An analysis of undecided voters in U.S. House of Representative elections

MISSING IMAGE

Material Information

Title:
An analysis of undecided voters in U.S. House of Representative elections
Physical Description:
vii, 169 leaves : ; 29 cm.
Language:
English
Creator:
Rossi, Deborah Charnes
Publication Date:

Subjects

Subjects / Keywords:
Elections -- United States   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1993.
Bibliography:
Includes bibliographical references (leaves 160-168).
Statement of Responsibility:
by Deborah Charnes Rossi.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001923477
notis - AJZ9311
oclc - 30541371
System ID:
AA00003256:00001

Full Text














AN
IN U.S.


ANALY
HOUSE


OF UNDECIDED VOTERS
REPRESENTATIVE ELECTIONS


DEBORAH


CHARNES


ROSSI


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
















ACKNOWLEDGMENTS


Although a


single


name appears


on the


title


page of


this di

effort.


.ssertation,


cannot


is most


adequately


certainly a


cooperative


express my gratitude


to my


committee


for not


only their


counsel but


also


their time,


energy,


and patience.


presented


usual


problems


of a candidate


to their


committee


along with


unique


ones


an older


student


had been


school


in the practical


years


applications of


and who was


voter


behavior


interested


research.


am extremely grateful


to Dr.


Margaret


Conway,


chair of my

measures of


dissertation


advice,


committee,


encouragement,


who offered

and reading


equal

lists.


virtue


Martinez


of being


suffered


in town


through my


summer


first


of 1992,


experiences


Michael D.


with SPSS.


Martinez


also provided the


insights to examine the


differences among undecided


voters


on the


basis


of political


involvement


and interest.


Richard K.


Scher


offered


counsel


to make


every paper


wrote


class


relate


to my


overall


research


interest


undecided


voters


that


might


graduate before


was


eligible


for


senior


citizen










William R.


Hamilton


guided my polling


efforts and


provided proof


that


practitioners,


such as


could


exist


in an academic environment.


John


Sutherland gave


friendship,


guidance,


and support.


Outside of


to many people.


dissertation


Alfonso


committee,


owe gratitude


Damico provided the


constant


assurance


that


could make


through


the doctoral


program


and that


requirements


were


reasonable.


My work


qualitative


direction


and Dr.


research


of Dr.


Stephen


would not


Sandra


Craig


have been


Damico.


initially


possible


without


Kenneth D.


convinced me


Wald


that


political


hack


could both benefit


from and contribute


to the


field of voter


behavior.


frequently


consulted Dr.


Wayne


Francis


regarding the appropriateness


of my


statistics.


am also very grateful


to Mrs.


Debbie


Wallen.


Without


knowledge of


forms,


deadlines,


and procedures


would


have been


able


to complete


this


program.


None


this


would have been


possible


without


love


understanding


of my husband,


Bill.




















TABLE OF CONTENTS


CHAPTERS


The Problem . . . .
The Significance of the Study.
Definition of Terms .......
Scope of the Study .. ..


*. . f. S. S .f S .
*.. . ..*. . ..tft S t
. .t t. .. f.. .t ft. .. .t . .f
* f. .S ft ft f. .ft f S f


PRIOR RESEARCH AND ITS IMPLICATIONS......


Research on Cross-Pressures.....................
Research on the Influence of Short-Term Factors.
Research on Mass Communication..................
Research on the Psychology of Voter
Decision Making .................... ........
Democratic Theory and Voter Decision Making....
Summary .. ... .. .. ... ... .. .. .... ...


Context of the 1982 Election.
Context of the 1992 Election.
Methodology .........a .. .
Quantitative Research Design.
Qualitative Research Design..


* f. S. S t. ft. .. . .ft
. .f t. .. ft ft. .. ft. t.. . ~
S. S. .f ~ .. ft S. f S .t.
. .ft . .f ~ ft.. 5 t. .. ft .
*.. ft S S 5. t. t .f f


RESULTS .. ..


Quantitative Results. .. ...
Other Results .. .. .. .....
Discriminant Analysis... ........
Discussion of Quantitative Results


. .. ..tt t..


Pac_


ACKNOWLEDGMENTS .. .. .. ..


ABSTRACT .. .. ..


INTRODUCTION .. .. .. ..... ............ .....


METHODOLOGY. ...................................












CONCLUSIONS ... .. ..... .. .. ................. 112


Research Question.. .... .... ...
Reduced Model. .. . .
Full Mode .. .. .. .. .. .... ...
Relationship of Findings to Previous
Implications for Practitioners......
Future Research ... . . .


*. .. C. .
*. . . C
. . ..C *
Research.
. . ..C *
. . ..C *


APPENDICES


FOCUS GROUP RESULTS . . . . . .. 123

ALACHUA COUNTY HIGH INTEREST/LOW INVOLVEMENT
GROUP .. .. .. 131

ALACHUA COUNTY HIGH INTERES/HIGH INVOLVEMENT
GROUP .... .... .... 135

SUMMIT COUNTY HIGH INTEREST/MIXED INVOLVEMENT
GROUP ... .. .. .. .. ... ... .. .. 141

TELEPHONE INVITATION--FLORIDA............ ..... 146

TELEPHONE INVITATION--OHIO .. 148

SELF-ADMINISTERED QUESTIONNAIRE--FLORIDA ......... 150

SELF-ADMINISTERED QUESTIONNAIRE--OHIO. ........... 153

FOCUS GROUP DISCUSSION OUTLINE--FLORIDA.......... 156

FOCUS GROUP DISCUSSION OUTLINE--OHIO.. 157

POSTELECTION SURVEY. .. .. .. . 158

INFORMED CONSENT FORM................C. ......... 159


REFERENCES . . . . . . 160

BIOGRAPHICAL SKETCH . . . .. . 169
















Abstract


of Dissertation Presented to


University


of Florida


the Graduate


in Partial Fulfillment


School
of the


Requirements


for the


Degree


of Doctor


of Philosophy


AN ANALYSIS OF UNDECIDED


IN U.S


VOTERS


HOUSE OF REPRESENTATIVE ELECTIONS


Deborah


Charnes


Rossi


August


1993


Chairman:


Margaret


Conway


Major


Department:


Political


Science


primary


research


question


is as


follows:


are


some


voters


still


undecided


less than


one month


prior to


election

utilizes


for the

data fr


U.S.


om five


House


of Representatives?


telephone


surveys


This


Virginia,


study

Ohio,


and Florida


conducted


during the


1982


1992


elections


three


focus


groups


in Ohio and Florida


held during the


1992


election.


data


from the


focus


groups


indicate


that


there


are


differences between


undecided voters


in an


open


incumbent


congressional


district.


A high


level


of political


involvement


general


political


information


does


not


preclude


voter


from being undecided.


The differences


the 1


evel of political


information


among undecided


voters










seat.


Where


information


available,


undecided


voters


tend


to be


cross-pressured between


party


affiliation and the


opinion


incumbent.


This


research


shows


that


undecided


voters


had paid attention


to the election,


most


had decided


their presidential


vote.


decision


was


finally made on


the basis


of party


affiliation


and issue


positions.


Two of


focus


group participants


did not


vote


this


race because


candidates


that


they


was


could not


sufficient


obtain


to make a


information about


decision.


This


indecision


research suggests


in presidential


that


theory


elections


lack


linking


information


also applies


U.S.


undecided voter


House of Representative


less


information


elections.


concerning the


election but


is neither unaware of


nor unconcerned about


this


lack


information.


The

results


theory that


indecision


two elections


cross-pressure


three


also


states


exerted upon


supported by the


this


data


study.


voter

from the


can


conclude

elections


that

but


this

also


theor

U.S.


y


applies to not


only presidential


House of Representative


races.


qualitative


research has been


successful


establishing


a link between


information


flow theories


cross-pressure


theories.
















CHAPTER


INTRODUCTION


Problem


This


research


focuses


upon


elections


for the


U.S.


House of Representatives.


examines


the question


of when


people decide


for whom to vote,


what


factors


affect


timing


this


decision,


whether there


are


differences


in the


attitudes


or demographic profiles


between


decided and


undecided voters.


Previous


research addresses


this


question


only


presidential


level,


indicates


that


long-


term


factors


such


as party


identification


and political


attitudes


are key


influencers


(Berelson


et al.,


1954;


Campbell


al.,


1960)


The quantitative


phase


examines


two major


questions.


First,


what


are


the differences between


the decided and


undecided


voter


in campaigns


characterized by


lower


levels


of information,


terms


such as


of demographics,


U.S.


House of


issues,


Representatives,


levels


involvement?


Secondly,


what


are


differences


among


undecided


voters?


Cross-pressure


theory


examined to determine


applicability as


we move


from the


high


information


level


I


~











believes


crucial


that


information


difference


between


about


candidates


presidential


congressional


elections.


questions


of when


electoral


cycle


people make


up their minds


for whom to


vote


and what


factors


influence


much


timing


attention


as vote


this

choice


decision have not


received as


field of political


behavior.


Yet,


undecided voters


least


informed


voters


hold the balance of power


the election.


can


even be argued


"that


less


involved and less


informed


voters


are


disproportionately


represented"


(Converse,


1962,


579)


research


question


follows:


are


some


voters


still


undecided


less


than


one month prior to


election


for the


U.S.


House of Representative?


Three


theories


guide


the quantitative


phase of


this


research.


Lack


information


distinguishes


congressional


elections


from presidential


ones.


"The


relative paucity


information


voters


have about


congressional


candidates


crucial


because


of the


connection between


information


about


candidates


and voting


decision


congressional


elections"


(Jacobson,


1980,


The voter


undecided because


s/he


not


paying


attention


yet.


essence,


a presidential


election


rI -n r1 ,-i y l- -


*"^ !


I 1 .


1


C- f^. fc\ ^


4- 1^ cl+


- C t^ i/ 4- <-


1-S f /n n











less


involved


elements


of the


electorate"


(Converse,


1962,


584)


Cross-pressures


exerted


upon


voter


results


indecision.


Voters


can be


pulled in


conflicting


directions


by their political


party,


their


friends,


family,


religion,


candidate


evaluations,


or attitudes


(Berelson


et al.,


1954;


Campbell


al.,


1960;


Kelley


& Mirer,


1974;


Flanigan


Zingale,


1987).


specific questions


that


guide


the qualitative


phase


research


are as


follows


there differences between


voters


in an open


district


and an


incumbent


congressional


district?


levels


involvement


political


activities


impact


undecided


voters?


there differences


level


of political


information


among the


undecided voters?


undecided voters


cross-pressured?


how?


Are undecided


voters paying


attention


to the


election?


the decision


finally made?


some


undecided


voters


vote?


The qualitative


research


provides


insights


into


process of


vote decision


rather


than


product


of that


A ^ Q- n---hb


A


/*^ th< -r k ki


I rr\ r r.n r\ 1 fl a i 1" tt


S1^


I I I t I


1


nE /


t7n


I I I











that


information and


involvement


undecided


voters


are not


paying


attention.


Sicrnificance of


Study


significance


this


study


that


it addresses


undecided


voter


in a


congressional


election.


Theories


regarding undecided voters


have been based only


upon


research


in presidential


elections.


This


research


focuses


upon


voters


in U.S.


House of Representative elections


determine


theories


hold.


This


research


examines


the differences between


decided and undecided


voter.


likely,


however,


that


undecided


voters may not be a homogenous


group


(although


they may


have more


in common


with


each


other than


they


with


decided


voters)


This


classification by


information


involvement


levels


is a


first


attempt


to define


differences


among undecideds.


This


research moves


the question


voter


indecision


from a


high


information


election


of president


lower


information


level


election


House of Representatives


where historically


winner.


incumbency


an attempt


prime


to establish a


determinant


link between


information


flow theories and


cross-pressure


theories using


Jacobson's


work to


support


the assumption


I hnrn: n^
^^ / i~s/"* ^^^/^ ^ i' ^ / <^\^^


1 1 >C


* 1 I


- a I


rnt-IS-


1 A











Definition


of Terms


Level


information


the extent


to which


voter


aware of the


candidates and


issues


the election.


Cross-pressure


from opposing


forces


input


that


psychological


voter


basis


receives


for this


phenomenon


is referred


to as


avoidance


reaction.


action


taken


in this


case,


not


taken)


to avoid or


reduce


amount


anxiety


or tension


caused by


conflict.


It has


roots


in cognitive


dissonance.


conflicting


pressures


conflict,


can be


resolved by


reinforcing


denying that


or bolstering


one


there


of beliefs over


another,


differentiating the beliefs


so that


a conflict


longer exists


, or withdrawing from the


situation


that


causes


conflict


(Abelson,


1959)


Voters,


subjected


cross-pressures,


hold party


identification


more


important


than


their


evaluation


incumbent.


They may


be motivated to alter


their


opinion


resolve


inconsistency.

The voter's


they may not


level


vote.


involvement


measured by the


number


of political


activities


in which


s/he participates.


Scooe of


Studv


Some


cautionary notes


are


in order


this


point


There


are


issues


which


this


research


does


not


address.


L.. -. -, L.. A.. I, ....- A..4.


__ *


I -


-V 1


II


~













evidence


in political


psychology


studies


that


undecided


voters are


uninvolved and/or


exposed


little


political


information


(McGuire,


1968;


Zaller,


1987)


Some


voters may


decide


upon a


candidate


and,


subsequently,


change


their


minds prior to the


actual


vote.


These


people may be more


like


the undecideds


than


decided.


This


question,


however,


was


addressed.


Panel


data


would be necessary


to measure


when


switching


occurs


and which group the


switchers


most


closely resemble.


The models


developed


in previous


research


were designed


to predict


vote


profile of


choice,


undecided


timing


voter.


This


the decision or

research builds


upon


that


work and also moves


research


question


into


area


of undecided voters


congressional


elections.


is also


likely that


undecided


voters


are


homogeneous


group.


The quantitative


phase of


this


research


addresses


the question


the differences


between


decided and


undecided


voters.


focus


groups


address


differences


within


undecided


group.
















CHAPTER


PRIOR RESEARCH AND


IMPLICATIONS


Research


in political behavior primarily


offers


cross-


pressures


theory evolves


Elmira


explanation


from the


to attitudinal


social


for undecided


voters.


cross-pressures


cross-pressures


to emotion.


identified in

nal ones.


Short-term factors


timing


campaign


vote decision


itself


and cause


can


to vary


affect


election


election.


Research


communications


field attributes


indecision


primarily to


lack


of media


exposure


and,


therefore,


lack


information.


Psychological


studies


cite


avoidance


resolving the


the decision

conflict


itself


A discussion


a means o

n of each


f avoiding


body


research


follows.


Research


on Cross-Pressures


Research


in political behavior


suggests


that


voters


experience c

delaying the


ross-pressure

presidential


will

vote


resolve the conflict by

decision. Cross-pressure


theory


evolved


include


various


types of pressures.


trace


with


the evolution


the citizens of Elmir


cross-pressure

a during the e


research,


election


we begin

1948











coworkers,


or from friends


and family


and,


therefore,


were


more


likely to arrive


their vote decision


later than


citizens


from homogeneous


groups.


Furthermore,


the more


intense


one-fifth


conflict,


to one-fourth of


later the decision.


respondents


Approximately


resolved conflict


by not v

to this

avoid or


oting at


reduce


all.


avoidance

the amo


Sociologists

e reaction;


unt


of anxiety


and psychologists


an action


or tension


refer


taken


caused by


conflict.


Subsequent


pressures


research not


delay the


only


vote decision


confirms


theory


that


cross-


also add


to the


strength


this


theory.


Campbell,


Converse,


Miller,


Stokes


(1960)


examine


role


that


cross-pressures have


timing


"attitude


pressures


Respondents


vote decision,


conflict.


social


were


although


they term it


They added these attitudinal


ones


asked about


cross-


identified in Elmira.


partisanship and six


attitudinal


dimensions:


domestic policy,


A measure


foreign


of consistency


Stevenson,


policy, and

or conflict


Eisenhower,


success


of attitudes


groups,


parties.


was


then


developed.


deciding


They


late


conclude


ALWAYS


that


percentage of


greater


degree


voters


of attitude


conflict


increases.


-'


I I a^ fl% a .-% C1 a L


-- S


1 -. \ F-. a i


-- -


I a yy RA-' .-a-n .-. I In I^ I i .


rh


^


n...rr*ln r~C


II \


Ih


I











expresses


a net


equal


number


favorable


responses


candidates


and/or parties,


the greater the


likelihood


that


s/he


would be


undecided or not


vote


at all.


1984


campaign,


Reagan


was


perceived as


both


effective


subjected


and popular.


cross


Therefore,


pressures,


voters


and Reagan's


were not


lead had been as


high as


26 points according to


NBC and Gannett


polls.


first


October debates brought


Mondale


closer


cutting


the President's


Independents,


lead by one


young voters,


half.


Mondale had picked


and Democrats.


second


debate,


however,


Reagan


appeared refreshed and


in control.


voters


responded by


staying with the


familiar.


Mondale


never


a chance after that


(Ladd,


1985,


Reviewing


1948


1984


elections


through data available


from the


Survey


Research


Center


the Center


Political


Studies,


Flanigan and


Zingale


(1987


conclude


that


reasonable


to expect


conflicting political


attitudes


to be


associated


with


indecisiveness


in voting


behavior"


1952


1956,


they


concur that


attitudes


favorable


Democrats


and Eisenhower were


associated


with nonvoting and


indecision.


Conflicting


attitudes


or values


are


termed


dissonant.


Psychologists


assume


people will


behave


manner to eliminate or


reduce dissonance,


those


* -- - - I


t *


1 1 ;


.1 _


11











Using


data


spanning the


1952-1960 elections


from


separate


surveys


over


130,000


interviews,


Pool,


Abelson,


and Popkin


(1965


create a


simulated model


1960


election.


They particularly


examine


cross-pressures


that

the


result

effects


from having


a Catholic nominee.


cross-pressure on


They examine


pro-Catholic Republican


anti-Catholic Democrat.


They


do not


address


timing of


vote decision,


although


they do present


findings


of Berelson


(1954)


"People


under


cross-


pressure make


their minds


late


campaign period.


People


under unified


pressures


are


to have


their minds


made


up from the beginning"


(Pool


al.,


1965,


p.13


Pool


et al.


differ with Berelson


et al.


(1954) ,


however,


that


citizens


with


the pro-


or anti-Catholic


cross-pressure


resolve


it by not


voting.


Pool


discuss a


theory


of push and pull


pressures.


A voter


being pulled


to a


candidate


Kennedy)


will


vote,


rather than a


voter who


being pushed away


(Nixon)


They


interpret


this


to mean


that


voter who


is attracted


to a


candidate


will


have


a greater


propensity to


vote


than


one


is pushed


away.


rate,


we can


say with


confidence


that


simple


theory that


postulates


longer


nonvoting


be maintained"


(Pool,


outcome of

et al., 1


cross-pressure


can no


965,


-4 -J i f (_


g q


1 I


I


__


I


, ^ -' ^ _










electoral


support


came


from people


were voting


against


Jimmy Carter


than


Reagan


333)


They argue


that


anticandidate


voting


is a device by which a


person


can


resolve


the cross-pressure conflict.


Gant


and Sigelman also


claim to


have


supported their


hypothesis


that


those who


decide


late


in a


campaign


are more


likely to cast


anticandidate


vote


than


those


who decide


early,


particularly


1980.


Research on


Influence of


Short-Term Factors


In addition


to cross-pressures,


short-term factors,


such as


campaign,


affect


timing of the


vote


decision.


This


is particularly true


when


voter


still


undecided


Mirer,


late


in the


1974)


campaign

elections


(Campbell,


1952


1960;


Kelley


1956 Campbell,


Converse,


Miller,


Stokes


(1960)


report


that


44%,


respectively,

would vote.


those


interviewed knew all


Another one-third decided at


along


how they


time


conventions.


Only approximately


decided


within


weeks of


the election.


They


interpret


these data


to mean


that


psychological


forces


work prior to


the beginning


of the


campaign


are


the key


factors


that


influence


vote.


"In both


the elections


1952


1956 the


proportion


variance of


voting


choice explained by these


nfn*^ ma^ a TnnT oa'o 1-rn T


etOrhh l nr! r l


th- n n


+- ^mo


o rr r *tr'QahQT


o otron


aMnnrrnr










decided


final


days


of the


campaign"


(Campbell


et al.,


1960,


p.79)


Short-term forces


campaigns


are


also among the


factors analyzed


in the


five


presidential


elections


(1952


1968)


Kelley


and Mirer


(1974)


their


attempt


to explain


how and why the electorate


votes as


does.


They accept


Campbell's


work


on attitudes


and parties


accounting


of the


explanation but move on


to explain


remaining


15%.


their work,


Kelley


and Mirer


address


complications


voters


changing their minds during


campaign period.


"Respondents


whose


decisions


were


close


ones


were more


likely


than


other


respondents


to have


changed


their minds and

favorable to the


to have ch

candidate


anged them in

favored in t


a direction


flow


information


in each


election"


(Kelley


& Mirer,


1974,


583)


They


find


this


to be


case


particularly


in the


1968


election.


They


claim


that


nearly


voters


changed


their minds


about


whom they would


vote during the


campaign


period.


Hinckley


(1983


claims


that


issues


also play


role


congressional


elections


because even a


few people


can make


difference


in the outcome of


election.


cites


evidence


fact


that


roll


call


votes


can hurt


1CI 4 a I a I


V^\j^ I I "*


II~nn f f~^ ^\1 C^/11/- t^+ ^ 1\^


HA 1 A Ulr. 1











providing


spending


(for or


against


a candidate),


which,


turn,


affects


visibility


vote


There


some


evidence


in the


research


that


timing


vote decision


the decline of party


is occurring


allegiance


later


(Ladd


in the


process


Hadley,


1973,


1975)


"There has been a


long-term decline of


party


allegiance,


and a


dramatic drop-off


over the


last


decade"


(Ladd


Hadley,


1975,


300)


They attribute


this


decline


to the


higher


level


of education,


higher


level


information


increased


concerning


issue


both


orientation,


issues


candidates,


an increase


in confidence


voter'


own


ability to


assess


candidates


and programs,


rise


in independents and ticket-splitting,


and reliance


electronic media


for political


information.


Ladd


(1981)


cites


1980


NBC News


and Associated Press polls


campaign period as


evidence of


taken


later


during


decision.


Fifty-two


percent


had not


made


up their minds


in early


August.


late October,


those


likely to vote


still


had not


decided


for whom.


Republicans


generally


decided;


was


the Democrats


Independents


that


were


still


not


sure.


Ladd also


reports


that


proportion


likely voters


who were


still


undecided in


last


week


1980


campaign


was


the highest


in American history.


41 4- -- I-.J -I L1


nnr-


* I


~\ L1* r -L


-^ -~~ ~ ~ -3 ^ 3 -_ T


U .


'* I l











week


24%.


to the


"The


campaign;

weakness of


candidates


in 1956,


voters'


1980


was


1972,


ties to


also apparent


and in

parties


in both


1976,

s and


the high


measure of


uncertainty


about


how to


vote


unusually


late


decision by many voters"


concludes


that


(Ladd,


the Carter-Reagan


1981,


debate


Ladd


week before


election had an


impact


the Reagan


win.


He dismisses the


theory


had not


"closet


correctly


Republicans"


anticipated


reason


size of


that


win.


polls


Rather,


Ladd attributes


last minute


decides.


He cites


the Democratic daily polls


support


of his


theory.


There


was


a rush


of undecideds moving


over to


Reagan between


November


and November


timing


of the


vote decision


varies


from election


to election.


"The


high


degree


of volatility


the electorate's


expressed presidential


preferences


throughout


1980


campaign


occurred


part


from the


weakness


of partisan


attachments"


(Ladd,


1985


, p.


Voters


perceived


their


president


ineffectual but


were not


sure


alternative.


result


was


that


many


did not


decide


they would


vote


until


the election.


1984


Ladd


states


"Party ties


were


still


weak,


referendum on


incumbent


president


went


decisively


way.


As a


result,


- -- n-~-4 -- ~ -, ..- 2 1 .-% -- u4 4 n- -. 4


13%


* t- /


I 1 1 1 -,,


,L,, L


~mA











voters


had made


their minds


vote


Reagan before


first


primaries


in February.


Using the


time of


the deci


sion


question


from the


National


Election


Studies,


1948


-1984,


Flanigan


and Zingale


(1987)


establish


that


generally the


decision


was made at


same


time


for the


same


proportion


voters


over the


entire


period.


About


one-third made


their minds


prior to


conventions,


another third decided during the conventions,


and the


1956


final


1984


third decided


elections,


during the


popularity


campaign.


president


increased

convention


1976


number of voters


to one-half.


1980 made


who decided before


In contrast


their minds during


the

voters


campaigns.


They


also


claim that


recent


elections,


independents and


weak


partisans


decide during the


campaign


while


strong


partisans

widely re


line


ported


up behind


that


their


the most


candidate

interested


early.

voters


remain


undecided


during the


campaign"


(Flanigan


Zingale,


1987,


167).


Research


on Mass


Communication


Communication


studies attribute


voter


indecision


primarily to

information


lack o

(Kimsey


f media


Hantz


exposure

, 1978;


and

Lucas


therefore a


Adams,


lack


1978;


n - I flfl C' ..r -----Ia A I -- -I--I- -


r\\


1 r\


_ f .^ 1 1 -


-L J_ I 1











days


prior to the election


to those


had decided prior to


campaigns


to those


who decided


during the


campaign.


Chaffee and Choe


determined


that


late


undecideds


relied


upon


latent


party


identification


to make


their vote choice.


Another


study


1980 presidential


election


attempt


build a model


which


could predict


vot e


choice of


undecided based


upon attitudinal


and demographic variables


candidate evaluations


(Fenwick,


Wiseman,


Becker,


Heiman,


1982)


Using


discriminant


analysis,


they


found


that


single


best


predictor of


vote choice


candidate


evaluation.


The model


correctly


classified


voters


were


undecided at


time of


survey,


2-1/2


weeks prior to


the election.


Watters


(1988


analyzes


undecided


voter


1980


presidential


election


using


communication


theory,


particularly


rhetorical


vision.


"The


research


examines


linkages between


audience


interpretation


of public decisions


of undecided voters


compares


undecided and committed


voters

decision


in terms

n within


each"


rhetorical

(Watters, 1


communities and


988,


voting


A rhetorical


vision


is a shared view of


social


reality


usually


symbolized


a slogan.


rhetorical


community


the group of


individuals


respond


to messages


in line


with


their


- -,,, rL -.- A1~- .-T- -.- .---- 1


I


A


__


T.T


LL


L










these data


that


undecided voters


do participate


rhetorical


visions,


and,


therefore,


are


part


rhetorical


community.


also finds


that


undecided


voters


are


not


the uninvolved and


uninformed


citizens.


They


did not


express a


candidate


preference


prior


to election


day,


they


expressed


clear


perceptions of


the campaign,


concerns


about


particular


issues,


In short,


and biases
undecided


toward
voter


candidates.


was


revealed


this


campaign
rhetoric


study


process
1 vision


candidates.


an active


participant


and participated


S


the president


(Watters,


1988,


in distinctive
tial campaign
156)


Cross-pressures


could


cause


probable


voters


to decide


late.


Flow


information and


political


involvement


theory


relates


to this


phase


research


design.


Converse


(1962)


was


concerned about


the consequences


flow of


information


upon


vote


choice


without


regard to


content


of the


information


or the


type of media


utilized.


hypothesis


that


uninformed


voter


the most


stable


voter.


The partisan


choice


cannot


be deflected


flow


information


information,


discussed


weak.


vote


paradox


a pure


that


voter


receives


party vote.


voter with


no new


Converse


least


information about


candidates


and issues


the one most


easily


changed--if


information


can


to him or


her.


-t0 r\ 0-o-h mm cQ 0


moH i


Pffort


Prtmro7 r P


fr'l nl t na


haH Tinn


Q-v^/ Qn 911 r'










during


a campaign,


early


stable


or do not


campaign


(Converse,


1962) .


pay any attention


their decision


tends


voter who has


decide


remain


moderate


exposure


likely to


change


mind.


Milburn


(1991


discusses the


level


of interest


that


person has


subject


as another variable affecting the


ability


support


a message


Converse


to effect


that


change.


decreasing


conclusions


interest


voters


in politics


results


little


information


getting


through and,

elections ar


therefore,


id campaigns,


short-term situations


attitudes


will be


such as


susceptible


change,


large


swings


in opinion


happened in


1988


election)

Zaller (3


will be


987)


fairly


combines


likely"


the models


(Milburn,


1991,


of Converse


(1962)


129).

and


McGuire


(1968)


Converse


finds


that


the higher


levels


political


involvement


are


associated with heavier


exposure


to partisan appeals and greater


amounts


stored


information


Therefore,


highly


involved are


stable


their


opinions,


are


uninvolved.


McGuire


adds


intelligence,


self-esteem,


and freedom from anxiety to the


variables


that


affect


persuasion.


finds


these


to be


positively


associated


with


exposure


the messages but


negatively


associated


with acceptance of


the messages.


Mr. Cii ra 1 or' a r F tZ\ ry


: nt 1~n 1r


C+- r\f- n nj- rt^i/r -


-i~ ^/-3k n\/*


T- n \/"


/"\-r--











follows:


"The highly


involved have high-exposure and


low-


acceptance


probabilities,


those


involvement


have


low-exposure and high-acceptance


probability"


(Zaller,


1987,


823)


Zaller then modified the model


include


partisanship.


Using


data


collected


from the


1964-1986 NES,


Zaller


confirmed his model.


The amount


type of political


information


to which


voters


are


exposed


can be


heavy


voter


paying


attention


campaign


or moderate even


they are


paying


attention


to the


campaign


(Berelson,


Lazarsfeld,


& McPhee,


1954)


"The more


that


people


read about


listen


the campaign


the mass


media,


the more


interested


they


become


election and


the more


strongly


they


come


feel


about


their


candidate"


245)


Presumably then,


they would have decided


upon


their vote


choice.


Bereleson


also


claim that


"media


exposure


gets


vote at


same


time


that


solidifies


preferences.


crystallizes and


reinforces more


than


converts"


248)


This


would only


true,


however,


individual


conflicting


is not


subjected to cross-pressures


information and attitudes.


Bereleson


claim further that


a "deviant"


intends


vote


candidate not


in line


with


choice of


peers,


will


oT7 4-nh t


-yr flj rn a 4- 4i 4-h r1 .


I *n


nlnn ^ ~ -^---i^ ^


------ -


T


-1-V^ a 4


I-- p^ fn


n" r^ n~











more


likely to be


unstable"


282)


They


categorize


sources


cross-pressures


social


structure,


party


identification,


campaign


appeals.


sources


pull


in a


direction


opposing the others,


voter


experiences


cross-pressures.


The options


open


to a


cross-


pressured voter


are


switch


their vote


intention,


decide


late,


or not


vote at


all.


Using Gestalt


theory


borrowed


from psychology,


Berelson


explains


that


the most


likely


option


to switch


the deviant


position


or vote


choice


maintain


consistency


in beliefs


and a homogeneity with


their


reference


fitness


than


groups.


is a more


reason and


short,


striking


appears


feature


calculation"


that


of political


311)


sense


preference


vote


choice


that


stable


the majority


of voters,


then


undecideds become even more


important.


Research on the Psychology of Voter Decision Making


A study


a psychologist


voters


used the emotion-


eliciting


appeal


of candidates


to classify


decided and


undecided


voters


Texas


gubernatorial


race


(Christ,


1985)


Christ


found


that


the more


positive


arousal


responses


elicited by


a candidate,


the more


their preference


could be discriminated.


The mean


scores


the discriminant


function


then,


used to predict


undecided


voters'


r


- I _


I i


A











Democratic Theory


Understanding the


and Voter Decision


undecided voter


Making


important


democratic


theory.


According to Converse


(1962),


paradox


that


the highly


interested,


involved,


informed


voter


decides


early


and,


in effect,


takes


himself


of the


game.


the attention and appeals


are


then


made


to the


studies


undecided voter,


as not


interested in


is described


election and not


in previous


informed.


Thus,


in a


close


election,


undecided


voter


holds


balance.


ideal


citizen


is not


even


a player.


If he


exists


the question


that


should


concern


political


scientists.


Even


if he does


exist,


he does


not


matter.


important


to know


if the


undecided voter


different


than


the decided


ones,


as well


whether


or not


highly


interested,


Berelson


informed,


comments


upon


involved voter


can


importance


also be


this


undecided.


undecided


voter:


For those


readily


who change


are th


ose


political


are


least


preferences
interested


most


are


subject


to conflicting


social


pressures,


have


inconsistent beliefs


and erratic voting


histories


Without


them--if


the decision


were


left


only to


integrated,


the deeply


concerned,


consistently-principled


well-
ideal


citizens--the


political


system might


easily prove


too


rigid


to adapt


to changing


domestic and


international


conditions.


They may


least


partisan and the


least


interested


voters,


they perform a


valuable


function


-. -N A


-N -S *( t


I n^ 4- 4v r I n I t' rN r% t II


A I


1n t r


'I


Srr-< T ^ -j^


I IIL. 71I J











fool


or not,


or whether the


voter


changed,


unchanging,


is not


vital


as understanding the


undecided


voter.


undecided voter


holds


the balance of power


the attempts to gain her/his


attention and vote may alter


the American agenda--at


least


for the duration


of the


campaign.


significance of


this


research


that


"undecideds"


decide


the election.


Elections


are generally


won at


the margin.


important


to understand the


undecided


voters'


thought


processes,


what


moves


them to a


decision and when.


Converse's


(1962)


"Information Flow and


Stability


of Partisan Attitudes" presents


foundation


this


research.


People


who are


the most


susceptible


the media


effects


of persuasion,


agenda


setting,


or priming


are


the people


who are


the most


difficult


to reach and the


least


likely to vote.


to decide


When


the election.


they

work


vote,


to date


however,


they tend


on undecideds


been


done


that


presidential


information


level


level,


is higher.


where


Are


is documented


conclusions


valid


lower-level


information


campaigns,


such as


U.S.


House of


Representatives?


Zaller


(1987)


suggests


high


involvement


equals


high


exposure


information


low acceptance


of messages


-i -1 A_ -/- - -- I


/*J A -^ -. L J !


__J 1_


* 1 --


~ I.~ UL


L


'I -- --L










Jacobson


(1980)


notes


that


although


there


less


information,


important


to the


vote


decision.


"Although


congressional


voters


have


relatively


little


information about


candidates,


both


the extent


content


of information


they


do have has a


decisive effect


how they vote"


This


research is


an attempt


to establish


link between


information


flow theories


the cross-pressure


theories


using


Jacobson's


work


to support


the assumption


that


congressional


elections


are at


a lower


level


information.


Summary


summary,


strong partisans


have


tended


to decide


early to


vote


for their party's


nominee.


Increasingly,


however,


candidate evaluations


and other


campaign


factors


are becoming more


important


as party affiliations


have


weakened.


Cross-pressures


are


linked to


the undecided


voter


(Campbell,


al.,


1960;


Gant


Sigelman,


1985;


Kelley


Mirer,


1974;


Pool


al.,


1965)


Education


level,


information


level,


media


use,


and,


1980,


debates


were


significant


in determining whether


or not


voters


had made


their


decision


for whom to vote


for president


(Ladd,


1981;


Light


Lake,


1985) .


Media


use and


interest


in the


election


t.Yf r 4- r, rvw -: n rJ A


1 -


31)


-I n in -


4- ^ ^


T-T-^ V /










Cross-pressures


can be


from religion


and social


conflict,


family


and friends,


family


coworkers


(Berelson


al.,


1954


attitude


conflict


based


upon


issues


(Campbell


1960


or conflicting


candidate


and party


evaluations


(Flanigan


Zingale,


1987;


Kelley


Mirer,


1974) .


In order to experience


cross-pressure,


however,


voter must


receive


information


(Converse,


1962)


level


of involvement


and party


identification affect


level


exposure and


acceptance of


the messages


(Zaller,


1987).


et al.,
















CHAPTER 3
METHODOLOGY


Context


of the


1982 Election


Voters


the Fourth


Congressional District


of Virginia


in April


1982


were equally divided about


the direction


United States,


with


indicating that


country was


moving in the

were off track


right


direction and


and moving


45%

wrong


saying that

direction.


things

They were


concerned about


economic problems,


specifically unemployment


inflation.


Dissatisfaction


with


Reagan


economic


program was


evident,


particularly among


traditional


Democratic


supporters--blue


collar


workers,


blacks,


women,


and retired persons--were evident.


These


same


voters,


however,

Gove rnmen


reported a sat

t and Governor


isfaction

Robb. R


with


obert


Virginia State


Daniel


received


positive performance and personal


ratings


from


voters.


Yet,


this


very popular


Republican


incumbent


was


defeated by the


relatively unknown


Democratic challenger.


In April


only


challenger,


one-fourth


Norman Sisisky.


voters


In his


area


could rate


of Petersburg,


about half


voters


were able


to rate Sisisky's











By the August


30th through September


2nd poll,


favorable


ratings


of President


Reagan had declined


percentage points


on performance and


percentage points


personal


remained


evaluation


same.


The direction


Unemployment


country


and social


responses


security were


issues


most


often


selected from a


list


the country' s


most


important


problem.


About


half


of the


voters


felt


that


Reagan's


economic programs


would eventually work,


plurality


upon


indicated


that


them personally.


would have a

incumbent Con


negative affect


gressman's


performance evaluation


remained high,


about


half


of the


voters


in this


predominantly


Democratic district


did not


know that


was


a Republican.


Awareness


of Norman


Sisisky


increased


11 percentage points


between


the April


August/September polls.


By October


Daniel's


favorable


rating


dropped another


percentage


points.


The Sisisky


campaign


concentrated on


identification Daniel


as an incumbent


Republican


tied to


a Republican


President's economic program,


rising


unemployment


, and


intended Social


Security


cut s.


message


was


targeted


to blacks,


retired persons,


women,


blue


collar workers,


as well


as the geographic Democratic


strongholds


of Portsmouth,


Chesapeake,


and Norfolk.


mL.- AO--CO -- -----


/. A F' n


n"


m


m A ---


^ ^ ^t


m i


_


C


n n











Context


of the


1992


Election


Never


before had


undecided


voter


been


so much


news.


Voters


were


unhappy with


their


choices;


"undecided"


won a


few Democratic primaries,


and an


independent,


personally financed campaign


for president


was


placed on


the ballot


through a


petition


drive


in all


states.


The

research


two congressional


were


districts analyzed in


both affected by reapportionment.


this

In Florida,


the Fifth


Congressional District


was


one


of the


four new


congressional


seats given


Florida because


of population


gains.


In addition


the majority


of Alachua


County,


district


includes


part


of Pasco and Marion


Counties;


most


Levy County;


and all


of Hernando,


Sumter,


Citrus,


Dixie,


Gilchrist


Counties.


University


of Florida


this


district.


Sixth


Cliff


District,


Stearns


was


which no


incumbent


longer


Congressman


includes Alachua


County.


was


successful


securing a


third term


in his new


district.


Democratic primary


on September


Karen


Thurman


of Dunnellon beat


challenger


Mario


Rivera


in a


landslide.


Tom Hogan


won


of the


vote


the Republican


primary


against Andy Martin and Gene


Keith.


In November,


Cindy


NA..n1 4, C 4nl.


- ~ ~ ~ ~~,,, n L- ~---4


^ -3 -^ rt, ^ 3 T ^ T


T~


1











campaigned


with


"Cindy Means


Business.


Noah


ran a


low-key,


low-budget


campaign.


Thurman


outspent


Hogan


to 1.


this


four-person


race


in the


general


election,


Thurman


won


with


vote districtwide and


vote


Alachua


County.


Thurman


stressed her


years


of experience


as a State


Senator


and her work


in education,


veterans


affairs,


and agriculture.


In Ohio


Fourteenth


Congressional


district


lost


northernmost


tip of


Summit


County


in the


redistricting.


district


also added


western


one-third


of Portage


County


the northern


tip of


Stark County.


University


Akron


included in


the district.


Democrat


Tom Sawyer,


seeking his


fourth term,


won


the June


primary with


vote against


two opponents.


In November


faced


Republican Robert


Morgan.


Sawyer


ran


on his


record.


Morgan


ran


against


Sawyer's


record but


did not


offer


a plan


issues of his


own.


Sawyer outspent


Morgan by


about


to 1


margin.


Democrat


incumbent


Tom Sawyer


defeated Republican


challenger


Robert Morgan.


Methodoloarv


an advocate of methodological


pluralism,


researcher utilized both


qualitative and


quantitative


techniques.


There are


benefits


using


two approaches


- -r j-. t -


L


L





J


L


J-- .. L


mt i











population.


This


research describes


differences


between


populations


of decided and


undecided voters.


strength


of focus


groups


insight


into


process


voting


decision.


individuals


analyzing


comparing


specific comments of


comments


within and between


groups,


this


technique


provides a means


of understanding and


explaining why


some


voters


are


undecided.


It also goes


behind the


numbers


explains why those


voters


are


undecided.


Survey


research


contextual


data,


and focus


groups


were


utilized


to explore


the nature of


undecided


voters


theoretically


link their decisions


or the


lack


one


to the


broader political


theory


vote


choice.


The quantitative


phase


tests


hypotheses


about


undecided voter


Congressional


elections,


using the


1982


Fourth


Congressional


District


District


of Virginia,


1992


of Ohio,


1992


Fourteenth


Fifth


Congressional


Congressional District


of Florida


as case


studies.


The qualitative phase explores


research questions


related


to the process


of the


vote


decision


rather than


vot e


choice


itself.


focus


groups


understanding


"numbers"


reported in


first


phase


project.


The data


sets


of the


1982


Virginia


election and


1992


Ohio and Florida


elections


-~1 -- -


-- --


. ... 1_ -- t ...- .1


____


1ir \r











districts


were


selected to


represent


an incumbent


and an


open


seat.


Ouantitative Research De


sicrn


Data


for the


Fourth


Congressional


District


of Virginia


from the Florida


Institute


Research


on Elections


(made


available by William R.


Hamilton and Staff,


Inc.)


were


analyzed


for the


race between


Republican incumbent


Robert


Daniel


Democrat


challenger


Norman


Sisisky.


Three


sets


of data


were analyzed:


a cross


section


telephone


poll


400 probable


voters


conducted April


16th


through


23rd,


1982;


a cross


section


telephone poll


349 probable


voters


conducted during August


and September of


1982;


and a


telephone


tracking poll


1,183


probable


voters


conducted


in eight


respondent


waves during


screen


September and


a probable


October


voter


1982.


the April


August/September polls


was


a self-report


that


s/he


was


registered


vote and had voted


in either


or both


1980


presidential


election


or the


1981


gubernatorial


election.


September/October


respondent


screen


included the


above,


and the


respondent must


have


indicated that


was


almost


certain,


or had a


50-50


likelihood,


that


would


vote


the November


election


for Congress.


Data


for the


1992


Fourteenth


District


of Ohio


was


-- A- J-- m- m


--.~-I


m-. n- '< ^ _


T.


L










utilized in


September/October


1982


survey.


Only


gender


vote


choice


were obtained


the Ohio poll.


Therefore,


the primary use of this data


was


to identify


respondents


for the


focus


group.


Data

obtained


telephone

September


for the

from the


poll


1992.


1992

Karen


400

The


Fifth District


Thurman


of Florida


campaign.


registered voters


respondents


was


A cross-section


was


were not


conducted


screened for their


likelihood of


voting


the November


election.


Models


The model


undecided voter proposed in


this


research describes


differences


among the


undecided


voters


classifies


undecided


voter


levels of


information and


involvement


Figure


represents


the undecided voter who


involved in

information


political


regarding


process and has


the election.


little or no


Previous


research


attribute

election


this


to not


(Jacobson,


paying


1987b;


attention this


Converse,


1962) .


early

Figure


"ideal


citizen"--involved,


informed,


decided upon


her/his


vote


choice.


Figure


proposes an


undecided voter


informed and


involved but


also


cross-pressured


and,


therefore,


unable


to decide


upon a


candidate.


_ I





















Party

\


Newspapers


Friends








No Link


Groups




Television


Family








- Political information cannot get to the voter because there is no


involvement.


Vote choice is not possible unless driven by party.


Vote
Choice

















Political involvement
goes out from the voter


Political information
comes back to the voter


Party


Newspapers


Friends


Groups





Television

Candidate


Evaluations


Family


If the information received is consistent, then vote choice is possible.





















Television Favors Candidate A


Group Favors Candidate A--

Group Favors Candidate B-

Family Favors Candidate A -

Family Favors Candidate B->

Friends Favor Candidate A -

Friends Favor Candidate B

Party Favors A


Newspaper Favors Candidate B


--Group Unfavorable to A

4-Group Unfavorable to B

<--Family Unfavorable to A

<--Family Unfavorable to B

<-Friends Unfavorable to A


Friends Unfavorable to B

Party Favors B


Vote Choice
Not Possible


Voter










groups, demographics,

political information


interest


the election,


newspapers,


reading


watching political


programs


television,


involvement


political


process.


Therefore,


Timing


of the


Vote


Decision


= a


+ xl
+ x2
+ x3
+ x4


PID
candidate evaluations
opinions


group


influences


+ x5 demographics


+ x6 PID


* candidate


evaluations


political


discussions


reading political


information


+ y3


in newspapers


watching political


programs


interest


on television

in the election


candidate contact
participation in
campaigns.


reduced model


of Vote Decision


= xl


was


tested


qualitative


the quantitative phase of


phase


this


explored the effects


research.


variables


full


model


upon


the process of


vote


decision.


Hypotheses


Models


predict

decision


developed


vote choice,

or a profile


in prior


research


rather than


of the


were developed


timing


undecided


voter.


This


vote

research


builds


upon


that


work


extends


research


question











Based upon


prior


researcher tested,


research


using the


undecided voters,


Fourth


Congressional


this


District


of Virginia and the Fifth


Congressional


District


of Florida


survey


data,


hypothese s


that


have been


validated


presidential


voters.


first hypothesis


that


a high


evaluation


incumbent's performance and party


identification


that


same


as the


incumbent


will


result


in an


early


decision


regarding


for whom to


vote


(Campbell


al.,


1960


Kelly


Mirer,


1974)


Therefore,


Timing of


a + bl


+ b2


+ b3


the Congressional


evaluation


party


Vote Decision


of candidates)


identification)


evaluation


Republican


identifiers who


have


a favorable


opinion


Robert

As a c


Daniel


orollary,


would have


decided


the Democrats


for whom they will


with a


favorable


vote.


opinion


Republican


incumbent


are


likely to be


cross-pressured


undecided.


interaction


term of


evaluation


times


party


identification


was


added


to test


the effect


cross-


pressures.

measured by


dependent


responses


variable

"And in


of vote


choice


the election


for U.S.


Congress

candidate

n1CL 1 f~ j'i


from the


are Robert


4- rh a flrv n, r' +-


Fourth District


Daniel,


% rt


I


of Virginia,


the Republican


k- n -C fl


I-. *4n*- r% A r C-.


and Norman

.- t1P, -: rn Ar


* party)


11 I











Sisisky,


or undecided.


independent


variables


were


operationalized by,


"Now


I'd like


to ask


your


impression


some


people


in public


life.


read


each


one,


just


tell me


person,


whether yo

a somewhat


u have a v

favorable


ery


favorable opinion


opinion,


a somewhat


that


unfavorable


opinion,


or a very unfavorable opinion.


don't


recognize


them,


just


so.


Robert


Daniel


" and


"Generally


speaking,


consider yourself


a Democrat,


a Republican,


or an


Independent?


INDEPENDENT)


Well,


which


party


normally


support


in elections--the Democrats or the


Republicans?"


the Fifth House


District


of Florida,


the dependent


variable


is measured by


responses


the election


U.S.


Congress


were


held


today,


would


vote


Democrat


Karen


Thurman,


Republican


Tom Hogan,


or Independent


Cindy Munkittrick?


independent


variables were


operationalized by,


"First,


like


to know your


impressions of


some public


figures.


read


each name,


please


tell


you have


very


favorable,


somewhat


favorable,


somewhat


unfavorable,


or very unfavorable opinion


of each


one.


State Senator


Karen


Thurman.


" and


"Are


registered


vote as


a Democrat,


Republican,


Independent?"


.Innr, rt iNn i' ^ A~ r' V ni 1-. n^ cn 4 A


cE,-,, Ar r%


c1- -L +


1


r


'I'V i


/^l k-* n lf ^ ^ ^











voters


(Converse,


1962;


Zaller,


1987


This


information


interest


in the


election.


Therefore,


Timing


+ bl


the Congressional


(number


Vote Decision


of opinions)


In all


five data


sets


opinion


is operationalized by the


number


issue


the answer,


and opinion


unless


was


questions


"don't


answered regardless


know.


In addition,


there


are hypotheses


developed


from the


research on


congressional


voters,


in general,


that


are


tested in


this


research


to determine


their


applicability to


undecided voters


incumbency


specifically.


prime determinant


finding


of whether


that


a candidate


will


elected


is well


documented and


will not be measured


this


research.


Past


studies


conclude


that


candidate


evaluations are


significant


in explaining the congressional


vote


choice


(Flanigan


Zingale,


1987;


Hinckley


1980


This


research also addresses


question


of whether the


evaluations differ


is hypothesized that


by the


timing


late deciders


vote decision.


have a


low evaluation


candidates


or no evaluation.


Thus,


Timing


Congressional


Vote Decision


- a


(candidate evaluations)


This


is operationalized by


opinion


Sisisky and opinion


- -'


II. -I


-t i


**9 r~~~ ~~ 1 -I ~~ ri Il Sr 4- .c ,a nnam aS 4-n I- b% a 4-I- .a--I- .


T\~n ^ ^ /


rn


I










impressions


of Robert


Daniel--personally--and/or things


done on his


done


job?


very much


What


else?";


to help the


"Congressman


farmers


this


Daniel


district";


"Robert


Daniel


really


cares


about


people


like me.


April


survey did not ask


evaluations about


Sisisky because


was an


unknown and


variation


in opinions


was


not


anticipated.


the August


October tracking polls


survey

variable


and the

s are,


September

"How would


rate


Robert


Daniel


handling


job as


Congressman--excellent,


good,


not


so good,


or poor?"


August


Survey,


"What


are


some of


things


like


about


Norman Sisisky--personally--and/or things


he has done


in his


job?"


"What


are


some


of your unfavorable


impressions of


Norman


Sisisky--personally--and/or things


he has


done


in his


job?"


are


also analyzed.


Fifth District


of Florida


is operationalized


"Overall,


how would


rate


job performance


of State


Senator


Karen


excellent,


Thurman?


good,


only


Would


fair,


say that


or poor


is doing an


job?"


Other


research


concludes


that


issues


are


significant


explaining the congressional


vote


choice


(Hinckley,


1980)


This


study


expands


research


to address


the question


whether there


are


issue differences by the


timing


- I- i _











hypothesis


that


there are


differences


in issue


orientations between


the decided and


undecided


voter.


In Virginia


this


hypothesis


operationalized by,


"Now


like


and have


to read


tell


several


different


me whether you


issues


favor or


or proposals


oppose each


one


delaying the cuts


scheduled


July


in individual


corporate


federal


income


taxes


to help


balance


federal


budget


S .do


favor


or oppose


this


proposal?


strongly


favor/oppose


or only moderately


favor/oppose


that?"


"A constitutional amendment


banning


abortions.


"The


Equal


Rights


Amendment--the


ERA.


"Federal


impact


aid--the


program that


provides


federal


money


local


school


districts


containing many


federal


employees


"Making Martin Luther


King's


birthday a


national


holiday.


"Having the


federal


government


provide more


funds


public


job training.


"Now


I'd like


to read


a few


statements


with


which


some


people agree


opinion.


Here


others


first


disagree.


one.


like


President


your


Reagan's


economic pl

expense of


favors big


the average


business


and wealthy people


working person.


strongly


- I.


1 J


1 I


1 I


______


I











should


elect


congressmen


who can be


counted


on to


support


President


Reagan.


"Democrats


are needed


in Congress


serve


a balance


to the Reagan Administration and to prevent


Reagan's


policies


from going too


"Democrats mean


far.


jobs;


when unemployment


Democrats


know how to get


people back


to work.


In Florida


the hypothesis


is operationalized by,


"Generally,


would


things


this


country


are


headed


in the


right


direction


or are off


on the


wrong track.


"Overall,


how would


rate


job performance of the


U.S.


Congress?


Would


they


are


doing


an excellent,


good,


only


fair,


or poor


job?"


"And how would you


rate


job performance of


Florida


State


Legislature?


Would


they


are doing an


excellent,


good,


only


fair,


or poor


job?"


"When


considering whether or not


for public office,


or her position


to support


a candidate


on abortion


one of


the most


important


issues


consider,


somewhat


important


important


as other


issues,


or not


really


important


concern


to you?"


"I am going to


read


a few


statements.


each


one,


please


tell me


strongly


agree,


somewhat


agree,


-- .12


-,,.,


.I


_ J


.. |


i


- .-. I











"Tom Hogan


spent


years


as a criminal


prosecutor.


This makes


me think


he would be


tough


on crime


if he


was


elected to Congress.


"The


state


legislature has


done


such a


poor


job;


should not


reward members


like


Karen


Thurman by


electing


them to higher


office.


"State Senator


Karen


Thurman


state


funding


home


care


for the elderly,


to help


senior


citizens


prescription


drugs


and meals


on wheels.


This


shows


really


cares


about


elderly


in Florida.


bothers me that


years


Tom Hogan


ran


negative


campaign


against


Karen


Thurman and said some


things


newspapers


said were untrue.


"State Senator


Karen


Thurman


worked


to pass


SWIM


bill


to clean


Florida' s


rivers and


lakes and


sponsored a


bill


to clean


leaking


underground storage


This makes me


think


tanks


really


that


cares


were


about


protecting the


environment.


"State Senator


Karen


Thurman


took


trips


lobbyists


expense


and should not


be elected


to office because


shows


cares


more


about


special


interest


groups


than


average


person.


a state


senator,


Karen


Thurman has


a strong


record


S1 | 1 1 -


_- -11~ 1 1 -


* t_


r










"Since he


never


held political


office,


Tom Hogan


would be


kind


of Congressman


that


could help make


changes


in Congress


that


we need.


"State Senator


funding


Karen


Thurman


for public schools


rural


passed a bill


counties.


to get more


That makes me


think


really


cares


about


children


getting a


good


education.


"State Senator


Karen


Thurman


is a professional


politician


that


the kind


person


we need


Congress.


"State


Senator


Karen


Thurman


passed a bill


prohibiting


other


states


from taxing the


pensions of people


retire


in Florida.


That makes me


think


cares


about


senior


citizens.


"Please


tell me


strongly


favor,


somewhat


favor,


somewhat


oppose,


strongly


oppose


each


one.


Terms


Congressmen


should be


limited


years


less


Statistical Analysis


frequency


were


distribution


the number of


undecided for whom they would vote,


by month


voters


of the


survey--April,


August,


September,


and October--was


calculated.


Cross


tabulations


were


run by the


independent


t?~~~, V zV %C .- V1r a .r A n c .


I"


r *^ lS1 n


Fh^ r] h^/ ~ 4- l* ^ ^^^/^*^/3^ ^











significance.


chi-square


test


is based on


a comparison


between


frequencies


that


are observed


the cells


classifications


tables and those


that


would be expected


if the null hypothesis


independence


were


true


(Agresti


Finlay,


1986).


An analysis


variance


was


calculated for the


independent


variables


the dependent


variable.


The F


test


was


used


to determine


their


significance


independent


variables


to select


independent


variables


to be


used


discriminant


analysis.


analysis


of variance


is an exploratory test


to detect


differences


between


group means


categorical


and interval


variables


was


an F test


null hypothesis that


means


are


equal


against


the alternative hypothesis


that


means


are


unequal.


The P-value


the probability


obtaining


an F-value at


least


large


observed one


if the null


hypothesis


were


true


(Agresti


Finlay,


1986)


Cross-tabulations


were


run


for the evaluation


candidate's performance


opinion


of the


candidate by


party


identification


to measure


effects


cross-


pressure.


square


was


used to


test


significance.


Interaction


terms


opinion


incumbent


times


party


identification,


rating


incumbent


times


party


- 21 L. J r 2 2


- 21 2 2 _- r


I-'--










A discriminant


analysis


is performed


using the


variables


that


were


significant


step


to determine


their


importance


understanding


undecided


voter


(all


variables

analysis


research


are


included


statistical


question because


he Ohio an

tool best


alysis)


suited to


identifies


Discriminant


this


variables


that


best


describe


the differences


between


categories--undecided


decided voters


(Norusis,


1990).


Oualitative


Research Desirn


participants


in the


focus


groups


were


contacted


five


times


from July through November


1992.


summary of


the qualitative


research


design


ollows:


Technique


Number of


Participants


Telephone


Survey


Total


Florida


Ohio


Likely
7/92 a


Telephone


Voters


8/92


Invitation


Classification


Undecided
10/6-7 an


Self
Unde


10/14


Focus


Voters
d 10/11


Administered Questionnaire
cided Voters


10/21


Groups


Undecided Voters


10/14


10/21


Post-Election


Interview


Focus


Group


Participants


m 1 I... -.


J 1 _- 1










first


contact


was made by the


candidates'


polling


firm.


The names


were


randomly


drawn


from


registered voters


lists


of the Congressional District.


researcher then


obtained


the names


and phone numbers


of the


voters


who were


undecided


in the


House of Representative


election.


Three


attempts


in Summit


were made


(Ohio)


to reach


or Alachua


each


these


(Florida)


people


county to


lived


invite


them


to participate


focus


groups.


Immediately prior to


focus


group,


the participants


completed a


self


administered


demographic and


candidate evaluation


questionnaire.


Three


focus


groups


were


primary


research


tool,


although


telephone


interviews and self-administered


questionnaires


were also utilized.


This


research


examined


relationship


between


levels


information


and levels


involvement


and the vote


choice of


uncommitted voter by


classifying the


voters


prior to


focus


group.


analyzed


candidate evaluations and


demographic variables


modified by


It begins


level


information and


to understand


process


level


that


involvement.


undecided voter


uses


to make


his/her


decision.


"Focus


groups


are useful


getting participants'


interpretation


results


from earlier


studies"


(Morgan,


1988,


-4 -


- A J- 4 A A^ -I A^ IJ -l 1 n .


,L1~! '_ _A


..t,..











Focus


groups also enable


researcher to observe


interaction process


as the


voters


discuss


their


decisions.


strength of


focus


groups


"is the ability to observe


interaction on a


topic"


(Lazarsfeld,


1972,


In this


case


research topic


to understand


vote


process,


not


the product--the


vote


choice.


"use


of focus


groups


as a means


to observe


this


process"


is one of


strengths


technique


(Morgan,


1988,


11) .


can


be used to


explore


focus


issues


groups


after the


interviews.


is to understand


primary purpose of


the numbers


obtained


quantitative


phase.


ideal


size


a focus


group


six to eight


people


(Morgan,


1988)


Specifically,


this


research


a qualitative


data


analysis of


Alachua


County,


Florida,


registered,


likely


voters


were


undecided regarding their vote decision


U.S.


House


of Representatives


three


weeks


prior to


election and


Summit


County,


Ohio,


registered,


likely


voters


were


undecided


weeks


prior to


the election.


screens to


Florida


select


registered to
(Ohio) at you


sample are as


vote


r present


follows:


state


address?


know there


November


3rd.


will be an


things


election
stand now,


for Congr


ess


likely


are


to vote


this


election--almost


certain


about
vote?


50-50,


or might


not


a chance


...1 --- i -- -


II_ _


I -


L.. LL


mt, .. -- .. ->










Democrat,


and XX,


Republican,


as things


stand now,


would


vote


X or


(IF UNDECIDED)


Well,


which one


lean


toward at


this


t ime ?"


For the


purposes


of this


research


leaners


were


classified as


undecided.


classification


variables were operationalized by


National Election


Studies


level


information


and


involvement


questions


listed


in Appendix


Information


levels were


very much

campaigns


considered high


interested


newspapers,


respondent s


campaign;

watched


said


read about

television


they were


the

programs


about


campaigns,


and discussed politics


with family


friends;


they


used any


one


source


frequently.


Involvement


was


considered high


if respondents


or more


contacts


with


incumbent


Congressman


they


performed two of


following


campaign


activities:


persuade


someone


how to


vote,


attend political


rallies or


meetings,


wear


a campaign button,


display


a bumper


sticker


or yard


sign,


or work


a political


party


or candidate.


Cross-pressures


were defined after the


focus


groups and


analyzed at


individual


level.


focus


group


discussions


concerning


respondents'


feelings


about


each


candidate,


their


family


friends


' feelings


about


each


candidate,


and their


organizations'


feelings


about


each


nn^ A"V A-I A^\ 4- +- r' n^Tr 1 *, r -1 fl- f-


n"l"In ^ / -1 "


/n T^> nn *


T-T/~ r 1n ^


**^inn /^ ^ /p- -Tn -^/^< ^^ -\r










There


seat


were


in the Fifth


voters undecided


Congressional


(23%)


District


race


of Florida


the

were


poll


conducted on


located


in Alachu


September

a County.


Thirty-four of these


Twenty-nine


invitation


telephone


interviews


were


conducted


October


6 and


Thirteen


voters


agreed


to participate


focus


group,


refused,


could not


reached


after


three


call-backs.


voters


stated that


they were


very much


interested


campaigns


this


year,


they


read about


them in


newspaper,


watched television


discussed politics


with


programs about


family


the elections,


or friends once or twice


a week


or more.


Therefore,


participants


were


classified as


focus groups


in the


having


were


5:30 pm group,


high


conducted


however,


interest


in Alachua


did not


campaign.


County.


part


Two


Respondents


cipate


campaigning process.


They answered no


to at


least


three out


of four questions


concerning political


activity


such


wearing


a button


or attending


a rally.


Participants


7:30


pm discussion


were


politically


active.


They answered


to at


least


two of


four questions


concerning


political


activity.


During the Alachua


County


invitation


interviews,


was


apparent


that


a combination


the Ross Perot


entry


into the


A-- r' ^ ,-.. 4 1 1 ^


4-


-J I


r^~ ^A / ^/ -


Lk











claimed to


very


interested


the political


campaigns


this


year.


As a result,


during the October


invitation


telephone


interviews


to Summit


County,


Ohio,


no attempt


was


made


to separate


participants


into two


sessions.


There


were


(22%)


undecided voters


in the Ohio


Fourteenth


Congressional


District


according to a


poll


taken July


these


voters


lived in Summit


County.


Fifty-seven


telephone


interviews


were


conducted


with Summit


County


residents.


Thirty-two


participate


focus


were


no longer undecided,


group,


refused,


agreed to


18 could not


reached after three


call-backs.


Three


focus


groups


were


conducted with a


total


of 18


participants.


There was


one


no-show


in each


group.


Informed


consent


forms


were


signed by each participant.


Participants


were


paid an


honorarium of


$25.


sessions


were


audio


recorded and ran


approximately


90 minutes


each.


discussions


began


with


the broad


questions


direction


in which


country was


headed and


was


blame.


The discussions


then moved


to the


U.S.


House


Representative election


their district.


voters


talked about


their


feelings


concerning the


candidates,


they were


still


undecided,


how their


family and


friends


felt


about


each


candidate,


and how groups


to which


they


belonged


r~~ a -


r


1


1


^


I *











additional


information


they needed


to make


a decision,


whether


or not


they expected


vote


this


race.


three


focus


groups


were moderated by Deborah Rossi,


researcher.


On Wednesday,


October


five


people


attended a


5:30


pm high


interest/low


involvement


focus group


held in


conference


room


in Weimer


Hall,


University


of Florida.


people attended


7:30 pm focus


group classified as


high


interest/high


involvement


held at


same


location.


participants


were


unaware of


classification


groups.


They were only asked to attend at


a specific time.


On Wednesday,


October


seven


people attended a


5:30


pm focus


group


held in


a conference


room in


Gardener


Student


Center,


University


Akron.


As did the Alachua


County participants,


these voters


expressed a high


interest


campaigns


political


the newspaper,


campaigns.


watched


They


read about


television


programs


about


elections,


and discussed


politics


with


family


friends


least


once


or twice a


week.


Summit


County


focus group

involvement


there

and f


were


ive


two participants


were classified as


classified as


high


involvement.


involvement


voters participated in


discussion


same degree


as the high


involvement


people.


only


1 I


I


I .I


rl













After the election,


telephone


interviews


(Appendix K)


were


conducted with


focus


group participants


determine


whether they voted


U.S.


House


Representative


election,


for the decision.


their vote


of the


18 part


choice,


cipants


reasons


were


interviewed by telephone.


After three attempts


to reach


other eight


partic


ipants,


questionnaire


was


mailed to


them enclosed with an


addressed,


stamped


envelope.


Responses


from


15 of


18 participants


have been


obtained.
















CHAPTER
RESULTS


Prior


research has determined


that


in presidential


elections

their vote


however,


cross-pressures


choice.


cause


In order


voter must


receive


voters


to experie


to be undecided in

nce cross-pressures,


information.


Using the


methodology


outlined in


Chapter


this


study


concludes


that


cross-pressures


also


cause


indecision


U.S.


House of


Representative elections.


We would also expect


to find


fewer


cross-pressures


among the


undecided


voters


these


elections


than


was


found


in presidential


elections


fact


that


congressional


elections


are


lower


information


level


elections


results


presented in


this


chapter


confirm both


these


hypotheses.


There

from party


evidence


that


identification


cross-pressures


candidate


are experienced


evaluations


congressional


elections.


Unlike


previous


research,


however,


this


study


does


present


evidence


that


voters


experience


cross-pressures


from family,


friends,


religion,


groups


issues.


Another


significant


finding of


this


study


that











to have


groups


a low or no opinion


respondents


candidates.


attributed their


focus


"no opinion"


responses


to the


fact


that


they were


unable


to obtain


information regarding the congressional


election.


analyzing the


results


from four


of the


five


surveys,


we are


able


describe


demographic


differences


between


decided and


undecided voter


Representative elections.


House


Demographic data


were not


available


the Ohio


survey.


There


were


significant


differences


between


two groups


in particular


surveys;


none of


results,


however,


were


significant


over time


all


four


surveys.


4/82


Virginia


poll and the


9/92


Ohio


poll more undecided voters


were


female


older than


were decided voters.


9/92


Florida


poll


undecided


voters


reported a


lower


income


than


did decided voters.


Income


was


significant


Virginia


polls.


Race


was


significant


Virginia


results;


was


tested


Florida.


Religion


was


significant


4/82


Virginia


poll


not


8/82


one.


Party


affiliation


was


significant


three


Virginia


polls and


the one


in Ohio.


was


significant


in Florida.


Detailed results


follow.


Ouantitative


Results


The percentages


are computed on


the basis


1. 1 1 I


I I 1 ^I


* _


_


I











Hypothesis:


Virginia Decideds


are More


Likely to be


Republican and have a


Favorable Opinion


of Daniel.


Florida Decideds


are More Likely to


be Democrat


and have a


Favorable Opinion of


Thurman.


This


hypothesis


is confirmed by Tables


shows


that


more Republicans


with a


favorable


opinion


of the


Republican

they will


incumbent


vote


Daniel


than have not


have made


a decision


decided.


for whom


Feelings about


incumbent


, however,


do appear to cause


cross-pressure


among


some


undecided


voters.


More Democrats


in all


three polls


have


a favorable opinion


or rating


Republican


incumbent


are


undecided.


This


potential


cross-pressure,


and this


cross-pressure may


the cause


indecision.


Table


indicates


party


affiliation


voters


their


rating


incumbent's


performance.


This


table


shows


that


there


are more Democrats


than Republicans


who are


undecided,


although


the gap decreases


election


gets


closer.


Although


Florida


seat


was


open,


the Democrat


Karen


Thurman


had been


the State Senator


for most


the newly


drawn


Congressional District


and more


closely


fits


role


an incumbent


than


does


Hogan.


In Florida


was


- S


1 I


1 1


_

































a oo


rl'


c-rn


C' r rl


r- -


(C (N]


(0 0


i-l C


I- r-


A'


tll



































































r- --


C 4r4


1-4 -
Cri 4-1

Sc 4J
0 C1


o Qr
O o



O Oi r-
0) (




-p (0 -
0) aC 0
o


to a 0
m m -p
0)-d g







































(N r-


mo ()


CN~



































































rI- r-I











Hypothesis:


Undecideds


have


fewer opinions on


candidates


Table


issues


compares


than


the decided


the number of


voters.


issue


and opinion


questions


that


were answered by undecided and decided


voters.


The data


this


table are


significant


that


there


were more


probable voters


who answered the


questions


in the decided


group than


undecided group.


Virginia


there


were


second,


such

the


question

third.


In Ohio


first

there


poll,

were 8


questions


and in Florida,


Respondent s


in the decided group offered more opinions


than


undecided respondents,


giving evidence


support


the hypothesis


that


undecided


voters


congressional


election


do not


have


enough


knowledge on


issues


campaign


to make an


informed decision


regarding


vote


choice.


This


is because


they are not


paying


attention


interest


to the


lack


campaign


either because of


of participation


lack


electoral


process.


Since


respondents


are


registered voters


who do


vote


intend to


vote


November


election,


lack


of interest


the electoral


process


in general


likely to be a


causal


factor.


Table


compares


undecided and


decided


voters


regarding


I-- I--1


-1 1 J _


A


1 -1


rr 11 1 1


I


r^











Table


Comoari


son


Unde


cided


Decided


Voters


Distribution


Freauencyv


of Havinar


Opinions


Fourth


Congressional


District


of Virginia


1982


Aoril


Auaust/Se


otember


Sect/Oct


Answers


Undecided Decided


Undec


ided Decided


Undecided Decided


11.9


10.0
18.9
18.9
14.9
14.7


13.2
17.9
25.0
28.0


11.9


11.9
16.9


14.5
12.9
12.9
17.7
11.3


10.7
14.5
20.3
16.9
15.5


17.5
23.4
25.6


Total


100%


100%


100%


100%


100%


Sample


Size


P <.0001


P <.0001


P <.0001


Mean


20.4


21.3


11.3


13.9


100%











Table


3--continued.


14th


Ohio


5th CD of Florida


. Of


July


1992


SePtember


1992


Answers


Undecided


Decided


Undec


ided


Decided


19.1%


15.7
18.0
21.3


16.9
16.5
19.3
17.3
20.9


Total


100%


NA
NA
NA
NA
NA
NA

100%


11.7
20.2


12.7
14.7
16.1
20.1
13.0


100%


100%


Sample


Size


P <.0001


P <.0001


Mean


15.6


18.3


Source:


William


Hamilton


Staff,


. 1982


Thurman
Sawyer


Campaign
Campaign


1992
1992


Computed by


author






















0o\
O 1>

M1 Z


*0

ro rcsrH


* Z
rH 0


r-NO
CO r H
0
<-1 C\] 0
Cr) in r-


- .o r->O c- -o\o


rHN H N-rl-i


.0


(c r O Cr M O O o\o

.....


* cI
rH


r"- rl
..
* S
r rHl


toO oW'
SO
.0
C O10
Cf lD rl


rN C o flC NLrHO
rHl H H


rtN CD r--IO r-- \ DO


























r-r-l r-
itr-rH


..ic4
JCNZ


r-t O fO
* *

C. .J
Nam)


SoHr H>


N r


o r-M


W r-4


M S

%. Or


-I r-


CN


W Hl O)


co O o XCM


(C rHl OUCN
r-Nl C


rH L 4O .0O0
* S S S S S S
61 C'. (f rra0O~
c co


o\

* S S S S S S

C y T- COCN
rl m I 9 A


-. .-


*u *


rH oo


















rHl
C S S


Hl U H r ( CN
*r N


) C


- 1 0 0
* *


CM t-orA t"
* *
ID lrA
TCL rt H
to a


H 0 rdC',
H O d 0*
0 C 0


OOJ
-0 -
Q0 M 0 0

-H aA (0
rl (f Tk


CC>

a a, >i
-HOM
-HGo)



O O
00
0) SO
a (u 0


>


MG)
0 ,L)


a) )
TC >

OO
OH E


-HG)
r-J




- 0)
l CO
H 3
-n


OA O
ro O 0 d

- .
d 0l 5


0-


C o


(0 4^ <
h, 0
MrO
-H 4-)
CO 0) f
-rI-H S -
U)3


0


rH C
-H tn





O
*drd c
E 0 ,C
P^0 ^


-H Ei
HQir
k dk
.-)











data


show that


the majority


respondents


in both


categories


do not


even


recognize


challenger's


name.


campaign


progresses,


however,


there are differences


between


the decided and the


undecided groups


with


regard


their opinion


the challenger.


Virginia


August/September


and September/October polls,


well


as the


Florida


poll,


there are more


voters


who do not


recognize


challenger's


name


undecided


category.


In the


express


case


of the


an opinion


incumbent,


significant


ability to


from the


first


form and


poll


April.


In all


four polls


percentage of


voters


no opinion


could not


rate


incumbent


was


higher


among


undecided


voters


than


was


decided group.


Table


compared


undecided voters and decided voters


regard to


their


ability to


correctly


identify


party


affiliation


incumbent.


lack


of information among


undecideds


further


supported by the


fact


that


significantly


higher proportion


undecided group could


correctly


identify


the political


party


incumbent.


Hypothesis:


A higher proportion


undecideds


voters


have a


low or no opinion


candidates.


Table


supports


this


hypothesis.


In April


undecided


voters


decided


voters


expressed a


Ir I r -. I


-I~ ~ r


l i t


1


1CL










Table


Comparison
Proportion


of Undecided and Decided Voters


of Respondents


Who Do Not


Know


Regarding
Incumbent's


Partv


Correct,


Incorrect,


Don't


Know


Fourth CD of VA,
Aug/Sept 1982

Decided
Undecided


Total
Sample


size


.0201


Fourth C
Sept/Oct


D of VA,
1982


Decided


Undecided


Total
Sample


size


.001


Questions:


do you happen


to know which


political


party


Robert


Daniel belongs


(IF YES)


Which


party


that?


Source:


William R.


Hamilton


and Staff,


Inc.


1982


Computed by author










Florida


89.4%


undecided


voters


compared


to 75


the decided


voters


had a


low or no opinion


of Hogan,


newcomer.


Forty-eight


percent


undecideds


decided


voters


had a


low or no opinion


officeholding


candidate


in April.


August


this


had fallen


to 28%


the decided voters,

In September/October


but

36%


remained at


for the


undecideds


undecideds.


still had a


or no opinion


decided


compared


of the


category.


to 46.6%


incumbent


In Florida


compared


85.1%


the decided voters


to 27%


undecideds


had a


low or


opinion


of Thurman.


rating


of the


performance


of Daniel


was


not


good,


decided


were


poor,


or none


voters


of the


in April.


given


undecided and


In August/September those


undecided and 20%


of the


ratings


decided;


in September/October those


ratings


were


cited by


undecided and 31


decided;


in Florida


performance of Thurman


was


rated as


either


not


so good,


poor,


or don't


know by


88.3%


of the


undecided


voters


compared


decided


voters.


Hypothesis:


There


are differences


in the


issue


orientations of


the decided and


undecided


voter.


Sn^ A 4 r-I4 ". A 4


rr 1 /T


n l th r, >


+~n n7r^ ~ v \+"1 *


F* I: nrnnnnn/^^ ^B/-


/-, +- 4->./











research,


however,


are mixed.


This


hypothesis


was


not


supported o

probability


n every measure.


from


Table


the analysis of


reports the F


variance


undecided/decided with


independent


variables.


the April


poll


there were


significant


differences


between t

abortion,


;he groups


size


on delaying the


the military


proposed


budget,


cut,


security


Israel,


constitutional


on abortions,


ERA and


the Middle East


policy.


Opinions


on direction


country,


nuclear


freeze,


weapons


freeze,


voting


rights


act,


Reagan's


economic


program,


Martin Luther


King


holiday,


impact


aid,


public


jobs


training,


however,


were not


significantly


different between


undecided and


decided


voters.


the August/September


and September/October polls


specific


issue


questions


were not


asked.


Those


relating to


opinions


issues


regarding the


were asked and


incumbent's performance


they were


on specific


significant.


September/October poll


responses


to questions


concerning


challenger


and specific


issues


were


not


significantly


different


between


undecided and decided


voters.


In Ohio


vote


choice


county prosecutor,


county


engineer,


county


clerk


of courts,


state


representative,


U.S.


Senator were


significant


between


two groups.


i*l -1 J -


A- -


- -1 2 -_ j


- p_ 1


t I


1













Table


Analysis


Variance


Decided/Undecided


Independent


with


F Probability
Virginia 1982


Variable


April Aug/Sept Sept/Oct


Inte


action


Term


Opinion


of Incumbent


* PID


.0000


.0000


.0000


Interac


tion


Term:


Rating


of Incumbent


* PID


Opini


Party


Opinion
Opinion
Opinion
Opinion


Issue:


on/Issue


Questions


Answered


Affiliation


Incumbent


Cares


About


the Incumbent/Thurman


Military
Security


Budget


of Israel


Ban Abortions


Race


Issue:
Rating


Opinion:
Opinion:
Opinion:
Issue:


Support


ERA


the Incumbent/Thurman


Congress


Middle


Should


Support


Pres


East


Abortion


Delay


Tax Cut


.0000
.0000
.0000
.0001
.0002
.0003
.0008
.0017
.0029
.0033
.0069
.0185
.0210
.0437
.0350


.0000
.0000
.3690
.0031
.0001


.0118


.0000
.0000
.0064
.0128
.0000


.4123


.0000
.1560


.0000
.5943


Incumbent
Incumbent
Incumbent
Incumbent
Incumbent


Performance


Prov


ides


Fair


Performance


Perf
Perf


Identification


ormance
ormance


on SS


Repre


sentation


on Fed Budget
on Agriculture
on Unemployment


of Incumbent'


Party


.0000
.0000
.0001
.0002
.0003
.0292


.0000












Table


6--continued.


Decided/Undecided


Independent


with


Variable


F Probability
Virginia 1982


April Aug/Sept Sept/Oct


Opinion


Direction
Opinion:
Religion
Age/Gender
Opinion:
Opinion of
Ideology


Issue:


Congress


Should


Balance


Pres


the Country


Nuclear


Weapons


Freeze


Freeze


the Challenger/Hogan


Voting


Rights


.0519
.0663
.0748
.1043
.1047
.1222
.1338
.2330
.2339


.7405
.7063


.9500
.0767


.0603


.4808


.0000


Interaction


Term:


Opinion


of Challenger


* PID


Opinion:
Opinion:


Reagan'
Democrat


s Economi
ts Mean J


Program


obs


Employment


Issue:
Income
Issue:
Issue:


Support


Support
Public


Union Member


Important


Issue:


Opinion:
Opinion:


M L King


Holiday


Impact


Jobs


Training


ship


.2715
.3571
.3744
.3594
.4382
.4651
.7184
.8038
.8943


Issues


Challenger


Challenger


Social


Business


Experience


.1559


.2831
.3022
.7770
.3492


.8430


.8344


.2523
.3391
.5119


too Liberal


Security


.5216
.1708
.4180













Table


6--continued.


Decided/Undecided


with


F Probability


Independent


Variable


Virginia


1982


April Aug/Sept Sept/Oct


Decided/Undecided


with


F Probability


Independent


Variable


Ohio,


July


1992


Ques


County
County
State


tions


Prosecutor
Engineer V


Represent


Senator


County Clerk


Vote


Anwered


Vote


ote


tive


Choice


Choice


Vote


Choice


Choice


of Courts


Vote


Choice


.0000
.0000
.0003
.0024
.0025
.0027


Pres


identical


County
Gender
Party


Exec


Vote


utive


Choice


Vote


Choice


.0685
.0786


Affiliation


Decided/Undecided


.5635


with


F Probability


Independent


Variable


Florida


1992


Opinion/Issue
of the Incumb


Opinion
Rating


Interaction
Interaction


Opinion
Opinion
Opinion
Opinion
Opinion
Opinion
Opinion
Opinion


Opinion:


Questions


Answered


ent/Thurman


Incumbent/Thurman


Term:
Term:


Thurman
Thurman


Need


new


Thurman


Rating
Opinion


cares
cares


face


would


the Chall


Hogan


Thurman
Senator


Hogan


Thurman


of Thurman


about
about


* PID


* PID


elderly


Congress


help


environment


enger


campaign


would


tough


too negative


help


elderly


on crime


.0000
.0000
.0000
.0000
.0000
.0000
.0001
.0010
.0010
.0014
.0010
.0010
.0030
.0130












Table


6-continued.


Decided/Undecided


with


F Probability


Independent


Variable


Florida


1992


Direction


Rating


of the Country


of Congress


is to blame


Important
Opinion:


for the


economy


Issues


Do not need


professional


.0751
.1086
.1279
.1580


politician


Congress


Term Limits


Opinion:
Age/Gender


Opinion
Party A
Opinion
Opinion
Income
Opinion


Thurman


cares


about


special


interest


President


affiliation


Need more
Write-in


women


in Congress


Candidate


Abortion


.2930
.3020
.3512
.5709/
.5842
.6674
.7039
.7186
.7710
.8750


.9938


Source:


William R


Hamilton


Staff,


1982


Thurman
Sawyer


Campaign
Campaign


1992
1992


Computed by


author











Table


compares


answers of the


undecided and decided


voters


regarding the direction


the country


question.


Undecided


voters


Virginia


and Florida


were more


likely to


feel


that


country was headed


wrong


direction


than


were


decided


voters.


As discussed previously,


the most


important


factor


about


issues


was


whether


an opinion


was


expressed,


rather than


nature of


the opinion.


specific opinion held


was


significant.


Whether


an opinion


was


held


was


significant.


The more


"no answers"


reported in


survey,


the greater the


likelihood


that


respondent


was


undecided as


congressional


vote


choice.


Other


Results


percentages


of undecideds


did not


move


throughout


1982


did not


election.


know who


In April


they would


the probable


vote


voters who


election


House of Representatives,


in August/September


voters


were


undecided,


in September/October there


were


16%.


the Ohio


1992


Fourteenth


Congressional


District


election


there


were


undecided voters


in July;


Florida


1992


Fifth


Congressional District


election


there


were


undecided voters


September.


terms


of demographics


there are


several


significant


*1 9 1 1 -


r F -


1 I


1


^ ^


m












Table


Comparison
Recrardinac


of Undecided


Direction


Decided


Voters


Their


ODinion


Country


Fourth


Congressional


District


of Virginia


1982


April


August/September


Sept/Oct


Direction


Undecided


Decided


Undecided Decided


Undecided Decided


Right


17.5%


36.1%


28.8%


34.5%


Not Available


Leaning


Right


Mixed


Leaning


Wrong
Don't


Wrong


16.1
11.3
38.7


13.6


39.6


33.9


34.5


Know


Total


100%


100%


100%


100%


Sample


Size


=.0119


Fifth


Congressional


District


of Florida


Sept


1992


Direction


Undecided


Decided


Right
Wrong
Don't

Sample


20.1%


81.9


71.6


Know

Size


.0418


Question:


think


that


the United


States


moving


in the right


direct


wrong


or do


section?


think
MIXED


that


things


IS VOLUNTEERED)


are off track


Well,


and moving


are more


things


in the
moving


- h-_ 1


/ p I I


1 I


* .


*


* *


1 *











Table 8

Comparison of Undecided and Decided Voters Democaranhics


Fourth Congressional District of Virginia 1982


April


August / September


Sept/Oct


Age/Sex


Undecided Decided


Undecided Decided Undecided Decided


18-24 M
25-34 M
35-49 M
50-64 M


1.6%


11.3
11.3
11.3


16.3
18.4


16.9
10.2


10.4
15.2
14.2


14.4
11.7
10.1


14.4
12.7


18-24 F
25-34 F
35-49 F
50-64 F


11.3
12.9
25.8


10.1
10.4
15.7


27.1
15.3


13.5
12.8


11.7
17.0
12.2
11.2


14.1
13.3


Total
Sample


100%


100%


100%


100%


100%


100%


Size


.0459


Employment


Private


21.3%


39.9%


37.3%


33.8%


Local Govt
State Govt
Federal Govt
Unemployed


11.5
11.5
19.7


19.7


Retired
Student


11.9


53.2%


57.9%


10.1


20.4


16.9


23.4


11.1
25.8


22.7


Housewife


13.1


12.8


10.6


Total
Sample


100%


100%


100%


100%


100%


100%


Size


=.0135


Union
Member


17.7%


16.3%


HH Member


Not a


Member


77.5


74.5


20.3%
11.9


67.8


14.5%
11.4
74.1


14.2%
11.1
74.7


14.7%


76.4











Table


8--continued.


Questions


Could


ease


tell


me--is


your


between


18-24,


25-34,


35-49,
status?
student


50-64,
Are


65 and
you empl


over?
oyed,


or a housewife?


How would
unemployed


(IF EMPLOYED)


describe


cooking


If that


with


your


employment


for a job,


private


retired,
industry,


local


member


your


government,


a labor


ouseho


state
union


id a member


government,
or organize


a labor


or f


federal


labor


union


government?


group?


or organize


(IF NO)
ed labor


anyone


group?


Source


William R


Hamilton


Staff,


1982


Thurman


Campaign


1992


Computed by


the author











Table


Comparison


of Undecided and Decided


Voters


Demoaraohics


Florida.


September


1992


Age Undecided Decided


18-29 4.3% 5.7%
30-45 24.5 19.1
46-60 20.1 20.1
Over 60 51.0 55.0

P = .0478



Gender Undecided Decided


Male 47.9% 47.8%
Female 52.1 52.2

Sample Size 93 297

P = 0.497


Source:


William R.


Hamilton and Staff,


Inc.


1982


Thurman Campaign 1992

Computed by the author












Table 10


Comparison of Undecided and Decided Voters Demographics


Fourth Congressional District of Virginia 1982


April


August/September


Sept/Oct


Religion


Undecided Decided


Undecided Decided Undecided Decided


Mainline Prot


21.3%


Fundamentalist 68.9


31.5%


57.1


29.6%


64.8


37.4%
53.2


Not Available


Catholic


Jewish
Other


Total


100%


100%


100%


100%


Sample


Size


P <.0425


Income
Under


$8,000


11.0%


7.6%


Not Available


Not Available


8,000-11,999
12,000-17,999


18,000-23
24,000-29
30.000-35


36,000-41,999


11.0
29.1


12.7
10.9


16.8
26.0
17.8
12.8


42.000+


Total


100%


100%


Sample


Size


Race
White
Black
Other


54.1%


45.9


72.6%
27.1


59.6%
40.4


74.8%
24.5


66.8%
31.6


72.4%
27.3


. a ao 0


m -A


i r%/n


4 f% f r\


4 f\ r\ n


hnA r


SAI n











Table


10--continu


Questions
Catholic.


you your


or Jewish?


religious


(IF PROTESTANT)


Wha


preference--i
t particular


Protestant,


denomination


that?


foll


Just


owing


just


for statistical


categories


to make


sure


your


purposes,


total


we have


family


can you
income


a representative


fell


falls


me which


one of the


into--before


sample--could


taxes?
tell me


your


race?


Source:


William R


Hamilton


Staff,


Inc. 1982


Thurman


Campaign


1992


Computed by the


author











Table


Comparison


of Undecided and Decided


Voters


Demographics.


Florida,


Sent


1992


Income Undecided Decided


Under $20,000 26.1% 28.7%
$20,000-$35,000 43.2 36.9
$35,000-$50,000 12.5 20.9
Over $50,000 18.2 13.5

Sample Size 93 297

P = .0478


Source:


William R


Hamilton and Staff,


Inc.


1982


Thurman Campaign 1992

Computed by the author










significant


only


the August/September


respondents


with


undecided


group


tending to be


older


and more


female


than


decided


group.


problem,


however,


that


these two


independent


variables


are


grouped together


cannot


analyzed separately.


and gender


could be


analyzed


separately,


likely that


finding


in the


August/September


poll


would hold


for the


other


two


studies.


In Florida


Employment


differences


1992

t was


between


age and gender were

only significant i


the decided and


significant.


n describing the


undecided respondents


the April


Virginia


poll.


that


time more


respondents


employed by the


government


were


undecided


than


those


employed


by private


industry,


unemployed,


retired,


students,


housewives.


This


finding


did not


hold in


subsequent


polls.


Income was


significant


1992 Florida


poll.


Religion


was


important


first


poll


and not


in the


second.


More Fundamentalists were


undecided group


than


other


religions.


Union membership was


not


significant.


Race


was


significant


in all


three


polls


with


more


blacks


falling


into


Table


versus decided


undecided category than


indicates the


voters.


party


the decided


affiliation


Party affiliation


one.


undecided


Virginia


was


significant


in describing the differences


between


I I I9 -


i I


* 11 1 J I











Table


Comparison


of Undecided and Decided Voters Party


Affiliation


Fourth Congressional District of Virginia 1982


April


August/September


Sept/Oct


Party


Undecided Decided


Undecided Decided Undecided Decided


Democrat


Leaning Dem
Independent
Leaning Rep
Republican


43.5%
11.3


17.7


38.8%
10.7
20.4


10.4
18.5


33.9%


11.9
32.2


10.2


41.7%


18.6
13.1
14.8


29.5%


25.8


38.4%


15.6


15.1


Other


Don't Know


12.9


13.8


Total


100%


100%


100%


100%


100%


100%


Sample


Size


P <.0001


=.0010


P <.0001


Fifth Congressional District of Florida 1992


Party


Undecided


Decided


Democrat


58.9%
10.0


Independent
Republican


59.3%


33.7


31.1


Total


100%


100%


Sample


Size











Table


12--continued.


Fourteenth


Congressional


District


of Ohio


1992


Party


Undecided


Decided


Democrat


Independent
Republican


Don't


Know


17.7%
46.2
14.9
21.3


52.8


34.8


Total


100%


100%


Sample


Size


.0011


Question:
Republican


Generally


speaking


, or an Independent?


consider


(IF INDEPENDENT)


yourse


Well,


a Democrat,


which


party


normally


support


in elections--the


Democrats


or the Republicans?


Source:


William R.


Hamilton


and Staff,


Inc.


1982


Thurman


Campaign


1992


Computed


the author











undecided


group.


Party


affiliation


in Florida,


however,


was


not


significant.


Direction


of the


country was


significant


first


Virginia


poll


with more


respondents


felt


that


country


headed


the wrong


direction being undecided.


There


no significant


difference between


decided and


undecided


with regard


direction


country


second Virginia


poll.


was


significant


in Florida


(see


Table


The data


indicate


that


there are differences


among the


undecided


voter


that


some


are


faced with


cross-pressures


that


result


their


indecision.


In Virginia


challenger was prochoice,


incumbent 's


position,


however,


not


available.


There


was


a question


regarding


constitutional ban against


abortion


and a


question


regarding


voter's


position


on abortion


first


poll


only.


that


first


poll


most


undecided


voters


did not


even


recognize


the challenger's


name,


unlikely they would


subjected to


cross-pressure by having a


favorable


opinion


and being


antiabortion.


Table


is a cross-tabulation


of undecided


voters by


party affiliation by their opinion


Democratic


challenger


Virginia.


There evidence of


cross-pressure


* *


q


I L


m


*











Table


Evidence of Cross-Pressures.


Party


Affiliation


and ODinion


Nonofficeholder


Republican


Democrat


Favorable


Unfavorable


Favorable


Unfavorable


Fourth CD of VA


Aug/Sept

Decided


1982


Undecided


Total
Sample


size:


.0363


Fourth CD


of VA


Sept/Oct 198

Decided
Undecided

Total


Sample


size:


Source:


William R.


Hamilton and Staff


Inc.


1982


Computed by the author










Table


their


is a


opinion


cross-tabulation


the direction


of undecided voters by


the country and their


rating


incumbent


Virginia.


direction


country


question


also


provides


insight


into


cross-


pressure


phenomenon


among undecided


voters,


although


assumptions may be


far-reaching.


Thirty-one


percent


(April


survey)


country was


undecided voters who


heading


wrong direction


thought


also


that


rated the


performance of their


incumbent


Congressman as


excellent


good.


This may


blame


for the


be either


wrong


cross-pressure


direction


or the


assignment


on someone other than


their


Congressman.


voter placed the blame


elsewhere,


s/he


would not


subjected


to cross-pressure.


Discriminant


Analysis


F probability


generated from the


analysis


variance


was


used to


select


independent


variables


for the


discriminant


analysis.


Virginia


results of


analysis


of variance between


the decided and


undecided groups


and each


variables,


respectively,


are


shown


Table


There


less


than


a .05


probability that


variation between


undecided and


decided probable


voters


is due


to chance


for the


independent


variables


of April--


number


of opinion/issue questions


answered,


party


- rftr f i .,.


- I -


_ r 1 _


I




















































ct) '











should support


President,


Middle


East


policy,


abortion


and tax


cuts;


August/September--number of


opinion/issue


questions


answered,


incumbent


cares


about me,


opinion


of the


incumbent,


race,


rating of the


incumbent,


incumbent


performance on


budget,


social


agriculture,


security, r

unemployment


presentation,


and the


federal


identification


incumbent's party;


September/October--number


opinion/issue questions


answered,


party


affiliation,


incumbent


cares


about me,


opinion


of the


incumbent,


rating


incumbent,


identification


of the


incumbent's party and


opinion


challenger.


1992


in Ohio


there


less


than a


.05 probability


that


the differences between


the decided and undecided voter


were


to chance on


following


independent


variables


number


of questions


answered,


vote


choice


for county


prosecutor,


county engineer,


state


representative,


U.S.


Senator


County Clerk


of Courts.


1992


in Florida


there


less


than


probability


that


differences


between


the decided and undecided voter


was


to chance


for the


following variables:


number of


opinion/issue


questions answered,


opinion


of Thurman,


rating


of Thurman,


opinion


of Hogan,


belief that


Thurman


cares


about


belief


that


Thurman


cares


about


the elderly,


belief that


n .h ,.A ~ -- --C. ----IE,


f,


^ _


1_ = f


1.


m


1 /"











the Senator,


belief


that


Hogan


would be


tough


on crime,


belief th

Congress,


iat a


state


rating


legislator


of the


state


should not

legislator,


be elected to

and presidential


vote


choice.


A discriminant


analysis


was


conducted next


using the


variables


listed above.


A discriminant analysis


appropriate mechanism to


statistically distinguish between


two groups


cases.


variables


discriminate between


groups


cases


and predict


into which


category


a case


falls,


based upon


values


these


variables.


Each


coefficient


represents


relative


contribution


the variable


to that


function.


group of


These


decided


characteristics


voters


are


the group of


ones


on which


undecided


voters are


expected


to differ.


Table

analysis.


reports


The pooled


results


within


of the discriminant


group correlation


interpreted


similar to multiple


to determine the


linear


contribution


regression analysis.


variable.


a way


index of


number of

important


issue


and opinion


independent


variab


questions

le in all


answered


five


was


the most


surveys.


April


Virginia


poll


party


identification


was the most


significant


variable.


Party


affiliation


did not


even make


to be


included


in the analysis


1992.


- . -- --


1 .. 1 -


i -1r 1


___I _


r-


I i m


I 1


1.











Table


Pooled Within Group Correlations--Virginia.


Ohio.


and Florida


Fourth CD of VA 1982


April


Aug/Sept


Sept/Oct


Party Affiliation


No. Of Answers to Opinion Questions


Opinion


of Incumbent


Rating of Incumbent


Incumbent


Opinion


Cares


* PID


* PID


About Me


of the Incumbent


Military Budget
Security of Israel


Ban Abortion


Race


ERA


Rating of the Incumbent


Congress


Should Support President


Delay Tax Cut
Mideast Policy


Position on Abortion
Incumbent Performance on


.68031
.65760
.57826
.54014
.41782
.40240
.38705
.33550
.33465
.31390
.31342
.28892
.26534
.23484
.23017
.21382


Incumbent Provides Fair Representation
Incumbent Performance on Agriculture
Incumbent Performance on Unemployment
Incumbent Performance on Federal Budget


Identification of Incumbent


s Party


.83058
.42817
.40219
.70058
.46696


-.24418
.94894
.55390
.53726
.57079
.46054


.34631


.56487


.66684
.60149
.50468
.47776
.41687
.16563


Opinion of Challenger


Sample


Size


.56165


.38761
.45235


1178


Percent


Cases


Correctly Classified


79.59%


76.6%


78.6%












Table


15--continued.


14th CD of 5th CD of
Ohio 1992 Florida 1992


Of Answers


Opinion
Opinion
Rating
Rating


S


to Opinion


Questions


* PID


.95857


of Incumbent/Thurman
* PID


O


.81774
.75280
.71002
.65447
.55840


Incumbent/Thurman


State


Legi


slator


Should


Not Be


Elected


Need New
Thurman
Thurman
Thurman


Opinion
Hogan C
Thurman
Hogan W


to Congress


Facess
Cares A
Cares A
Cares A


in C


bout
bout
bout


of Challenger/H


campaign


Would


would Be


Presidential


Rating


of State


congress
Education
Seniors


Environment


ogan


Too Negative


Help E
Tough


Vote


Legi


Iderly


on Crime


oice


slature


.49396
.45612
.44667
.39400
.32396
.30049
.28251
.24860
.23973
-.19022
.14037


State


representative


vote


choice.


State


representative


County prosecutor


Senator


County
County


vote


engineer


clerk


vote


vote


choice


choice


choice


vote


of courts


choice


vote


choice


.54525
.45966
.41818
.35748
.33521


Sample


Size


Percent


Cases


Correctly


Classified


70.1%


73.5%


F test


not significant


and variable


was not included


in the analysis


** Not available


in the


survey


Source:


William R.


Hamilton


Staff,


Inc.


1982


Thurman
Sawyer


Campaign
Campaign


1992


n -- -I1 1 -


I 1 1











groups


Virginia


and Florida.


These data


were not


available


in Ohio.


The model


correctly


classified


to 79%


of Virginia


probable


voters


into


the undecided


or decided


categories.


In Ohio


the probable


voters were


correctly


classified,


in Florida


73.5%


voters


were


correctly


classified.


Discussion of


Ouantitative


Results


There


are measurable differences


undecided voters


in elections


for the


between


decided and


House of


Representatives.


The most


significant


difference


that


undecided


voter


formed fewer opinions


regarding


issues


candidates


than


decided


voter.


some


instances


opinion matters,


e.g.,


incumbent


cares


about me.


Primarily,


however,


is more


significant


that


an opinion


held,


rather than


specific content


the opinion.


This


result


consistent


across


time


for the


1982


1992


elections.


It also


consistent


in all


three


states.


These


undecided


voters


do not


have as many


opinions


as do decided


voters,


focus


groups


pursue


reasons


for this.


The existence of


cross-pressures


also


is a significant


difference


between


decided and undecided


voters.


Undecided


Virginia


voters


1982


undecided Florida


voters


1992


* ,- 1- I -i p..-' 9 I .


_ r_ I i


mi 1 I


I


I