Ideological Deviation and Support for Reelection: Party Differences in Senate Races

• Date: 01 December 1991
• Author: Robert A. Bernstein
• Published date: 01 December 1991
• DOI: 10.1177/106591299104400411
• Subject Matter: Articles

IDEOLOGICAL

DEVIATION

AND

SUPPORT

FOR

REELECTION:

PARTY

DIFFERENCES

IN

SENATE

RACES

ROBERT

A.

BERNSTEIN

Auburn

University

key

empirical

claim

often

made

in

defense

of

elections

is

that

deviation

from

the

preferences

of

the

electorate

tends

to

be

jL

punished

at

the

polls.

The

traditional

hypothesis

is

that

the

more

incumbents

deviate

from

the

ideological

preferences

of

their

con-

stituencies,

the

less

support

they

tend

to

receive

in

reelection

races

(MacNeil

1963;

Erikson

1971;

Wright

1977, 1978, 1989;

Pomper

1980;

Johannes

and

McAdams

1981;

Davidson

and

Oleszek

1981;

Johannes

1984;

Erikson

and

Wright

1985;

Wright

and

Berkman

1986;

Whitby

and

Bledsoe

1986;

Hamilton

1987;

Abramowitz

1988).

Elections

are

said

to

&dquo;achieve

accountability&dquo;

and

&dquo;insure

responsiveness

to

constit-

uency

preferences&dquo;

by

linking

increased

deviation

from

those

prefer-

ences

to

increased

probability

of

defeat

(Wright

1989:

465).

Underlying

the

traditional

hypothesis

is

the

assumption

that

there

is

at

least

a

substantial

minority

of

the

electorate

whose

votes

are

influ-

enced

by

the

relative

ideological

proximity

of

candidates.

Those

voters

are

assumed

to

prefer

candidates

whose

ideological

positions

are

clos-

est

to

theirs.

Ideological

deviation

reduces

incumbents’

reelection

pros-

pects

because

as

incumbents

deviate further

from

their

constituencies’

preferences,

they

decrease

the

probability

that

they

will

be

seen

as

ideologically

closer

than

their

challengers

(Downs

1957;

Hinich,

Ledyard,

and

Ordeshook

1973;

Fiorina

1981).

The

party

of

the

incumbent

and

direction

of

ideological

deviation

are

irrelevant

in

the

statement

of

the

traditional

hypothesis.

The

fur-

Received :

May

29,

1990

Revision

Received:

November

8,

1990

Accepted

for

Publication:

November

29,

1990

NOTE:

A

revision

of

a

paper

presented

to

the

Northeastern

Political

Science

Asso-

ciation

meetings

in

Providence,

November

1990.

I

am

very

grateful

to

Alan

Abramowitz

for

providing

me

with

most

of

the

data

used

in

the

regression

analyses,

and

to

the

Center

for

Political

Studies of

the

University

of

Michigan

and

the

ICPSR

for

the

data

used

in

the

crosstabular

analysis.

I

am

also

grateful

to

Carl

Richard,

Bruce

Reed,

and

James

Gundlach

for

help

in

shifting

university

computing

systems,

and

to

Cynthia

Bernstein,

Dean

Mann,

and

anonymous

referees

for

this

journal

for

comments

on an

earlier

draft

of

this

study.

988

ther

to

their

right

or

left

that

voters

see

incumbents,

the

higher

the

probability

that

challengers

will

be

seen

as

closer

than

incumbents-

regardless

of

whether

the

incumbents

are

Democrats

or

Republicans.

Thus

tests

of

the

traditional

hypothesis

have

almost

always

measured

deviation

in

terms

of

absolute,

not

directional,

distance

(see,

e.g.,

Johannes

and

McAdams

1981;

Whitby

and

Bledsoe

1986;

Abramowitz

1988).

However,

party

of

the

incumbent

and

direction

of

ideological

devi-

ation

are

irrelevant

only

if

one

assumes

that

party

neither

influences

voters’

perceptions

of

the

relative

ideological

proximity

of

candidates

nor

constrains

candidates’

ability

or

willingness

to

translate

proximity

into

votes.

That

would

be

the

case,

for

example,

if

Senate

races

were

contests

between

unconstrained

candidates

before

a

fully

informed

elec-

torate.

However,

in

a

world

in

which

party

both

influences

voters’

per-

ceptions

and

constrains

candidates,

party

of

the

incumbent

and

direc-

tion

of

deviation

are

relevant:

they

condition

the

extent

to

which

incumbents’

ideological

deviation

is

punished

at

the

polls.

To

the

extent

that

voters

(1)

use

party

labels

as

guides

to

candidates’

ideologies,

and

(2)

tend

to

assume

that

Republican

candidates

are

to

the

right

of

Democrats,

the

relative

proximity

of

some

incumbents

is

distorted.

Voters

see

too-conservative

Democratic

incumbents

as

closer

than

their

Republican

challengers

and

too-liberal

Republican

incum-

bents

as

closer

than

their

Democratic

challengers.

To

the

extent

that

voters

use

party

labels

in

this

fashion,

the

traditional

hypothesis

link-

ing

ideological

deviation

to

reduced

support

will

not

hold

for

Demo-

crats

who

deviate

to

the

right

and

Republicans

who

deviate

to

the

left

of

their

constituencies’

preferences.

To

the

extent

that

party

leaders,

activists,

and

contributors

influ-

ence

challengers

not

to

make

an

issue

out

of

a

particular

ideological

position

relative

to

that

of

the

opposition,

some

ideological

deviation

by

incumbents

tends

to

go

unpunished.

At

least

in

the

1980s,

it

has

been

harder

for

Republican

challengers

to

make

an

issue

out

of

being

more

liberal

than

their

Democratic

opponents

than

it

has

been

for

them

to

make

an

issue

out

of

being

more

conservative.

A

Republican

challenger

who

wishes

to

make

an

issue

out

of

being

more

liberal

than

the

incumbent

has

to

reject

the

support

of

an

incumbent

president

and

the

conservative

activists

and

contributors

within

the

party.

A

Repub-

lican

who

wants

to

make

an

issue

out

of

being

more

conservative

than

the

incumbent

can

lay

claim

to

their

support.

Hence,

Republicans

challenging

conservative

Democrats

tend

to

be

less

able

or

willing

to