHWX Golan Nuu pom fiuu noon on +~o mluoz Hma. u Hucwx 06-2n was 6:6 Hus £666 66 p-062 +26 mm m mm mm n ma ma 6 Ha mluoz m coma om Hm mm Hq mm «H ma q HH mluoz m coma «NH. n Huuwx Onsfle NHU “yam fin“ nuon um mluoz +~c $621.6 n Twex omIHN Nuu can Hnu noon um muuoz +Ho 255 .oauqanwAAH .muAamuu< mucmnHuaH AH >ufi>fluo< Hmsaflucoo how mommmooum umwpo vcoumm mo zuHHmsvm MOM mummy uuonoo mmouuwuu< Hmscwuaoo How mommwuoum pmvuo wcoumm mo zuwamscm How mume uuocoo mm0poHuo< Hmoawuaoo you mommmooum umpuo vnoumm mo zufiamsum you mummy phonoo mmouo